Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Male-Killing Bacteria in Insects: Mechanisms, Incidence, and Implications

Male-Killing Bacteria in Insects: Mechanisms, Incidence, and Implications Peer-Reviewed Journal Tracking and Analyzing Disease Trends Pages 2141–2404 EDITOR-IN-CHIEF D. Peter Drotman Associate Editors EDITORIAL BOARD Paul Arguin, Atlanta, Georgia, USA Timothy Barrett, Atlanta, Georgia, USA Charles Ben Beard, Fort Collins, Colorado, USA Barry J. Beaty, Fort Collins, Colorado, USA Martin J. Blaser, New York, New York, USA Ermias Belay, Atlanta, Georgia, USA Richard Bradbury, Atlanta, Georgia, USA David Bell, Atlanta, Georgia, USA Christopher Braden, Atlanta, Georgia, USA Sharon Bloom, Atlanta, Georgia, USA Arturo Casadevall, New York, New York, USA Mary Brandt, Atlanta, Georgia, USA Kenneth C. Castro, Atlanta, Georgia, USA Corrie Brown, Athens, Georgia, USA Benjamin J. Cowling, Hong Kong, China Charles Calisher, Fort Collins, Colorado, USA Vincent Deubel, Shanghai, China Michel Drancourt, Marseille, France Christian Drosten, Charité Berlin, Germany Paul V. Effler, Perth, Australia Isaac Chun-Hai Fung, Statesboro, Georgia, USA Anthony Fiore, Atlanta, Georgia, USA Kathleen Gensheimer, College Park, Maryland, USA David Freedman, Birmingham, Alabama, USA Rachel Gorwitz, Atlanta, Georgia, USA Peter Gerner-Smidt, Atlanta, Georgia, USA Duane J. Gubler, Singapore Stephen Hadler, Atlanta, Georgia, USA Richard L. Guerrant, Charlottesville, Virginia, USA Matthew Kuehnert, Edison, New Jersey, USA Scott Halstead, Arlington, Virginia, USA Nina Marano, Atlanta, Georgia, USA Katrina Hedberg, Portland, Oregon, USA Martin I. Meltzer, Atlanta, Georgia, USA David L. Heymann, London, UK David Morens, Bethesda, Maryland, USA Keith Klugman, Seattle, Washington, USA Takeshi Kurata, Tokyo, Japan J. Glenn Morris, Gainesville, Florida, USA S.K. Lam, Kuala Lumpur, Malaysia Patrice Nordmann, Fribourg, Switzerland Stuart Levy, Boston, Massachusetts, USA Ann Powers, Fort Collins, Colorado, USA John S. MacKenzie, Perth, Australia Didier Raoult, Marseille, France John E. McGowan, Jr., Atlanta, Georgia, USA Pierre Rollin, Atlanta, Georgia, USA Jennifer H. McQuiston, Atlanta, Georgia, USA Frank Sorvillo, Los Angeles, California, USA Tom Marrie, Halifax, Nova Scotia, Canada David Walker, Galveston, Texas, USA Nkuchia M. M’ikanatha, Harrisburg, Pennsylvania, USA J. Todd Weber, Atlanta, Georgia, USA Frederick A. Murphy, Bethesda, Maryland, USA Managing Editor Barbara E. Murray, Houston, Texas, USA Byron Breedlove, Atlanta, Georgia, USA Stephen M. Ostroff, Silver Spring, Maryland, USA Marguerite Pappaioanou, Seattle, Washington, USA Copy Editors Kristina Clark, Dana Dolan, Karen Foster, Johann D. Pitout, Calgary, Alberta, Canada Thomas Gryczan, Michelle Moran, Shannon O’Connor, Mario Raviglione, Geneva, Switzerland Jude Rutledge, P. Lynne Stockton, Deborah Wenger David Relman, Palo Alto, California, USA Production Thomas Eheman, William Hale, Barbara Segal, Guénaël Rodier, Saône-et-Loire, France Reginald Tucker Connie Schmaljohn, Frederick, Maryland, USA Editorial Assistants Kristine Phillips, Susan Richardson Tom Schwan, Hamilton, Montana, USA Rosemary Soave, New York, New York, USA Communications/Social Media Sarah Logan Gregory, P. Frederick Sparling, Chapel Hill, North Carolina, USA Tony Pearson-Clarke, Deanna Altomara (intern) Robert Swanepoel, Pretoria, South Africa Founding Editor Phillip Tarr, St. Louis, Missouri, USA Joseph E. McDade, Rome, Georgia, USA Duc Vugia, Richmond, California Emerging Infectious Diseases is published monthly by the Centers for Disease Control John Ward, Atlanta, Georgia, USA and Prevention, 1600 Clifton Rd NE, Mailstop H16-2, Atlanta, GA 30329-4027, USA. Jeffrey Scott Weese, Guelph, Ontario, Canada Telephone 404-639-1960, fax 404-639-1954, email Mary E. Wilson, Cambridge, Massachusetts, USA The conclusions, findings, and opinions expressed by authors contributing Use of trade names is for identification only and does not imply to this journal do not necessarily reflect the official position of the U.S. Depart - endorsement by the Public Health Service or by the U.S. Department of Health ment of Health and Human Services, the Public Health Service, the Centers for and Human Services. Disease Control and Prevention, or the authors’ affiliated institutions. Use of EMERGING INFECTIOUS DISEASES is a registered service mark of trade names is for identification only and does not imply endorsement by any the U.S. Department of Health & Human Services (HHS). of the groups named above. ∞ Emerging Infectious Diseases is printed on acid-free paper that meets the All material published in Emerging Infectious Diseases is in the public requirements of ANSI/NISO 239.48-1992 (Permanence of Paper) domain and may be used and reprinted without special permission; proper citation, however, is required. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Zoonotic Infections December 2018 Terrestrial Bird Migration and West Nile Virus Circulation, United States D. Swetnam et al. 2184 Capnocytophaga canimorsus Capsular Serovar and Disease Severity, Helsinki Hospital District, Finland, 2000–2017 E. Hess et al. 2195 Related material available online: article/24/12/17-2060_article Crimean-Congo Hemorrhagic Fever Virus, Mongolia, 2013–2014 On the Cover M.A. Voorhees et al. 2202 Related material available online: Hendrick Avercamp (1585-1634), Winter Landscape with Ice Skaters (c. 1608). Oil on panel, 30.4 in x 52 in/77.3 cm x 131.9 cm. Digital image courtesy of article/24/12/18-0175_article Rijksmuseum, Amsterdam. About the Cover p. 2389 Novel Type of Chronic Wasting Disease Detected in Moose (Alces alces), Norway Restaurant Inspection Letter Synopses L. Pirisinu et al. 2210 Grades and Salmonella Infections, New York, Related material available online: Outbreak of HIV Infection Linked New York, USA to Nosocomial Transmission, article/24/12/18-0702_article M.J. Firestone, C.W. Hedberg 2164 China, 2016–2017 Genomic Characterization X. Pan et al. 2141 of β-Glucuronidase–Positive Research Escherichia coli O157:H7 Strain Producing Stx2a Spatial Analysis of Wildlife Tuberculosis Based on a Y. Ogura et al. 2219 Reemergence of St. Louis Serologic Survey Using Dried Related material available online: Encephalitis Virus in Blood Spots, Portugal the Americas N. Santos et al. 2169 article/24/12/18-0404_article Reemerging virus in the western United States may have been mediated via Survey of Ebola Viruses in Rat Lungworm Infection in migrating infected birds from southern Frugivorous and Insectivorous Rodents across Post-Katrina South America, where it reemerged most Bats in Guinea, Cameroon, and New Orleans, Louisiana, USA recently in 2002. the Democratic Republic of R.C. Rael et al. 2176 A. Diaz et al. 2150 the Congo, 2015–2017 Related material available online: H.M. De Nys et al. 2228 Autochthonous Human Case Related material available online: article/24/12/18-0056_article of Seoul Virus Infection, the Netherlands article/24/12/18-0740_article C. Swanink et al. 2158 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Rat Hepatitis E Virus as Cause of Persistent Hepatitis after Liver Transplant S. Sridhar et al. 2241 Related material available online: December 2018 article/24/12/18-0937_article Influences of Community Interventions on Zika Prevention Behaviors of Pregnant Women, Puerto Rico, July 2016–June 2017 G. Earle-Richardson et al. 2251 Related material available online: article/24/12/18-1056_article Emerging Multidrug-Resistant Hybrid Pathotype Shiga Toxin– Producing Escherichia coli O80 and Related Strains of Clonal Complex 165, Europe A. Cointe et al. 2262 Related material available online: article/24/12/18-0272_article Comparison of 2016–17 and Previous Epizootics of Highly Pathogenic Avian Influenza H5 Guangdong Lineage in Europe Prevalence of Avian Influenza Vaccinia Virus among A(H5) and A(H9) Viruses in Live P. Alarcon et al. 2270 Domestic Dogs and Wild Coatis, Bird Markets, Bangladesh Brazil, 2013–2015 Related material available online: Y. Kim et al. 2309 G. Barbosa Costa et al. 2338 article/24/12/17-1860_article Related material available online: Related material available online: CTX-M-65 Extended-Spectrum article/24/12/18-0879_article article/24/12/17-1584_article β-Lactamase–Producing Salmonella enterica Serotype Human Exposure to Novel Infantis, United States Highly Pathogenic Avian Bartonella Species from Contact Influenza Virus A(H5N6) in A.C. Brown et al. 2284 with Fruit Bats Domestic Cats, South Korea Related material available online: Y. Bai et al. 2317 K. Lee et al. 2343 article/24/12/18-0500_article Related material available online: Historical Review article/24/12/18-0290_article Emergent Sand Fly–Borne Candidatus Cryptoplasma Phleboviruses in the Substance Use and Adherence to Associated with Green Lizards Balkan Region HIV Preexposure Prophylaxis for and Ixodes ricinus Ticks, N. Ayhan, R.N. Charrel 2324 Men Who Have Sex with Men Slovakia, 2004–2011 Substance and alcohol use were not B. Kočíková et al. 2348 Dispatches associated with decreased adherence. M. Hoenigl et al. 2292 Excess Mortality and Causes Isolation of Burkholderia Associated with Chikungunya, Related material available online: pseudomallei from a Pet Puerto Rico, 2014–2015 Green Iguana, Belgium article/24/12/18-0400_article A.R.R. Freitas et al. 2352 T. Hellebuyck et al. 2331 Highly Pathogenic Clone of Shiga Borrelia miyamotoi Infections Neglected Hosts of Small Toxin–Producing Escherichia coli in Small Mammals, Ruminant Morbillivirus O157:H7, England and Wales California, USA C. Schulz et al. 2334 L. Byrne et al. 2303 D.J. Salkeld et al. 2356 Related material available online: Related material available online: article/24/12/18-0507_article article/24/12/18-0409_article Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 In-Host Adaptation of Salmonella enterica Serotype Dublin during Prosthetic Hip Joint Infection F. El Sayed et al. 2360 Related material available online: article/24/12/18-0214_article December 2018 Risk Factors for Acquiring Scrub Typhus among Children in Deoria and Gorakhpur Districts, Uttar Pradesh, India, 2017 J.W.V. Thangaraj et al. 2364 Related material available online: article/24/12/18-0695_article Using PCR-Based Sequencing Lung Involvement in Chronic Mycoplasma ovipneumoniae to Diagnose Haycocknema Schistosomiasis with Bladder in Wildlife Species beyond perplexum Infection in Human Squamous Cell Carcinoma Subfamily Caprinae Myositis Case, Australia A. Saade et al. 2375 M.A. Highland et al. 2384 A.V. Koehler et al. 2368 Related material available online: Strongyloidiasis and Culture- article/24/12/18-0632_article Negative Suppurative Meningitis, Research Letters Japan, 1993–2015 M. Mukaigawara et al. 2378 Locally Acquired Leptospirosis Chuzan Virus in Yaks, in Expedition Racer, Related material available online: Qinghai-Tibetan Plateau, China Manitoba, Canada article/24/12/18-0375_article M. Wang et al. 2371 S.S. Kassim et al. 2386 Related material available online: Use of Next-Generation About the Cover Sequencing for Diagnosis of article/24/12/17-1414_article West Nile Virus Infection in Patient Returning to Belgium Conservation of White An Icy Vista from a Golden Age from Hungary by Rhinoceroses Threatened by B. Breedlove 2389 Bovine Tuberculosis, E. Wollants et al. 2380 South Africa, 2016–2017 Etymologia Disseminated Spiroplasma M.A. Miller et al. 2373 Capnocytophaga canimorsus apis Infection in Patient with Related material available online: R. Henry 2201 Agammaglobulinemia, France article/24/12/18-0293_article N. Etienne et al. 2382 Reviewer Appreciation 2391 Related material available online: article/24/12/18-0567_article Thank You EID Reviewers We couldn’t do it without you. We only maintain high standards because of your support. EID’s 2018 Impact Factor of 7.42 ranked it 1st among open-access infectious disease journals and 4th out of 88 infectious disease journals. The Google Scholar h-Index is 86; 2nd of top 20 publications in Epidemiology and 2nd among open-access journals; ranked 3rd among top 20 publications in Communicable Diseases and 1st among open-access journals. The electronic table of contents goes to 1,262,951 subscribers each month. All articles published in the Emerging Infectious Diseases journal are peer-reviewed by volunteers from around the globe, enabling us to bring you high-quality content about new and emerging infectious diseases and trends world-wide. A list of reviewers is posted at http: // Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 RESEARCH SYNOPSIS Outbreak of HIV Infection Linked to Nosocomial Transmission, China, 2016–2017 1 1 Xiaohong Pan, Jianmin Jiang, Qiaoqin Ma, Jiafeng Zhang, Jiezhe Yang, Wanjun Chen, 2 2 Xiaobei Ding, Qin Fan, Zhihong Guo, Yan Xia, Shichang Xia, Zunyou Wu On January 25, 2017, a physician from ZC Hospital in system, LIT is a category III medical service, meaning Hangzhou, China, reported to the Zhejiang Provincial that each hospital regulates itself (11). Center for Disease Control and Prevention that a poten- On January 24, 2017, a woman receiving LIT at ZC tial HIV outbreak might have occurred during lympho- Hospital in Hangzhou, China, called a hospital staff mem - cyte immunotherapy (LIT) performed at the hospital on ber, Dr. X, asking if she had risk for HIV infection. She December 30, 2016. We immediately began investigat- explained that her husband had just received a confirmed ing and identified the index case-patient as an LIT pa - diagnosis of HIV infection and that on December 30, tient’s husband who donated lymphocytes for his wife’s 2016, she had received LIT using lymphocytes her hus- LIT and later screened HIV-reactive. Subsequent con- band donated. Dr. X immediately reported this informa- tamination by a technician resulted in the potential ex - tion to the hospital’s deputy director, who informed the posure of 34 LIT patients. Acute HIV infection was diag- nosed in 5 persons. Phylogenetic analysis confirmed that clinical medical laboratory director, Dr. Y. At ≈4:00 pm the HIV-1 gag, pol, and env gene sequences from the the same day, Dr. Y informed the responsible laboratory index and outbreak-related cases had >99.5% similar- technician, Dr. Z, and requested that she stop LIT. One ity. Rapid investigation and implementation of effective hour later, Dr. Z voluntarily reported to Dr. Y that she had control measures successfully controlled the outbreak. deviated from protocol on December 30 and that other This incident provides evidence of a lapse in infection patients who received LIT on the same day might have control causing HIV transmission, highlighting the need been exposed. At 5:30 pm, the director of ZC Hospital for stronger measures to protect patients from infectious called an emergency meeting with department directors, disease exposure. who decided to request help from the Zhejiang Provin- cial Center for Disease Control and Prevention (Zhejiang ymphocyte immunotherapy (LIT) to treat recurrent CDC). On January 25, 2017, Zhejiang CDC epidemiolo- Lmiscarriage involves receipt of lymphocytes to a gists began investigating a possible HIV outbreak among patient from a donor, usually the patient’s male partner. LIT recipients at ZC Hospital. We report on the inves- Although the European Society of Human Reproduction tigation conducted, control measures implemented, and and Embryology (1), the Royal College of Obstetricians outcomes observed. (2), and the American College of Obstetricians and Gyne- cologists (3) have issued clear guidance against LIT, sup- Methods ported by a 2014 Cochrane review (4), more recent meta- The potential HIV outbreak at ZC Hospital was declared analyses support its use (5,6), as do 4 newer intervention a public health emergency, and a formal investigation be- control studies conducted in China (7–10). Although the gan on January 25, 2017, supported by provincial (Zheji- number of LIT recipients in China is estimated to be large, ang Health Commission and Zhejiang CDC) and national no statistics are available. Within China’s healthcare (National Health Commission and National Center for AIDS/STD Control and Prevention, Chinese Center for Author affiliations: Zhejiang Provincial Center for Disease Control Disease Control and Prevention [China CDC]) authori- and Prevention, Hangzhou, China (X. Pan, J. Jiang, Q. Ma, J. ties and resources. Neither institutional review board ap- Zhang, J. Yang, W. Chen, X. Ding, Q. Fan, Z. Guo, Y. Xia, S. Xia); proval nor individual informed consent was required for National Center for AIDS/STD Control and Prevention, Beijing, the investigation. Routine informed consent for HIV, China (Z. Wu); University of California, Los Angeles, California, USA (Z. Wu) These first authors contributed equally to this article. These senior authors contributed equally to this article. DOI: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2141 SYNOPSIS hepatitis B virus (HBV), hepatitis C virus (HCV), and syph- compliance specialists and 3 public health officials from ilis testing (oral or written) and for contact tracing (oral) independent institutions. They thoroughly reviewed all was obtained. relevant records: staffing, training, qualification, certifica - tion, security, inventory, equipment, LIT protocol, compli- Case Definition, Case Finding, and Contact Tracing ance, supervision, and infection control procedures, as well We defined an outbreak-related case as a newly diagnosed as records generated during the execution of LIT-related laboratory-confirmed HIV infection, with evidence of procedures. The audit also included private interviews acute infection suggesting occurrence of transmission on with all laboratory staff and direct observation of staff re- December 30, 2016, among women who had received LIT hearsing LIT procedures; investigation of other potential at ZC Hospital that day or their secondary contacts, with violations of protocol that might have caused nosocomi- HIV gene sequence highly related to that of the index case- al transmission; and a check of baseline laboratory tests patient. Initial case finding began among all women who for HIV, HBV, HCV, and syphilis for all 34 women and received LIT at ZC Hospital on December 30. A trained their husbands. public health specialist conducted interviews on HIV risk behavior during December 30, 2016–January 25, 2017, to Molecular Phylogeny Analysis assess the possibility that HIV infection had been acquired All HIV sequencing was performed at the Zhejiang CDC by means other than LIT and that HIV already had been laboratory using plasma specimens. Two technicians in transmitted to others. separate laboratory areas extracted HIV RNA from speci- mens, each using a different method: one used the QIAamp HIV, HBV, HCV, and Syphilis Testing Viral RNA Mini Kit (QIAGEN, Hilden, Germany), the All potential outbreak-related case-patients and their other a viral RNA/DNA extraction kit on an automatic ex- contacts were provided free testing and counseling at ZC traction platform (Suzhou Tianlong, Suzhou, China). Par- Hospital. Persons in whom HIV infection was diagnosed tial sequences for the HIV genes gag, pol, and env were were referred to treatment. For HIV, serologic screening amplified by reverse transcription PCR and nested PCR us- was conducted at ZC Hospital’s laboratory using the ing GUX/GDX primers for gag, 5 different pairs of primers Anti-HIV (1+2) 4th-generation antigen/antibody enzyme for pol, and M13F/M13R primers for env. PCR products immunoassay (EIA) kit (Shanghai Kehua Bio-Engineering, were confirmed by 1% agarose gel electrophoresis and then Shanghai, China) and the HIV 1/2/O Tri-Line HIV Rapid purified and sequenced. Test Device (ABON Biopharm, Hangzhou). If reactive, We analyzed sequences with Sequencher v5.0 (Gene new venous blood specimens were collected and sent to Codes Corporation, Ann Arbor, MI, USA), examined them the Hangzhou Center for Disease Control and Prevention for similarity, and aligned them to reference sequences us- laboratory for confirmatory serologic testing by Western ing BioEdit v7.2.0 (Ibis Therapeutics, Carlsbad, CA, USA). blot (WB; MP Biomedicals, Singapore). In parallel, Two sets of reference sequences for each gene were selected plasma specimens were sent to the Zhejiang CDC, where for comparison to outbreak-related consensus sequences. HIV nucleic acid testing was conducted using COBAS The first set was international reference sequences obtained AmpliPrep/COBAS TaqMan HIV-1 Test v2.0 kits (Roche, from the Los Alamos National Laboratory (https://www. Branchburg, NJ, USA). The second was representative of strains cir- HBV, HCV, and syphilis testing were performed at ZC culating in the area at the time of the outbreak. We used the Hospital’s laboratory. For HBV, samples were screened neighbor-joining tree method (Kimura 2-parameter model) for 5 indicators (i.e., hepatitis B surface antigen, hepatitis to determine HIV subtype and phylogenetic relationships B surface antibody, hepatitis B e-antigen, hepatitis B and genetic distance between sequences. Two technicians e-antibody, hepatitis B c-antibody) using EIA kits (InTec blindly and independently analyzed 2 specimens from each Products, Xiamen, China). For HCV, samples were screened patient. Neighbor-joining phylogenetic trees were con- for antibodies using an EIA kit (Zhuhai Livzon Diagnostics, structed using MEGA 6.0 ( Zhuhai, China). For syphilis, samples were initially with 1,000 replicate bootstrap alignments. We defined a screened by Toluidine Red Untreated Serum Test (TRUST, transmission cluster as having a bootstrap value >90% and Shanghai Rongsheng Biotech, Shanghai, China). Reactive a mean genetic distance of <0.015. samples were confirmed by Treponema pallidum particle agglutination assay (Fujirebio Inc., Nagasaki, Japan). Results Laboratory Audit Epidemiologic Investigation An audit of the hospital laboratory began immediately on Mrs. P0, age 36, is the wife of the index-case patient, P0, January 25 and lasted 6 days. It was conducted by 3 trained and had been enrolled in LIT starting June 21, 2016. At 2142 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 HIV Infection Linked to Nosocomial Transmission enrollment, P0 and Mrs. P0 both underwent behavioral Mrs. P0’s baby was born on July 9, 2017, and Mrs. P0 health screening, physical examination, and HIV testing stopped all use of antiretroviral therapy (ART) on the same (both tested HIV negative). Mrs. P0 received LIT on July day. Her most recent HIV screening test was performed on 19, August 16, September 13, October 14, November 11, March 9, 2018, at 9 months since she stopped ART and >14 December 2, and December 30, 2016, at ZC Hospital, months since her exposure. Her results remained negative. each time with lymphocytes donated by her husband 3 days before her LIT dates. Later in the day after her Index Case-Patient December 30 treatment, Mrs. P0 learned that her husband P0 was the 40-year-old husband of Mrs. P0. Public had screened HIV reactive. On December 31, Mrs. P0 went health workers constructed a 2-month timeline related to XX Hospital in Hangzhou, which treats persons living to his HIV exposure and testing (Figure 1). His most with HIV, where she informed the physician she had just recent negative HIV screening result was on November discovered she was pregnant and had great concern about 27, 2016. The residual specimen was retested using a possible exposure to HIV by LIT because her husband had 4th-generation antigen/antibody test during the investi- just screened HIV-reactive, which she worried she might gation and confirmed negative. The exposure event that transmit to her unborn baby. She was strongly encouraged probably led to P0’s HIV infection was traced to Decem- to immediately begin postexposure prophylaxis (PEP), ber 1, 2016, when he had condomless, receptive anal sex using a regimen of 3 antiretroviral medications (tenofovir, with a man he did not know at a gay bathhouse. Approxi- lopinavir/ritonavir, and lamivudine) for 4 weeks. She mately 2 weeks later, fever developed, and he began to started PEP the same day. suspect HIV infection. Upon attending antenatal care shortly thereafter, Mrs. P0 went to the Yiwu CDC for HIV testing on Friday, P0 tested negative for HIV, HBV, HCV, and syphilis. On December 23, but his specimen was not tested until De- January 24, 2017, after her husband received a confirmed cember 29 (HIV testing by both antibody-only and anti- diagnosis of HIV infection, Mrs. P0 alerted staff at ZC body/antigen EIA is only performed on Thursdays at Yiwu Hospital that she was concerned she was exposed to HIV CDC). He was not informed of his reactive result until on via LIT on December 30, 2016; this report was the initiating December 30, at which time he was encouraged to return event that sparked the outbreak investigation. As part of the for confirmatory testing. However, P0 instead sought re- investigation, Mrs. P0 was followed and regularly tested for screening at another facility. He returned for confirmatory HIV, HBV, HCV, and syphilis. All results were repeatedly testing on January 3, 2017. His first WB result was indeter- negative. Her PEP regimen was converted to a prevention minate (gp160/p24). His second result (using a new speci- of mother-to-child transmission (PMTCT) program (same men obtained on January 23) was positive (gp160/gp120/ medicines extended to 8 weeks after delivery). p66/p31/p24), and his viral load (VL) was 121,000 copies/ Figure 1. Timeline of HIV exposure and HIV diagnosis of the index case-patient, P0 (blue), and the HIV exposure of his wife, Mrs. P0 (orange), Hangzhou, China, November 27, 2016–January 24, 2017. CDC, Center for Disease Control and Prevention; LIT, lymphocyte immunotherapy; P, patient; WB, Western blot. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2143 SYNOPSIS mL. P0 was informed of his HIV diagnosis on January 24 or other failures that could have similarly resulted in and started treatment the next day. He tested negative for nosocomial transmission of HIV, was found before De- HBV, HCV, and syphilis. Investigators determined that P0 cember 30 or during December 31–January 25. made no other donations of fluids, cells, or tissues after his The auditors concluded that lymphocyte processing exposure on December 1, 2016. deviated from the protocol on December 30 and that the technician responsible contaminated an unknown number Laboratory Investigation of patients’ prepared lymphocytes on December 30 with The audit of the LIT laboratory resulted in 5 main find - lymphocytes from the index case-patient. Thus, all patients ings. First, although the appropriate protocol was used who received LIT on December 30 should be tested as if and requires that each disposable sterile tube for lympho- they had potentially been exposed. Auditors recommended cyte processing is used only once, the laboratory expe- that all the women and their husbands be tested for HIV, rienced a shortage of these tubes for 1 day on December HBV, HCV, and syphilis. 30. To provide LIT to 34 women, >136 tubes were need- ed (34 for moving lymphocytes from culture containers Contact Tracing to washing plates and 102 for washing the 34 cultures 3 Along with Mrs. P0, 33 other women received LIT at ZC times each). Approximately 100 tubes were available on Hospital on December 30. A medical records review found December 30. Second, instead of stopping and calling that all 33 had tested negative for HIV, HBV, HCV, and this issue to the attention of a supervisor, the laboratory syphilis before beginning LIT in 2016. None reported technician processing donated lymphocytes for LIT on any HIV risk behavior other than sexual contact, and December 30 deviated from protocol and reused tubes. none reported sexual contact with anyone other than their Deviations occurred in 2 procedures: tubes were used re - husbands. All 33 women (possible primary contacts) and peatedly for moving lymphocytes from culture contain- their husbands (possible secondary contacts) were tested ers to washing plates and for washing the lymphocyte for HIV, HBV, HCV, and syphilis. Five cases of HIV were cultures. Third, the technician failed to properly docu- found (Figure 2). ment the work performed and upon interview, admitted to reusing disposable tubes “a few times,” but could not Case-Patient 1 remember how many times or for which couples. Fourth, P1, age 35 (not pregnant) at the time of the investigation, no deviation occurred on December 27 that could have received LIT at ZC Hospital on November 4, December caused contamination during blood specimen collection, 2, and December 30, 2016. She reported having fever, lymphocyte separation, or lymphocyte culturing. Final- sore throat, and other symptoms, for which she had been ly, LIT was performed at ZC Hospital only ≈1 time each given penicillin at a local clinic for a suspected bacterial month. No LIT had yet been conducted during December infection. HIV serologic results were nonreactive, but 30, 2016–January 25, 2017. No evidence of these failures, virologic results were positive with VL of 756,000 copies/ Figure 2. Timeline of HIV exposure, symptoms, diagnosis, and treatment initiation for the 5 HIV-infected women during nosocomial HIV outbreak, Hangzhou, China, December 30, 2016–February 9, 2017. ART, antiretroviral therapy; LIT, lymphocyte immunotherapy; P, patient; WB, Western blot. 2144 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 HIV Infection Linked to Nosocomial Transmission mL on January 26, 2017, suggesting acute infection. Later P5 was informed of her diagnosis and initiated ART on confirmatory WB result was positive (gp160/gp120/p41/ February 9. p24). P1 was informed of her diagnosis on January 29, and she initiated ART the next day. Other Contacts The remaining 29 women, including Mrs. P0, the wife of Case-Patient 2 the index case-patient, all had multiple negative serologic P2, age 28 and pregnant (10 weeks’ gestation) at the time and virologic results during follow-up (Table 1, https:// of the investigation, received LIT at ZC Hospital on July 22, August 19, September 16, October 14, November None had HBV, HCV, or syphilis. 11, December 9, and December 30, 2016. She reported Although all 33 husbands were tested initially, 3 symptoms including headache, dizziness, and fever but husbands of the 5 women with newly diagnosed HIV sought no care. HIV screening results were nonreactive, infection, all of whom who reported sexual contact with but virologic results were positive with VL of 8,920,000 their wives after December 30, 2016, were followed and copies/mL on January 25, 2017, suggesting acute infection. provided HIV serologic and virologic testing, as well as Confirmatory WB result was positive (gp160/gp120/p24). HBV, HCV, and syphilis testing. All results were negative. P2 was informed of her HIV diagnosis, and she enrolled in Additionally, the infants of Mrs. P0 (born July 9, 2017), P2 PMTCT and initiated ART on January 30. (born August 14, 2017), and P4 (born July 3, 2017) were tested for HIV by early infant diagnosis (EID; by PCR) and Case-Patient 3 were HIV-negative. P3, age 34 (not pregnant) at the time of the investigation, received LIT at ZC Hospital on October 10, November Phylogenetic Investigation 4, December 2, and December 30, 2016. She had muscle The HIV sequences derived from the index case-patient soreness and fever but did not seek care. HIV screening and the 5 women with newly diagnosed HIV infection results were nonreactive; however, virologic results were shared a very high degree of similarity: mean of 99.95% positive with VL of 841,000 copies/mL on January 26, for gag, 99.48% for pol, and 99.92% for env (Table 2). 2017, suggesting acute infection. Confirmatory WB result The gag and pol sequences were consistent with HIV- was positive (gp160/gp120/p24) on January 31, and she 1 subtype CRF01_AE, and env sequences were subtype was informed of her diagnosis on the same day. She began as C, indicating that all 6 persons were infected with a ART on February 9, 2017. recombinant CRF01_AE/C strain. Phylogenetic trees of gag, pol, and env sequences (Figure 3) indicate a very Case-Patient 4 close genetic relationship between the virus present in the 5 P4, age 28 and pregnant (17 weeks’ gestation) at the women with newly diagnosed infection and the index case- time of the investigation, received LIT at ZC Hospital on patient. The sequences of all 3 genes map to monophyletic March 4, April 1, April 29, May 27, June 24, July 29, clusters in 100% of bootstrap replicates with genetic August 19, November 18, December 9, and December 30, distances of <0.015. 2016. She reported symptoms including feeling cold and red allergy-like spots on her chest but did not see a doctor. Summary of Outbreak Response HIV screening results were nonreactive, but virologic In response to this outbreak, the National Health results were positive with VL of 3,000,000 copies/mL Commission immediately suspended all LIT services on January 31, 2017, suggesting acute infection. First nationwide, and all 34 couples involved were provided confirmatory WB result was indeterminate (gp160/p24); counseling and support. The epidemiologic investigation second was positive (gp160/gp120/p24). P4 was informed found that a lymphocyte donor (P0) had become infected of her diagnosis, enrolled in PMTCT, and initiated on with HIV before donation on December 27 and that ART on February 2. laboratory contamination occurred on December 30, which together caused 5 women to become infected Case-Patient 5 with HIV. HIV phylogenetic investigation confirmed the P5, age 34 (not pregnant) at the time of the investigation, causal relationship. All 5 women had initiated ART as received LIT at ZC Hospital only on December 30, 2016. of February 9, 2017, only 15 days after the investigation She reported no symptoms. HIV screening results were began. All remaining 29 women who initially screened nonreactive, but virologic results were positive with VL of nonreactive and the husbands of the 5 infected women 6,460,000 copies/mL on January 26, 2017, suggesting acute were followed up for 6 months; no additional HIV infection. First confirmatory WB result was indeterminate infections were found. The 3 pregnant women were (gp160), but the second was positive (gp160/gp120/p24). provided PMTCT; their newborn infants were followed Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2145 SYNOPSIS Table 1. Interval between HIV exposure and follow-up HIV tests for the 37 persons followed up after initially screening HIV-nonreactive in investigation of nosocomial HIV outbreak, Hangzhou,China, 2016–2017* Days between HIV exposure and follow-up HIV tests Potential contact 1st test 2nd test 3rd test 4th test 5th test 6th test 7th test 8th test Primary† Mrs. P0‡ 24 29 45 68 103 130 191 464 Q1 28 48 82 Q2 28 58 86 186 Q3 27 29 59 84 182 Q4 28 58 86 192 Q5 28 45 86 192 Q6 28 56 94 188 Q7 27 28 57 99 211 Q8 29 58 100 189 Q9 27 57 45 85 Q10 27 28 60 93 187 Q11 28 64 106 202 Q12 29 55 83 185 Q13 27 29 58 89 Q14 27 35 62 92 199 Q15 27 33 64 93 212 Q16 27 45 93 Q17 27 33 Q18 27 34 59 90 189 Q19 27 33 63 91 187 Q20 28 60 90 194 Q21 30 55 80 192 Q22 31 45 93 186 Q23 31 53 81 188 Q24 33 55 89 187 Q25 32 63 92 196 Q26 33 63 95 189 Q27 33 47 97 193 Q28 33 60 Secondary§¶ P3’s husband 7 30 57 97 182 P4’s husband 11 28 52 98 180 P5’s husband 2 18 25 32 60 196 Infant P0 3 42 90 242 Infant P2 3 43 Infant P4 1 44 *First HIV test was conducted using both nucleic acid testing and EIA. All subsequent HIV tests were conducted using EIA only. Blank cells indicate no further HIV test. EIA, enzyme immunoassay; LIT, lymphocyte immunotherapy; P, patient infected with HIV; Q, women who might have been exposed to HIV but were not infected. †Beginning of the time interval was counted from contaminated LIT at ZC Hospital on December 30, 2016. ‡Mrs. P0 is the wife of the index case-patient, P0. She is continuing to be followed once a year for at least 3 years since she discontinued antiretroviral therapy. Her most recent HIV test, on March 9, 2018, was again negative. §For husbands, beginning of the time interval was counted from the most recent sexual contact with the wife from the wife’s exposure on December 30, 2016, through the start of the investigation on January 25, 2017. Husbands received both serologic (4th-generation Ag/Ab EIA, Shanghai Kehua Bio- Engineering, Shanghai, China) and virologic testing (HIV-1 RNA, COBAS AmpliPrep/COBAS TaqMan HIV-1 Test v2.0, Roche, Branchburg, NJ, USA). ¶For infants, beginning of the time interval was counted as the date of birth. The first 3 HIV tests for infants were early infant diagnosis tests by a standard nucleic acid testing protocol. The fourth test for infant P0 was a 4th-generation Ag/Ab EIA. up, and no HIV infection was found. The laboratory Discussion technician was sentenced to 2.5 years in prison. The These epidemiologic and phylogenetic investigations used hospital director, deputy director, and division chief techniques similar to those used during HIV outbreak in- accountable for the laboratory were dismissed. LIT vestigations including a famous case of a Florida, USA, services were suspended until a new guideline was dentist (13–16); several criminal cases (17–20); a prison issued on December 22, 2017 (12). outbreak in Scotland, UK (21); 2 nosocomial outbreaks Table 2. Similarity of HIV genetic sequence of viral nucleic acid from the index case-patient and the 5 women infected by during nosocomial HIV outbreak, Hangzhou, China, 2016–2017 Sequence similarity, % Region Case-patient 1 Case-patient 2 Case-patient 3 Case-patient 4 Case-patient 5 Mean gag 99.70 100.00 100.00 100.00 100.00 99.95 pol 99.50 99.40 99.50 99.50 99.50 99.48 env 100.00 100.00 100.00 100.00 99.60 99.92 2146 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 HIV Infection Linked to Nosocomial Transmission Figure 3. Phylogenetic trees showing relationships between HIV-1 gene sequences from index case-patient and 5 women infected during nosocomial HIV outbreak, Zhejiang Province, China, 2016–2017, and reference sequences. Bootstrap values >90% only are shown for gag sequences (A), pol sequences (B), and env sequences (C). Triangles indicate index case-patient (P0) and 5 women found to have HIV infection (P1–5); dots indicate international reference sequences. Scale bars indicate nucleotide substitutions per site. P, patient. (22,23); and a recent outbreak in Indiana, USA, associated unacceptably long process of HIV diagnosis in China di- with injection drug use (24). These investigations identi- rectly contributed to this outbreak. Ample evidence of the fied P0 as the index case-patient for this nosocomial HIV substantial benefit of streamlining and accelerating Chi - outbreak and demonstrate that deviation from protocol and na’s HIV care continuum already exists (27–29). How- lapse in infection control during LIT were the cause. We ever, although a rapid 1-visit testing, diagnosis, clinical have yet to detect HIV infection in Mrs. P0, the wife of the staging, and ART initiation protocol has been adopted, index case-patient, suggesting that her immediate initiation China must accelerate the pace at which these changes are of PEP might have averted infection. However, we are un- implemented if it is to avoid another, similar outbreak. able to definitively determine Mrs. P0’s HIV status because Second, China must implement more frequent and she has been followed for only 9 months since she discon- thorough training for medical professionals on the risks of tinued ART, and evidence of viremic rebound nearly 30 nosocomial HIV transmission. The finding that Mrs. P0’s months after ART cessation was observed in the case of a attending physician at XX Hospital was concerned enough child in Mississippi, USA (25,26). These results underscore to start her on PEP but not to alert public health officials the critical importance of quickly investigating a suspected and the finding that the laboratory technician did not con- outbreak. Among the 34 women potentially exposed, only sider reusing sterile tubes to be unsafe both indicate that 5 acquired infections, and potential onward transmission of education about the risks for nosocomial HIV transmission HIV to their husbands and infants was averted. is still lacking. China must act quickly to fill this gap. This study was subject to at least 2 limitations. First, Finally, laboratories in medical settings must be follow-up HIV testing for the 29 potentially exposed women placed under stricter controls. Immediate supervision and was voluntary, and some declined to have third and fourth monitoring and thorough and frequent laboratory audits HIV tests. For example, Q17 was followed up at 27 days should be implemented immediately in China’s medical (with nucleic acid testing and antigen/antibody EIA) and 33 laboratories. A high level of vigilance in the medical days (antigen/antibody EIA only) and, although unlikely, it laboratory setting is critical if China is to prevent similar is possible that she had undetected HIV infection. Second, as future nosocomial outbreaks. noted, we were unable to definitively ascertain Mrs. P0’s HIV status within the scope of this study. Hence, HIV infection Acknowledgments linked to this outbreak might not yet have been diagnosed. We are very grateful to Jennifer McGoogan for her invaluable The results of this outbreak investigation offer im - comments, suggestions, and English editing of many revisions portant lessons that China must not ignore. First, the of the paper; Roger Detels and Julio Montaner for their helpful Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2147 SYNOPSIS 8. Yang Z, Zhou C, Li X, Wang M, Tan Y. Clinical observation of comments; the national and Zhejiang provincial panels for this lymphocyte active immunization combined with dydrogesterone emergency response; and the staff of Yiwu Municipality CDC, in women with recurrent spontaneous abortion caused by blocking Hangzhou Municipality CDC, Shangcheng District CDC, ZC antibody-negative. J Reprod Med. 2014;23:128–30. Hospital, and Xixi Hospital for their participation and support 9. Liang X, Qiu T, Qiu L, Wang X, Zhao A, Lin Q. Female third party lymphocytes are effective for immunotherapy of patients in the epidemiologic investigation, as well as HIV testing, with unexplained primary recurrent spontaneous abortion: a counseling, and treatment. retrospective analysis of outcomes. Eur J Contracept Reprod Health Care. 2015;20:428–37. The study was funded by the Emergency Response Funds 13625187.2015.1046593 from the National Health and Family Planning Commission, 10. Chen JL, Yang JM, Huang YZ, Li Y. Clinical observation of People’s Republic of China; the Emergency Response Funds lymphocyte active immunotherapy in 380 patients with from the Zhejiang Health and Family Planning Commission; unexplained recurrent spontaneous abortion. Int Immunopharmacol. 2016;40:347–50. China National Technical Support and Operational Research 11. The National Health and Family Planning Commission of the for HIV/AIDS Prevention, Treatment and Care (grant no. 131- People’s Republic of China. The announcement of calling for 13-000 105-01); National Science and Technology Major cancellation of approval requirement for clinical use of third Project on Prevention and Treatment of Major Infectious category of medical services in China. Announcement no. 71 in 2015. The National Health and Family Planning Commission of Diseases Including AIDS and Viral Hepatitis (grant no. People’s Republic of China. June 29, 2015 [cited 2018 May 17]. 2012ZX10001-007); and Zhejiang Provincial Key Programs for - Science and Technology (grant no. 2013C03047-1). The 09883ae417256b3c49.shtml funding agencies were actively involved in the field and 12. The Center for Drug Evaluation, State Food and Drug Administration. The guiding principles for cell therapeutic laboratory investigations, data collection, analysis, and products used in scientific research and technical evaluation. interpretation. They did not contribute to manuscript development. The Center for Drug Evaluation, State Food and Drug Administration. December 22, 2017 [2018 Jun 25]. About the Author 13. Ou CY, Ciesielski CA, Myers G, Bandea CI, Luo CC, Korber BT, Dr. Pan is the director of the Division of HIV/STD at the et al. Molecular epidemiology of HIV transmission in a dental practice. Science. 1992;256:1165–71. Zhejiang Provincial CDC. Her research interests are monitoring science.256.5060.1165 and responding to HIV, sexually transmitted diseases, and HCV 14. Smith TF, Waterman MS. The continuing case of the Florida epidemics. Her current research priority is the HIV epidemic dentist. Science. 1992;256:1155–6. among men who have sex with men in China. science.256.5060.1155 15. DeBry RW, Abele LG, Weiss SH, Hill MD, Bouzas M, Lorenzo E, et al. Dental HIV transmission? Nature. 1993;361:691. References 1. The ESHRE Guideline Group on RPL, Atik RB, Christiansen OB, 16. Hillis DM, Huelsenbeck JP. Support for dental HIV transmission. Elson J, Kolte AM, Lewis S, et al. ESHRE guideline: recurrent Nature. 1994;369:24–5. pregnancy loss. Hum Reprod Open. 2018; 2018:1–12. 17. Machuca R, Jørgensen LB, Theilade P, Nielsen C. Molecular investigation of transmission of human immunodeficiency virus 2. Royal College of Obstetricians and Gynaecologists. The type 1 in a criminal case. Clin Diagn Lab Immunol. 2001; investigation and treatment of couples with recurrent first-trimester 8:884–90. and second trimester miscarriage. Green-top Guideline No. 17. 18. Metzker ML, Mindell DP, Liu XM, Ptak RG, Gibbs RA, 2011 [cited 2018 Oct 17]. Hillis DM. Molecular evidence of HIV-1 transmission in a documents/guidelines/gtg_17.pdf criminal case. Proc Natl Acad Sci U S A. 2002;99:14292–7. 3. American College of Obstetricians and Gynecologists. Practice bulletin: early pregnancy loss. 2015; Number 150 [cited 2018 19. Lemey P, Van Dooren S, Van Laethem K, Schrooten Y, Oct 17]. Derdelinckx I, Goubau P, et al. Molecular testing of multiple Committee-on-Practice-Bulletins----Gynecology/Public/pb150.pdf HIV-1 transmissions in a criminal case. AIDS. 2005;19:1649–58. 4. Wong LF, Porter TF, Scott JR. Immunotherapy for recurrent miscarriage. Cochrane Database Syst Rev. 2014;(10):CD000112. 20. Birch CJ, McCaw RF, Bulach DM, Revill PA, Carter JT, Tomnay J, 5. Cavalcante MB, Sarno M, Araujo Júnior E, Da Silva Costa F, et al. Molecular analysis of human immunodeficiency virus strains Barini R. Lymphocyte immunotherapy in the treatment of associated with a case of criminal transmission of the virus. recurrent miscarriage: systematic review and meta-analysis. Arch J Infect Dis. 2000;182:941–4. Gynecol Obstet. 2017;295:511–8. 21. Yirrell DL, Hutchinson SJ, Griffin M, Gore SM, Leigh-Brown AJ, s00404-016-4270-z Goldberg DJ. Completing the molecular investigation into the HIV 6. Liu Z, Xu H, Kang X, Wang T, He L, Zhao A. Allogenic outbreak at Glenochil prison. Epidemiol Infect. 1999;123:277–82. lymphocyte immunotherapy for unexplained recurrent spontaneous abortion: a meta-analysis. Am J Reprod Immunol. 2016;76:443–53. 22. Blanchard A, Ferris S, Chamaret S, Guétard D, Montagnier L. Molecular evidence for nosocomial transmission of human 7. Liu X, Tian L. Level of the blocking antibodies in women with immunodeficiency virus from a surgeon to one of his patients. recurrent spontaneous abortion and the effect of immunotherapy J Virol. 1998;72:4537–40. with lymphocytes. Chinese Journal of Healthy Birth & Child Care. 23. de Oliveira T, Pybus OG, Rambaut A, Salemi M, Cassol S, 2012;18:276–8. Ciccozzi M, et al.; Benghazi Study Group. Molecular epidemiology: 2148 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 HIV Infection Linked to Nosocomial Transmission HIV-1 and HCV sequences from Libyan outbreak. Nature. 2006;444: and after a structural intervention. PLoS Med. 2015;12:e1001874. 836–7. 24. Peters PJ, Pontones P, Hoover KW, Patel MR, Galang RR, 28. Wu Z, Tang Z, Mao Y, Van Veldhuisen P, Ling W, Liu D, et al. Shields J, et al.; Indiana HIV Outbreak Investigation Team. HIV Testing and linkage to HIV care in China: a cluster-randomised infection linked to injection use of oxymorphone in Indiana, trial. Lancet HIV. 2017;4:e555–65. 2014–2015. N Engl J Med. 2016;375:229–39. S2352-3018(17)30131-5 10.1056/NEJMoa1515195 29. Ma Y, Dou Z, Guo W, Mao Y, Zhang F, McGoogan JM, et al. 25. Persaud D, Gay H, Ziemniak C, Chen YH, Piatak M Jr, Chun TW, The human immunodeficiency virus care continuum in China: et al. Absence of detectable HIV-1 viremia after treatment cessation 1985–2015. Clin Infect Dis. 2018;66:833–9. in an infant. N Engl J Med. 2013;369:1828–35. 10.1093/cid/cix911 10.1056/NEJMoa1302976 Address for correspondence: Shichang Xia, Zhejiang Provincial Center 26. Luzuriaga K, Gay H, Ziemniak C, Sanborn KB, Somasundaran M, Rainwater-Lovett K, et al. Viremic relapse after HIV-1 remission in for Disease Control and Prevention, 3399 Binsheng Rd, Hangzhou, a perinatally infected child. N Engl J Med. 2015;372:786–8. 310051, Zhejiang, China; email:; Zunyou Wu, National Center for AIDS/STD Control and Prevention, 155 Changbai Rd, 27. Wu Z, Zhao Y, Ge X, Mao Y, Tang Z, Shi CX, et al. Simplified HIV Beijing, 102206, China; email: testing and treatment in China: analysis of mortality rates before June 2018 Zoonoses • Ferrets as Models for Influenza Virus Transmission Studies • Novel Parvovirus Related to Primate Bufaviruses in Dogs and Pandemic Risk Assessments • Novel Poxvirus in Proliferative Lesions of Wild Rodents in • Occupation-Associated Fatal Limbic Encephalitis Caused by East Central Texas, USA Variegated Squirrel Bornavirus 1, Germany, 2013 • Rickettsia parkeri in Dermacentor parumapertus Ticks, • Use of Bead-Based Serologic Assay to Evaluate Chikungunya Mexico Virus Epidemic, Haiti • Foot-and-Mouth Disease in the Middle East Caused by an • Widespread Treponema pallidum Infection in Nonhuman A/ASIA/G-VII Virus Lineage, 2015–2016 Primates, Tanzania • Novel Salmonella enterica Serovar Typhimurium Genotype • Genomic Epidemiology of Global Carbapenemase- Levels as Herald of Seasonal Salmonellosis Epidemics Producing Enterobacter spp., 2008–2014 • Urban Wild Boars and Risk for Zoonotic Streptococcus suis, • Influenza D Virus Infection in Feral Swine Populations, Spain United States • Human Endophthalmitis Caused b y Pseudorabies Virus • Prion Disease in Dromedary Camels, Algeria Infection, China, 2017 • Frequent Implication of Multistress-Tolerant Campylobacter • Pulmonary Infections with Nontuberculous Mycobacteria, jejuni in Human Infections Catalonia, Spain, 1994–2014 • Bioclinical Test to Predict Nephropathia Epidemica Severity • Westward Spread of Highly Pathogenic Avian Influenza at Hospital Admission A(H7N9) Virus among Humans, China • Hepatitis E in Long-Term Travelers from the Netherlands to • Importation of Human Seoul Virus Infection to Germany Subtropical and Tropical Countries, 2008–2011 from Indonesia To revisit the June 2018 issue, go to: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2149 SYNOPSIS Reemergence of St. Louis Encephalitis Virus in the Americas Adrián Diaz, Lark L. Coffey, Nathan Burkett-Cadena, Jonathan F. Day In support of improving patient care, this activity has been planned and implemented by Medscape, LLC and Emerging Infectious Diseases. Medscape, LLC is jointly accredited by the Accreditation Council for Continuing Medical Education (ACCME), the Accreditation Council for Pharmacy Education (ACPE), and the American Nurses Credentialing Center (ANCC), to provide continuing education for the healthcare team. Medscape, LLC designates this Journal-based CME activity for a maximum of 1.00 AMA PRA Category 1 Credit(s)™. Physicians should claim only the credit commensurate with the extent of their participation in the activity. All other clinicians completing this activity will be issued a certificate of participation. To participate in this journal CME activity: (1) review the learning objectives and author disclosures; (2) study the education content; (3) take the post-test with a 75% minimum passing score and complete the evaluation at; and (4) view/print certificate. For CME questions, see page 2400. Release date: November 14, 2018; Expiration date: November 14, 2019 Learning Objectives Upon completion of this activity, participants will be able to: • Examine the history of St. Louis encephalitis virus epidemiology in the United States • Evaluate the emergence and reemergence of St. Louis encephalitis virus in South America • Assess the reemergence of St. Louis encephalitis virus in the western United States CME Editor Deborah Wenger, MBA, Copyeditor, Emerging Infectious Diseases. Disclosure: Deborah Wenger, MBA, has disclosed no relevant financial relationships. CME Author Laurie Barclay, MD, freelance writer and reviewer, Medscape, LLC. Disclosure: Laurie Barclay, MD, has disclosed the following relevant financial relationships: owns stock, stock options, or bonds from Pfizer. Authors Disclosures: Adrián Diaz, PhD; Lark L. Coffey, PhD; Nathan Burkett-Cadena, PhD; and Jonathan F. Day, PhD, have disclosed no relevant financial relationships. We summarize and analyze historical and current data re- Americas. After the introduction of West Nile virus in 1999, garding the reemergence of St. Louis encephalitis virus activity of SLEV decreased considerably in the United States. (SLEV; genus Flavivirus) in the Americas. Historically, SLEV During 2014–2015, SLEV caused a human outbreak in Ari- caused encephalitis outbreaks in the United States; however, zona and caused isolated human cases in California in 2016 it was not considered a public health concern in the rest of the and 2017. Phylogenetic analyses indicate that the emerging SLEV in the western United States is related to the epidem- Author affiliations: Universidad Nacional de Córdoba, Córdoba, ic strains isolated during a human encephalitis outbreak in Argentina (A. Diaz); University of California, Davis, California, Córdoba, Argentina, in 2005. Ecoepidemiologic studies sug- USA (L.L. Coffey); University of Florida, Vero Beach, Florida, USA gest that the emergence of SLEV in Argentina was caused by (N. Burkett-Cadena, J.F. Day) the introduction of a more pathogenic strain and increasing populations of the eared dove (amplifying host). DOI: 2150 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Page 1 of 1 Reemergence of St. Louis Encephalitis Virus he disease known as St. Louis encephalitis (SLE) is Texas (1964). There was also a 1990 SLE epidemic in Flor- Tcaused by St. Louis encephalitis virus (SLEV), identi- ida (5), and, most recently, a 2015 epidemic in the Phoenix fied as the causative agent of a mosquitoborne viral epi- area of Arizona (Figure 1). demic in St. Louis, Missouri, USA, during the summer Epidemics of SLE are promoted by environmental of 1933 (1). SLEV is transmitted by numerous mosquito factors including summer temperature, rainfall, snowmelt, species in the genus Culex and is amplified by passerine and surface water conditions (7,8). One of the most notable and columbiform avian species (1). Phylogenetic analysis environmental drivers for SLEV activity is the cycling based on the full-length E gene sequences grouped SLEV of rainfall and drought. The wet–dry cycle can affect the strains into 8 genotypes (2). Genotypes I and II are preva- epidemiology of SLEV by forcing gravid floodwater Culex lent in the United States and genotype V is widely distrib- vectors to delay oviposition long enough to complete viral uted in South America. Other genotypes have limited dis- development (extrinsic incubation) in a single gonotrophic tribution: genotype III is in southern South America, IV is cycle, thus making them capable of viral transmission limited to Colombia and Panama, VI is in Panama, VII is in during their second blood meal (8). Drought has also been Argentina, and VIII has been detected only in the Amazon linked to urban SLE and West Nile virus (WNV) epidemics region of Brazil (2). involving vectors in the Cx. pipiens complex (9). A retrospective analysis revealed that 38 human cas- es and 14 deaths were caused by SLEV in Paris, Illinois, Transmission Cycles USA, during the summer of 1932 (3). A 1933 SLE epidem- The 4 primary vectors of SLEV in the United States are ic resulted in 1,095 clinical human cases and 201 deaths Cx. pipiens pipiens Linnaeus, Cx. pipiens quinquefasciatus (3). Because subclinical cases are not identified or reported, Say, Cx. tarsalis Coquillett, and Cx. nigripalpus Theobald retrospective serosurveys were conducted to determine the mosquitoes (1). Vector species distribution determines the ratio of subclinical to clinical infections, which was deter- geographic distribution of SLEV, and affects whether epi - mined to be 300:1 (3). Using this ratio, the actual number demics are urban or rural. Urban SLE epidemics usually of SLE cases during the 1933 SLE epidemic was ≈328,500, involve Cx. pipiens pipiens and Cx. pipiens quinquefas- affecting nearly 40% of the city’s 821,960 inhabitants, ciatus (Cx. pipiens complex) mosquitoes, species that ovi- based on US census data for 1930 (4). posit in the permanent aquatic habitats provided by storm Since SLEV was first identified, 4 human SLE trans- drains, sewage treatment facilities, and wastewater reten- mission scenarios have been reported (5). First, during tion ponds. Rural SLE epidemics usually involve floodwa - most years, no human SLE infections are reported. How- ter species, such as Cx. tarsalis and Cx. nigripalpus mos- ever, SLEV transmission to sentinel animals and virus iso- quitoes (1). Urban and rural epidemic transmission patterns lation from mosquito pools is documented in the absence are best demonstrated by the history of SLE in Florida, of human cases. Second, small numbers of spatially and where urban human SLEV cases were detected in Miami temporally isolated human SLE cases occur. For example, in 1952 and 1954, followed by St. Petersburg in 1959, in 1993, 8 human SLE cases were reported in Lee and Col- Tampa in 1961, and Sarasota in 1962. In 1977, SLEV epi- lier counties, Florida, USA; 5 of the cases reported onset demic transmission in Florida shifted from urban to rural during October (6). Third, sporadic transmission occurs as areas (7). The 1977 and 1990 Florida SLE epidemics both widely dispersed (temporally and spatially) individual hu- started in Indian River County and then spread throughout man cases. For example, in 1997, 9 human SLE cases were the Florida peninsula (7). The temporal shift from urban reported from 6 Florida counties, ranging from Brevard to rural epidemic transmission in Florida was facilitated County on the central Atlantic Coast south to Lee County by changes in Cx. nigripalpus oviposition preference and on the southern Gulf Coast. Onset for these cases ranged behavior. The Cx. nigripalpus mosquito is a widespread from July through late October (6). Finally, epidemic trans- subtropical species and a highly opportunistic blood feeder mission occurs as focused (in space and time) clusters of that oviposits in freshly flooded temporary aquatic habitats. human clinical cases: for example, an extensive 1975 epi- In urban habitats, Cx. nigripalpus mosquitoes oviposit in demic that occurred along the Mississippi and Ohio River wastewater retention ponds and open wastewater outflow basins from Ontario, Canada; Cleveland, Ohio; and Chi- ditches (10). The shift to rural transmission in 1977 and cago, Illinois, in the north to Birmingham, Alabama, and 1990 was facilitated by an increase in citrus farming, as cit- Mississippi in the south. Well-documented SLE outbreaks rus groves were designed to be maintained by flood irriga - include the 1933 St. Louis epidemic and the 1959, 1961, tion (10). In the 1970s through the 1990s, rural citrus grove 1962, 1977, and 1990 epidemics in southern Florida. Other drainage furrows became the preferred oviposition site for epidemics of note occurred in St. Louis (1937); Hidalgo Cx. nigripalpus mosquitoes (11). The proclivity of Cx. County, Texas (1954); High Plains, Texas; Louisville, Ken- nigripalpus mosquitoes to blood-feed on birds as well as tucky; and Cameron County, Texas (1956); and Houston, mammals and the tendency of females to delay oviposition Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2151 SYNOPSIS Figure 1. Geographic distribution of historical St. Louis encephalitis human cases reported in the Americas through November 2017. Dot size represents the number of human cases reported in each episode. Colors represent year of detection. until the proper aquatic oviposition habitats are created infection after rechallenge with WNV or SLEV (12). In by heavy summer rainfalls make it an excellent vector of contrast, house finches first exposed to SLEV showed low- SLEV (11). er subsequent viremias after rechallenge with WNV. This The introduction of WNV into the United States in suggests that WNV could competitively exclude SLEV 1999 promoted debate about how the presence of WNV from amplification in shared avian hosts like house finches would affect the continued transmission of SLEV, given (13). Indeed, transmission of SLEV in Florida decreased the serologic cross-reactivity of the 2 viruses in avian notably following the introduction and establishment of hosts. Laboratory studies demonstrated that house finches WNV (13). For competitive exclusion to suppress SLEV (Haemorhous mexicanus) experimentally infected with transmission, considerable host overlap between SLEV and WNV developed neutralizing immunity that prevented WNV must occur. In Florida, avian species in the families 2152 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Reemergence of St. Louis Encephalitis Virus Cardinalidae (northern cardinal, Cardinalis cardinalis), to have Argentinean hemorrhagic fever. Human encephali- Columbidae (mourning dove, Zenaida macroura), Corvi- tis cases were reported sporadically until 2005, when an un- dae (blue jay, Cyanocitta cristata), and Icteridae (common precedented outbreak of SLE was reported in Córdoba City, grackle, Quiscalus quiscula) are frequently exposed to Argentina (22). During this outbreak, signs and symptoms SLEV (14). For WNV, house sparrows (Passer domesti- associated with a neurologic infection, including headache, cus) and blue jays are highly infectious to mosquitoes; in sensory depression, temporal–spatial disorientation, tremors, contrast, mourning doves were found to be 1 of the least and changes in consciousness, were reported. A correlation competent WNV hosts of 25 bird species examined ex- between age and sign or symptom severity was detected perimentally (15). Even if mourning doves are poor WNV (Spearman coefficient = 0.74) (22). The frequency of en- hosts, being exposed to WNV may provide cross neutral- cephalomyelitis varied from 80% of the cases in patients <20 ization to SLEV and preempt amplification of SLEV, pro- years of age to 95% in those >60 years of age (22). A total of vided that doves are exposed to WNV early in the transmis- 47 probable and confirmed human cases were reported. Of sion season. In California, sparrows, finches, and corvids these, 45 patients were hospitalized; 9 died, 1 25 years of age (house finch, house sparrow, purple finch [Haemorhous and 8 >50 years of age (22). purpureus]); song sparrow (Melospiza melodia); western The SLEV strains CbaAr-4006 and CbaAr-4006 were scrub jay (Aphelocoma californica); and white-crowned isolated from Cx. pipiens quinquefasciatus mosquitoes sparrow (Zonotrichia leucophrys) are considered major collected during the outbreak in the backyard of the index amplifying hosts of WNV and SLEV (1,16). As in Florida, case-patient in Córdoba City (23). Molecular classification host overlap of WNV and SLEV likely contributed to the and phylogenetic analyses indicated that the strains isolated initial disappearance of SLEV from the western United during the outbreak were closely related to a genotype III States (17). Of these avian hosts, only house sparrows, SLEV strain (79V-2533) that had been isolated in Santa Fe house finches, American crows (Corvus brachyrhynchos), Province, Argentina (23). A 3-year retrospective phyloge- and western scrub jays were found to be amplifying hosts netic analysis of SLEV genomes from mosquitoes collect- capable of developing viremias above the threshold for in- ed in Córdoba City, designed to find an enzootic progenitor fection of Cx. tarsalis, the principal vector of WNV and of the outbreak strains, indicated no circulation of genotype SLEV throughout much of California (16). Further work III before the 2005 outbreak (24). The extensive sampling is needed to characterize the specific avian hosts fed on by effort detected low levels of SLEV transmission of geno- Cx. pipiens quinquefasciatus and Cx. tarsalis mosquitoes types other than genotype III. The CbaAr-4005 strain was in epidemic and nonepidemic settings to determine whether more virulent and produced higher viremias in avian and additional host species may be involved in amplification of murine models. In house sparrows, CbaAr-4005 produced SLEV and to what degree host overlap contributes to the viremias that were 2.2 logs higher and lasted for 2 days lon- competitive exclusion or reemergence of SLEV in areas of ger than its closest relative, the 1978 79V-2533 strain. Be- the United States outside California. cause mosquitoes show a dose response to infection, house Since the introduction of WNV into the United States, sparrows inoculated with SLEV strain CbaAr-4005 will human SLE cases continue to occur throughout the coun- theoretically produce viremia sufficient to infect 10 times try. During 2004–2013, 92 clinical SLE cases were report- more mosquitoes than those inoculated with a nonepidemic ed (18). Most cases were located in the Gulf Coast states of strain (25). Louisiana (10 cases), Mississippi (13 cases), and Texas (16 In adult Swiss mice, the CbaAr-4005 strain of SLEV cases). During 2014–2016, a total of 32 human SLE cases resulted in 100% (10/10) illness and death compared with were reported (19). Most of these cases were reported in sympatric nonepidemic SLEV strains isolated previously the Phoenix area, where a SLE epidemic resulted in 23 con- in Córdoba and Santa Fe Province, Argentina (26). Inocu- firmed human cases, including 1 fatality (19). SLEV has lation of only 1 plaque-forming unit in 10-day-old mice remained endemic throughout much of the United States or 2 plaque-forming units in 21-day-old mice caused a despite the introduction and establishment of WNV. 50% death rate with the strain CbaAr-4005 (lethal dose 50 [LD ] in 10-day-old mice = 0.02), proving it to be the most Emergence and Reemergence in South America lethal strain compared in the study (78V-6507 = 1.75; LD50 Historically, SLEV has not been considered a major public CorAn-9275 = 3.90). That evidence supports the hy- LD50 health threat in the Americas, other than in the United States. pothesis that a more virulent SLEV strain was introduced Human SLE cases have been reported sporadically through- into Córdoba City in early 2005 from Santa Fe (Argenti- out Latin America, but no human epidemics were reported na) or São Paulo (Brazil) (Figure 2). Serologic studies of until 2005. In Argentina, Charosky et al. (20) reported a neu- wild birds sampled before and during the 2005 outbreak rologic human SLEV case in 1968. Two years later, Mettler indicated that 99% (434/437) of the avian population et al. (21) isolated SLEV from a febrile human suspected lacked a SLEV-neutralizing antibody and were therefore Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2153 SYNOPSIS Figure 2. Phylogeny and spread of St. Louis encephalitis virus (SLEV) in the Americas. A) Multiple sequence alignment of 44 complete envelope protein SLEV sequences obtained from GenBank. The orange highlighted cluster contained the emerging SLEV strains isolated in Argentina, Brazil, and western United States. Alignment was performed by ClustalX, followed by tree generation using a neighbor-joining algorithm using MEGA 6 software ( Sequences are labeled by their GenBank accession number. Sequences belonging to lineage III also contain place and year of isolation data. Bootstrap support values are given for each node. Scale bar represents nucleotide substitutions per site. B) Geographic spread of SLEV. A discrete Bayesian phylogeographic reconstruction for SLEV lineage III was made using 11 envelope protein sequences (highlighted cluster in Figure 2, panel A). We applied a constant-size coalescent tree before the phylogeny and a TNF93 nt substitution model. The Monte Carlo Markov chain model was obtained after 30 million iterations and subsampling every 20.000 iterations. Analyses were made using Beast version 1.8.3 software ( Numbers over the arrows indicate the probability for the postulated viral dispersion in that pathway. Color of arrows indicates the location origin for the strain introduced. susceptible to infection with SLEV, potentially serving as alert_ epi_2010_31_marz_encefalitis_san_luis.pdf), but amplification hosts (27). The absence of natural immunity no ecologic studies focused on these outbreaks. Phyloge- in the wild avian population may have promoted the 2005 netic analysis of SLEV genomes associated with human SLE outbreak in Córdoba City. A host competence study in cases in Buenos Aires confirmed the presence of SLEV avian species showed that the eared dove (Zenaida auricu- genotype III (30). lata) and picui ground-dove (Columbina picui) produce SLE was diagnosed in a febrile human in São Paulo high levels of viremia compared with other avian species State, Brazil, in 2004, by viral isolation and molecular tested (28). Moreover, that study confirmed the role as am - detection (31). Molecular characterization classified the plifying host of these 2 dove species during the 2005 SLEV isolate as SLEV genotype III. During a dengue virus out- outbreak in Córdoba. Additional studies indicate that the break in São José do Rio Preto (São Paulo State, Brazil) population of eared doves has been increasing in the central in 2006, Mondini et al. (32) reported the first outbreak of region of Argentina during the past 10 years because of SLE in Brazil. Human cases were diagnosed by molecular agricultural geographic expansion (29). detection of SLEV RNA in serum or cerebrospinal fluid. After the 2005 outbreak, additional SLE outbreaks All SLEV-infected patients (6) had an initial diagnosis of in Argentina occurred in Parana (2006), Buenos Aires dengue fever or viral encephalitis; 3 cases were diagnosed (2010), and San Juan (2011) (Pan American Health Or- as viral meningoencephalitis, and the other 3 patients had ganization, signs of hemorrhagic disease (32). This finding was the 2154 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Reemergence of St. Louis Encephalitis Virus first reported link of SLE infection and human hemorrhag - Instead, the absence of SLEV activity suggested extirpa- ic disease (32). A blastn analysis (https://blast.ncbi.nlm. tion from California. comparing genomes of different SLEV Beginning in July 2015, SLEV activity was detect- strains, which we carried out for this review, indicates that ed by the presence of viral RNA in mosquito pools and the SLEV strains from Brazil are closely related to SLEV sentinel chicken seroconversions in the Coachella Val- genotype V. This finding indicates that genotype V SLEV ley of Riverside County, California (43). In 2016, SLEV strains are also pathogenic for humans, as shown in pre- spread northward to 6 additional counties in California vious studies in murine models (26). Because genotype V (44); in 2017, a total of 15 California counties reported strains are widespread in South America, human SLEV activity (45). A clinical study used unbiased clinical cases could be misdiagnosed as dengue virus infection. A testing by metagenomic next-generation sequencing to detailed analysis of South America human SLE cases is diagnose a fatal case of meningoencephalitis from SLEV shown in Figure 1. in a patient from Kern County, California, in September 2016 (46). Reemergence in Western United States In Arizona, SLEV detection was historically less SLEV was recognized in 1937 in California; cases were frequent than in California; low enzootic activity was documented during 1940–1950 and in Kern County reported most years during 1972–2006 and only a single during a 1952 outbreak of Western equine encepha- human case during 2010–2014 (47). In Maricopa County litis virus (33). SLEV caused periodic epidemics in (which includes Phoenix), a human SLEV outbreak dur- humans and epizootics in equines during the 1950s– ing July–October 2015 resulted in 23 confirmed cases 1990s. However, expansive epidemics similar to those (48). Three patients in that outbreak were organ trans- observed in Missouri and Texas were not detected in plant recipients who experienced fever, rigors, diarrhea, California, although concurrent activity in mosquitoes headache, and confusion; all developed meningoen- and seroconversions in birds were repeatedly docu- cephalitis, and 1 patient died (48). SLEV infection in mented in the San Joaquin Valley, Los Angeles Basin, the 2 surviving transfusion recipients was confirmed by and Imperial Valley (1,34–36). An outbreak resulting plaque reduction neutralization tests (48). Retrospective in 26 human cases occurred in 1984 in Los Angeles, testing of archived mosquito pools from Phoenix col- (34), leading to avian studies (37), vector competence lected in 2014 revealed a single SLEV isolate, indicating experiments (38), and vectoral capacity studies (39). that SLEV was present in Arizona the summer before the A SLEV serosurvey following the Los Angeles out- 2015 outbreak (48). break found that 1.6% of 1,803 serum samples tested To define the genetic relatedness of reemerging SLEV from residents of Los Angeles were seropositive. A from Arizona and California to SLEV from elsewhere to subsequent outbreak of 26 confirmed SLEV cases was infer a possible origin and pattern of spread, we performed centered in Kern County but disappeared the follow- phylogenetic analyses of genomes from mosquitoes in ing summer (35,36). Envelope gene sequencing from 2014–2016 and the fatal case in 2016 in Kern County. The mosquito isolates suggested that different SLEV strains 2014 and 2015 California and Arizona SLEV isolates share were most likely responsible for these outbreaks (40). >99% nucleotide identity with each other and also with Sequencing of longitudinal SLEV isolates from Cali- their closest published relative isolated from Cx. pipiens fornia showed that isolates were genetically similar quinquefasciatus mosquitoes collected in the 2005 epi- from 1952 through the 1970s, after which genetic demic in Córdoba City (49). The SLEV genome sequence changes were observed (41). from the fatal case in Kern County from September 2016 Even in the absence of epidemics since 1989, SLEV was >99% identical with 2014–2015 SLEV isolates from activity was documented continuously in California mosquitoes in California and Arizona, suggesting that the from the 1990s through 2003 by human cases, detections patient was infected by the reemergent genotype circulat- in mosquitoes, or seroconversions in sentinel chickens ing in the southwestern United States (46). The 2014 and (42). In 2003, WNV activity was first detected in the 2015 SLEV isolates are genetically distinct from the 2003 state (17). During 2003–2015, no SLEV activity was de- Imperial Valley, California, strain that was isolated before tected in California, despite ongoing SLEV surveillance the 11-year absence of SLEV activity in the state (49). in Culex mosquitoes and sentinel chicken seroconver- These results suggest there was likely a single introduction sions (17). The absence of SLEV activity during that of SLEV into the United States from South America, and period was likely not the result of a lack of surveillance possibly Argentina, no later than November 2014, the earli- because the invasion of WNV led to a ≈6-fold statewide est dated sample from which SLEV was isolated in Arizona increase in mosquito pool testing, from ≈5,000 pools and that the virus spread in the summer of 2015 from Ari- in 2003 to 30,000–35,000 pools annually since 2007. zona to California (49). Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2155 SYNOPSIS References Conclusion and Future Perspectives 1. Reisen WK. Epidemiology of St. Louis encephalitis virus. Arthropodborne virus infections are emerging and re- Adv Virus Res. 2003;61:139–83. emerging infectious diseases worldwide. Mosquitoborne S0065-3527(03)61004-3 viruses, including dengue (emerged in 1990), WNV 2. Rodrigues SG, Nunes MR, Casseb SM, Prazeres AS, Rodrigues DS, Silva MO, et al. Molecular epidemiology of Saint (emerged in 1999), chikungunya (2013), and Zika vi- Louis encephalitis virus in the Brazilian Amazon: genetic ruses (2015), have emerged as public health threats divergence and dispersal. J Gen Virol. 2010;91:2420–7. in the Western Hemisphere. SLEV is maintained in a bird–mosquito transmission network involving multiple 3. Chamberlain RW . History of St. Louis encephalitis. In: Monath TP, editor. St. Louis encephalitis. Washington (DC): American Public species in a broad range of ecosystems encompassing a Health Association;1980. p. 680. wide geographic distribution that ranges from southern 4. Monath TP, Tsai TF. Flaviviruses. In: Richman DD, Whitley RJ, Canada to southern Argentina. Biotic factors, including Hayden FG, editors. Clinical Virology. New York: Churchill- vector and host abundance and population age structure, Livingstone;1997. p. 1133–86. 5. Day JF, Stark LM. Frequency of Saint Louis encephalitis virus as well as abiotic factors, including rainfall and drought in humans from Florida, USA: 1990–1999. J Med Entomol. dynamics and elevated summer temperatures, combine 2000;37:626–33. to produce conditions favorable for transmission of 6. Day JF, Curtis GA. Blood feeding and oviposition by Culex SLEV. A better understanding of how SLEV circulates nigripalpus (Diptera: Culicidae) blood feeing and oviposition before, during and after a widespread St. Louis encephalitis between enzootic transmission cycles in nature and epi- epidemic in Florida.J Med Entomol. 1999;36:176–81. demic transmission in human populations is needed to more accurately predict where and when human SLEV 7. Day JF. Predicting St. Louis encephalitis virus epidemics: epidemics will emerge. lessons from recent, and not so recent, outbreaks. Annu Rev Entomol. 2001;46:111–38. The reemergence of SLEV in central Argentina is annurev.ento.46.1.111 associated with several factors, including the recent in- 8. Shaman J, Day JF, Stieglitz M. Drought-induced amplification troduction of a more virulent strain of SLEV into a na- of Saint Louis encephalitis virus, Florida.Emerg Infect Dis. ive bird community and increased populations of eared 2002;8:575–80. 9. Epstein PR, Defilippo C. W est Nile virus and drought Glob doves, a highly susceptible amplification species in Change Hum Health. 2001;2:105–7. Argentina. Argentina has experienced intense land use 10.1023/A:1015089901425 changes primarily because of the expansion of agricul- 10. Curtis GA. Habitat selection strategies of mosquitoes tural and urbanized habitats. More research is needed inhabiting citrus irrigation furrows.J Am Mosq Control Assoc. 1985;1:169–73. to define the effects of environmental change on avian 11 Day JF, Curtis GA. When it rains, they soar—and that makes reservoir and vector populations to clarify the dynam- Culex nigripalpus a dangerous mosquito.Am Entomol. ics of SLEV transmission, introduction, reintroduction, 1994;40:162–7. and reemergence in susceptible habitats throughout the 12. Fang Y, Reisen WK. Previous infection with West Nile or St. Louis encephalitis viruses provides cross protection during Western Hemisphere. reinfection in house finches. Am J Trop Med Hyg. 2006;75:480–5. The reemergence of SLEV in California and Arizona resulted from introduction of a South American strain of 13. Ottendorfer CL. Impact of the West Nile virus on the natural SLEV. The genetic diversity of SLEV in the Americas is history of St. Louis encephalitis virus in Florida [ddissertation]. Tampa (FL): University of South Florida; 2008 [cited 2018 Oct 2]. spatially influenced, with wide genetic variation across the space, but some SLEV strains from North and South Amer- ica show high genetic similarity, indicating long-range 14. Day JF, Stark LM. Avian serology in a St. Louis encephalitis dispersal. Similar to WNV, long-range SLEV dispersal is epicenter before, during, and after a widespread epidemic in south Florida, USA.J Med Entomol. 1999;36:614–24. likely mediated by migrating SLEV-infected birds. A bet- 10.1093/jmedent/36.5.614 ter understanding of SLEV in wild birds and avian host mi- 15. Komar N, Langevin S, Hinten S, Nemeth N, Edwards E, Hettler D, gration patterns is necessary to predict the spread of SLEV. et al. Experimental infection of North American birds with the New York 1999 strain of West Nile virus.Emerg Infect Dis. 2003; 9:311–22. About the Author 16. Reisen WK, Chiles RE, Martinez VM, Fang Y, Green EN. Dr. Diaz, a biologist, is an adjunct researcher at the National Experimental infection of California birds with western equine Council for Scientific and Technical Researches (CONICET) encephalomyelitis and St. Louis encephalitis viruses. and associate professor at the Instituto de Virología Dr. J.M. J Med Entomol. 2003;40:968–82. 0022-2585-40.6.968 Vanella, Universidad Nacional de Córdoba, Argentina, where 17. Reisen WK, Lothrop HD, Wheeler SS, Kennsington M, he is in charge of the Arbovirus Laboratory. His research Gutierrez A, Fang Y, et al. Persistent West Nile virus transmission interests include ecology and phylogeography of viral and the apparent displacement St. Louis encephalitis virus in zoonoses and biological interactions among vectors, hosts, southeastern California, 2003–2006.J Med Entomol. 2008; 45:494–508. and viruses. 2156 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Reemergence of St. Louis Encephalitis Virus 18. Centers for Disease Control and Prevention. Saint Louis 35. Reisen WK, Meyer RP, Milby MM, Presser SB, Emmons RW, encephalitis; 2015[cited 2018 Jul 27]. Hardy JL, et al. Ecological observations on the 1989 outbreak technical/epi.html of St. Louis encephalitis virus in the southern San Joaquin 19. Centers for Disease Control and Prevention. Saint Louis Valley of California.J Med Entomol. 1992;29:472–82. encephalitis virus: epidemiology and geographic distribution; 2017 [cited 2018 Jul 27]. 36. Tueller JE. An outbreak of illness due to St. Louis encephalitis 20. Charosky L, Baldassari EL, KorenF, Mettler NE, Loizaga C. virus in the southern San Joaquin Valley, California, 1989.Proc Síndrome encefálico a posible etiología por virus San Luis. Calif Mosq Vector Control Assoc.1990;58:12. Rev Asoc Med Argent. 1968;82:267–9. 37. McLean RG, Webb JP, Campos EG, Gruwell J, Francy DB, 21. Mettler NE, Casals J. Isolation of St. Louis encephalitis virus from Womeldorf D, et al. Antibody prevalence of St. Louis encephalitis man in Argentina.Acta Virol. 1971;15:148–54. virus in avian hosts in Los Angeles, California, 1986. J Am Mosq 22. Spinsanti LI, Díaz LA, Glatstein N, Arselán S, Morales MA, Control Assoc. 1988;4:524–8. Farías AA, et al.Human outbreak of St. Louis encephalitis 38. Hardy JL, Meyer RP, Reisen WK, Presser SB. A further evaluation detected in Argentina, 2005.J Clin Virol. 2008;42:27–33. of Culex mosquitoes in the greater Los Angeles area for their ability to vector St. Louis encephalitis virus. Proc Calif Mosq 23. DiazL A, Ré V, Almirón WR, Farías A, Vázquez A, Vector Control Assoc.1986;54:9–10. Paz Sanchez-Seco M, et al.Genotype III Saint Louis encephalitis 39. Reisen WK, Milby MM, Presser SB, Hardy JL. Ecology of virus outbreak, Argentina, 2005.Emerg Infect Dis. 2006; mosquitoes and St. Louis encephalitis virus in the Los Angeles 12:1752–4. Basin of California, 1987-1990. J Med Entomol. 1992;29:582–98. 24. Díaz LA, Albrieu Llinás G, Vázquez A, Tenorio A, Contigiani MS. Silent circulation of St. Louis encephalitis virus prior to an 40. Reisen WK, Lothrop HD, Chiles RE, Cusack R, Green EG, encephalitis outbreak in Cordoba, Argentina (2005).PLoS Negl Fang Y, et al.Persistence and amplification of St. Louis Trop Dis. 2012;6:e1489. encephalitis virus in the Coachella Valley of California, journal.pntd.0001489 2000–2001.J Med Entomol. 2002;39:793–805. 25. Diaz LA, Nemeth NM, Bowen RA, Almiron WR, Contigiani MS. 10.1603/0022-2585-39.5.793 Comparison of Argentinean Saint Louis encephalitis virus 41. Kramer LD, Presser SB, Hardy JL, Jackson AO. Genotypic and non-epidemic and epidemic strain infections in an avian model. phenotypic variation of selected Saint Louis encephalitis viral PLoS Negl Trop Dis. 2011;5:e1177. strains isolated in California. Am J Trop Med Hyg. 1997;57:222–9. journal.pntd.0001177 26. Rivarola ME, Tauro LB, Llinás GA, Contigiani MS. Virulence 42. Reisen W, Lothrop H, Chiles R, Madon M, Cossen C, Woods L, variation among epidemic and non-epidemic strains of Saint Louis et al. West Nile virus in California.Emerg Infect Dis. 2004; encephalitis virus circulating in Argentina. Mem Inst Oswaldo 10:1369–78. Cruz. 2014;109:197–201. 43. Feiszli T, Padgett K, Simpson J, Barker CM, Fang Y, Salas M, 0074-0276130475 et al. Surveillance for mosquito-borne encephalitis virus 27. Diaz LA, Quaglia AI, Konigheim BS, Boris AS, Aguilar JJ, activity in California, 2015.Proc Papers Mosq Vector Control Komar N, et al. Activity patterns of St. Louis encephalitis and Assoc Calif. 2016;84:124–9. West Nile viruses in free ranging birds during a human encephalitis 44. Feiszli T, Padgett K, Simpson J, Barker CM, Fang Y, Salas M, outbreak in Argentina.PLoS One. 2016;11:e0161871. et al. Surveillance for mosquito-borne encephalitis virus activity in California, 2016. Proc Papers Mosq Vector Control Assoc Calif. 28. Díaz A, Flores FS, Quaglia AI, Contigiani MS. Evaluation of 2017;85:9–14. Argentinean bird species as amplifying hosts for St. Louis 45. Saint Louis encephalitis virus (SLEV) in California counties 2017 encephalitis virus (Flavivirus, Flaviviridae). Am J Trop YTD; updated Jan 2018[cited 2018 Jul 27]. Med Hyg. 2018; 99:216–21. downloads.php?download_id=3839&filename=2017_slev_ ajtmh.17-0856 county_map.pdf 29. Calamari NC, Dardanelli S, Canavelli SB. Variaciones en la 46. Chiu CY, Coffey LL, Murkey J, Symmes K, Sample HA, abundancia poblacional de palomas medianas a lo largo del tiempo. Wilson MR, et al. Diagnosis of fatal human case of St. Louis INTA EEA Paraná.Serie Extensión. 2011;64:23–8. encephalitis virus infection by metagenomic sequencing, 30. Valinotto LE, Barrero PR, Viegas M, Alvarez López MC, California, 2016.Emerg Infect Dis. 2017;23:1964–8. Mistchenko AS. Molecular evidence of St. Louis encephalitis virus infection in patients in Buenos Aires, Argentina.J Clin Virol. 47. Arizona State Public Health Laboratory. Epidemiology and disease 2012;54:349–51. control.2014 [cited 2018 Jul 27]. - 31. Santos CL, Sallum MA, Franco HM, Oshiro FM, Rocco IM. ness/epidemiology-disease-control/index.php. Genetic characterization of St. Louis encephalitis virus isolated 48. Venkat H, Krow-Lucal E, Hennessey M, Jones J, Adams L, from human in São Paulo, Brazil.Mem Inst Oswaldo Cruz. 2006; Fischer M, et al. Concurrent outbreaks of St. Louis encephalitis 101:57–63. virus and West Nile virus disease—Arizona, 2015. MMWR Morb 32. Mondini A, Cardeal IL, Lázaro E, Nunes SH, Moreira CC, Mortal Wkly Rep. 2015;64:1349–50. Rahal P, et al.Saint Louis encephalitis virus, Brazil.Emerg Infect 49. White GS, Symmes K, Sun P, Fang Y, Garcia S, Steiner C, Dis. 2007;13:176–8. et al. Reemergence of St. Louis encephalitis virus, California, 33. ReevesWC, Hammon WM, Longshore WA Jr, McCLUREHE, 2015.Emerg Infect Dis. 2016;22:2185–8. Geib AF. Epidemiology of the arthropod-borne viral encephalitides eid2212.160805 in Kern County, California, 1943-1952. Publ Public Health Univ Calif. 1962;4:1–257. Address for correspondence: Adrián Diaz, National University of 34. Murray RA, Habel LA, Mackey KJ, Wallace HG, Peck BA, Córdoba–Arbovirus Laboratory, Institute of Virology Dr. J.M. Vanella, Mora SJ, et al. Epidemiological aspects of the 1984 St Louis School of Medicine, Enfermera Gordillo Gómez s/n, CPX5016GCA, encephalitis epidemic in southern California. Proc Calif Mosq Vector Control Assoc.1985;53:5–9. Córdoba, Argentina; email: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2157 SYNOPSIS Autochthonous Human Case of Seoul Virus Infection, the Netherlands Caroline Swanink, Johan Reimerink, Jet Gisolf, Ankje de Vries, Mark Claassen, Liesbeth Martens, Toos Waegemaekers, Harry Rozendaal, Stasja Valkenburgh, Tabitha Hoornweg, Miriam Maas Orthohantaviruses are a group of rodentborne viruses with cases of SEOV in wild and pet rats have been described a worldwide distribution. The orthohantavirus Seoul virus in the United Kingdom, Belgium, Sweden, and France (SEOV) can cause hemorrhagic fever with renal syndrome (3–6). Additional outbreaks have been reported in Canada in humans and is distributed worldwide, like its reservoir host, and the United States (7). The first non–laboratory-related the rat. Cases of SEOV in wild and pet rats have been de- human infections with SEOV in Europe were reported in scribed in several countries, and human cases have been the United Kingdom and France in 2012, although retro- reported in the United Kingdom, France, Canada, and the spectively earlier cases might have occurred (8–10). From United States. In the Netherlands, SEOV has previously been 2013 on, multiple additional human cases were reported in found in wild brown rats. We describe an autochthonous hu- the United Kingdom and France (3,11,12). In the United man case of SEOV infection in the Netherlands. This patient States, domestic cases of HFRS attributable to SEOV have had nonspecific clinical symptoms of an orthohantavirus infec - been described since 1994 (13). In the Netherlands, SEOV tion (gastrointestinal symptoms and distinct elevation of liver enzymes). Subsequent source investigation revealed 2 po- has been reported in wild rats (14,15). In this article, we tential sources, the patient’s feeder rats and a feeder rat farm. describe an autochthonous human case of SEOV infection At both sources, a high prevalence of SEOV was found in the in the Netherlands and the subsequent source investigation. rats. The virus closely resembled the Cherwell and Turckheim SEOV strains that were previously found in Europe. Case Description In September 2016, a 28-year-old man sought medical care at Rijnstate Hospital (Arnhem, the Netherlands); he report- rthohantaviruses are a family of rodentborne viruses ed having fever, vomiting, abdominal cramps, and diarrhea Owith a worldwide distribution. Human infection can of 7 days’ duration. He had a history of gamma-hydroxubu- occur when virus-contaminated aerosols of rodent excreta tyrate (GHB) addiction and tobacco use (10 cigarettes/d). are inhaled while entering or cleaning rodent-infested ar- He denied drinking alcohol. He had no history of travel. eas (1). Infection can also be transmitted by rodent bites or The patient mentioned that he had been bitten regularly when orthohantavirus contaminated materials are directly while handling live rats at his work at a rat breeding farm introduced into broken skin or conjunctiva. The clinical and by live rats that he kept for his reptiles at home. He syndromes that are associated with severe disease are hem- also mentioned he had been swimming in the Rhine River, orrhagic fever with renal syndrome (HFRS) and orthohanta- in which rats can be found. Physical examination revealed virus cardiopulmonary syndrome (HCPS). HFRS cases are a sweating, obese, ill patient with normal blood pressure, found in large parts of Europe and Asia, whereas HCPS cas- tachycardia (113 beats/min), and a temperature of 38.6°C. es are found in North America and South America (2). The No abnormalities on auscultation of heart and lungs or orthohantavirus Seoul virus (SEOV) causes HFRS of me- lymphadenopathy were found. Abdominal examination dium severity and was originally found in Asia. In Europe, revealed a painful enlarged liver. No abnormalities of the Author affiliations: Rijnstate Hospital, Arnhem, the Netherlands skin were recorded. We have summarized the patient’s (C. Swanink, J. Gisolf, M. Claassen); National Institute for Public laboratory test results (Table 1). Ultrasound revealed nor- Health and the Environment, Bilthoven, the Netherlands mal aspect of liver and gall bladder and a slightly enlarged (J. Reimerink, A. de Vries, T. Hoornweg, M. Maas); Municipal spleen (17.2 cm). The patient was hospitalized with an Health Service Gelderland-Midden, Arnhem (L. Martens, initial diagnosis of gastroenteritis, colitis, or leptospirosis. T. Waegemaekers); Dutch Food and Consumer Products Safety Antibiotic treatment with cefuroxime, metronidazole, and Authority, Utrecht, the Netherlands (H. Rozendaal, S. Valkenburgh) doxycycline was started. Blood cultures remained negative. A serum sample from the patient, taken at admission, tested DOI: 2158 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Human Seoul Virus Infection, the Netherlands Table 1. Laboratory test results for a patient diagnosed with Seoul virus infection, the Netherlands, September 2016* Laboratory test Reference range Day 1† Day 3‡ C-reactive protein, mg/L <10 32 59 Leukocytes, 10 cells/L 4.0–11.0 5.0 12.3 Lymphocytes, 10 cells/L 1.0–3.5 NT 8.27 Atypical lymphocytes – NT + Platelets, 10 cells/L 150–400 72 79 Creatinine, µmol/L 60–110 78 72 Alanine aminotransferase, U/L <45 114 211 Aspartate aminotransferase, U/L <35 123 283 Lactate dehydrogenase, U/L <250 753 1906 Bilirubin, µmol/L <17 12 10 Creatinine kinase, U/L <170 NT 677 *NT, not tested; –, negative; +, positive. †Day 1 of hospitalization, 7 days after onset of symptoms. ‡Day 3 of hospitalization, 9 days after onset of symptoms. negative for hepatitis A, B C, and E viruses; HIV; Trepo- collected by vortexing and centrifugation of the rat hearts, nema pallidum (syphilis); cytomegalovirus; Epstein-Barr as described previously (15). Antibodies in rat serum were virus; and Leptospira spp. The patient did not have signs of detected by using a human SEOV ELISA (Hantavirus Do- acute kidney injury and showed only a mild proteinuria of brava/Hantaan IgG Elisa; Progen Biotechnik GmbH, Hei- 0.25 g/L in a single urine sample. He was tested for ortho- delberg, Germany), which was adapted to enable detection hantavirus infection because he mentioned that he was bitten of IgG in rats. Rabbit-α-rat horseradish peroxidase-labeled by rats regularly. The father of the patient, who took care of IgG (Sigma-Aldrich Chemie B.V., Zwijndrecht, the Neth- the reptiles occasionally, and the patient’s partner, who did erlands) was used as conjugate at a 1:5,000 dilution. A cut- not have any contact with the reptiles, were not feeling ill. off value was based on the average OD of negative control rat serum + 3 × SD (in this case, a value of 0.2–0.3). Material and Methods For euthanized and frozen rats, lung tissue was col- lected in RNAlater (Applied Biosystems, Foster City, CA, Human SEOV Diagnostics USA) and stored at −80°C. Lung tissue was disrupted in For detection of hantavirus IgG and IgM, we used an im- MagNA Pure 96 External Lysis Buffer (Roche) by using munofluorescent assay (IFA) with mosaic slides containing Lysis matrix D (MP Biomedicals, Santa Ana, CA, USA) SEOV and other orthohantaviruses Puumala virus, Sin Nom- and Fast Prep FP120 homogenizer (Thermo Savant, Carls- bre virus, Hantaan virus, Dobrava-Belgrade virus, and Saa- bad, CA, USA). Total nucleic acid was isolated by auto- remaa virus (Euroimmun, Lübeck, Germany), according to mated nucleic acid extraction by using the MagNA Pure the manufacturer’s instructions (16). The titer was defined as 96 system (Roche). As a first screening, a hantavirus ge- the last sample dilution for which the fluorescence was iden - nus–specific rRT-PCR was performed on lung tissue, as tifiable, and a titer >1:32 was considered positive. Serum described previously. Subsequently, a selection of the sam- samples of the patient and 2 close contacts were tested for ples was confirmed with a nested rRT-PCR assay of the hantavirus antibodies. Total nucleic acid was extracted from large (L) segment, as described by Klempa et al. (18). The patient serum by using the MagNAPure 96 system (Roche, resulting fragments were purified with ExoSAP-IT PCR Basel, Switzerland) and tested for SEOV RNA by using a clean-up (Isogen Life Science, Utrecht, the Netherlands) hantavirus genus–specific real-time reverse transcription and sequenced by Baseclear (Leiden, the Netherlands). For PCR (rRT-PCR), as described by Kramski et al. (17). clarity, details of the selection of rats and the subsequent experimental procedures are summarized in Figure 1. Investigation of Feeder Rats Owned by the Patient Because the patient kept feeder rats at home and these Investigation of Rats from the Feeder Rat Breeding Farm rats are a known source of SEOV infection, the rats were The feeder rats from the patient’s home likely originated collected for source investigation. At the time of investi- from the feeder rat breeding farm where the patient vol- gation, the patient had 5 live and 5 frozen feeder rats at unteered. At the time of the investigation, the feeder rat home. The rats were housed in a domestic residence and breeding farm housed 8,000–9,000 rats, 7 rabbits, and ap- were 7–13 months old. All likely originated from a feed- proximately 30 gerbils, 100 mice, and 60 snakes (constric- er rat breeding farm, where the patient worked regularly tors). Adult rats at the farm were housed in 40 open boxes as a volunteer, although the patient gave contradicting that measured 1 m × 1 m. About 30 rats were kept per box, information about this. All available rats were tested for of which ≈5–10 were male and 20–25 were female. Smaller SEOV virus. The 5 live rats were euthanized, and serum and boxes were available for pregnant female rats, female rats lung tissues were collected. For the frozen rats, serum was with pups, and juvenile rats. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2159 SYNOPSIS Figure 1. Flowchart depicting the selection and subsequent testing of feeder rats in a source investigation following detection of a human case of Seoul virus infection, the Netherlands, September, 2016. *Rats were randomly picked; †rats were randomly picked from among Seoul virus–positive animals. RT- PCR, reverse transcription PCR. Sixty rats, of which 40 were adults (age 7–13 months) days, platelet count and liver enzyme test improved, and he and 20 were juveniles (age 4–6 weeks), were collected. One was discharged. A second serum sample for detection of an- adult rat per box was picked randomly out of each box for tibodies to orthohantaviruses was taken 3 weeks later when this study. Five juvenile rats were collected randomly from the patient had made a full recovery. 4 different boxes that contained the juvenile rats. When The patient was found to be positive in the orthohan- handling the rats, researchers used face masks (protective tavirus IFA with the highest IgG and IgM titers to SEOV class FFP-2), disposable gloves, and coveralls. Serum and (Table 2). Orthohantavirus RNA was not detected in the lung samples of the adults and juveniles were collected and serum sample. Two close contacts of the patient tested neg- analyzed as described previously. ative in serologic testing for orthohantaviruses. The first The owner of the rat farm did not keep a record of where close contact was the father of the index patient. While the he bought and sold his rats. He reported that he regularly sold index patient was hospitalized, his father had fed several rats to several feeder rat breeding farms within the Nether- rats to the reptiles. The second close contact was the cohab- lands. These farms were subsequently contacted, but their iting partner of the index patient and reported no contact management would not cooperate with the investigation. with the rats. Sequencing Feeder Rats from Patient From 1 rat owned by the patient and 1 rat from the breeding Of the 10 rats collected from the patient’s home, 6 (2/5 farm, the complete SEOV genome (i.e., the small, medium fresh and 4/5 frozen) rats were found positive by serologic [M], and L segments) was sequenced. Primers were devel- testing and rRT-PCR (Table 3). All 5 fresh rats were tested oped based on published sequences and are available on in the nested rRT-PCR assay of the L segment, and again, request. All fragments were purified with ExoSAP-IT PCR the same 2 rats were positive. clean-up (Isogen Life Science) and sequenced by Baseclear. Feeder Rats Breeding Farm Results Of 60 rats purchased from the rat breeding farm, 4 juveniles died from poor condition before they could be euthanized. Patient and Close Contacts The remaining 40 adults and 16 juveniles were tested using Testing found high antibody titers to the orthohantaviruses, ELISA and rRT-PCR. All 40 adult rats were seropositive especially SEOV, in the patient. The patient improved in 4 for orthohantaviruses. Lung tissues of all adult rats tested 2160 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Human Seoul Virus Infection, the Netherlands Table 2. Serologic orthohantavirus results for the patient with Seoul virus infection and 2 close contacts, the Netherlands, September 2016* Age, SEOV SEOV DOBV DOBV SAAV SAAV HNTV HNTV PUUV PUUV SNV SNV Sample y IgG IgM IgG IgM IgG IgM IgG IgM IgG IgM IgG IgM Patient 28 1:16,384 1:4,096 1:2,048 1:2,048 1:512 1:1,024 1:16,384 1:512 1:64 1:32 1:512 1:64 sample 1† Patient 1:32,768 1:2,048 1:8,192 1:512 1:2,048 1:256 1:16,384 1:1,024 1:512 1:64 1:512 1:128 sample 2‡ Close 59 – – – – – – – – – – – – contact 1 (father) Close 27 – – – – – – – – – – – – contact 2 (partner) *DOBV, Dobrava-Belgrade orthohantavirus; HNTV, Hantaan orthohantavirus; PUUV, Puumala orthohantavirus; SAAV, Saaremaa orthohantavirus; SEOV, Seoul orthohantavirus; SNV, Sin Nombre orthohantavirus; –, negative. †Taken 7 days after onset of symptoms. ‡Taken 31 days after onset of symptoms (reconvalescent). positive for SEOV RNA by rRT-PCR. A selection of 5 liver enzymes. Although the patient did not develop acute adult rats was tested with the nested rRT-PCR assay of the kidney injury, we found mild proteinuria and thrombocyto- L segment, and all 5 rats were positive. penia, which might also be found in SEOV infection. Be- Of the juveniles, 1 of the 16 was found to be sero- cause SEOV infections are related to a mild form of HFRS, positive. However, all 16 were orthohantavirus negative by with low incidence of hemorrhagic manifestations and low rRT-PCR on lung tissues (Table 3). mortality rates, previous cases might have been missed or misdiagnosed as viral infection or gastroenteritis. Elevation Sequencing of liver enzymes has been reported to occur to a greater The 7 SEOV-positive rats (2 from the patient and 5 from extent in SEOV infections but in this patient might have the rat breeding farm) tested by nested rRT-PCR showed been attributable to use of medication (acetaminophen) or identical sequences of the L segment. The complete SEOV gamma-hydroxubutyrate. Diagnosis of HFRS attributable genome was sequenced from 1 of the patient’s rats and 1 to SEOV infection is largely based on serologic testing. breeding farm rat. Sequences were submitted to GenBank The patient had high SEOV IgG and IgM titers (accession nos. MG764078–83). The phylogenetic tree of (1:16,384 for IgG and 1:4,096 for IgM) by IFA. Cross-reac- the small segments (Figure 2) shows the strains are 100% tions exist within 2 defined serogroups of orthohantaviruses identical. Furthermore, these strains are 100% identical to but are limited between the serogroups. Antibodies against the Turckheim strain isolated from pet rats in France (11) Puumala virus and Sin Nombre virus cross-react, whereas and 99.6% identical to the Cherwell strain (3) isolated from SEOV as member of the other serogroup cross-reacts with pet rats in the United Kingdom. Also, the M segment were Hantaan virus, Dobrava-Belgrade virus, and Saaremaa vi- identical to each other and 99.8% identical to the Cherwell rus. This investigation found a clear link from the patient strain. The L segment was 99.6% identical to the Cherwell to the SEOV-infected rats at home and at the breeder farm. strain. The sequence of the M and L segments of the Turck- Combined with the high antibody titers against SEOV, we heim strain were not available for comparison. concluded that the patient’s positive IFA result was due to a recent SEOV infection. Discussion The patient’s rats probably originated from the breed- SEOV was detected in wild brown rats in the Netherlands ing farm where the patient worked as a volunteer, which is in 2013, but no human cases had been reported. In this also suggested by the sequence results. Management at the article, we report an autochthonous case of SEOV infec- farm facilitated spread of SEOV by frequently moving rats tion in the Netherlands. The case-patient had nonspecific from one box to the other and randomly returning female clinical symptoms of an orthohantavirus infection, show- rats to boxes after they had weaned their pups. No regis- ing gastrointestinal symptoms and distinct elevation of tration of these movements was recorded. The tested rats Table 3. Results of Seoul virus tests in rats from the patient’s residence and the rat breeding farm, the Netherlands, September 2016 No. (%) rats found to be Source Tested rats No. (%) seropositive rats positive by rRT-PCR Feeder rats of the patient, n =10 5 fresh adults 2 (40) 2 (40) 5 frozen adults 4 (80) 4 (80) Feeder rats from the farm, n = 8,000–9,000 40 adults 40 (100) 40 (100) 16 juveniles 1 (6) 0 (0) *rRT-PCR, real-time reverse transcription PCR. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2161 SYNOPSIS Figure 2. Phylogenetic tree calculated for the coding region (1,290 bp) of the small segment of the nucleocapsid protein in the Seoul virus strain implicated in a human case infection in the Netherlands, September 2016, compared with reference viruses. Boldface indicates isolates from this study; GenBank accession numbers are provided for reference viruses. were in a poor body condition, and many had bite wounds notification of orthohantavirus infection in humans is on their ears, providing ideal transmission conditions for mandatory but not in animals. Therefore, no legislation SEOV (19). Several health risks (including public health on control measures (e.g., enforced quarantine of infected risks) were identified with regard to the SEOV-positive animals, a ban on selling rats, or a forced closure of the feeder rat breeding farm, including the possibility of infec- breeding farm) was in effect. These limitations complicated tion of personnel and visitors to the farm and the potential source investigation because testing of the breeder rat farm spread of the pathogen by regular trade and exchange with was based on voluntary cooperation of the farm owner; they other feeder rat breeding farms and occasionally with un- also severely complicated our efforts to contain this virus. known parties, possibly including pet stores. Also, SEOV Duggan et al. showed that the seroprevalence of an- might spread from the feeder rats to wild rats living around tibodies to SEOV in persons with a high contact rate with the farm through direct contact or through contaminated rats, such as rat owners, is 34%, compared with 3% in con- materials (e.g., bedding material). Anyone entering the trols with occupational exposure to pet fancy rats or wild breeding farm was advised to wear disposable gloves, shoe rats (12). Naturally, this probability depends on the spread sleeves, an apron, and a face mask (protective class FFP-2) of SEOV in the domestic rat populations in the Nether- and to wash their hands immediately after leaving the barn. lands. To what extent SEOV is present in rat populations in They were also warned to avoid being bitten by the rats. All the Netherlands is unknown. Hantavirus-infected rodents employees and volunteers were offered a blood test to see do not show any overt symptoms and might spread ortho- if they had been infected; no one accepted. Anyone buying hantavirus for a prolonged period, possibly lifelong (20– or otherwise taking rats from the breeding farm was given 23). However, the presence in feeder rats and anecdotal a letter, signed by the Municipal Health Services, inform- information about exchange between rat populations sug- ing them of the possibility of the rats being infected. Pest gest that SEOV might be present in captive rat populations control measures were set up, preventing contact of wild in the Netherlands. Also, exchange of pet and feeder rats rats with feeder rats. The distribution of the used sawdust between countries in Europe might be extensive, which is over nearby farm lands was discontinued. Used sawdust supported by the close resemblance of the SEOV strain in was thereafter brought to the municipal facility to be burnt the Netherlands to the Cherwell strain in the United King- in a waste disposal facility. By law in the Netherlands, dom and the Turckheim strain in France. In Europe, SEOV 2162 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Human Seoul Virus Infection, the Netherlands 9. Macé G, Feyeux C, Mollard N, Chantegret C, Audia S, Rebibou JM, has been detected in pet rats in England and Wales (13), et al. Severe Seoul hantavirus infection in a pregnant woman, Sweden (5), and France (11). France, October 2012. Euro Surveill. 2013;18:20464. This case illustrates the importance of clinical aware- 10. McElhinney LM, Marston DA, Pounder KC, Goharriz H, Wise EL, ness for orthohantavirus infections after contact with ro- Verner-Carlsson J, et al. High prevalence of Seoul hantavirus in a breeding colony of pet rats. Epidemiol Infect. 2017;145:3115–24. dents, including in patients with nonspecific symptoms, and the challenges that arise when source investigation and 11. Reynes JM, Carli D, Bour JB, Boudjeltia S, Dewilde A, implementation of control measures are hampered by lack Gerbier G, et al. Seoul virus infection in humans, France, 2014– of legislation. The source investigation and implementation 2016. Emerg Infect Dis. 2017;23:973–7. eid2306.160927 of control measures required multidisciplinary, constructive 12. Duggan JM, Close R, McCann L, Wright D, Keys M, McCarthy N, cooperation between research institutions and authorities. et al. A seroprevalence study to determine the frequency of Future studies to assess the extent of SEOV infection in the hantavirus infection in people exposed to wild and pet fancy rats in domestic rat populations in the Netherlands are needed to England. Epidemiol Infect. 2017;145:2458–65. 10.1017/S0950268817001480 inform the general public concerning the risk for contracting 13. Glass GE, Watson AJ, LeDuc JW, Childs JE. Domestic cases of this virus by handling rats and the related health risks. hemorrhagic fever with renal syndrome in the United States. Nephron. 1994;68:48–51. About the Author 14. Goeijenbier M, Verner-Carlsson J, van Gorp EC, Rockx B, Koopmans MP, Lundkvist Å, et al. Seoul hantavirus in brown rats in Dr. Swanink is a clinical microbiologist at Rijnstate Hospital, the Netherlands: implications for physicians—epidemiology, clinical Arnhem and Velp, the Netherlands. Her primary research aspects, treatment and diagnostics. Neth J Med. 2015; 73:155–60. interests are clinical virology, hepatitis, HIV, immune- 15. Verner-Carlsson J, Lõhmus M, Sundström K, Strand TM, compromised patients, and epidemiology. Verkerk M, Reusken C, et al. First evidence of Seoul hantavirus in the wild rat population in the Netherlands. Infect Ecol Epidemiol. 2015;5:27215. References 16. Lederer S, Lattwein E, Hanke M, Sonnenberg K, Stoecker W, 1. de St Maurice A, Ervin E, Schumacher M, Yaglom H, Lundkvist Å, et al. Indirect immunofluorescence assay for the simultaneous detection of antibodies against clinically important VinHatton E, Melman S, et al. Exposure characteristics of hantavirus pulmonary syndrome patients, United States, 1993– old and new world hantaviruses. PLoS Negl Trop Dis. 2015. Emerg Infect Dis. 2017;23:733–9. 2013;7:e2157. eid2305.161770 17. Kramski M, Meisel H, Klempa B, Krüger DH, Pauli G, 2. Peters CJ, Simpson GL, Levy H. Spectrum of hantavirus Nitsche A. Detection and typing of human pathogenic hantaviruses by real-time reverse transcription-PCR and pyrosequencing. infection: hemorrhagic fever with renal syndrome and hantavirus pulmonary syndrome. Annu Rev Med. 1999;50:531–45. Clin Chem. 2007;53:1899–905. 2007.093245 3. Jameson LJ, Taori SK, Atkinson B, Levick P, Featherstone CA, 18. Klempa B, Fichet-Calvet E, Lecompte E, Auste B, Aniskin V, van der Burgt G, et al. Pet rats as a source of hantavirus in England Meisel H, et al. Hantavirus in African wood mouse, Guinea. Emerg Infect Dis. 2006;12:838–40. and Wales, 2013. Euro Surveill. 2013;18:20415. 4. Plyusnina A, Heyman P, Baert K, Stuyck J, Cochez C, Plyusnin A. 19. Glass GE, Childs JE, Korch GW, LeDuc JW. Association of Genetic characterization of Seoul hantavirus originated from Nor- intraspecific wounding with hantaviral infection in wild rats way rats (Rattus norvegicus) captured in Belgium. (Rattus norvegicus). Epidemiol Infect. 1988;101:459–72. J Med Virol. 2012;84:1298–303. 20. Heyman P, Vaheri A, Lundkvist A, Avsic-Zupanc T. Hantavirus jmv.23321 5. Lundkvist A, Verner-Carlsson J, Plyusnina A, Forslund L, infections in Europe: from virus carriers to a major public-health Feinstein R, Plyusnin A. Pet rat harbouring Seoul hantavirus in problem. Expert Rev Anti Infect Ther. 2009;7:205–17. Sweden, June 2013. Euro Surveill. 2013;18:20521. 10.2807/1560-7917.ES2013.18.7.20521 21. Meyer BJ, Schmaljohn CS. Persistent hantavirus infections: characteristics and mechanisms. Trends Microbiol. 2000;8:61–7. 6. Dupinay T, Pounder KC, Ayral F, Laaberki MH, Marston DA, Lacôte S, et al. Detection and genetic characterization of Seoul virus from commensal brown rats in France. Virol J. 2014;11:32. 22. Easterbrook JD, Klein SL. Immunological mechanisms mediating hantavirus persistence in rodent reservoirs. PLoS Pathog. 7. Kerins JL, Koske SE, Kazmierczak J, Austin C, Gowdy K, 2008;4:e1000172. 23. Voutilainen L, Sironen T, Tonteri E, Bäck AT, Razzauti M, Dibernardo A, et al.; Seoul Virus Working Group; Canadian Seoul Virus Investigation Group (Federal); Canadian Seoul Virus Karlsson M, et al. Life-long shedding of Puumala hantavirus in Investigation Group (Provincial); Contributors. Outbreak of Seoul wild bank voles (Myodes glareolus). J Gen Virol. 2015;96:1238– virus among rats and rat owners—United States and Canada, 2017. 47. MMWR Morb Mortal Wkly Rep. 2018;67:131–4. 10.15585/mmwr.mm6704a5 Address for correspondence: Caroline Swanink, Rijnstate Hospital, 8. Jameson LJ, Logue CH, Atkinson B, Baker N, Galbraith SE, Department of Medical Microbiology and Immunology, President Carroll MW, et al. The continued emergence of hantaviruses: Kennedylaan 100, 6883 AZ, Velp, the Netherlands; email: isolation of a Seoul virus implicated in human disease, United Kingdom, October 2012. Euro Surveill. 2013;18:4–7. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2163 SYNOPSIS Restaurant Inspection Letter Grades and Salmonella Infections, New York, New York, USA Melanie J. Firestone, Craig W. Hedberg Rates of Salmonella infection in the United States have not cause of foodborne illness hospitalizations and death (1). changed over the past 20 years. Restaurants are frequent Reducing Salmonella infections is a Healthy People 2020 settings for Salmonella outbreaks and sporadic infections. objective (5), yet rates have not substantially changed over Few studies have examined the effect of posting letter the last 20 years (6). Inspection of food service establish- grades for restaurant inspections on the incidence of food- ments to protect food safety is a core function of state and borne illness. We compared Salmonella infection rates in local health departments; the inspections help to identify New York, New York, USA (NYC), with those in the rest of risk factors for foodborne illness, such as those associated New York state before and after implementation of a letter with Salmonella transmission, and to correct them, thus grade system for restaurant inspections in NYC. We calcu- protecting consumers and industry. lated a segmented regression model for interrupted time se- New York does not have a statewide policy for report- ries data. After implementation of letter grading, the rate of Salmonella infections decreased 5.3% per year in NYC ver- ing restaurant inspection results. In 2005, the Department sus the rest of New York state during 2011–2015, compared of Health and Mental Hygiene (DOHMH) in New York, with the period before implementation, 2006–2010. Posting NY, USA (NYC), began using a point-scoring system for restaurant inspection results as letter grades at the point of food service establishment inspections to weight violations service was associated with a decline in Salmonella infec- to reflect risk factors for foodborne illness. In 2010, the tions in NYC and warrants consideration for broader use. NYC DOHMH implemented a letter grade program that converted these scores into categorical rankings of A, B, or C, or grade pending, in an effort to improve restaurant ach year, an estimated 48 million persons get sick, food safety, increase transparency of inspection informa- E128,000 are hospitalized, and 3,000 die from foodborne tion, and reduce the risk for foodborne illness transmission illness in the United States (1). Consumption of food pre- in restaurants (7). The DOHMH required establishments pared away from home, such as at restaurants and for take- to post a sign with the letter grade in its window so that out, has continuously increased, accounting for 50% of food consumers could see it before entering. Public disclosure expenditures in 2014, up from 33% in 1970 ( 2). Restaurants programs like this one seek to provide information to con- are frequent settings for transmission of foodborne illness; in sumers when and where they need it so they can make in- 2015, 60% of foodborne illness outbreaks were associated formed decisions about potential risks (8). These consumer with restaurants (3). There is also evidence that commercial decisions encourage restaurant operators to improve and food service establishments, such as restaurants, play a role maintain sanitary standards, thus improving sanitary con- in sporadic (nonoutbreak) cases of foodborne illness (4). ditions in restaurants. The letter grade program has already Given our dependence on food prepared away from home, been shown to lead to improvements in sanitary conditions reducing the risk for foodborne illness from commercial in NYC: 35% more restaurants earned A grades in the 3 food service establishments is of critical importance. years after grading, compared with the 3 years before (9). Food service establishments play a role in the epidemi- The value of posting restaurant inspection ratings at ology of Salmonella infections; Salmonella may contami- the point of service has been the subject of considerable de- nate a wide range of raw ingredients, infect food workers, bate. Few studies have looked at the impact of posting poli- survive on contaminated surfaces, and grow in improperly cies on the incidence of foodborne illness. Two studies of held food items. Nontyphoidal Salmonella infections are a letter grade program in Los Angeles County, California, the second most common foodborne illness and the leading USA, showed a reduction in foodborne-illness hospitaliza- Author affiliations: University of Minnesota School of Public tions (10,11). In NYC, a preliminary analysis of letter grad- Health, Minneapolis, Minnesota, USA; Minnesota Integrated Food ing at 18 months suggested a decline in Salmonella infec- Safety Center of Excellence, Minneapolis tions. The goal of our study was to compare the incidence of Salmonella infections in NYC with incidence in the rest DOI: 2164 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Inspection Grades and Salmonella, New York of the state before and after the implementation of posting and region by trend before and after policy implementation, letter grade placards at the point of service. to determine if the average Salmonella infections in NYC changed after policy implementation versus NYS (14). We Methods also conducted a subanalysis using data from 2000–2015 to compare NYC to surrounding counties (NYC metropoli- Data tan area) and NYS. We analyzed data using Stata version Salmonella infections are nationally notifiable (12). We 14.2 (StataCorp LLC, College Station, TX, USA) and de- obtained yearly laboratory-confirmed case counts from termined statistical significance as p<0.05. 1994–2015 from the NYC DOHMH and the New York State Department of Health. Cases are reported by year of Results diagnosis and county of residence. The annual rate of Salmonella infections decreased in both NYC and NYS during 1994–2015 (Table 1; Figure). The period-to-period percent change after letter grading imple- Statistical Analyses mentation was a decline of 32.6% in NYC, compared with a To account for population changes, we calculated annual decline of 14.1% in NYS (Table 1). Mean Salmonella infec- rates using intercensal population estimates for 1994–2015 tion rates in NYC between 1994 and 2010 were significantly from the US Census Bureau. We calculated the percent higher (p<0.01) than in NYS (Table 2). In the period after change from year to year. We compared mean rates of letter grading was implemented (2011–2015), the mean rate Salmonella infection before and after implementation of a of Salmonella infection was no longer significantly differ - point scoring system in 2005 and after implementation of ent (p = 0.37) in NYC (mean 12.6 cases/100,000 persons; grade cards in 2010 using t-tests for NYC and the rest of the 95% CI 10.9–14.4) compared with NYS (mean 12.0 cas - state (NYS). We considered the year of implementation to es/100,000 persons; 95% CI 11.4–12.6). be a part of the before period in both analyses because the In NYC and NYS, Salmonella infections were de- policies were not implemented on January 1 of each year. creasing before either policy was implemented in NYC We used segmented regression to determine the trend (IRR 0.95; 95% CI 0.94–0.96; p<0.01). The interaction before implementing the policy in Salmonella infections, an term for trend by region after letter grade implementation immediate change at the time of policy implementation, and was statistically significant (p<0.01), which indicates that the long-term trend after policy implementation (13). We Salmonella infection rates declined on average in NYC hypothesized that the long-term trend would decline after versus NYS in the years after letter grading was imple- policy implementation. We expected a delayed effect be - mented, compared with the period after the point scoring cause restaurants are not all on the same inspection cycle and system was implemented (Table 3). After letter grading because underlying improvements in sanitation driving the was implemented, the rate of Salmonella infections de- decline are not likely to be immediate. Because there were creased 5.3% per year in NYC versus NYS (IRR 0.95; 95% indications of overdispersion and heteroskedasticity, we CI 0.92–0.98; p<0.01). used a negative binomial regression model with robust SEs In a subanalysis of Salmonella infections from 2000– to quantify the effect of letter grade placards on Salmonella 2015 comparing number of infections in NYC with that of infections. We used an offset term to account for population other counties in the NYC metropolitan area and NYS sep- changes across the period. Examination of autocorrelation arately, Salmonella infections declined in both NYC and and partial autocorrelation functions confirmed that the out - the NYC metropolitan area compared with those in NYS. come was not autocorrelated. We calculated incidence rate In contrast to the findings in NYS, in the period after letter ratios (IRR) comparing Salmonella infections in NYC to grading was implemented, NYC Salmonella infections de- NYS before and after the implementation of the point scor- clined 8.8% per year and Salmonella infections in the NYC ing system and posting of letter grade placards. metropolitan area declined 7.5% per year compared with Key variables included year, coded as a continuous the period between the implementation of a point scoring variable starting with 1994 = 1; variables representing the system and the letter grade program. 2 policy periods (prepolicy = 0, postpolicy = 1); 2 variables representing the trends after policy implementation, coded Discussion as 0 before the policy and a continuous numerical function Overall, Salmonella infections declined in NYC and NYS after the policy was implemented; and a variable (region) between 1994 and 2015. Although NYS had declines in representing NYC versus NYS. The model also included Salmonella infection rates after 2010, NYC saw declines 5 interaction terms: region by year, to control for regional greater than those in NYS. Inspection processes were secular trends; region by policy for each policy (before largely unchanged in NYC with the implementation of and after letter grade was implemented), to account for the the point scoring system in 2005 (15). The letter grade mean level change in NYC after the new policy took effect; Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2165 SYNOPSIS Table 1. Confirmed Salmonella infection case counts, rates, and percent changes for New York, NY, USA, and the rest of the state, 1994–2015* NYC NYS Rate per 100,000 Year-to-year Rate per 100,000 Year-to-year Year No. cases population change, % No. cases population change, % 1994 1,890 25.0 – 1,977 18.2 – 1995 2,166 28.4 13.7 1,912 17.6 3.3 1996 1,927 25.0 11.8 1,940 17.8 1.5 1997 1,772 22.8 1,649 15.2 8.9 14.9 1998 1,751 22.3 1,680 15.4 1.7 2.3 1999 1,508 19.0 1,516 13.9 14.8 10.1 2000 1,215 15.2 20.1 1,293 11.8 15.1 2001 1,386 17.2 13.5 1,397 12.7 7.7 2002 1,458 18.1 5.0 1,613 14.6 15.0 2003 1,307 16.2 1,282 11.5 10.3 20.8 2004 1,273 15.8 1,291 11.6 0.5 2.3 2005† 1,203 15.0 -5.1 1,427 12.8 10.6 2006 1,272 15.9 6.0 1,423 12.8 0.2 2007 1,304 16.3 2.3 1,476 13.3 3.7 2008 1,268 15.7 3.4 1,491 13.4 0.8 2009 1,236 15.2 3.3 1,370 12.3 8.4 2010‡ 1,304 15.9 4.7 1,448 12.9 5.4 2011 1,125 13.6 14.7 1,423 12.7 1.9 2012 1,171 14.0 3.1 1,395 12.4 2.0 2013 1,124 13.3 1,300 11.6 4.8 6.9 2014 987 11.6 1,320 11.7 1.6 12.8 2015 918 10.7 1,314 11.7 0.0 7.8 Period-to-period change, 1994–2005 vs. 2006–2010 22.8 10.6 Period-to-period change, 2006–2010 vs. 2011–2014 20.0 7.1 Period-to-period change, 1994–2010 vs. 2011–2015 32.6 14.1 *Percentages were calculated from unrounded values; the values shown may be different when calculated from the rounded values in the table. NYC, New York City; NYS, rest of New York state. †Year of implementation of point scores for food safety inspections. ‡Year of implementation of letter grades for food safety inspections. placard program begun in 2010 did not change the point implemented at the end of a hyperendemic period of Salmo- scoring system but rather used the points to create a readily nella enterica serotype Enteriditis infections. In the 1980s, comprehensible ranking system accessible at the point of high rates of Salmonella Enteriditis infections primarily as- service. Our analysis supports the hypothesis that having sociated with shell eggs were recognized, and a variety of a point scoring system was not associated with declines in prevention and control measures were put in place to combat Salmonella infections but having a simple way to publicly these rising rates (16). The Northeast was particularly af- disclose the results of the inspection was. fected by this outbreak; New York state reported the high- Although it appears that the implementation of a point est number of outbreaks during 1985–1999 (16). Prevention scoring system in NYC was associated with a leveling measures appeared to have the greatest success in reducing off of declines in Salmonella infections, the system was rates in the Northeast compared with others (16). Further- more, 2 notable changes in NYC during this time period led to improved surveillance, which typically results in better detection and reporting. In 2006, the NYC Board of Health mandated electronic laboratory reporting of notifiable diseas - es (17); in 2009, NYC became a Foodborne Diseases Centers for Outbreak Response Enhancement (FoodCORE) center with the goal of improving surveillance for Salmonella infec- tions (18). To the extent that these efforts may have improved surveillance in NYC, they may also have been expected to increase case detection in NYC versus NYS. The implementation of letter grading in 2010 marks the beginning of the current declining trend in Salmonella Figure. Confirmed Salmonella infection cases per 100,000 infections. By improving sanitation conditions in NYC population in NYC and the rest of the state, 1994–2015. Dashed food service establishments, the letter grade program can lines indicate implementation of point scores in 2005 and letter be expected to benefit other areas as well. In this study, grades in 2010. NYC, New York City; NYS, rest of New York state. 2166 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Inspection Grades and Salmonella, New York Table 2. Mean Salmonella infection rates for New York, NY, USA, and the rest of the state, by policy implementation period, 1994–2015* 1994–2005 2006–2010 2011–2015 Location Mean (SE) 95% CI p value Mean (SE) 95% CI p value Mean (SE) 95% CI p value NYC 20.0 (1.3) 17.1–22.9 Ref 15.8 (0.2) 15.3–16.3 0.06‡ 12.6 (0.6) 10.9–14.4 <0.01‡ NYS 14.4 (0.7) 12.9–16.0 Ref 12.9 (0.2) 12.4–13.5 0.21‡ 12.0 (0.2) 11.4–12.6 0.01‡ NYC vs. NYS† <0.01 <0.01 0.37 *NYC, New York City; Ref, referent; NYS, rest of New York state. †Comparison of mean rate between NYC and NYS within time period. ‡Comparison of mean rate within region to the preceding time period. Salmonella infection were reported by county of residence, Salmonella transmission serves as a good indicator of over- but NYC sees its population change daily due to commut- all restaurant food safety practices. ing and tourism. Manhattan, one of the 5 NYC boroughs, Despite limitations, NYC’s experience provides a useful sees its population nearly double during the workday. Of case study of the beneficial effect of letter grading programs. its commuters, 36% (≈550,000 persons) travel from outside Although the relationship between restaurant inspections and the other 4 boroughs (19). Additionally, NYC is a popu- risk for foodborne illness is not well understood (23) and in- lar place for tourists; >60 million persons visited in 2016 spections represent a snapshot in time that may not represent (20). Our study showed that Salmonella infections in NYS the overall sanitary conditions in restaurants, factors related declined after the letter grade program was implemented. to food handling and preparation practices and food worker This finding may be in part because improved sanitary health and hygiene are frequent contributors to outbreaks in conditions in NYC restaurants after the implementation of restaurants (24). These factors probably contribute to trans- letter grades reduced risk for Salmonella exposure among mission of sporadic infections in restaurants, which are much NYS residents who commuted to or visited NYC. more common; outbreak cases represent <10% of all Salmo- This study had several limitations. First, this was a nella infections (4,6). The NYC restaurant letter grade pro- quasi-experimental, ecologic study that represents an as- gram has been shown to be associated with sustained improve- sociation and not a causal relationship. Second, the NYC ments in sanitary conditions in restaurants, including several restaurant letter grade program involved multiple changes factors associated with outbreaks (9). Furthermore, that study to sanitation enforcement in addition to letter grade post- showed that after 18 months, 81% of adults in NYC had seen ing; changes included inspection frequency, greater risk letter grade placards, and 88% of those persons considered for fines, improvements to online resources, and additional the letter grades in their dining decisions (9). This finding sug- training opportunities (21). As a result, we could not deter- gests that consumer behavior helped support the program goal mine which factors contributed the most to the reduction of driving improvements in sanitary conditions. Although fu- in Salmonella infections. Furthermore, we were not able to ture studies are needed to parse which restaurant inspection assess whether Salmonella-infected persons had a known results may contribute most to declines in Salmonella infec- exposure to restaurants in the period before illness. tions, our findings support the hypothesis that the successful This study supports findings from an earlier NYC study (21) and previous studies of Los Angeles County that showed Table 3. Regression for interrupted time series analysis of Salmonella infection rates, New York, NY, USA, and the rest of a decline in foodborne illness hospitalizations. Hospitaliza- the state, 1994–2015* tions represent a subset of foodborne illnesses that may be Robust caused by a variety of agents, such as Campylobacter, an- Predictor variables IRR SE 95% CI p value other leading cause of foodborne illness in the United States Year 0.95 0.01 0.94–0.96 <0.01 Region 1.50 0.08 1.35–1.68 <0.01 (1). However, Campylobacter rarely causes outbreaks in Region × year 0.99 0.01 0.97–1.00 0.16 restaurant settings because its biology limits transmission Point scoring policy 1.20 0.06 1.08–1.32 <0.01 to inadequate cooking of contaminated poultry or meats or implementation Trend after point scoring 1.04 0.01 1.02–1.07 <0.01 cross-contamination from raw to ready-to-eat foods (22). As implementation a result, improvements in restaurant sanitary conditions are Region × point scoring 0.92 0.07 0.80–1.07 0.29 unlikely to affect Campylobacter transmission in restaurants. implementation In contrast, the selection of Salmonella infections is Region × trend after 1.02 0.02 0.99–1.05 0.28 points a strength of this study because the biology of Salmonella Letter grade policy 1.01 0.04 0.93–1.10 0.81 makes it uniquely suited for transmission in food service implementation establishments. In restaurants, Salmonella can cause illness Trend after grading 0.98 0.01 0.96–1.00 0.09 Region × grading 0.99 0.06 0.88–1.11 0.85 from contaminated raw ingredients, through cross-contam- Region × trend after 0.95 0.02 0.92–0.98 <0.01 ination, or from infected food workers. Improper cooling grading of inadequately cooked foods, or failure to maintain cold or Intercept 0.00 0.00 0.00–0.00 <0.01 *IRR, incidence rate ratio. hot holding temperatures can amplify contamination. Thus, Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2167 SYNOPSIS 8. Fung A, Graham M, Weil D. Full disclosure: the perils and promise implementation of a letter grade program was associated with of transparency. Cambridge: Cambridge University Press; 2007. a reduction of Salmonella transmission in restaurants in NYC. 9. Wong MR, McKelvey W, Ito K, Schiff C, Jacobson JB, Kass D. In conclusion, in the United States, considerable re- Impact of a letter-grade program on restaurant sanitary conditions sources have been invested to prevent contamination of and diner behavior in New York City. Am J Public Health. 2015;105:e81–7. the food supply before the point of service. However, Sal- 10. Jin GZ, Leslie P. The effect of information on product quality: monella infections remain unchanged at the national level. evidence from restaurant hygiene grade cards. Q J Econ. Previous studies have shown improvements in sanitary 2003;118:409–51. conditions after the implementation of the NYC restaurant 11. Simon PA, Leslie P, Run G, Jin GZ, Reporter R, Aguirre A, et al. Impact of restaurant hygiene grade cards on foodborne-disease letter grade program, and our study suggests a beneficial hospitalizations in Los Angeles County. J Environ Health. effect on the incidence of foodborne illnesses. Implement - 2005;67:32–6, 56, 59–60. ing a letter grade program is a feasible and relatively inex- 12. Centers for Disease Control and Prevention. 2018 National pensive tool to reduce Salmonella infections that warrants notifiable conditions. [cited 2018 July 14]. nndss/conditions/notifiable/2018/ consideration for broader use. Other jurisdictions should 13. Wagner AK, Soumerai SB, Zhang F, Ross-Degnan D. Segmented consider adopting a letter grade program and decide on the regression analysis of interrupted time series studies in medication form and location of the placard, frequency of inspections, use research. J Clin Pharm Ther. 2002;27:299–309. and approaches to engage restaurant-industry and commu- 10.1046/j.1365-2710.2002.00430.x 14. Johns M, Farley SM, Rajulu DT, Kansagra SM, Juster HR. nity support to ensure program success. Smoke-free parks and beaches: an interrupted time-series study of behavioural impact in New York City. Tob Control. 2015;24:497– Acknowledgments 500. We thank Andrew Ryan and Joseph Servadio for analytic 15. New York City Department of Health and Mental Hygiene. Notice of adoption of an amendment to Title 24 of the rules of the City of consultations. New York adding Chapter 23 (“Food Service Establishment Inspection Procedures”). New York: The Department; 2005. About the Author 16. Patrick ME, Adcock PM, Gomez TM, Altekruse SF, Holland BH, Ms. Firestone is a doctoral candidate at the University of Tauxe RV, et al. Salmonella enteritidis infections, United States, Minnesota. Her primary research focuses on using surveillance 1985–1999. Emerg Infect Dis. 2004;10:1–7. data to inform policy targeted toward reducing foodborne illness. 10.3201/eid1001.020572 17. Nguyen TQ, Thorpe L, Makki HA, Mostashari F. Benefits and Dr. Hedberg is a professor at the University of Minnesota. barriers to electronic laboratory results reporting for notifiable diseases: the New York City Department of Health and Mental His primary research focuses on improving surveillance of food- Hygiene experience. Am J Public Health. 2007;97(Suppl 1): borne illness and food safety hazards with the goal of reducing S142–5. the burden of foodborne illness. 18. Centers for Disease Control and Prevention. FoodCORE Center: New York City. 2016 [cited 2018 March 26]. foodcore/centers/nyc.html References 19. McKenzie B, Koerber W, Fields A, Benetsky M, Rapino M. 1. Scallan E, Hoekstra RM, Angulo FJ, Tauxe RV, Widdowson MA, Commuter-adjusted population estimates: ACS 2006–10. 2013 Roy SL, et al. Foodborne illness acquired in the United States— [cited 2018 Sep 17]. major pathogens. Emerg Infect Dis. 2011;17:7–15. 2013/acs/2013_McKenzie_02.html 20. NYC and Company. NYC travel and tourism visitation statistics 2. US Department of Agriculture Economic Research Service. Food [cited 2018 Feb 22]. expenditures: table 10, food away from home as a share of food nyc-statistics-page expenditures 2016 [cited 2018 Sep 17]. 21. New York City Department of Health and Mental Hygiene. data-products/food-expenditures.aspx Restaurant grading in New York City at 18 months. New York: 3. Centers for Disease Control and Prevention. Surveillance for The Department; 2012. foodborne disease outbreaks, United States, 2014, Annual Report. 22. Silva J, Leite D, Fernandes M, Mena C, Gibbs PA, Teixeira P. Cam- Atlanta: US Department of Health and Human Services; 2016. pylobacter spp. as a foodborne pathogen: a review. Front 4. Jones TF, Angulo FJ. Eating in restaurants: a risk factor Microbiol. 2011;2:1-12. for foodborne disease? Clin Infect Dis. 2006;43:1324–8. 23. Lee P, Hedberg CW. Understanding the relationships between inspection results and risk of foodborne illness in restaurants. 5. Healthy People 2020. 2020 topics and objectives. 2018 Sep 21 [cited Foodborne Pathog Dis. 2016;13:582–6. 2018 Sep 25]; fpd.2016.2137 6. Centers for Disease Control and Prevention. Foodborne Disease 24. Angelo KM, Nisler AL, Hall AJ, Brown LG, Gould LH. Active Surveillance Network (FoodNet): FoodNet 2015 Epidemiology of restaurant-associated foodborne disease surveillance report (final data). Atlanta: US Department of Health outbreaks, United States, 1998–2013. Epidemiol Infect. and Human Services; 2017. 2017;145:523–34. 7. New York City Department of Health and Mental Hygiene. Address for correspondence: Melanie J. Firestone, University of Notice of adoption of amendments to Article 81 of the New York City Health Code. 2010 [cited 2018 Sep 25]. Minnesota School of Public Health, 420 Delaware St SE, Mayo Mail assets/doh/downloads/pdf/notice/2010/Article-81.pdf Code 807, Minneapolis, MN 55455, USA; email: 2168 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 RESEARCH Spatial Analysis of Wildlife Tuberculosis Based on a Serologic Survey Using Dried Blood Spots, Portugal Nuno Santos, Telmo Nunes, Carlos Fonseca, Madalena Vieira-Pinto, Virgílio Almeida, Christian Gortázar, Margarida Correia-Neves We investigated the spatial epidemiology of bovine tuber- TB is maintained in a multihost pathogen system in which culosis (TB) in wildlife in a multihost system. We surveyed M. bovis and M. caprae circulate between sympatric wild bovine TB in Portugal by serologic analysis of elutes of ungulates (mainly wild boar [Sus scrofa] and red deer [Cer- dried blood spots obtained from hunted wild boar. We mod- vus elaphus]) and free-ranging domestic ungulates (1). eled spatial disease risk by using areal generalized linear In Portugal, control of bovine TB has resulted in a mixed models with conditional autoregressive priors. Anti - low prevalence in livestock (4.5 cases/10,000 cattle and bodies against Mycobaterium bovis were detected in 2.4% 2.9 cases/1,000 herds) in 2017 (3). Nevertheless, disease (95% CI 1.5%–3.8%) of 678 wild boar in 2 geographic incidence has stabilized in recent years, and awareness of clusters, and the predicted risk fits well with independent wildlife hosts has fueled the discussion over their role as reports of M. bovis culture. Results show that elutes are an reservoirs of bovine TB. In 2011, the Portuguese Animal almost perfect substitute for serum (Cohen unweighted κ Health Directorate (Lisbon, Portugal) established a surveil- = 0.818), indicating that serologic tests coupled with dried blood spots are an effective strategy for large-scale bovine lance area for bovine TB in large game species, encom- TB surveys, using wild boar as sentinel species. Results passing regions where the disease was known to be present also show that bovine TB is an emerging wildlife disease in wild ungulate populations (Figure 1, panel B). and stress the need to prevent further geographic spread Bovine TB in wildlife shows spatial structuring in the and prevalence increase. Iberian Peninsula. There is a core area in the central–south- western region, in which the average prevalence of macro- scopic lesions is 59% in wild boar (4). At the periphery of ovine tuberculosis (TB) is a zoonotic disease caused this core area, prevalence decreases, and becomes low to Bby Mycobacterium bovis and other members of the M. undetectable in eastern, northern, and western regions of tuberculosis complex, whose natural hosts are wild and do- the Iberian Peninsula (5–7). Nevertheless, spatial analyses mestic mammals (1). Bovine TB is a disease of economic of bovine TB on wildlife in the Iberian Peninsula, other and public health relevance and is subjected to mandatory than disease mapping, are notably lacking. control programs in livestock in many countries. As a re- Large-scale disease surveys in wildlife require mass- sult of these programs, bovine TB has been eradicated in scalable and inexpensive diagnostic tests; serologic meth- regions such as Australia and Scandinavia. However, in ods are one of the most suitable techniques (8). An ELISA other regions, persistence of infection has been attributed for detecting antibodies against the M. tuberculosis com- to wildlife reservoirs, such as cervids in North America plex has been described and validated for use in wild boar (2). In the Iberian Peninsula (Figure 1, panel A), bovine (9,10) and showed a moderately high estimated sensitivity Author affiliations: University of Minho School of Medicine (ICVS), of 79.2% and an excellent specificity of 100% (10). Anoth- Braga, Portugal (N. Santos, M. Correia-Neves); ICVS/3B’S, er improvement for large-scale disease surveys in wildlife Portugal Government Associate Laboratory, Braga/Guimarães, is a sampling protocol that might be used by nonspecial- Portugal (N. Santos, M. Correia-Neves); University of Lisbon, ized personnel, such as hunters. One example is the dried Lisbon, Portugal (T. Nunes, V. Almeida); University of Aveiro, Aveiro, blood spot technique, which was originally developed for Portugal (C. Fonseca); University of Trás-os-Montes e Alto Douro, human sampling but has been increasingly used for wildlife Vila Real, Portugal (M. Vieira-Pinto); Instituto de Investigación en disease surveys (11). Recursos Cinegéticos, Ciudad Real, Spain (C. Gortázar) Current affiliation: Research Centre in Biodiversity and Genetic DOI: Resources, University of Porto,Vairão, Portugal. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2169 RESEARCH Figure 1. Choropleth maps for spatial study of bovine tuberculosis (TB) in wildlife, Portugal. A) Iberian Peninsula. B) Official surveillance area for bovine TB in large game species. Red numbers indicate historical population refuges of wild ungulates: 1) Gerês, 2) Montesinho, 3) Malcata, 4) São Mamede, and 5) left bank of the Guadiana River. C) Distribution of serologic samples analyzed per county. D) Distribution of bovine TB–positive samples. Black circles indicate the 2 clusters identified. Wild boar have been shown to be a maintenance host the distribution of wildlife bovine in Portugal on the basis of for bovine TB in the Iberian Peninsula (12). Furthermore, serologic methods, investigate spatial clustering of bovine wild boar have been used as a sentinel for bovine TB in TB in wildlife, and model the risk for bovine TB in Portugal. wildlife because of their high susceptibility to infection with M. bovis and M. caprae and extensive exposure to Methods these pathogens through direct contact, necrophagic hab- its, or fossorial habits (13–15). Wild boar populations in Collection of Samples Portugal reached their nadir in the middle of the 20th cen- Biological samples were obtained from 678 wild boar hunt- tury, when the species survived in only 5 populations or ed during 2006–2013. Of these samples, 107 were serum and historical refuges (16,17) (Figure 1, panel B). Since that 571 were dried blood spots collected on Protein Saver (PS) time, wild boar populations have increased markedly and 903 cards (Whatman, Maidstone, UK) (n = 308) or Flinders are currently found throughout Portugal (16,17). Technology Associates (FTA) paper (Whatman) (n = 263). To assess the spatial epidemiology of bovine TB in Serum samples were obtained from blood collected wildlife, we analyzed the multihost pathogen system in the from the thoracic or abdominal cavities of hunted wild boar Iberian Peninsula by using the wild boar as a sentinel species. and stored at –20°C. Absorbent papers were distributed The main aims of this study were to assess the performance to hunters with instructions for the papers to be soaked in of dried blood spots as an alternative sample collection and blood available from the thoracic cavity. Papers were dried storage technique for serologic surveys of bovine TB, map in the shade and kept at room temperature in zipper-lock 2170 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Spatial Analysis of Wildlife TB, Portugal bags with data on the location of collection. Dried blood Rv1510 genes after DNA extraction by using the standard spots were recovered at the end of the hunting season and phenol-chloroform method after bead-beating with 100 kept at –20°C until processed. Elutes were obtained by μL of 0.1-mm zirconia/silica beads (Biospec Products, cutting half a circle of PS card or a quarter of FTA cards. Bartlesville, OK, USA) in a FastPrep 24 Homogenizer (MP According to the manufacturer’s instructions, a PS card ab- Biomedicals, Santa Ana, CA, USA). sorbs 80 µL of blood, and an FTA card absorbs 125 µL of blood. These samples were further divided into 5 por- Data Analysis tions that were incubated overnight refrigerated in 200 µL To determine agreement between ELISA results (positive of phosphate-buffered saline (PBS), and the elute obtained or negative) from paired serum and elute samples, we com- was subjected to serologic analysis. Paired samples of se- puted the Cohen unweighted κ value (19) by using irr in R rum and dried blood spots were collected from 22 wild boar software (20) (R Development Core Team, Vienna, Austria). that had macroscopic lesions compatible with bovine TB. We obtained serologic data for 92 of 278 counties in Portu- In 1 region of southeastern Portugal, tissue samples gal (Figure 1, panel C). The area considered for each county were collected from 340 hunted wild boar with either bo- excluded regions classified as urban or water bodies in the vine TB–compatible lesions or pooled lymph nodes when CORINE database (21). We created choropleth maps, which lesions were absent. These samples were obtained during use differences in shading, coloring, or placing of symbols 2009–2014 and kept at –20°C until bacteriological cultures within predefined areas to indicate average values of a prop - were performed. erty or quantity in those areas, of regions with bovine TB and performed spatial interpolation by using QGIS version Laboratory Analysis 2.6.1 Brighton software ( - We tested serum samples by using an ELISA and bovine volved/governance/governance.html). To detect spatial ag- purified protein derivative (bovine PPD) as antigen and pro - gregation of bovine TB, we performed cluster analysis on tein G–horseradish peroxidase as conjugate (10). In brief, we the basis of Kuldorff spatial scan statistics by using the Ber - coated wells of ELISA plates with 100 µg of bovine PPD noulli distribution and setting the maximum cluster size at for 18 hours at room temperature, washed the plates with 50% with SatScan version 9.3.1 software (22). PBS containing 0.05% Tween 20 (PBST), and incubated the We assessed the association between detection of bo- plates for 1 hour at 37°C with 140 µL/well of 5% skimmed vine TB in wild boar in each county and independent vari- milk in PBST to block potential free binding sites. We added ables (Table 1) by using spatial generalized linear mixed serum or elute samples to plates (10 µL/well) at dilutions of modeling of areal data with the localized conditional au- 1:200 in PBS for serum and 1:50 in PBS for elutes and incu - toregressive priors proposed by Lee and Mitchel (23). We bated samples for 1 hour at 37°C. Protein G–horseradish per- included the bioregions of the Iberian Peninsula (Atlantic/ oxidase conjugate was added (100 µL/well) at a dilution of Mediterranean) (24) as local conditional autoregressive 2.5 µg/mL in PBST and incubated at 37°C for 80 min. A to- priors because they have been shown to have distinct bo- tal of 100 µL of substrate (SigmaFast OPD; Sigma-Aldrich, vine TB epidemiologic scenarios (7). We implemented St. Louis, MO, USA) was added to each well and incubated models by using the CARBayes version 5.0 package (25) in at room temperature in the dark. The reaction was stopped R software. We based inference on 20,000 Markov Chain after 20 min by the addition of 50 µL/well of 3N H SO . Monte Carlo iterations (200,000 iterations with a thin fac- 2 4 We measured optical density (OD) by using a spectro- tor of 10 to reduce autocorrelation) after an initial burn-in photometer at 450 nm. Blanks and positive and negative of 40,000 iterations. Taking into consideration the home controls were tested in duplicate in each plate, and samples ranges of wild boar (26), we decided to include counties were tested in triplicate. Results were calculated as mean <25 km apart in the neighborhood matrix of each other. sample OD divided by 2 times the mean negative control We estimated domestic host absolute densities on the OD; the cutoff for positivity was 1 ( 10). basis of data from the Portuguese Livestock Movement Da- We prepared bacteriological cultures of tissue samples tabase ( with the following inclusion criteria: in a Biosafety Level 3 facility according to a described pro- we considered only animals >6 months of age among free- tocol (18). In brief, 3 g of tissue were homogenized and range pigs, extensively reared sheep and goats, and meat decontaminated for 2 hours with 0.75% hexa-decylpyri - production cattle. We excluded intensively reared animals dinium chloride, centrifuged at 2,566 × g for 30 min, and to provide more realistic estimates of the livestock popu- the supernatant collected. We inoculated 2 tubes contain- lation at potential risk from contact with wildlife. Meat- ing Coletsos medium (bioMerieux, Marcy l’Étoile, France) production cattle were selected as a proxy for extensive with 250 µL of supernatant–sediment interface and incu- rearing because this is the predominant beef cattle produc- bated these tubes at 37°C for 15 wks. Isolates were identi- tion system in Portugal, and dairy cattle herds are almost fied by PCR for 16S rRNA, insertion sequence 1561, and exclusively housed indoors. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2171 RESEARCH Table 1. Independent variables included in initial models of bovine tuberculosis, Portugal* Variable Specific variable Unit Wild host density Wild boar hunting bag Wild boar hunted/km † Red deer hunting bag Red deer hunted/km † Fallow deer hunting bag Fallow deer hunted/km † Game management Intensity of management as proportion of area as tourist or Proportion national hunting zones Domestic host density Cattle density >6 mo of age, meat Cattle/km Sheep density >6 mo of age, extensive Sheep/km Goat density >6 -mo of age, extensive Goats/km Pig density >6 mo of age, free range Pigs/km Bovine tuberculosis incidence in cattle Bovine tuberculosis incidence in cattle Proportion Historical population dynamics Distance to the nearest historical refuge km *Spatial unit for all variables in county. †Average no. of each wild game species hunted per county per year/area of hunting areas for which data were available per county per year. We selected hunting bag (number of animals hunted We detected antibodies against bovine PPD in 16/678 per square kilometer in 1 year) as a proxy for wild ungulate wild boar (2.4%, 95% CI 1.5%–3.8%). Antibody-posi- density and calculated this density as the annual average of tive wild boar originated from 4/92 counties tested (Fig- the number of hunted ungulates (wild boar, red deer, and ure 1, panel C). Kulldorff spatial scan statistics identified fallow deer) from those hunting areas for which >2 years 2 clusters of bovine TB in wildlife. Cluster 1 was found of data were available for 2008–2012. Because hunting bag in 1 county that had 8/36 positive wild boar (relative risk data were not available for 65/278 counties, we performed = 17.83, p<0.001), and cluster 2 was found in 5 counties an inverse distance weighted interpolation with power set (radius = 43 km) that had 8/65 positive wild boar (relative at 3 to obtain estimates for the entire territory. risk = 9.43, p = 0.011) (Figure 1, panel D). We selected the proportion of the area of each county The selected nonspatial logistic regression model dedicated to commercial hunting (tourism and national hunt- with the presence of bovine TB as the dependent variable ing areas usually intensively managed for maximizing profit) showed a McFadden pseudo-R value = 0.656 (Table 2). as a proxy for intensity of game management. Other types of The spatial generalized linear mixed model (deviance in- hunting areas are dedicated to recreational hunting (i.e., usu- formation criterion = 24.141) that included the variables ally no fencing, restocking, or large-scale artificial feeding of distance to historical refuge, bovine TB incidence in cattle, large game species). We included the historical presence of red deer hunting bag, intensity of management, and red wild boar as the distance from the centroid of each county to deer hunting bag times intensity of management (Table 2) the nearest historical refuge (Figure 1, panel B). We calcu- explained 45.5% of the deviance. The posterior probability lated the incidence of bovine TB in cattle as the average of of the presence of bovine TB in wildlife (Figure 2) was the annual incidence rate for each county during 2008–2012 significantly higher for those counties with independent on the basis of data from the bovine TB eradication program. reports of isolation of M. bovis from free-ranging wildlife We estimated a variance inflation factor to assess mul- (Mann-Whitney–Wilcoxon W = 320; p<0.001) (26,29–32). ticollinearity with a threshold of 5. We used a nonspatial We further investigated bovine TB in southeastern Por- logistic regression with backward stepwise elimination and tugal. This investigation included cluster 1, where M. bovis selected the final model on the basis of Akaike information (n = 51) and M. caprae (n = 2) were isolated from 53/340 criterion corrected for small samples and model weights wild boar in 6/17 hunting areas, which had an overall bac- (27). We calculated the McFadden pseudo-R value to teriological culture prevalence of 15.6% (95% CI 12.1%– measure the goodness of fit of each model. We used the 19.8%). In 1 nonfenced hunting area, culture prevalence posterior probabilities of the spatial model to predict the of wild boar bovine TB increased significantly (p = 0.049 risk for bovine TB in wild boar populations in all counties by Fisher exact test) from 46.2% (95% CI 26.6%–66.6%) in Portugal. We assessed model convergence on the basis during 2005–2006 (6) to 67.7% (95% CI 55.4%–78.0%) of the Geweke statistic (28). during 2009–2014 (Figure 3). Results Discussion We found that 11 paired serum samples and PS elutes were We report a spatial analysis of bovine TB in wildlife in positive for antibodies against bovine PPD, 9 were nega- the multihost system of the Iberian Peninsula on the basis tive for both, and 2 were positive in serum and negative in of serologic data for hunted wild boar. Serologic analysis elutes. There was an almost perfect agreement between se- was previously shown to have reasonably good sensitivity rologic results for both types of samples (mean ± SD Cohen and excellent specificity for detecting bovine TB in wild unweighted κ value = 0.818 ± 0.121). boar (10), and this species is a suitable sentinel for bovine 2172 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Spatial Analysis of Wildlife TB, Portugal Table 2. Variables included in nonspatial binomial general linear and spatial generalized linear mixed models of bovine tuberculosis in wild boar, Portugal Spatial generalized linear mixed model Logistic regression model Median coefficient Geweke Variable Coefficient (95% CI) p value (95% credible interval) statistic Intercept 8.561 25.992 to 2.644) 0.073 12.886 19.767 to 7.985) 1.3 Wild boar historical refuge 0.101 1.4 0.114 0.356 to 0.020) 0.088 0.170 to 0.022) Bovine tuberculosis incidence in cattle 73.719 (6.919 to 231.075) 0.126 35.797 4.396 to 82.742) 1.6 Management intensity 29.296 (6.167 to 101.249) 0.126 25.009 (6.510 to 49.979) 1.3 Red deer hunting bag 9.391 (1.884 to 34.953) 0.150 6.990 (1.823 to 11.814) 0.8 Red deer hunting bag × management intensity 30.922 118.421 to 7.887) 0.162 23.221 39.083 to 8.343) 0.9 TB (13–15). We provide evidence showing that dried blood also found in wild boar and red deer (17,30). No additional spots collected on PS cards by nonspecialist personnel are M. bovis isolates have been reported from this region, sug- appropriate for serologic surveys of bovine TB, as shown gesting that after the initial introduction and spillover into by the almost perfect agreement between serologic results local wild boar, the infection waned or persists at a low prev- for serum samples and elutes. These results were expected alence, in accordance with the predicted low risk for bovine because dried blood spots have been extensively validated TB in our model. for many other human or animal host–pathogen systems This observation suggests that relatively dense red and generally found to be suitable for serologic surveys, deer populations are needed to maintain bovine TB at a even when blood is collected in remote and demanding high prevalence in wild ungulate populations. This finding conditions (11,33). This combination of dried blood spots also contrasts with the situation reported from south-central and serologic analysis is a major advance for large-scale surveillance of bovine TB in wildlife anywhere wild boar are a suitable indicator species. Our results support previous data suggesting a strong spatial structure of bovine TB in wildlife. Two clusters were identified in southern (cluster 1) and cen - tral-eastern (cluster 2) Portugal, which are located at the periphery of the high-prevalence core area in the cen- tral-southwestern region of the Iberian Peninsula (4,5). A subsequent survey for bovine TB in cluster 1 relied on bacteriological culture of tissues collected from hunted wild boar, which is an established but more expensive and labor-intensive diagnostic technique than serologic analysis. This regional bacteriological survey confirmed bovine TB as an emerging disease in wildlife and docu- mented a 46.2% increase in prevalence in less than a de - cade at 1 nonfenced hunting area, similar to other popu- lations in the Iberian Peninsula (4). The spatial risk model we reported identified some pre - dicted high-risk counties not included in the surveillance area for bovine TB in large game species (Figure 1, panel B) and thus could be used for better allocation of resources for wild- life disease surveillance and public health protection. The spatial model of bovine TB risk in wildlife generally agrees with results of published independent surveys because most reported wildlife isolates of M. bovis and M. caprae overlap with predicted moderate-risk to high-risk areas. The single exception is M. bovis isolated from a wild boar in Coimbra Figure 2. Choropleth map of risk for bovine TB in wildlife, Portugal, showing the probability of the presence of bovine TB in in central-western Portugal (Figure 2), an area that has a low wildlife in counties based on the conditional autoregressive spatial predicted risk for bovine TB in wildlife (30). Red deer were generalized linear mixed model. Stars indicate counties in which introduced into this region during 1995–1999 and some of Mycobacterium bovis was isolated from free-ranging wildlife, the founder animals originated from bovine TB–infected ar- determined on the basis of independent published data (6,29–32). eas included in cluster 1, in which the same spoligotype was TB, tuberculosis. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2173 RESEARCH TB in livestock could be an indicator of local disease trans- mission in wildlife. In conclusion, we report a spatial analysis of bovine TB in wildlife in Portugal that used wild boar as a senti- nel species and assessed the relative performance of dried blood spots collected on PS cards as a new sampling tool for large-scale serologic surveys. We confirmed the strong spatial clustering of bovine TB in wildlife and identified risk factors related to red deer density, intensity of wild un- gulate management, historical population dynamics of wild boar, and incidence of bovine TB in cattle. The risk maps developed provide new tools for the targeted control of bo- vine TB in wild ungulate populations and identify areas at high risk for spread of disease. Figure 3. Temporal trend in prevalence of bovine tuberculosis in Acknowledgments wild boar in 1 nonfenced hunting area, Portugal, by bacteriological We thank the hunters for collecting blood samples from wild culture during 2005–2006 (6) and 2009–2014 (this study). Error bars indicate 95% CIs. boar; Mariana Boadella, Paqui Talavera, and Natacha Melo for providing assistance with ELISAs; the Direcção Geral de Alimentação e Veterinária for providing data on populations of Spain, where wild boar is considered the main maintenance domestic hosts and bovine TB incidence in cattle; and the host, probably because intensive game management is rare Instituto de Conservação da Natureza e Florestas for providing in Portugal, with similarities with what was described for data on the number of hunted wild ungulates. the Atlantic bioregion of Spain (14,17). Further studies in- volving spatially structured estimation of the abundance of This study was supported by Plan Nacional (grant CGL2017- wild ungulates and bovine TB prevalence along the edge 89866 from the Ministry of Economy and Competitiveness, of the infected area are warranted to identify the relative Spain, and Fondo Europeo de Desarollo Regional) and Programa roles of wild boar and red deer densities in maintenance of Operacional Regional do Norte (grant ON.2 O Novo Norte), bovine TB in this multihost system. Quadro de Referência Estratégico Nacional through the Fundo Red deer density and its interaction with the intensity of Europeu de Desenvolvimento Regional. N.S. was supported by game management were associated with bovine TB in wild PhD grant SFRH/BD/69390/2010 from Fundação para a Ciência boar in the spatial risk model. Red deer density is a better e Tecnologia. predictor of bovine TB in wild boar populations than wild boar density, further strengthening the role of red deer as a About the Author key reservoir of bovine TB in wildlife in Portugal. The his- Dr. Santos is a postdoctoral research scientist at the Research torical dynamics of wild boar populations are associated Centre in Biodiversity and Genetic Resources, Vairão, Portugal. with contemporary distribution of bovine TB. Starting in the His research interests are bovine tuberculosis epidemiology mid-20th century, wild boar populations expanded from his- in wild ungulates, disease ecology at the wildlife–livestock torical refuges, and bovine TB in wildlife also seems to be interface, and conservation physiology using carnivores and increasing but at a much slower pace. One possible explana- ungulates as research models. tion for this pattern is that as wild boar populations spread, densities at the front of the expansion wave were low (34). Thus, bovine TB could not be maintained even with the addi- References tion of infected animals from infected historical refuges. As 1. Gortázar C, Delahay RJ, Mcdonald RA, Boadella M, Wilson GJ, expansion continued and wild boar densities increased (34), Gavier-Widen D, et al. The status of tuberculosis in European wild mammals. Mammal Review. 2012;42:193–206. the range of M. bovis seems to be slowly increasing. The incidence of bovine TB in livestock was included 2. Palmer MV. Mycobacterium bovis: characteristics of wildlife in the selected model and highlighted the link between do- reservoir hosts. Transbound Emerg Dis. 2013;60(Suppl 1):1–13. mestic and wild epidemiologic cycles in wildlife, which 3. Directorate General for Food and Veterinary Affairs. Animal health was strongly suspected on the basis of molecular epidemi- technical report: bovine tuberculosis, 2017 [in Portuguese] [cited ology data (30). Although the directionality of such a link 2018 Jul 26]. could not be inferred in our study, spillover from wildlife DGV. was shown to partially explain incidence of bovine TB in 4. Vicente J, Barasona JA, Acevedo P, Ruiz-Fons JF, Boadella M, Diez-Delgado I, et al. Temporal trend of tuberculosis in wild cattle in south-central Spain (35), suggesting that bovine 2174 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Spatial Analysis of Wildlife TB, Portugal ungulates from Mediterranean Spain. Transbound Emerg Dis. 2012 [cited 2018 Sep 4]. - 2013;60(Suppl 1):92–103. ages/irr/ 5. Boadella M, Acevedo P, Vicente J, Mentaberre G, Balseiro A, 21. European Environmental Agency. Luxembourg NATLAN. Arnal M, et al. Spatio-temporal trends of Iberian wild boar contact Nature/land cover information package; 2006 [cited 2016 Sep 11]. with Mycobacterium tuberculosis complex detected by ELISA. Eco- Health. 2011;8:478–84. 22. Kulldorff M. A spatial scan statistic. Communications Statistics 6. Santos N, Correia-Neves M, Ghebremichael S, Källenius G, Theory Methods. 1997;26:1481–96. Svenson SB, Almeida V. Epidemiology of Mycobacterium bovis 03610929708831995 infection in wild boar (Sus scrofa) from Portugal. J Wildl Dis. 23. Lee D, Mitchell R. Boundary detection in disease mapping studies. 2009;45:1048–61. Biostatistics. 2012;13:415–26. 7. Muñoz-Mendoza M, Marreros N, Boadella M, Gortázar C, biostatistics/kxr036 Menéndez S, de Juan L, et al. Wild boar tuberculosis in Iberian 24. European Environmental Agency. Luxembourg. Biogeograhical Atlantic Spain: a different picture from Mediterranean habitats. BMC regions; 2016 [cited 2016 Sep 11]. Vet Res. 2013;9:176. maps/data/biogeographical-regions-europe-3#tab-gis-data 8. Gilbert AT, Fooks AR, Hayman DT, Horton DL, Müller T, 25. Lee D. CARBayes: an R package for Bayesian spatial modeling Plowright R, et al. Deciphering serology to understand the ecology with conditional autoregressive priors. Journal of Statistical of infectious diseases in wildlife. EcoHealth. 2013;10:298–313. Software. 2013;55:1–24. 26. Bosch J, Peris S, Fonseca C, Martínez M, De La Torre A, Iglesias I. 9. Aurtenetxe O, Barral M, Vicente J, de la Fuente J, Gortázar C, Distribution, abundance and density of the wild boar on the Iberian Juste RA. Development and validation of an enzyme-linked Peninsula, based on the CORINE program and hunting statistics. immunosorbent assay for antibodies against Mycobacterium bovis Folia Zoologica (Brno). 2012;61:138–51. in European wild boar. BMC Vet Res. 2008;4:43. 10.1186/1746-6148-4-43 27 Burnham KP, Anderson DR. Model selection and multimodel 10. Boadella M, Lyashchenko K, Greenwald R, Esfandiari J, Jaroso R, inference: a practical information-theoretical approach. 2nd ed. Carta T, et al. Serologic tests for detecting antibodies against New York: Springer; 2002. Mycobacterium bovis and Mycobacterium avium subspecies 28. Geweke J. Evaluating the accuracy of sampling-based approaches paratuberculosis in Eurasian wild boar (Sus scrofa scrofa). J Vet Diagn to the calculation of posterior moments. In: Berger JO, Bernardo JM, Invest. 2011; 23:77–83. Dawid AP, Smith AF, editors. Bayesian statistics. Oxford (UK): 11. Curry PS, Ribble C, Sears WC, Hutchins W, Orsel K, Godson D, Oxford University Press; 2002. p. 169–193. et al. Blood collected on filter paper for wildlife serology: detecting 29. Vieira-Pinto M, Alberto J, Aranha J, Serejo J, Canto A, Cunha MV, antibodies to Neospora caninum, West Nile virus, and five bovine et al. Combined evaluation of bovine tuberculosis in wild boar viruses in reindeer. J Wildl Dis. 2014;50:297–307. (Sus scrofa) and red deer (Cervus elaphus) from central-east Por- 10.7589/2012-02-047 tugal. European Journal of Wildlife Research. 2011;57:1189–201. 12. Naranjo V, Gortázar C, Vicente J, de la Fuente J. Evidence of the role of European wild boar as a reservoir of Mycobacterium 30. Cunha MV, Matos F, Canto A, Albuquerque T, Alberto JR, tuberculosis complex. Vet Microbiol. 2008;127:1–9. Aranha JM, et al. Implications and challenges of tuberculosis in wildlife ungulates in Portugal: a molecular epidemiology 13. Nugent G, Whitford J, Young N. Use of released pigs as sentinels perspective. Res Vet Sci. 2012;92:225–35. for Mycobacterium bovis. J Wildl Dis. 2002;38:665–77. j.rvsc.2011.03.009 31. Matos AC, Figueira L, Martins MH, Pinto ML, Matos M, Coelho 14. Gortázar C, Vicente J, Boadella M, Ballesteros C, Galindo RC, AC. New insights into Mycobacterium bovis prevalence in wild Garrido J, et al. Progress in the control of bovine tuberculosis in mammals in Portugal. Transbound Emerg Dis. 2016;63:e313–22. Spanish wildlife. Vet Microbiol. 2011;151:170–8. http://dx.doi. org/10.1016/j.vetmic.2011.02.041 32. Madeira S, Manteigas A, Ribeiro R, Otte J, Fonseca AP, Caetano P, 15. Yockney IJ, Nugent G, Latham MC, Perry M, Cross ML, Byrom AE. et al. Factors that influence Mycobacterium bovis infection in red Comparison of ranging behaviour in a multi-species complex of deer and wild boar in an epidemiological risk area for tuberculosis free-ranging hosts of bovine tuberculosis in relation to their use as of game species in Portugal. Transbound Emerg Dis. 2017;64:793– disease sentinels. Epidemiol Infect. 2013;141:1407–16. 804. 33. Parker SP, Cubitt WD. The use of the dried blood spot sample in 16. Lopes F, Borges J. Wild boar in Portugal. Galemys. 2004;16:243–51. epidemiological studies. J Clin Pathol. 1999;52:633–9. 17. Vingada J, Fonseca C, Cancela J, Ferreira J, Eira C. Ungulates and their management in Portugal. In: Appolonio M, Andersen R, 34. Holland E, Aegerter JN, Smith GC. Spatial sensitivity of a generic Putman R, editors. European ungulates and their management in population model, using wild boar (Sus scrofa) as a test case. the 21st century. Cambridge (UK): Cambridge University Press; Ecological Modelling. 2007;205:146–58. 2010. p. 392–418. j.ecolmodel.2007.02.026 18. Santos N, Geraldes M, Afonso A, Almeida V, Correia-Neves M. 35. LaHue NP, Baños JV, Acevedo P, Gortázar C, Martínez-López B. Diagnosis of tuberculosis in the wild boar (Sus scrofa): a comparison Spatially explicit modeling of animal tuberculosis at the wildlife– of methods applicable to hunter-harvested animals. PLoS One. livestock interface in Ciudad Real Province, Spain. Prev Vet Med. 2010;5:e12663. 2016;128:101–11. 19. Fleiss JL, Cohen J, Everitt BS. Large sample standard errors of kappa and weighted kappa. Psychological Bulletin. 1969;72:323–7. Address for correspondence: Nuno Santos, Research Centre in Biodiversity and Genetic Resources, University of Porto, Campus de 20. Gamer M, Lemon J, Fellows I, Singh P. Various coefficients of Vairão, 4485-661 Vairão, Portugal; email: interrater reliability and agreement. R package version 0.84; Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2175 RESEARCH Rat Lungworm Infection in Rodents across Post-Katrina New Orleans, Louisiana, USA Rosalyn C. Rael, Anna C. Peterson, Bruno Ghersi-Chavez, Claudia Riegel, Amy E. Lesen, Michael J. Blum Rat lungworm (Angiostrongylus cantonensis), a parasitic New Orleans (3,9). Infections have also since been reported nematode that can cause eosinophilic meningitis in hu- in nonhuman incidental mammal hosts (9,10), and 2 cases of mans, was first detected in New Orleans, Louisiana, USA, human eosinophilic meningitis from rat lungworm infection in the mid-1980s and now appears to be widespread in the were diagnosed in nearby areas of Louisiana (11,12). Rat southeastern United States. We assessed the distribution, lungworm also appears to have become widespread across prevalence, and intensity of A. cantonensis infection in Florida (9) and has been recently detected in Oklahoma (11). New Orleans by examining lung biopsy samples of rodents We trapped rodents across New Orleans to character- trapped at 96 sites in 9 areas in Orleans Parish and 1 area ize the current distribution, prevalence, and intensity of A. in neighboring St. Bernard Parish during May 2015 through cantonensis infection and to determine how these aspects February 2017. These areas were selected to capture con- vary according to organismal and ecological characteris- trasting levels of income, flooding, and post-disaster land - scape management after Hurricane Katrina in 2005. We tics of definitive hosts, including species co-occurrence. detected A. cantonensis in all areas and in 3 of the 4 rat This study enabled us to identify factors associated with species trapped. Overall prevalence was ≈38% but varied definitive host infection, which might affect transmission by area, host species, and host species co-occurrence. risk across the city and offer further insight into the global Infection intensity also varied by host species. These find- progression, surveillance, and control of rodent-associated ings suggest that socioecological analysis of heterogeneity infectious disease. in definitive and intermediate host infection could improve understanding of health risks across the city. Methods oncern is increasing about the spread of rat lungworm Study Animals C(Angiostrongylus cantonensis), especially in the south- We collected rats during May 2015 through February 2017 eastern United States (1–5). A parasitic nematode carried (following Tulane University [New Orleans, LA, USA] In- by intermediate mollusk hosts and definitive rat hosts ( 6,7), stitutional Animal Care and Use Committee [IACUC] pro- rat lungworm can cause eosinophilic meningitis in humans tocol #0451) during a quantitative population survey across who become infected by ingesting intermediate hosts or 96 city blocks in 8 neighborhoods in New Orleans, a natural paratenic hosts, such as freshwater shrimp and frogs (6,7). area in Orleans Parish, and an area in neighboring St. Ber- A. cantonensis was first reported in North America from nard Parish. These areas were selected to capture contrasting Norway rats (Rattus norvegicus) trapped in New Orleans, levels of income, flooding, and postdisaster landscape man - Louisiana, USA, along the Mississippi River during April agement after Hurricane Katrina in 2005 (Figure 1; online 1986 through February 1987 (8). Later surveys suggest the Technical Appendix Table 1, parasite has since become more widespread in Louisiana. article/24/12/18-0056-Techapp1.pdf) (13,14). We selected Surveys of intermediate apple snail (Pomacea canaliculata) 8–10 sites in each study area by random stratification across hosts, for example, detected the parasite in suburban areas of a 1-km resolution grid spanning the city. Each site was vis- ited 4 times during May 2015 through January 2017: once Author affiliations: Tulane University, New Orleans, Louisiana, during summer and once during winter months each year, USA (R.C. Rael, A.C. Peterson, B. Ghersi-Chavez, A.E. Lesen, except for sites in St. Bernard Parish, which were visited M.J. Blum); University of Tennessee, Knoxville, Tennessee, USA only twice (summer 2016 and winter 2016–17). During each (A.C. Peterson, B. Ghersi-Chavez, M.J. Blum); The City of New trapping period, 30 Tomahawk Live traps (Hazelhurst, WI, Orleans Mosquito, Termite, and Rodent Control Board, New USA) were set in locations with potential or evident rodent Orleans (C. Riegel) activity for a minimum of 3 consecutive nights. Trapping DOI: efforts continued at each site until no additional rats were 2176 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Rat Lungworm in Rodents, New Orleans captured. Sherman traps (H.B. Sherman Traps, Inc., Talla- parity. We categorized all Norway rats and roof rats (R. rat- hassee, FL, USA) were also placed at a subset of 48 sites to tus) into 3 age classes (juvenile, subadult, adult) according to capture smaller rodents following the same approach (Tu- body weight (15,16). Urine, lung, liver, spleen, kidney, and lane University IACUC protocol #0460). tail tissues were sampled and archived in –80°C freezers. All rodents were necropsied after euthanasia following We visually screened lung tissues for parasites, which were Tulane University IACUC protocols #0451 and #0460. We isolated, counted, and preserved in 95% ethanol. Representa- recorded standard weight and length measurements, as well tive lung parasites were identified through PCR ( 17) (online as species identity; sex; sexual maturity; and, in females, Technical Appendix; online Technical Appendix Table 2). Figure 1. Prevalence of rat lungworm (Angiostrongylus cantonensis) in rodents, New Orleans, Louisiana, USA, May 2015–February 2017. A) Roof rats (Rattus rattus); B) Norway rats (R. norvegicus). Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2177 RESEARCH Statistical Analyses necropsied, 160 (36.0%) were positive for A. cantonensis, We report on the distribution and prevalence of A. cantonen- whereas 101 (44.1%) of 229 Norway rats, 4 (21.1%) of 19 sis according to all species trapped, but additional statistical S. hispidus, and 0 of 4 O. palustris were positive (Table analyses considered data only from Norway rats and roof 1). Prevalence differed between Norway rats and roof rats because of small sample sizes or because the parasite rats (χ = 3.810, p = 0.051). Median site-level prevalence was not detected in other species. Generalized linear models estimates for roof rats (33.3%) and Norway rats (47.2%) (GLMs) were constructed with a quasibinomial error distri- (online Technical Appendix Figure) were similar to those bution to determine whether sex or age class was a signifi - estimated by pooling rats by species and area (Table 1). cant predictor of infection status (i.e., infected, not infected) A total of 488 house mice (Mus musculus) were col- in Norway rats and roof rats. We ran 3 GLMs: 1 with both lected at 48 locations with Sherman traps. Lungworms Rattus species together and 1 for each species. The same pre- were not detected in any house mice, affirming that they do dictors were used in 3 GLMs with a quasi-Poisson distribu- not serve as definitive hosts ( 18). tion to examine relationships with infection intensity (i.e., number of parasites per infected rodent) in Norway rats and Geographic Variation in Prevalence roof rats together and separately by species. Excluding the St. Bernard Parish sites because of small sam- We used χ tests to determine whether infection preva- ple sizes (n = 5), we found that overall prevalence in both lence in Norway rats and roof rats differed among study ar- Norway rats and roof rats differed among the sampled areas eas, and among sites with 1 versus >1 species present. We (χ = 81.21, p<0.001). Rats from the Bywater area exhibited ran a subset of pairwise tests to compare prevalence among the highest overall prevalence of A. cantonensis infection study areas, correcting for multiple comparisons. (71%), whereas rats from the French Quarter exhibited the We used a Kruskal-Wallis test to compare infection lowest (8%) (Table 1; online Technical Appendix; online intensity among areas, combining data from Norway rats Technical Appendix Table 3). Most median site-level preva- and roof rats. We used Mann-Whitney U tests to compare lence values (Figure 2, panel A) were similar to those esti- infection intensity between Norway rats and roof rats and mated by pooling rats by species and area (Table 1). to compare intensity among sites with 1 or >1 species pres- Considering roof rats and Norway rats separately in ent. We conducted all statistical analyses using R version areas with >10 samples, we found that prevalence dif- 3.4.2 ( fered among areas for both species (roof rats, χ = 46.755, p<0.0001; Norway rats, χ = 43.62, p<0.0001). Prevalence Results in roof rats was lowest for rats trapped in the Lakeview area (10%) and highest for those from the Bywater area Rodent Trapping and Overall Prevalence of (70%), although pairwise comparisons showed that prev - A. cantonensis alence differed only among a subset of the study areas A total of 696 rats were trapped at 78 of the 96 sampling (online Technical Appendix; online Technical Appendix sites. Both Norway rats and roof rats were found in all 10 Table 4). Prevalence in Norway rats was lowest for rats areas, whereas hispid cotton rats (Sigmodon hispidus) were trapped in the French Quarter (2%) and highest among found in the natural area and the Lower 9th Ward, and rice those from the Gentilly area (80%). As with roof rats, rats (Oryzomys palustris) were found only in the natural area. prevalence in Norway rats differed only among a subset of We detected A. cantonensis in all 10 areas (Figure 1) the study areas (online Technical Appendix; online Tech- and in 3 of the 4 rat species sampled. Of the 444 roof rats nical Appendix Table 5). Table 1. Prevalence of rat lungworm (Angiostrongylus cantonensis), New Orleans, Louisiana, USA, May 2015–February 2017* No. positive/no. trapped (%) Area Roof rats Norway rats Cotton rats Rice rats Total Uptown 16/40 (40) 3/4 (75) – – 19/44 (43) Lakeview 7/72 (10) 1/1 (100) – – 8/73 (11) Lakeshore 27/49 (55) 0/1 (0) – – 27/50 (54) Gentilly 29/64 (45) 8/10 (80) – – 37/74 (50) French Quarter 3/6 (50) 1/47 (2) – – 4/53 (8) Bywater 14/20 (70) 6/8 (75) – – 20/28 (71) Upper 9th 14/55 (25) 10/23 (43) – – 24/78 (31) Lower 9th 42/112 (38) 71/133 (53) 0/2 (0) – 113/247 (46) Natural area 5/22 (23) 0/1 (0) 4/17 (24) 0/4 (0) 9/44 (20) St. Bernard Parish 3/4 (75) 1/1 (1) – – 4/5 (80) Total 160/444 (36) 101/229 (44) 4/19 (21) 0/4 (0) 265/696 (38) *Prevalence (in parentheses) was computed for each species in each area as the total number of positive rodents divided by the number trapped. Totals represent overall prevalence and number trapped, pooled by neighborhood, species, or both (total positive/total trapped). Roof rats, Rattus rattus; Norway rats, R. norvegicus; cotton rats, Sigmodon hispidus; rice rats, Oryzomys palustris. Dashes indicate no rats were trapped. 2178 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Rat Lungworm in Rodents, New Orleans Figure 2. Boxplots of rat lungworm (Angiostrongylus cantonensis) prevalence (A) and intensity of infection (no. lungworms per infected rat) (B) showing summary statistics across sites for each area, New Orleans, Louisiana, USA, May 2015–February 2017. Tops and bottoms of boxes indicate 25th and 75th percentiles, horizontal lines within boxes indicate medians, and error bars indicate minimum and maximum values (excluding outliers). Plots were created by using the R statistical software package ( Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2179 RESEARCH Infection prevalence was significantly lower in sites (df = 240, coefficient = 0.0595, p = 0.82) (online Tech - where only 1 rat species was trapped (30%, n = 267, 48 nical Appendix Table 7). Excluding the St. Bernard and sites) than in sites that harbored multiple rat species (44%, n French Quarter areas, each of which had only 4 infected = 429, 30 sites) (χ = 11.654, p<0.001). However, sites with rodents, we did not detect significant differences in infec - 1 species had significantly fewer rodents (mean 5.6) than tion intensity across the sampled areas (Kruskal-Wallis sites with multiple species (mean 14.3) (Mann-Whitney U test, p = 0.484). Also, we found no differences in infec - test, p<0.0001). tion intensity between sites with 1 species (8.75, n = 79) versus multiple species present (11.74, n = 186) (Mann- Likelihood of Infection Whitney U test, p = 0.072). Some median site-level in - Considering Norway rats and roof rats together, the likeli- tensities within areas (Figure 2, panel B) differed from hood of infection did not differ by sex (df = 656, coefficient intensities estimated by pooling all rats of each species = 0.13133, p = 0.42), but adults were significantly more within each area (Table 2). When we considered species likely to be infected than juveniles (df = 656, coefficient = separately, infection intensity did not differ according to –1.26201, p<0.0001) and subadults (df = 656, coefficient = sex or age class for Norway rats (all p>0.05) or roof rats –0.42601, p<0.05) (online Technical Appendix Table 6). (all p>0.05) (online Technical Appendix Table 7). When we considered species separately, we also de- tected differences in the likelihood of infection according Discussion to age class. The likelihood of infection in Norway rats We assessed the current distribution and prevalence of A. did not differ by sex (df = 224, coefficient = –0.04621, p cantonensis in definitive rat hosts across New Orleans, = 0.86), but adults were more likely to be infected than where the parasite has been present since at least 1986 either juveniles (df = 224, coefficient = –1.88906, p = (8). Our overall estimate of 38% infection prevalence in 0.003) or subadults (df = 224, coefficient = –0.82707, p New Orleans is comparable to count-based estimates re- = 0.007) (online Technical Appendix Table 6). Among ported for other areas where A. cantonensis is considered roof rats, likelihood of infection did not differ by sex (df = endemic (19,20, but see 21). We also found A. canto- 427, coefficient = 0.2305, p = 0.26) or between adults and nensis in rats across New Orleans and in neighboring St. subadults (df = 427, coefficient = –0.1541, p = 0.54), but Bernard Parish, which contrasts with the patchy distribu- adults were more likely than juveniles to be infected (df = tions exhibited by other rodent-associated pathogens in 427, coefficient = –0.9550, p = 0.004) (online Technical cities (22–25). Although rat lungworm is present across Appendix Table 6). New Orleans, infection prevalence varied according to geography and rodent host species, suggesting the risk Infection Intensity for transmission to humans might be mediated in part Infection intensity (Table 2) significantly differed be - by geographically variable landscape features that affect tween roof rats and Norway rats (Mann-Whitney U test, commensal rats (13,14). It is also likely, however, that the p<0.01). Considering Norway rats and roof rats together distributions of intermediate hosts and human population across all areas, infection intensity did not differ by sex densities moderate transmission risk. (df = 240, coefficient = 0.01331, p = 0.972). Subadults The first record of A. cantonensis in New Orleans re- had lower infection intensity than adults (df = 240, co- ported lower prevalence in rat hosts than those observed in efficient = –0.44999, p = 0.0124), but juveniles did not our study. An overall count-based prevalence of 18% was Table 2. Intensity of infection by rat lungworm (Angiostrongylus cantonensis), New Orleans, Louisiana, USA, May 2015–February 2017* Intensity (no. positive) Area Roof rats Norway rats Cotton rats Rice rats Total Uptown 13.7 (19) 19.5 (4) – – 14.7 (23) Lakeview 8.0 (7) – – – 8.0 (7) Lakeshore 8.7 (23) – – – 8.7 (23) Gentilly 8.5 (29) 27.9 (7) – – 12.3 (36) French Quarter 23.3 (3) 4.0 (1) – – 18.5 (4) Bywater 11.1 (12) 12.2 (5) – – 11.4 (17) Upper 9th 7.1 (15) 12.3 (7) – – 8.7 (22) Lower 9th 7.7 (42) 13.3 (65) – – 11.1 (107) Natural area 6.4 (5) – 13.5 (4) – 9.6 (9) St. Bernard Parish 6.3 (3) 1 (1) – – 5.0 (4) Total 9.2 (158) 14.3 (90) 13.5 (4) – 11.1 (252) *Intensity was computed for each species and each area as the sum of all lungworms counted, divided by the total number of lungworm-positive rodents for which lungworms were counted (in parentheses). Totals represent overall intensity and numbers positive pooled by neighborhood, by species, or both (total lungworms/total infected rats). Roof rats, Rattus rattus; Norway rats, R. norvegicus; cotton rats, Sigmodon hispidus; rice rats, Oryzomys palustris. Dashes indicate no rats were positive, except in the Lakeview area, where no count data were available for the 1 positive rat trapped. 2180 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Rat Lungworm in Rodents, New Orleans found for Norway rats and roof rats trapped in 1986–1987 We intentionally diagnosed infection through a visual (8). However, evidence of infection was found only in Nor- and count-based survey to draw comparisons to histori- way rats; A. cantonensis was detected in 20 (21%) of 94 cal records, but implementing complementary approaches trapped Norway rats and in 0 of 19 trapped roof rats (8). In could have provided further understanding of A. cantonen- comparison, we found overall prevalence >18% in 8 of the sis infection. As has been noted in prior studies (19,20), 10 trapping areas in our study (Table 1), and we detected count-based approaches probably yield conservative esti- infected roof rats in all trapping areas. However, it is un- mates relative to PCR-based approaches of infection. For clear whether the distribution of rodent host infections has example, in Hawaii, a count-based approach yielded a changed, because the 1986–1987 surveys were limited to prevalence estimate of 54%, whereas PCR yielded an es- trapping on wharves along the Mississippi River (8). Thus, timate of 100% (20). More extensive use of PCR-based although we can affirm that infection prevalence varies by approaches in our study probably would have afforded geography and definitive host (8), we cannot conclude that additional perspective on parasitism and could also have A. cantonensis became more broadly distributed across the excluded possible errors due to misidentification ( 31–33). city during the past 3 decades. Our results indicate that cross-disciplinary analysis of Evidence of greater prevalence and infection intensity A. cantonensis infection could shed further light on the risk of A. cantonensis in adult rats most likely reflects the in - for transmission to humans (34). As has been found with creasing probability over time that an individual rat will other zoonoses, human risk might correspond to socio- consume an infected intermediate host and that infection ecological disparities in habitat and resources favored by becomes more evident in lung tissue. Waugh et al. (19) infected definitive and intermediate hosts (22,23,35–38). similarly reported differences in intensity and prevalence For example, Rael et al. (13) detected a positive correla- in rats from Jamaica according to size but not sex, although tion between land abandonment and rat abundance in New an earlier study found that female Norway rats were more Orleans across only low-income neighborhoods. The con- likely to be infected (26). Evidence that A. cantonensis in- trasting landscapes (14) and rodent assemblages (Table 1) fection differs by host age contrasts with findings for other found in the Lower 9th Ward in New Orleans and in adja- urban rodent–associated pathogens, including flea-vec - cent St. Bernard Parish highlight the possibility that public tored Bartonella bacteria (25). The finding of distinct Bar- health risks have been shaped by differences in postdisaster tonella species in Norway rats and roof rats (25) suggests (i.e., Hurricane Katrina) land management policies (29). that co-occurrence does not facilitate pathogen transmis- Although too few rats were trapped in St. Bernard Parish (n sion, whereas our results indicate otherwise. Contrasting = 5) to confidently estimate rat lungworm prevalence (Ta- patterns in the demography of definitive host infection may ble 1), the observed differences in rat abundance suggest reflect pathogen-specific differences in transmission path - that transmission risk sharply differs between the Lower ways. Further study is therefore warranted to determine the 9th Ward and St. Bernard Parish. Thus, further study of roles of definitive host abundance and diversity in patho - associations between definitive and intermediate host in- gen transmission. fection and socioecological factors probably would better It has been more than a decade since the last diagnosed define transmission risk across the city ( 14,39). case of rat lungworm infection in the New Orleans area Although rat lungworm is just 1 of many pathogens (12), which suggests that factors unrelated to rodent hosts harbored by urban rats (40,41), citywide estimates of host mitigate the risk for transmission to humans. Although infection like those presented here can provide an epidemio- the most recent case in Louisiana resulted from consump- logic baseline that can improve understanding of infectious tion of a paratenic host (12), work elsewhere suggests that disease dynamics in cities. Baselines are particularly infor- transmission to humans most likely occurs through acci- mative for cities where zoonotic pathogens are likely to (re) dental consumption of raw or undercooked infected snails emerge because of shifting climate conditions (30,40,42) or on produce (7). Accordingly, the distribution and abun- extreme events, such as hurricanes, that can foster disease dance of infected intermediate hosts are probably key fac- outbreaks (13). New Orleans unenviably straddles both sets tors affecting transmission risk, especially in cities such as of circumstances (43–45). Accordingly, further study of rat New Orleans, where interest in urban agriculture is on the lungworm could help inform public health policies, surveil- rise (27,28). It is also possible that risk is influenced by lance programs, and intervention to safeguard the well-being climate-driven spread of invasive mollusks (29,30), such of vulnerable communities in New Orleans and elsewhere. as apple snails, that can serve as reservoirs (1,3). Further study of infection prevalence in intermediate hosts would Acknowledgments thus probably improve understanding of transmission risk We thank the investigators of the Coupled Natural and Human across New Orleans and other cities that are vulnerable to Systems–New Orleans project team for collaborations that climate change. supported this work. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2181 RESEARCH EcoHealth. 2016;13:450–5. The Tulane-Xavier Center for Bioenvironmental Research, the s10393-016-1157-1 Tulane ByWater Institute, the National Science Foundation 14. Lewis JA, Zipperer WC, Ernstson H, Bernik B, Hazen R, (BCS-0948993, BCS-1313703), and the Louisiana Board of Elmqvist T, et al. Socioecological disparities in New Orleans Regents provided funding for this work. following Hurricane Katrina. Ecosphere. 2017;8:e01922. 15. McGuire B, Pizzuto T, Bemis WE, Getz LL. General ecology of a About the Author rural population of Norway rats (Rattus norvegicus) based on inten- Dr. Rael is a mathematical biologist involved in research on sive live trapping. Am Midl Nat. 2006;155:221–36. http://dx.doi. coupled human and natural systems at the ByWater Institute org/ 10.1674/0003-0031(2006)155[0221:GEOARP]2.0.CO;2 16. King CM, Innes JG, Gleeson D, Fitzgerald N, Winstanley T, of Tulane University. Her research interests include modeling O’Brien B, et al. Reinvasion by ship rats (Rattus rattus) of forest ecological interactions and infectious disease epidemiology. fragments after eradication. Biol Invasions. 2011;13:2391–408. 17. Eamsobhana P, Lim PE, Solano G, Zhang H, Gan X, Yong HS. References Molecular differentiation of Angiostrongylus taxa (Nematoda: 1. Stockdale Walden HD, Slapcinsky JD, Roff S, Mendieta Calle J, Angiostrongylidae) by cytochrome c oxidase subunit I (COI) gene Diaz Goodwin Z, Stern J, et al. Geographic distribution of sequences. Acta Trop. 2010;116:152–6. Angiostrongylus cantonensis in wild rats (Rattus rattus) and j.actatropica.2010.07.005 terrestrial snails in Florida, USA. PLoS One. 2017;12:e0177910. 18. Yamashita T, Sato Y, Shiraki T, Otsuru M, Suzuki T. Development of Angiostrongylus cantonensis in rats, mice and guinea-pigs. 2. Al Hammoud R, Nayes SL, Murphy JR, Heresi GP, Butler IJ, Japanese J Parasitol. 1975;24:114–21. Pérez N. Angiostrongylus cantonensis meningitis and myelitis, 19. Waugh CA, Lindo JF, Lorenzo-Morales J, Robinson RD. An Texas, USA. Emerg Infect Dis. 2017;23:1037–8. epidemiological study of A. cantonensis in Jamaica subsequent 10.3201/eid2306.161683 to an outbreak of human cases of eosinophilic meningitis in 3. Teem JL, Qvarnstrom Y, Bishop HS, da Silva AJ, Carter J, 2000. Parasitology. 2016;143:1211–7. White-McLean J, et al. The occurrence of the rat lungworm, S0031182016000640 Angiostrongylus cantonensis, in nonindigenous snails in the Gulf of 20. Qvarnstrom Y, Bishop HS, da Silva AJ. Detection of rat lungworm Mexico region of the United States. Hawaii J Med Public Health. in intermediate, definitive, and paratenic hosts obtained from 2013;72(Suppl 2):11–4. environmental sources. Hawaii J Med Public Health. 2013; 4. York EM, Creecy JP, Lord WD, Caire W. Geographic range 72(Suppl 2):63–9. expansion for rat lungworm in North America. Emerg Infect Dis. 21. Zhang RL, Chen MX, Gao ST, Geng YJ, Huang DN, Liu JP, et al. 2015;21:1234–6. Enzootic angiostrongyliasis in Shenzhen, China. Emerg Infect Dis. 5. Lv S, Zhang Y, Steinmann P, Yang GJ, Yang K, Zhou XN, et al. 2008;14:1955–6. The emergence of angiostrongyliasis in the People’s Republic of 22. Costa F, Ribeiro GS, Felzemburgh RDM, Santos N, Reis RB, China: the interplay between invasive snails, climate change and Santos AC, et al. Influence of household rat infestation on Leptospira transmission dynamics. Freshw Biol. 2011;56:717–34. transmission in the urban slum environment. PLoS Negl Trop Dis. 2014;8:e3338. 6. Alicata JE. Biology and distribution of the rat lungworm, 23. Ko AI, Galvão Reis M, Ribeiro Dourado CM, Johnson WD Jr, Angiostrongylus cantonensis, and its relationship to eosinophilic Riley LW; Salvador Leptospirosis Study Group. Urban epidemic meningoencephalitis and other neurological disorders of man and of severe leptospirosis in Brazil. Lancet. 1999;354:820–5. animals. Adv Parasitol. 1965;3:223–48. S0065-308X(08)60366-8 24. Himsworth CG, Bai Y, Kosoy MY, Wood H, DiBernardo A, 7. Cowie RH. Biology, systematics, life cycle, and distribution of Lindsay R, et al. An investigation of Bartonella spp., Rickettsia Angiostrongylus cantonensis, the cause of rat lungworm disease. typhi, and Seoul hantavirus in rats (Rattus spp.) from an inner-city Hawaii J Med Public Health. 2013;72(Suppl 2):6–9. neighborhood of Vancouver, Canada: is pathogen presence a 8. Campbell BG, Little MD. The finding of Angiostrongylus reflection of global and local rat population structure? Vector cantonensis in rats in New Orleans. Am J Trop Med Hyg. Borne Zoonotic Dis. 2015;15:21–6. 1988;38:568–73. vbz.2014.1657 9. Kim DY, Stewart TB, Bauer RW, Mitchell M. Parastrongylus 25. Peterson AC, Ghersi BM, Alda F, Firth C, Frye MJ, Bai Y, et al. (=Angiostrongylus) cantonensis now endemic in Louisiana Rodent-borne Bartonella infection varies according to host wildlife. J Parasitol. 2002;88:1024–6. species within and among cities. EcoHealth. 2017;14:771–82. 0022-3395(2002)088[1024:PACNEI]2.0.CO;2 10. Gardiner CH, Wells S, Gutter AE, Fitzgerald L, Anderson DC, 26. Waugh CA, Lindo JF, Foronda P, Ángeles-Santana M, Harris RK, et al. Eosinophilic meningoencephalitis due to Lorenzo-Morales J, Robinson RD. Population distribution and Angiostrongylus cantonensis as the cause of death in captive non- zoonotic potential of gastrointestinal helminths of wild rats Rattus human primates. Am J Trop Med Hyg. 1990;42:70–4. rattus and R. norvegicus from Jamaica. J Parasitol. 2006;92:1014– 8. 11. New D, Little MD, Cross J. Angiostrongylus cantonensis infection 27. McMillan T. 5 Urban farms reshaping the food world in New from eating raw snails. N Engl J Med. 1995;332:1105–6. Orleans. The Atlantic. 2010 Nov 12 [cited 2018 Jan 1] 12. Louisiana Office of Public Health. Angiostrongylus cantonensis farms-reshaping-the-food-world-in-new-orleans/66473/ annual report 2016 [cited 2017 Nov 30]. http://new.dhh.louisiana. 28. Langenhennig S. Seeds and the city: urban lots are sprouting farms gov/assets/oph/Center-PHCH/Center-CH/infectious-epi/ across New Orleans. 2015 Jun 17 [cited 2018 Jan 01]. Annuals/Acantonensis_LaIDAnnual.pdf 13. Rael RC, Peterson AC, Ghersi BM, Childs J, Blum MJ. Distur- the_city_a_network_o.html bance, reassembly, and disease risk in socioecological systems. 2182 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Rat Lungworm in Rodents, New Orleans 29. Morand S, Bordes F, Chen HW, Claude J, Cosson JF, Galan M, et al. Global parasite and Rattus rodent invasions: The consequences for rodent-borne diseases. Integr Zool. 2015;10:409–23. EID Podcast 30. Mazza G, Tricarico E, Genovesi P, Gherardi F. Biological invaders are threats to human health: An overview. Ethol Ecol Evol. Rat Lungworm Expands 2014;26:112–29. 31. Kinsella JM. Studies on the life cycle and host specificity of Parastrongylus schmidti (Nematoda: Angiostrongylidae). Proc into North America Helminthol Soc Wash. 1987;54:245–8. 32. Spratt DM. Species of Angiostrongylus (Nematoda: Metastrongyloidea) in wildlife: A review. Int J Parasitol Parasites Wildl. 2015;4:178–89. 33. Miller CL, Kinsella JM, Garner MM, Evans S, Gullett PA, Schmidt RE. Endemic infections of Parastrongylus (=Angiostrongylus) costaricensis in two species of nonhuman The rat lungworm, Angiostrongylus primates, raccoons, and an opossum from Miami, Florida. J Parasitol. 2006;92:406–8. (Parastrongylus) cantonensis, causes 34. Grisotti M, Avila-Pires FD. Worms, slugs and humans: the medical and popular construction of an emerging infectious disease. Hist Cienc Saude Manguinhos. 2011;18:877–91. eosinophilic meningitis in humans 35. Mills JN, Childs JE. Ecologic studies of rodent reservoirs: their relevance for human health. Emerg Infect Dis. 1998;4:529–37. and various disease symptoms in 36. Macpherson CNL. Human behaviour and the epidemiology of atypical host species, including parasitic zoonoses. Int J Parasitol. 2005;35:1319–31. 37. Neiderud C-J. How urbanization affects the epidemiology of wildlife and captive animals. emerging infectious diseases. Infect Ecol Epidemiol. 2015;5:27060. 38. Bradley CA, Altizer S. Urbanization and the ecology of wildlife diseases. Trends Ecol Evol. 2007;22:95–102. Emily York, integrated pest 10.1016/j.tree.2006.11.001 39. Gulachenski A, Ghersi BM, Lesen AE, Blum MJ. Abandonment, management specialist at the Sam Noble ecological assembly and public health risks in counter-urbanizing cities. Sustainability. 2016;8:491. Museum of Natural History, discusses su8050491 40. Jones KE, Patel NG, Levy MA, Storeygard A, Balk D, Gittleman JL, et al. Global trends in emerging infectious diseases. the rat lungworm expansion in Nature. 2008;451:990–3. 41. Lloyd-Smith JO, George D, Pepin KM, Pitzer VE, Pulliam JRC, North America. Dobson AP, et al. Epidemic dynamics at the human-animal inter- face. Science. 2009;326:1362–7. science.1177345 42. Gubler DJ, Reiter P, Ebi KL, Yap W, Nasci R, Patz JA. Climate variability and change in the United States: potential impacts on vector- and rodent-borne diseases. Environ Health Perspect. 2001;109(Suppl 2):223–33. 43. Dorn PL, Perniciaro L, Yabsley MJ, Roellig DM, Balsamo G, Diaz J, et al. Autochthonous transmission of Trypanosoma cruzi, Louisiana. Emerg Infect Dis. 2007;13:605–7. 10.3201/eid1304.061002 Visit our website to listen: 44. Allen T, Murray KA, Zambrana-Torrelio C, Morse SS, Rondinini C, Di Marco M, et al. Global hotspots and correlates of emerging zoonotic diseases. Nat Commun. 2017;8:1124. player.asp?f=8640172 45. Hotez PJ, Jackson Lee S. US Gulf Coast states: The rise of neglected tropical diseases in “flyover nation”. PLoS Negl Trop Dis. 2017;11:e0005744. journal.pntd.0005744 Address for correspondence: Rosalyn C. Rael, The ByWater Institute, 627 Lindy Boggs Center, 6823 St. Charles Ave, Tulane University, New Orleans, LA 70118, USA; email: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2183 RESEARCH Terrestrial Bird Migration and West Nile Virus Circulation, United States Daniele Swetnam, Steven G. Widen, Thomas G. Wood, Martin Reyna, Lauren Wilkerson, Mustapha Debboun, Dreda A. Symonds, Daniel G. Mead, Barry J. Beaty, Hilda Guzman, Robert B. Tesh, Alan D.T. Barrett Host migration and emerging pathogens are strongly asso- The rapid expansion of WNV in the United States ciated, especially with regard to zoonotic diseases. West probably cannot be attributed to the movement of humans Nile virus (WNV), a mosquitoborne pathogen capable of because humans are dead-end hosts. However, in nature, causing severe, sometimes fatal, neuroinvasive disease in WNV is maintained in an enzootic transmission cycle in- humans, is maintained in highly mobile avian hosts. Using volving mosquito vectors and highly mobile avian reser- phylogeographic approaches, we investigated the relation- voirs. Hyalomma marginatum ticks have also been impli- ship between WNV circulation in the United States and the cated in the transmission of WNV (8). flight paths of terrestrial birds. We demonstrated southward Although evidence of WNV infection has been identi- migration of WNV in the eastern flyway and northward mi- fied in many species of birds, deaths and disease among gration in the central flyway, which is consistent with the birds vary greatly, ranging from asymptomatic to fatal in- looped flight paths of many terrestrial birds. We also identi - fied 3 optimal locations for targeted WNV surveillance cam - fections; peak viremia potentially reaches >10 PFU/mL paigns in the United States—Illinois, New York, and Texas. (9). WNV RNA has been detected in bird spleen and kid- These results illustrate the value of multidisciplinary ap- neys as long as 36 weeks after infection (10) and in brains proaches to surveillance of infectious diseases, especially of Nestor notabilis kea up to 72 months after infection (11). zoonotic diseases. Although phylogenetic evidence of geographic cluster- ing by location is limited, a recent study reported that WNV isolates clustered according to avian flyway ( 12). Because est Nile virus (WNV) is a mosquitoborne virus that birds are the primary reservoirs for WNV, this finding was Wcan cause severe and even fatal disease in humans. not surprising, but it is relevant because bird migration has After WNV introduction into New York, NY, USA, its geo- also been implicated in the movement of influenza A virus graphic range expanded quickly, reaching the West Coast (13), Borrelia burgdorferi (Lyme disease agent) (14), other in 2003. Previous studies have shown that the spread of pathogenic organisms (1), and even invasive invertebrate or- WNV occurred faster than could be explained by contigu- ganisms (15). Several serologic studies (e.g., ELISA, plaque ous diffusion (1–4) and demonstrated that its expansion oc - reduction neutralization test) have been used to determine curred heterogeneously, consisting of contiguous diffusion the direction of WNV movement within the Atlantic, Missis- and long distance translocations (2,5). Since then, phylo- sippi, and Pacific flyways and demonstrated WNV in birds geographic studies have reported frequent mixing of WNV migrating southward, whereas evidence of the virus in birds strains from local and distant locations. The most notable during northward migration is limited (16,17). exception is California, where several genetic studies have Studies of virus movement associated with avian hosts shown limited movement into and out of the state (6,7). in the United States have concentrated on the migration of waterfowl and excluded terrestrial birds, largely because Author affiliations: University of California at Davis, Davis, the migratory patterns of waterfowl have been thorough- California, USA (D. Swetnam); University of Texas Medical ly characterized by banding studies. However, passerine Branch, Galveston, Texas, USA (D. Swetnam, S.G. Widen, birds, the primary reservoir for WNV, are terrestrial birds, T.G. Wood, H. Guzman, R.B. Tesh, A.D.T. Barrett); Harris County not waterfowl. Terrestrial birds and waterfowl fly along Public Health, Houston, Texas, USA (M. Reyna, L. Wilkerson, similar but distinct flyways. Although waterfowl follow M. Debboun); Chesapeake Mosquito Control Commission, regular paths bounded by mountains and rivers, terres- Chesapeake, Virginia, USA (D.A. Symonds); University of trial birds often follow looped routes that enable them to Georgia, Athens, Georgia, USA (D.G. Mead); Colorado State maximize tail winds, avoid head winds, and correlate with University, Fort Collins, Colorado, USA (B.J. Beaty) seasonal fluctuations in food availability (18 ,19). Although DOI: looped migration paths have been described for several 2184 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Terrestrial Bird Migration and West Nile Virus species of birds (Selasphorus rufus hummingbirds [20], We generated libraries with a TruSeq RNA version 2 kit (Il- Circus aeruginosus western marsh harriers [21], Falco ele- lumina, San Diego, CA, USA) and samples sequenced by onorae Eleonora’s falcons [22], Cuculus canorus common the University of Texas Medical Branch at Galveston Next cuckoos [23]), the general flyways of terrestrial birds have Generation Sequencing Core on an Illumina 1500 Hi-Seq been inadequately studied. However, in 2014, La Sorte et platform. Adaptor sequences and poor quality reads (Q score al. provided a general description of terrestrial bird flyways <20) were removed with Trimmomatic (27). Reads were in North America (18). They defined 3 flyways: the single aligned with Bowtie2 (28) under the sensitive local param- distinct Western flyway and 2 overlapping flyways, the eter against the prototypical strain of WNV (NY99 flamingo Central and Eastern flyways. A similar 3-flyways system 382–99, GenBank accession no. AF196835). Consensus se- (Pacific, Central, and Atlantic flyways) has been described quences were generated by using SAMtools (29). for waterfowl (24); however, most studies have relied on the more common 4-flyways system (Pacific, Central, Mis- Phylogeny sissippi, and Atlantic flyways). In this study, we used phy- To evaluate temporal structure, we generated a time-naive logeographic approaches to investigate the relationship be- phylogeny (i.e., a maximum-likelihood phylogeny) to enable tween WNV circulation in the United States and the flight determination of the patristic distance between all isolates on paths of terrestrial birds. the phylogeny. We generated maximum-likelihood trees with RAxML-HPC Black Box on Cyberinfrastructure for Phylo- Methods genetic Research version 3.3 (30) and determined automatic halting by bootstrapping. We determined the root-to-tip dis- Generation of Alignments tance, which is a phylogenetic measure of genetic distance, We identified all unique sequences of natural and labora - for each isolate of the maximum-likelihood phylogenies by tory WNV strains by using the Virus Variation Resource using TempEst (formerly Path-o-gen) (31). We evaluated the (25). Virus sequences meeting the following criteria were correlation (Pearson method) between root-to-tip distance included in this study: 1) the nucleotide sequence spanned and collection date in R ( the complete open reading frame, 2) the sequence was de- We used a Bayesian Markov chain Monte Carlo rived from natural isolates and not laboratory strains, 3) the (MCMC) approach to infer phylogeographic relationships sequence was unique (i.e., all sequences differed by >1 nt), and selected the most appropriate phylogenetic model by and 4) the sequence contained no degenerate nucleotides. using standard path sampling and stepping-stone approach- All sequences were manually aligned in BioEdit version es. XML files were generated in BEAUti version 1.8.3 or 7.1.3 ( ) or 1.8.4 and run with BEAST version 1.8.4 (32) on Cyberin- MEGA7 ( , and noncoding frastructure for Phylogenetic Research (30). We used the regions were removed when necessary (i.e., the open read- GTR+Γ+I (general time reversible with gamma rate dis- ing frame was used for analyses). tribution and invariable sites) model to infer nucleotide substitution frequencies, an uncorrelated lognormal clock model to infer the mutation rate, and a Bayesian Skyline Isolation of Viral RNA and Next-Generation Sequencing tree prior to model changes in population size. The evolu- We obtained additional WNV isolates from the World Ref- –4 –4 tion rate mean was restricted to 10 through 9 × 10 substi- erence Center for Emerging Viruses and Arboviruses at the tutions/site/year, consistent with previously reported rates University of Texas Medical Branch at Galveston (Galves- for WNV evolution (7,33). ton, TX, USA) (26). Isolates were originally collected from We ran trees with an MCMC length of 100 million Virginia, Georgia, Texas, and Colorado. We extracted viral and sampled every 5,000 steps. Log files were reviewed RNA from the supernatant of infected Vero cells by using in Tracer ( ) to deter- a QIAamp Viral RNA Mini Kit (QIAGEN, Germantown, mine burn-in, which ranged from 5% through 10%. We ran MD, USA) according to the manufacturer’s instructions. Table 2. Years in which West Nile virus sequences were available in study of terrestrial bird migration and West Nile virus circulation, United States* Location 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 NY x x x x x x x x x VA x x x x x x x x x x x GA x x x x x x x IL x x x x x x TX x x x x x x x x x x x CO x x x x x x x ND x x x x x x x SD x x x x x x x *x, available; blank cells, not available. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2185 RESEARCH Figure 1. Maximum-likelihood phylogeny generated with all West Nile virus sequences from New York, Virginia, Georgia, Illinois, North Dakota, South Dakota, Texas, and Colorado (n = 379) in study of terrestrial bird migration and West Nile virus circulation, United States. Sequence names include the 2-letter state abbreviation to indicate the origin of isolation, followed by the year. Multiple isolates collected from the same state within the same year are differentiated by letter. GenBank accession numbers are provided for all taxa that were not sequenced in this study. Scale bar indicates nucleotide substitutions per site. 2186 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Terrestrial Bird Migration and West Nile Virus Figure 2. Analysis of correlation between virus isolation date and genetic diversity in study of terrestrial bird migration and West Nile virus circulation, United States. Root-to-tip distances of all sequences were determined for each isolate by using the maximum- likelihood tree shown in Figure 1 (https://wwwnc article/24/12/18-0382-F1.htm) and plotted against the year. Dots are colored by location of isolation. The correlation between the root-to-tip distance and year of isolation was determined with linear regression shown in blue. 95% CIs are shown in gray. The equation of the linear regression line was used to estimate the year of the most recent common ancestor (MRCA) and the mutation rate (m): y = mx + MRCA. multiple independent MCMC chains until effective sample down-sampled such that the number of sequences used cor- size values exceeded 200. Log and tree files were combined related (p<0.05 by Pearson method) with the incidence of in LogCombiner version 1.8, and a maximum clade cred- WNND reported to the Centers for Disease Control and ibility tree was generated in TreeAnnotator (32). Locations Prevention (CDC) in a particular year (the most accurate were inferred by using ancestral state reconstruction with record of relative WNV activity). Our analysis ensured that an asymmetric discrete trait substitution model (34). the dataset was representative of the WNV activity of each region in a particular year. Analysis of Migration We calculated incidence by using the number of WNND After the XML files were generated in BEAUti, we manu- cases reported to CDC from each state during each year and ally edited them to enable counting of all Markov jumps dividing that number by the estimated population of each (MJ) (which described the relative magnitude of migration state. The population estimates were obtained from the Time out of source locations and into sink locations) for 2001 Series of Intercensal State Population Estimates available through 2009 (35). This method for evaluating migration, at the Population Division of the US Census Bureau (42). first described by Minin and Suchard (35), has been used to States with insufficient sequences to represent the WNND characterize migration of several major pathogens includ- incidence were excluded. Down-sampling was undertaken in ing rabies virus (36), dengue virus (37), HIV (38), influ- at least duplicate to ensure that reduction in sample size and enza virus (39,40), and Rift Valley fever virus (41). diversity did not remove important relationships. As expected for an emerging zoonotic disease, the annual West Nile neurologic disease (WNND) inci- Results dence and sample collection efforts varied dramatically among states over time, adding substantial complexity Sequence Collection to the model. To mitigate the effects of inconsistent sam - All previously published sequences of natural WNV iso- pling and to confirm the observed results, we applied a lates collected in the United States were obtained from Gen- stricter inclusion criterion to confirm the results obtained Bank on January 1, 2016. The number of WNV sequences by using the full dataset. The sequences were randomly varied substantially over time and among locations, which Table 3. Statistical support for phylogeny in study of terrestrial bird migration and West Nile virus circulation, United States* Variable Mean ESS 95% HPD interval Posterior 49722.50 1370 49803.21 to 49640.55 Prior 1110 3987.09 4051.92 to 3916.56 Likelihood 2179 45735.41 45780.00 to 45691.43 MRCA 11.92 3119 10.82 to 13.08 4 4 4 UCLD.mean 1604 3.92  10 3.55  10 to 4.49  10 *ESS, effective sample size; HPD, highest posterior density; MRCA, most recent common ancestor (years before 2009); UCLD.mean, evolution rate inferred with an uncorrelated clock model with lognormal distribution. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2187 RESEARCH presented statistical challenges. In particular, although multiple years was critical to the analysis. Only a few states GenBank has >900 WNV open reading frames, most come had sufficient numbers of WNV sequences available in from a few states where laboratories were actively involved GenBank to enable analysis for multiple consecutive years: in WNV surveillance and research: California, New York, New York, Connecticut, Illinois, North Dakota, South Da- and Texas. The ability to compare multiple isolates over kota, Texas, and California. To mitigate the influence of sampling bias, we ob- tained additional WNV isolates from the World Refer- ence Center for Emerging Viruses and Arboviruses for 3 states and sequences to support the analysis: Virginia (n = 39), Georgia (n = 20), and Colorado (n = 31) (Table 1, htm). Given that previous studies have demonstrated lim- ited WNV movement into or out of California (6,7), we did not include isolates from California in the analysis. Similar- ly, because of the proximity of New York and Connecticut, we chose New York to represent WNV in the Northeast be- cause Connecticut is a small state. Last, to ensure that each location was represented across a similar time frame, we included only isolates collected during 2001–2009 in the Bayesian phylogeny and migration analysis. Table 2 shows the states and availability of yearly isolates. Model Selection We compared 203 nucleotide substitution models by using the Bayesian and Akaike Information Criteria in JModel- Test2 (https://github/com/ddariba/jmodeltest2 ) and found the GTR+Γ+I model to be the most appropriate. For assess - ing temporal signature, we used a maximum-likelihood tree with sequences of WNV strains from New York, Virginia, Georgia, Illinois, North Dakota, South Dakota, Texas, and Colorado (n = 379) (Figure 1, article/24/12/18-0382-F1.htm). We identified a statistically significant positive correlation (r = 0.93, 95% highest poste- –16 rior density [HPD] = 0.92–0.94; p<2.2 × 10 ) between the root-to-tip distance and the date of isolation in Temp-Est (for- merly known as Path-O-gen) (Figure 2). The mutation rate –4 was estimated to be 4.05 × 10 substitutions/site/year, and the most recent common ancestor (MRCA) was in 1997. To- gether these results indicated a strong temporal signal in the dataset. Finally, we evaluated Bayesian tree priors (skyride, skygrid, and skyline) and uncorrelated clock models (lognor- mal and exponential) by using path-sampling and stepping- stone approaches. The uncorrelated lognormal clock model with the Bayesian skyline tree prior was the most appropriate. Phylogeographic Analysis for the United States Analysis of all WNV sequences collected from New York, Virginia, Georgia, Illinois, North Dakota, South Dakota, Texas, and Colorado during 2001–2009 provided estimates Figure 3. Bayesian phylogeny of West Nile virus isolates collected in representative regions along the Eastern and Central flyways of the introduction date of the MRCA and mean evolution between 2001 and 2009, United States. Maximum-clade credibility rate that were consistent with the estimates of the root-to- tree was obtained by using a Bayesian approach. The location tip distance analysis (Table 3; Figure 3, https://wwwnc. of each isolate and the inferred location of each ancestor are The date of depicted by color. 2188 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Terrestrial Bird Migration and West Nile Virus Table 4. Source and sink analysis in study of terrestrial bird migration and West Nile virus circulation, United States* Source Sink Markov jumps, mean ESS 95% HPD interval CO ND 1.368 8603 0–3 GA IL 1.31 31284 1–3 IL CO 8.376 3311 1–14 IL GA 8.226 3046 3–13 IL ND 10.43 15423 6–14 IL NY 29.965 1355 20–40 IL SD 6.691 3635 1–11 IL TX 22.872 4545 14–30 IL VA 11.449 1631 6–16 NY CO 4.362 2143 0–8 NY GA 7.039 2170 2–11 NY IL 1.012 3023 0–4 NY SD 2.177 2740 0–5 NY TX 4.564 2149 0–9 NY VA 4.24 1171 0–8 SD CO 1.328 3964 0–5 TX CO 9.775 6039 3–16 TX ND 5.177 7617 2–9 TX NY 1.606 9410 0–4 TX SD 7.557 5113 2–12 VA CO 1.16 9807 0–3 VA GA 3.616 14054 2–6 *Mean number of Markov jumps detected between each source (origin) and sink (destination) location indicates the minimum number of migration events observed from each source to each sink. Only Markov jumps >2 are shown. A summary of all Markov jumps is shown in Table 5 ( *ESS, effective sample size; HPD, highest posterior density. MRCA introduction was estimated as 1997, and the aver- of the total migration events observed (Table 4; Figure 4). –4 age evolution rate was 3.92 × 10 substitutions/site/year. Southward and westward movements were detected along We used MJ between reconstructed ancestral states the East Coast, but only northward movement was observed to estimate the magnitude of relative migration out of, within the central United States. A notable exception was or into, each of the 8 regions (Table 4; Table 5, https:// observed in Illinois, where evidence of WNV movement in all directions was demonstrated. Frequent migration (>2 MJ) was detected from Illinois to Colorado (8.38 MJ), Illinois to Georgia (8.23 MJ), Il- Incidence-Controlled Phylogeny linois to North Dakota (10.43 MJ), Illinois to New York To mitigate the effects of inconsistent sampling, we applied (29.97 MJ), Illinois to South Dakota (6.69 MJ), Illinois a stricter inclusion criterion to ensure that the dataset was to Texas (22.87 MJ), Illinois to Virginia (11.45 MJ), representative of WNV activity in each region in a particu- New York to Colorado (4.36 MJ), New York to Georgia lar year. In this approach, the sequences were randomly (7.04 MJ), New York to South Dakota (2.18 MJ), New down-sampled by using the sample command in R, such York to Texas (4.56 MJ), New York to Virginia (4.24 that the number of sequences was proportional to the inci- MJ), Texas to Colorado (9.78 MJ), Texas to North Da- dence of WNND reported to CDC (Table 6). Illinois, North kota (5.18 MJ), Texas to South Dakota (7.56 MJ), and Dakota, and South Dakota were not included in the down- Virginia to Georgia (3.62 MJ). sampled datasets because there were insufficient sequences Overall, 3 major sources of WNV circulation (New to represent WNND incidence in these states. To ensure York, Illinois, and Texas) seemed to be the origin of 88.5% that reduction in sample size and diversity did not remove Figure 4. Summary of source/sink analysis in study of terrestrial bird migration and West Nile virus circulation, United States. Minimum number of migration events detected from A) the Eastern flyway, B) Illinois, and C) the Central flyway. Only events that occurred at least twice are depicted. Red arrows, northward migration; black arrows, southward migration; green arrow, lateral migration; dotted arrows, migration that could not be confirmed by incident-controlled down sampling because of an insufficient number of sequences. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2189 RESEARCH Table 6. Incidence-controlled down-sampling strategy used in study of of terrestrial bird migration and WNV circulation, United States* Location, year WNV incidence Sequences available Sequences used GA 7 2001 5 3 7.16  10 6 2002 11 6 3.29  10 6 2003 3.13  10 1 1 6 2004 1.60  10 2 2 6 2005 1 1 1.01  10 7 2006 0 0 2.18  10 6 2007 5 5 2.46  10 7 2008 4.21  10 3 3 7 2009 4.16  10 3 3 NY 7 2001 6.81  10 10 3 6 2002 3.55  10 10 6 6 2003 2.97  10 15 5 7 2004 10 3 3.65  10 6 2005 7 4 1.57  10 7 2006 5 3 8.37  10 7 2007 8.36  10 9 3 6 2008 1.67  10 18 4 7 2009 2 2 3.11  10 VA 2001 0 6 2 6 2002 2.20  10 10 5 6 2003 2.58 10 6 5 7 2004 4 3 6.69  10 2005 0 3 2 2006 0 4 2 7 2007 3.87  10 2 2 2008 0 4 2 7 2009 6.31  10 5 3 *The numbers of West Nile virus sequences available and of sequences used in the down-sampled dataset are summarized. WNV, West Nile virus. important relationships, the down-sampling was indepen- (4.04 and 2.66 MJ); New York to Texas (2.66 and 2.73 dently performed twice. MJ); Virginia to Georgia (1.55 and 3.62 MJ); and, in data- According to the 2 incidence-controlled datasets, the set 2 only, Virginia to Colorado (1.31 MJ) (Figure 6). MRCA was ≈1997 in both down-sampling exercises (95% Together, the MJ analyses of the incidence-controlled HPD 1996.00 1998.52 and 95% HPD 1995.7 1998.25), dataset and the full dataset illustrate a consistent pattern and the overall mutation rates were estimated to be 4.02 × of WNV circulation. All southward movement originated -4 –4 10 and 3.83 × 10 substitutions/site/year (Table 7; Figure in the eastern United States (New York and Virginia), and 5, most of the northward movement originated in the central htm). As with the full dataset, the Markov analysis dem- United States (Texas) (Figure 7). onstrated that New York and Texas were strong sources of WNV circulation. Significant movement (mean >2 MJ) Discussion was detected from Texas to Colorado (20.42 and 20.44 MJ); In recent years, emerging zoonotic diseases caused by Texas to New York (12.36 and 11.77 MJ); Texas to Geor- Ebola, Zika, Nipah, Middle Eastern respiratory syn- gia (8.28 and 9.55 MJ); Texas to Virginia (7.14 and 7.732 drome, and influenza A viruses have become major pub - MJ); New York to Georgia (6.1 and 5.38 MJ); New York lic health problems, devastating communities and cost- to Virginia (4.95 and 3.65 MJ); New York to Colorado ing millions for public health interventions. Decisive, Table 7. Statistical support for the incidence-controlled phylogenies determined in study of terrestrial bird migration and West Nile virus circulation, United States* Dataset 1 Dataset 2 Variable Mean ESS 95% HPD interval Mean ESS 95% HPD interval Posterior 1385 1803 36338.60 36394.36 to 36280.37 36798.90 36853.19 to 36740.97 Prior 3141.96 1187 3189.54 to 3091.38 3153.08 1387 3201.84 to 3104.50 Likelihood 33196.64 1367 33229.23 to 33165.79 33645.82 2115 33678.80 to 33614.39 MRCA 11.66 4720 10.47 to 12.10 11.97 6985 10.75 to 13.30 4 4 4 4 4 4 UCLD.mean 4.02  10 2336 3.53  10 to 4.55  10 3.83  10 1749 3.31  10 to 4.39  10 *ESS, effective sample size; HPD, highest posterior density; MRCA, most recent common ancestor (years before 2009); UCLD.mean, evolution rate inferred with an uncorrelated clock model with lognormal distribution. 2190 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Terrestrial Bird Migration and West Nile Virus Figure 5. Incidence-controlled phylogeny of Eastern and Central flyways, United States. Sequences were down-sampled such that the number of sequences was proportional to the annual incidence of West Nile neurologic disease incidence for each location between 2001 and 2009. Down-sampling was undertaken twice (A and B) to ensure that the reduction in sequences did not result in a substantial loss of diversity. Illinois, North Dakota, and South Dakota were not included in the incidence-control analysis because too few sequences were available to support down-sampling. Bayesian approaches were used to generate maximum-clade credibility trees. Scale bars indicate nucleotide substitutions per site. evidence-based approaches are critical for managing exclusion of terrestrial birds because waterfowl are easily emerging infectious diseases, but effective and efficient tracked with banding; thus, their migration has been countermeasures will be possible only after the relation- thoroughly characterized. ships between pathogens and their hosts have been thor- The introduction and subsequent spread of WNV into oughly characterized. the Americas underscores the invasive potential of emerg- Bird migration has been implicated in the movement ing pathogens in the New World, as has been recently ex- of a variety of pathogens (1). In particular, character- emplified by Zika virus, another mosquitoborne flavivirus. ization of the relationship between avian influenza vi - Dramatic variations in the location, timing, and intensity rus movement and waterfowl migration has supported of WNV strain collection and sequencing has left the field surveillance and early warning programs (1,43). How- with a limited understanding of virus circulation patterns ever, studies of avian hosts in the Americas have main- and no reliable way of predicting the geographic spread of ly concentrated on the migration of waterfowl to the WNV outbreaks. We have addressed this knowledge gap Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2191 RESEARCH Figure 6. Summary of Markov jump analysis performed on the incident- controlled phylogeny. A, B) The results of the Markov jump analysis for each down-sampled dataset are summarized as box plots. Box tops indicate third quartiles, box bottoms indicate first quartiles; horizontal bars within boxes indicate medians; error bars indicate maximums and minimums. Red, northward movement; teal, southward movement; purple, movement that is neither north nor south; dotted arrows, movement that was not observed in the incident-controlled down-sampling because of an insufficient number of sequences. C, D) Movement originating in the eastern and central United States. Only Markov jumps that occurred >2 times are depicted. by characterizing the movement of WNV with regard to the seasonal shifts in terrestrial bird migration routes ensure migratory patterns of its natural hosts, terrestrial birds. We that Illinois supports birds from the Eastern and Central compiled 379 virus sequences for analysis, including 289 flyways during annual migrations. previously reported sequences from New York, Virginia, Of note, although mosquito and WNV activity occurs Georgia, Illinois, North Dakota, South Dakota, Texas, and earlier in the southern than in the northern United States, Colorado, plus 90 novel sequences from Virginia, Georgia, southward migration was detected along the East Coast and Colorado. during our sampling period, 2002–2009, indicating that the Phylogeographic analysis revealed that 3 locations— southeastern United States is probably a dead end for WNV New York, Illinois, and Texas—accounted for 88.5% of circulation. Indeed, low-level transmission probably oc- the total WNV MJ inferred. Because New York is the pre- curs during the winter in warmer locations such as Florida, sumed original introduction point for WNV into the United Texas, and Louisiana. This possibility is supported by iso- States, its role as a major source of WNV movement was lations of WNV from mosquitoes and birds during Decem- expected. However, 74.2% of the observed MJ originated ber and January in Harris County, Texas (44), and suggests in Illinois and Texas only. Of note, North Dakota and South that ecologic factors not related to mosquito abundance and Dakota, which are 2 of the states with the highest annual WNV activity drive WNV movement along the East Coast. WNND incidence, seem to be strong sinks for WNV mov- Instead, movement of WNV into the northeastern United ing out of Illinois and Texas. States (New York) from Illinois and Texas was observed The contributions of Illinois and Texas to WNV circu- (in the incident-controlled analysis). These results suggest lation are not surprising because both locations are situated that introduction of WNV into the northeastern United at major convergence points between the Eastern and Cen- States originated from the central United States. tral flyways. In the case of Texas, birds from both flyways Overall, we have defined the pattern of WNV circula - may avoid long-distance flights across the Gulf of Mexico tion in the United States (Figure 7) and demonstrated looped by traveling along the circa-Gulf route that follows the Gulf virus movement patterns in the Eastern and Central flyways Coast through Texas into Mexico. In the case of Illinois, that are bridged by Illinois, a region shared between the 2192 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Terrestrial Bird Migration and West Nile Virus these advancements support the construction of targeted surveillance and vector mitigation strategies to predict the annual flow of WNV strains and to enable public health of- ficials to anticipate changes in WNV circulation resulting from altered bird migration. This study was supported in part by National Institutes of Health contracts N01-AI 25489, N01-AI30027, and HHSN272201000040I/HHSN27200004/D04 to R.B.T. and grant AI 067847 to A.D.T.B. About the Author Dr. Swetnam is a postdoctoral fellow at the University of California at Davis. Her research interests include the evolution and Figure 7. Model summarizing the general patterns of West Nile molecular epidemiology of zoonotic and emerging diseases. virus movement in the United States. Red, northward movement; teal, southward movement; dotted arrows, relationships that could not be confirmed in incident-controlled datasets because of an References insufficient number of sequences. 1. Reed KD, Meece JK, Henkel JS, Shukla SK. Birds, migration and emerging zoonoses: West Nile virus, lyme disease, influenza A and enteropathogens. Clin Med Res. 2003;1:5–12. 2 flyways. This specific pattern correlates with the looped migration patterns of terrestrial birds. Although other geo- 2. Pybus OG, Suchard MA, Lemey P, Bernardin FJ, Rambaut A, graphic regions may contribute to virus movement, there Crawford FW, et al. Unifying the spatial epidemiology and were insufficient virus sequences available from other molecular evolution of emerging epidemics. Proc Natl Acad Sci U S A. 2012;109:15066–71. states to incorporate into this analysis. Thus, on the basis of pnas.1206598109 available information, 3 of the 8 locations considered (New 3. Lewis M, Rencławowicz J, van den Driessche P. Traveling waves York, Illinois, and Texas) seem to be the preferred sites for and spread rates for a West Nile virus model. Bull Math Biol. efficiently monitoring ongoing WNV evolution. 2006;68:3–23. 4. Maidana NA, Yang HM. Spatial spreading of West Nile Virus As new WNV sequences become available, similar described by traveling waves. J Theor Biol. 2009;258:403–17. phylogeographic methods can be used to develop more detailed information about WNV circulation in the United 5. Duggal NK, Reisen WK, Fang Y, Newman RM, Yang X, Ebel GD, States. For example, on the East Coast, WNV circulation et al. Genotype-specific variation in West Nile virus dispersal in California. Virology. 2015;485:79–85. occurs southward, so surveillance efforts in the Northeast j.virol.2015.07.004 are likely to be more informative than surveillance in the 6. Duggal NK, Bosco-Lauth A, Bowen RA, Wheeler SS, Reisen WK, Southeast. Conversely, WNV in the central United States Felix TA, et al. Evidence for co-evolution of West Nile Virus travels northward, so surveillance in the south-central and house sparrows in North America. PLoS Negl Trop Dis. 2014;8:e3262. United States is more likely than surveillance in the north- 7. Añez G, Grinev A, Chancey C, Ball C, Akolkar N, Land KJ, et al. central United States to be informative. Last, the region Evolutionary dynamics of West Nile virus in the United States, of overlap between the Eastern and Central flyways is the 1999–2011: phylogeny, selection pressure and evolutionary most likely location for deriving surveillance information time-scale analysis [cited 2013 Dec 13]. journal.pntd.0002245 because WNV in this area travels in multiple directions. 8. Formosinho P, Santos-Silva MM. Experimental infection of Collectively, the results of this study illustrate the val- Hyalomma marginatum ticks with West Nile virus. Acta Virol. ue of using multidisciplinary approaches to surveillance of 2006; 50:175–80. infectious diseases, especially zoonotic diseases. Animal 9. Komar N, Langevin S, Hinten S, Nemeth N, Edwards E, Hettler D, et al. Experimental infection of North American birds with the migration is shaped by a delicate balance of ecologic fac- New York 1999 strain of West Nile virus. Emerg Infect Dis. tors and anthropomorphic barriers. Natural and manmade 2003;9:311–22. events (e.g., climate change, atmospheric fluctuations, hab- 10. Wheeler SS, Langevin SA, Brault AC, Woods L, Carroll BD, itat destruction) can drastically alter host behavior, which Reisen WK. Detection of persistent West Nile virus RNA in experimentally and naturally infected avian hosts. Am J in turn affects the circulation patterns of infectious agents Trop Med Hyg. 2012;87:559–64. such as WNV. In this study, we defined the patterns of ajtmh.2012.11-0654 WNV circulation and key areas for surveillance and cor- 11. Bakonyi T, Gajdon GK, Schwing R, Vogl W, Häbich AC, related them with the migratory patterns of their primary Thaller D, et al. Chronic West Nile virus infection in kea (Nestor notabilis). Vet Microbiol. 2016;183:135–9. reservoir, terrestrial birds. Although this information does not enable prediction of the size of annual WNV outbreaks, Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2193 RESEARCH 12. Di Giallonardo F, Geoghegan JL, Docherty DE, McLean RG, 29. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Zody MC, Qu J, et al. Fluid spatial dynamics of West Nile virus in et al. 1000 Genome Project Data Processing Subgroup. The the United States: rapid spread in a permissive host environment. Sequence Alignment/Map format and SAMtools. Bioinformatics. J Virol. 2015;90:862–72. 2009;25:2078–9. 13. Bahl J, Krauss S, Kühnert D, Fourment M, Raven G, Pryor SP, 30. Miller MA, Pfeiffer W , Schwartz T. Creating the CIPRES Science et al. Influenza a virus migration and persistence in North Gateway for inference of large phylogenetic trees [cited 2018 American wild birds. PLoS Pathog. 2013;9:e1003570. Oct 5]. 31. Rambaut A, Lam TT, Max Carvalho L, Pybus OG. Exploring 14. Vollmer SA, Bormane A, Dinnis RE, Seelig F, Dobson ADM, the temporal structure of heterochronous sequences using Aanensen DM, et al. Host migration impacts on the phylogeography TempEst (formerly Path-O-Gen) [cited 2018 Oct 5]. of Lyme borreliosis spirochaete species in Europe. Environ Microbiol. 2011;13:184–92. 32. Drummond AJ, Suchard MA, Xie D, Rambaut A. Bayesian j.1462-2920.2010.02319.x phylogenetics with BEAUti and the BEAST 1.7. Mol Biol Evol. 15. Muñoz J, Amat F, Green AJ, Figuerola J, Gómez A. Bird 2012;29:1969–73. migratory flyways influence the phylogeography of the invasive 33. Gray RR, Veras NMC, Santos LA, Salemi M. Evolutionary brine shrimp Artemia franciscana in its native American range. characterization of the West Nile virus complete genome. Mol PeerJ. 2013;1:e200. Phylogenet Evol. 2010;56:195–200. 16. Dusek RJ, McLean RG, Kramer LD, Ubico SR, Dupuis AP II, j.ympev.2010.01.019 Ebel GD, et al. Prevalence of West Nile virus in migratory 34. Lemey P, Rambaut A, Drummond AJ, Suchard MA. Bayesian birds during spring and fall migration. Am J Trop Med Hyg. phylogeography finds its roots. PLOS Comput Biol. 2009;81:1151–8. 2009;5:e1000520. 17. Reisen WK, Wheeler SS, Garcia S, Fang Y. Migratory birds and 35. Minin VN, Suchard MA. Counting labeled transitions in the dispersal of arboviruses in California. Am J Trop Med Hyg. continuous-time Markov models of evolution. J Math Biol. 2010;83:808–15. 2008;56:391–412. 18. La Sorte FA, Fink D, Hochachka WM, Farnsworth A, 36. Tohma K, Saito M, Kamigaki T, Tuason LT, Demetria CS, Rodewald AD, Rosenberg KV, et al. The role of atmospheric condi- Orbina JRC, et al. Phylogeographic analysis of rabies viruses tions in the seasonal dynamics of North American migration flyways. in the Philippines. Infect Genet Evol. 2014;23:86–94. J Biogeogr. 2014;41:1685–96. 19. La Sorte FA, Fink D, Hochachka WM, DeLong JP, Kelling S. 37. Raghwani J, Rambaut A, Holmes EC, Hang VT, Hien TT, Spring phenology of ecological productivity contributes to Farrar J, et al. Endemic dengue associated with the co-circulation the use of looped migration strategies by birds. Proc Biol Sci. of multiple viral lineages and localized density-dependent 2014;281:20140984–20140984. transmission. PLoS Pathog. 2011;7:e1002064. rspb.2014.0984 10.1371/journal.ppat.1002064 20. Phillips AR. Migrations of Allen’s and other hummingbirds. 38. Ward MJ, Lycett SJ, Kalish ML, Rambaut A, Leigh Brown AJ. Condor. 1975;77:196–205. Estimating the rate of intersubtype recombination in early 21. Klaassen RHG, Strandberg R, Hake M, Olofsson P, Tøttrup AP, HIV-1 group M strains. J Virol. 2013;87:1967–73. Alerstam T. Loop migration in adult marsh harriers Circus aeruginosus, as revealed by satellite telemetry. J Avian Biol. 2010; 39. Lemey P, Rambaut A, Bedford T, Faria N, Bielejec F, Baele G, 41:200–7. et al. Unifying viral genetics and human transportation data to 22. Mellone U, López-López P, Limiñana R, Piasevoli G, Urios V. predict the global transmission dynamics of human influenza The trans-equatorial loop migration system of Eleonora’s falcon: H3N2. PLoS Pathog. 2014;10:e1003932. differences in migration patterns between age classes, regions journal.ppat.1003932 and seasons. J Avian Biol. 2013;44 [cited 2018 Oct 5]. 40. Nelson MI, Viboud C, Vincent AL, Culhane MR, Detmer SE, Wentworth DE, et al. Global migration of influenza A viruses in 23. Willemoes M, Strandberg R, Klaassen RHG, Tøttrup AP, swine. Nat Commun. 2015;6:6696. Vardanis Y, Howey PW, et al. Narrow-front loop migration in a ncomms7696 population of the common cuckoo Cuculus canorus, as revealed 41. Samy AM, Peterson AT, Hall M. Phylogeography of Rift Valley by satellite telemetry. PLoS One. 2014;9:e83515. fever virus in Africa and the Arabian Peninsula. PLoS Negl Trop 10.1371/journal.pone.0083515 Dis. 2017;11:e0005226. 24. Boere GC, Stroud DA. The flyway concept: what it is and what it journal.pntd.0005226 isn’t. In: Boere GC, Galbraith C., Stroud DA, editors. Waterbirds 42. US Census Bureau. State intercensal tables: 2000–2010 [cited around the World. Edinburgh: The Stationery Office; 2006. p. 40–4. 2018 Oct 3]. 25. Hatcher EL, Zhdanov SA, Bao Y, Blinkova O, Nawrocki EP, popest/intercensal-2000-2010-state. html Ostapchuck Y, et al. Virus Variation Resource–improved response 43. Global Consortium for H5N8 and Related Influenza V iruses. to emergent viral outbreaks. Nucleic Acids Res. 2017;45:D482–90. Role for migratory wild birds in the global spread of avian influenza H5N8. Science. 2016;354:213–7. 26. Health UTMB. World Reference Center for Emerging Viruses and 10.1126/science.aaf8852 Arboviruses [cited 2018 Feb 20]. 44. Tesh RB, Parsons R, Siirin M, Randle Y, Sargent C, Guzman H, home et al. Year-round West Nile virus activity, Gulf Coast region, 27. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer Texas and Louisiana. Emerg Infect Dis. 2004;10:1649–52. for Illumina sequence data. Bioinformatics. 2014;30:2114–20. 28. Langmead B, Salzberg SL. Fast gapped-read alignment with Address for correspondence: Alan D.T. Barrett, University of Texas Bowtie 2 [cited 2014 Mar 19]. Medical Branch, 301 University Blvd, Galveston, TX 77555-0436, USA; articlerender.fcgi?artid=3322381&tool=pmcentrez&rendertype= email: abstract 2194 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Capnocytophaga canimorsus Capsular Serovar and Disease Severity, Helsinki Hospital District, Finland, 2000–2017 1 1 1 Estelle Hess, Francesco Renzi, Panu Karhunen, Mélanie Dol, Adrien Lefèvre, 2 2 Jenni Antikainen, Elodie Carlier, Johanna Hästbacka, and Guy R. Cornelis We assembled a collection of 73 Capnocytophaga cani- with influenza-like symptoms and intestinal complaints morsus isolates obtained from blood cultures taken from (1), a disease severity not always reaching the threshold patients treated at Helsinki University Hospital (Helsinki, for a blood culture. Moreover, C. canimorsus is a fas- Finland) during 2000–2017. We serotyped these isolates by tidious and slow-growing organism, rendering its culture PCR and Western blot and attempted to correlate pathogen and isolation difficult ( 2). serovar with patient characteristics. Our analyses showed, Human exposure to a dog’s oral flora can occur through in agreement with previous research, that 3 C. canimorsus a bite or scratch or even through just being in close prox- serovars (A–C) caused most (91.8%) human infections, de- imity to the animal (1,5). Although splenectomy, asplenia, spite constituting only 7.6% of isolates found in dogs. The alcohol abuse, smoking, and advanced age are often de- 3 fatalities that occurred in our cohort were equally repre- scribed as predisposing factors for severe illness caused by sented by these serovars. We found 2 untypeable isolates, which we designated serovars J and K. We did not detect this bacterium, up to 40% of patients have no obvious risk an association between serovar and disease severity, im- factor (1); thus, C. canimorsus should not be considered mune status, alcohol abuse, or smoking status, but dog exclusively an opportunistic pathogen. bites occurred more frequently among patients infected with C. canimorsus is enveloped by a lipooligosaccharide non-A–C serovars. Future research is needed to confirm and a capsule consisting of units of the same O antigen serovar virulence and develop strategies to reduce risk for but assembled by different polymerases (6). The capsule these infections in humans. confers to C. canimorsus resistance to the bactericidal ef- fects of human serum and phagocytosis by macrophages apnocytophaga canimorsus is a gram-negative, (6). One study showed that despite the seemingly vast rep- Crod-shaped, usually commensal bacteria of dog and ertoire of capsular serovars among C. canimorsus isolates cat oral flora that causes rare but potentially severe in - from dog mouths, 3 serovars (A, B, and C) are associated fections in humans (1,2). Even with administration of with most human infections (7). However, this finding was adequate antimicrobial therapy, C. canimorsus–induced from a study carried out with just 25 isolates from patients septicemia can progress to a debilitating disease or sep- worldwide. To validate this finding, we evaluated the se- tic shock and can cause a mortality rate as high as 30%. rovars present in a collection of 73 isolates from patients Annual incidence of C. canimorsus infections has been treated at Helsinki University Hospital (Helsinki, Finland) estimated at 0.5–0.67 cases/1 million persons (3,4), but during 2000–2017. in a retrospective study, a prevalence of 4.1 cases/1 mil- lion persons was estimated (5); this discrepancy prob- Materials and Methods ably resulted from the choice of diagnostics. The clinical manifestation of C. canimorsus infection might be mild, Study Setting HUSLAB (Helsinki) is a central laboratory that offers Author affiliations: University of Namur, Namur, Belgium (E. Hess, microbiological services to the whole Helsinki Hospi- F. Renzi, M. Dol, A. Lefèvre, E. Carlier, G.R. Cornelis); University tal District, which encompasses the city of Helsinki and of Eastern Finland, Kuopio, Finland (P. Karhunen); University of surrounding municipalities. The laboratory maintains a Helsinki and Helsinki University Hospital, Helsinki, Finland (J. Antikainen, J. Hästbacka) These first authors contributed equally to this article. DOI: These authors were co–principal investigators. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2195 RESEARCH frozen archive of bacterial isolates obtained from pa- serotyping, we grew bacteria on heart infusion agar plates tient blood cultures. For the purposes of this study, we (BD Difco, Franklin Lakes, NJ, USA) supplemented with searched laboratory records for blood cultures positive 5% sheep blood (Oxoid, Basingstoke, UK) and 20 µg/mL for C. canimorsus during 2000–2017; a corresponding gentamicin (Sigma-Aldrich, Darmstadt, Germany) for 48 frozen bacteria isolate could be found for 78 patients. Of h at 37°C with 5% CO . these frozen isolates, we could grow and analyze 73. To correlate analyses with clinical data, we searched patient C. canimorsus Identification by 16S rDNA Sequencing journals, electronic patient records, and laboratory da- We extracted genomic DNA directly from blood culture tabases for patient characteristics, clinical information, bottles or by boiling of a single colony (online Tech- and laboratory data. We recorded patient age, sex, con- nical Appendix Table 1, current medical conditions, medications administered, article/24/12/17-2060-Techapp1.pdf). We used 4 different immune status, lifestyle factors, and type of contact amplification methods involving 8 different primers to se- with dogs (bitten, contact but not bitten, or not known), quence 16S rDNA from bacterial isolates (online Techni- whenever the information was available. Of the clini- cal Appendix Tables 2, 3). When >1 primer was used to cal data, we recorded the level of care, length of stay in sequence a PCR product, we obtained the consensus se- the hospital, complications, 30-day and 1-year mortality quence using Bioedit ( rates, and registered coagulation and fibrinolysis labora - com), and we analyzed sequences using BLAST (https:// tory variables. We analyzed partial thromboplastin time according to the Owren method (8). Antisera Production and Adsorption The Administrative Department of Helsinki Hos- The production of antisera to serovars A–I has previ- pital District and Helsinki City College of Social and ously been described (7). We produced rabbit polyclonal Health Care gave approval for obtaining this data from anti-J (against isolate H12) and anti-K (against isolate patient medical records. Because only data registers H24) likewise (7). Immunizations were carried out at the were used for acquiring data, obtaining informed con- Centre d’Economie Rurale (Aye, Belgium). The Centre sent from patients was waived. d’Economie Rurale animal welfare committee approved our animal handling protocols and procedures. We ad- Bacterial Isolates and Growth Conditions sorbed anti-J and anti-K sera with a mixture of 25 iso- We cultured C. canimorsus bacterial isolates (Table 1) lates from patients (Cc1–Cc25; online Technical Appen- obtained from HUSLAB, which were originally obtained dix Table 4) to obtain polyclonal antibodies specifically from blood samples of patients in Finland, as described recognizing J or K capsular serovars. We performed ad- previously (9). In brief, we incubated aerobic and an- sorptions by incubating 250 µL of antiserum with 6 × aerobic blood culture bottles with BacT/ALERT 3D (bio- 10 paraformaldehyde-fixed bacteria on a rotating wheel Mérieux, Marcy l’Etoile, France) for 6 days or until the at room temperature for >2 hours. We removed bacteria cultures became positive. We used Gram staining and by successive centrifugations. We repeated the incuba- cultivated all positive samples on chocolate agar, fas- tions and centrifugations 4 times. We performed capsular tidious anaerobe agar, or heart infusion agar plates. For Table 1. Capsular typing of 73 Capnocytophaga canimorsus isolates from patient blood samples, Helsinki Hospital District, Finland, 2000–2017* PCR typing† Western blot typing‡ Isolates ABC A B C D E A B C D E F G H I Serovar H11, H16, H23, H37, H39, H42, H48, H52, H56, + + + – – – + – – ND ND ND ND ND ND A H60, H62, H70, H74, H75, H76, H78, H80 H3, H4, H5, H6, H9, H14, H22, H25, H26, H30, H35, + – + – – – – + – ND ND ND ND ND ND B H38, H49, H50, H53, H55, H57, H58, H63, H65, H67, H68, H69, H71, H72, H73, H79 H27 + – + – – – + + – ND ND ND ND ND ND B H1, H7, H8, H10, H13, H15, H17, H18, H19, H20, + – – + – – – – + ND ND ND ND ND ND C H28, H29, H33, H34, H36, H43, H44, H45, H46, H47, H51, H59 H41, H64 – – – – + – – – – + ND ND ND ND ND D H31 – – – – – + – – – ND + ND ND ND ND E H21 – – – – – – – – – – – – – – + I H12, H24 – – – – – – – – – – – – – – – NT *ND, not done; NT, nontypeable. †We performed PCR capsular typing using the oligonucleotides given in online Technical Appendix Table 2 ( 2060-Techapp1.pdf). Results were interpreted as done previously (7): isolates positive for PCR ABC, A, and B were typed as A, isolates positive for PCR ABC and B were typed as B, and isolates positive for PCR ABC and C were typed as C. ‡Western blot analyses on polysaccharidic structures were performed by using specific polyclonal rabbit antisera. 2196 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 C. canimorsus Serovar and Disease Severity typing of C. canimorsus by Western blot, ELISA, and serovar A or B, but in agreement with the PCR results, PCR as previously described (7). we considered this isolate a B capsular serovar only. In short, 91.8% (67/73) of isolates tested were serovars A Statistical Analysis (n = 17), B (n = 28), or C (n = 22). We expressed categorical data as counts and percentages We then subjected isolates to PCR analyses for the and continuous data as medians and interquartile ranges. detection of capsular types D and E, which have previ- We compared categorical data between groups by Fisher ously been detected among C. canimorsus isolates from exact test. We assumed continuous data were nonnormally human infections (7). Two isolates were serovar D and 1 distributed and analyzed data using Mann-Whitney U-test serovar E (online Technical Appendix Figure 1), findings for 2 groups and Kruskal-Wallis nonparametric test for >3 that were confirmed by Western blot analyses (Table 1; groups. Because of the retrospective nature of the study, online Technical Appendix Figure 2). We tested the 3 re- many data points were unavailable for many cases (data maining nontypeable (non-A–E) isolates by Western blot were more complete for severely ill patients and less com- for capsular types F–I, which have only been detected in plete for mildly ill patients), so we provided the number of isolates obtained from dogs (7). Isolate H21 was typed as patients included in each analysis. We considered p val- serovar I, leaving only 2 strains (H12 and H24) not typed ues <0.05 statistically significant and performed analyses of the 73 tested. using SPSS version 22 ( We raised rabbit antisera against H12 and H24 bacte- spss-statistics-software). ria and adsorbed antisera with related bacteria strains. The 2 new antisera recognized only the capsule of the isolate Results against which they were raised, indicating the 2 isolates be- longed to 2 new serovars, which we named J and K (online Technical Appendix Figure 2). Thus, the 73 C. canimorsus Capsular Typing Collection of 73 Isolates from Finland isolates from the Helsinki University Hospital collection We identified the 73 isolates originating from Helsinki comprised 8 serovars (Figure 1, panel A); A, B, and C dom- University Hospital (Table 1) as C. canimorsus through inated (91.8%), consistent with the findings of the previous 16S rDNA sequencing (online Technical Appendix study involving 25 worldwide isolates (Figure 1, panel B) Tables 1, 3). We subjected isolates to a PCR designed (7). The distribution of serovars A (p = 0.071), B (p = 0.47), to detect capsular serovars A, B, and C (7); 67 of 73 C (p = 0.20), D (p = 0.27), E (p = 0.45), and I–K (p = 1) isolates were ABC positive (Table 1; online Technical was not significantly different between the 2 collections (all Appendix Figure 1). We also typed these 67 strains us- p values analyzed by Fisher exact test; Figure 1, panel C). ing A-, B-, and C-specific PCR tests (7 ). To validate the PCR typing results, we performed Western blot analy- Screening of Dog Isolates for Capsular Serovars ses with polysaccharide samples of the 73 isolates using J and K antiserum specifically recognizing A, B, or C capsular We next tested for the prevalence of the J and K capsu- serovars (online Technical Appendix Figure 2). This lar serovars in a previously described collection of C. analysis confirmed the PCR typing results and interpre - canimorsus isolates obtained from mouths of healthy dogs tation of all isolates tested, except H27. According to (7,10). We screened these 52 dog isolates by ELISA using Western blot analyses, isolate H27 could be considered Figure 1. Prevalence of capsular serovars among Capnocytophaga canimorsus isolates from patients and dogs. A) Prevalence among 73 isolates from patients in Helsinki, Finland, 2000–2017. B) Prevalence among 25 isolates acquired from patients worldwide. C) Prevalence among pooled samples (n = 98). D) Prevalence among 52 isolates from dog mouths, Switzerland and Belgium. Percentages do not add up to 100% because of rounding. A portion of the data presented in panels B and D were previously published (7). Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2197 RESEARCH the antisera we produced. Although no isolates reacted with serovars A, B, and C are all capable of causing severe dis- the anti-K serum, isolate CcD35 from a dog in Switzerland ease in humans. reacted with the anti-J serum (Figure 1, panel D; online We looked for an association between capsular serovar Technical Appendix Table 5). We confirmed this result and patient immune status or lifestyle factors but found no by Western blot analysis of the polysaccharidic structures statistically significant link between serovar and immune (online Technical Appendix Figure 2), which showed that compromised state (p = 0.682), alcohol abuse (p = 0.982), capsular serovar J is thus not limited to Finland. or smoking (p = 0.713) (Figure 2, panels B–D). We defined patients as immune compromised if they had been on im- Correlation between Disease Severity and mune suppressive medication or had recently received che- Capsular Type motherapy, had a concurrent medical condition associated We also tested the association between serovar and disease with impaired immunity or active cancer, or had undergone severity. For this investigation, the level of care was used splenectomy. One of the 2 splenectomized patients had a as a surrogate; patients treated in a regular ward or who severe course of disease, but both survived. had only visited the emergency department were regarded Severe C. canimorsus infections are often associated as having a mild course of disease, and patients treated in with purpura or petechiae, disseminated intravascular co- a high-dependency or intensive care unit were regarded as agulation, and gangrene of extremities (1). In particular, severely ill. No statistically significant difference could be coagulation disorders were found to be associated with found in the proportions of any serovar between patients 94% of patients having C. canimorsus–induced septic with mild and severe disease (p = 0.76; Figure 2, panel A). shock in a 10-year retrospective study in Helsinki (5). In Among the 73 cases of C. canimorsus infection included in our study, no statistically significant association could be this study, 3 were fatal (Table 2). The isolates from these found between coagulation and fibrinolysis laboratory 3 patients were serovars A (H80), B (H26), and C (H28). variables (platelet count, partial thromboplastin time, fi - Extensive amputations were reported in 6 cases, among brin D-dimers) and capsular serovars (Table 3). Given the which included the nonsurviving patient infected with the low number of cases associated with some serovars, we capsular B isolate H26. The 5 other capsular types associ- could assess only the 3 dominant serovars (A, B, and C). ated with amputations were A (n = 2, H48 and H56), B (n We compared coagulation and fibrinolysis disorder mark - = 1, H79), and C (n = 2, H46 and H59). Therefore, capsular ers between patients with mild and severe clinical course. Figure 2. Association between Capnocytophaga canimorsus capsular serovar and various patient factors, Helsinki, Finland, 2000–2017. A) Disease severity (n = 70); B) immune compromised (n = 73); C) alcohol abuse (n = 49); D) smoking status (n = 48); and E) contact with dogs (n = 73). Fisher exact test was used for statistical analysis. 2198 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 C. canimorsus Serovar and Disease Severity As expected, the analyzed variables were more affected Table 2. Patient demographics, clinical characteristics, and contact with dogs, Helsinki Hospital District, Finland, 2000–2017* in patients with a severe course of infection (Table 3). No. Of note, deviating coagulation and fibrinolysis variables Characteristic patients† Value were frequently present in patients with mild courses of Age, y, median (IQR) 73 55 (48.3– disease, further strengthening the previously reported 64.8) Sex 73 close association of C. canimorsus infection and coagula- M 38 (52.1) tion disorders. F 35 (47.9) Immune compromised 73 7 (9.6) Smoking 48 30 (62.5) Correlation between Type of Contact with Dog Alcohol abuse 49 18 (36.7) and Capsular Type Contact with dog 73 The type of contact with dogs did not differ among infec - Not known 21 (28.8) tions with any of the dominant serovars, but 4 of the 5 Contact but not bitten 15 (20.5) Bitten 37 (50.7) patients infected with serovar D, E, I, or J had been bitten Disease severity 70 (Figure 2, panel E). The contact type was not known for Regular ward or emergency 45 (64.3) the patient infected with the serovar K isolate, the fifth department High surveillance unit 11 (15.7) rare serotype. Intensive care unit 14 (20.0) Length of hospital stay, d, median (IQR) 62 6 (3–13.3) Discussion Deaths at day 30 73 3 (4.1) In this study, we analyzed 73 C. canimorsus isolates ob- Deaths at 1 y 61 4 (6.6) Amputation 73 6 (8.2) tained from patients treated at Helsinki University Hospi- *Values are no. (%) patients except as indicated. IQR, interquartile range. tal. All isolates were serotyped and found to be endowed †Because of missing data, number of patients in each category varied. with a capsular polysaccharide (CPS), further confirming serovars D and E were previously isolated from patients in the commonality of the presence of a CPS in C. canimor- the United States (n = 1), Belgium (n = 1), and Switzerland sus isolates (6,7). We confirmed the high prevalence of (n = 1) (7). Thus, although serovars D and E represent only capsular serovars A, B, and C among isolates from human 4.1% and 2%, respectively, of the total clinical isolates in infections; 67 (91.8%) of 73 isolates were typed as 1 of this study, these serovars should be considered virulent and these 3 serovars. No significant difference was found in the taken into account in prophylaxis. prevalence of these serovars between this collection of 73 One patient in our cohort was infected with a se- isolates from Finland and a previously studied collection of rovar I strain. This serovar had not been encountered 25 isolates obtained from cases worldwide (7). Among the before among humans but was found in dogs (1 in Bel- 98 C. canimorsus isolates from these 2 studies, 89 (90.8%) gium and 1 in Switzerland) (7). These findings suggest were capsular types A, B, or C. Our data confirmed that se- that not only serovars A–E but also rare serovars are rovars A, B, and C are significantly more common among widely distributed. clinical isolates than dog isolates (4/52; 7.6%), suggesting Last, we describe 2 new capsular serovars, J and K, these serovars are more virulent than the others. Our data each with a limited (1%) prevalence in human infections. also confirmed that serovars A, B, and C are present in dif- We tested these 2 new antisera against our collection of iso- ferent geographic areas. lates obtained from dogs in Switzerland and Belgium (10) Besides the A, B, and C serovars, the Helsinki collec- and found 1 C. canimorsus isolate had a J-type CPS. Thus, tion contained 2 other serovars: 2 isolates of serovar D and using the 11 antisera we have that are specific to serovars 1 of serovar E. This observation is of high interest because Table 3. Coagulation and fibrinolysis laboratory variables, by Capnocytophaga canimorsus serovar and disease severity, Helsinki, Finland, 2000–2017* Reference Serovar Severity of illness Variable range A, n = 17 B, n = 28 C, n = 22 p value Severe, n = 25 Mild, n = 45 p value Platelets, 10 /L 150–360 109 (29–137) [2] 109 (28–140) 93 (23–166) [1] 0.98 23 (9.5–89) [0] 117 (95–154.3) <0.001 [1] [4] PTT, %† 70–130 56 (24–71) [10] 78 (56–86) 58 (44.5–75) 0.284 54 (39–66) [2] 87 (70.5–109.5) <0.001 [11] [12] [31] FiDD, mg/L <0.5 47.9 (5.05–83.8) 9.1 (3.4–85.7) 14.5 (4.1–80.7) 0.888 32.9 (5.6–81) 1.6 (0.6–74.7) 0.057 [12] [19] [15] [5] [40] *Values are given as median (interquartile range) [no. missing values] except as indicated. The comparison between patients with mild and severe courses of disease was defined by the level of care they needed. Patients with mild disease were those who were treated in a regular ward or the emergency department, and patients with severe disease were those treated in high surveillance or intensive care units. FiDD, fibrin D-dimers; PTT, partial thromboplastin time. †PTT was analyzed according to the Owren method (8). PTT was calculated as the ratio of the result (in seconds) from normal plasma to the result (in seconds) from the patient sample x 100. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2199 RESEARCH A–K, which identified 98 human clinical isolates, we can that only a few C. canimorsus strains are virulent in hu- only type 36.5% (19/52) of our collection of dog isolates mans, and few dogs carry these dangerous strains. In- from Switzerland and Belgium. This finding, again, rein- deed, the 3 most prevalent serovars in human infection forces the hypothesis of the existence of a large repertoire (A–C), represent only 7.6% of the C. canimorsus iso- of CPS serovars in C. canimorsus among dog isolates. lates from dogs (7), suggesting that a minority of dogs Because C. canimorsus extensively deglycosylates hu- represent a risk for humans. This disease might be pre- man N-linked glycoproteins from cell surfaces (11–13), a ventable in humans by identifying the dogs that carry given blood group might be a predisposing factor for C. these dangerous serotypes and specifically vaccinating canimorsus infection, but further research is needed to in- them to eliminate the pathogen or drastically reduce vestigate an association between blood type and serovar. pathogen shedding. Blood group information was available for 55 patients This work was financed by grant SOC 1510582 from the Belgian in our cohort, and we found no enrichment in any blood Walloon Region. groups among patients infected with C. canimorsus com- pared with the blood group distribution of the population of (data not shown). About the Author The availability of clinical records associated with Dr. Hess is a researcher at the Research Unit in the Biology the isolates typed in this study gave us the opportunity to of Microoganisms at the University of Namur in Belgium investigate the link between capsular serovar and disease and an immunologist with a research interest in host– severity, patient immune status, lifestyle, or type of contact pathogen interactions. with dogs. When comparing the most prevalent capsular types (A, B, and C) found in these 73 clinical isolates, we References found no significant correlation between disease severity 1. Butler T. Capnocytophaga canimorsus: an emerging cause of and capsular type. In the previous C. canimorsus capsular sepsis, meningitis, and post-splenectomy infection after dog typing study, the authors suggested that strains belonging bites. Eur J Clin Microbiol Infect Dis. 2015;34:1271–80. 10.1007/s10096-015-2360-7 to capsular types of lower prevalence, like D and E, might 2. Butler T, Weaver RE, Ramani TK, Uyeda CT, Bobo RA, Ryu JS, preferentially infect immunocompromised patients (7); we et al. Unidentified gram-negative rod infection. A new disease of could not draw such a conclusion here. In addition, alcohol man. Ann Intern Med. 1977;86:1–5. abuse or smoking status could not be linked to infection 0003-4819-86-1-1 3. Pers C, Gahrn-Hansen B, Frederiksen W. Capnocytophaga by a specific capsular type. Alcohol abuse, smoking status, canimorsus septicemia in Denmark, 1982–1995: review of 39 and immune suppression all were not significantly associ- cases. Clin Infect Dis. 1996;23:71–5. ated with disease severity or the 30-day mortality rate (data clinids/23.1.71 not shown), although the relatively low sample size and 4. van Dam AP, Jansz A. Capnocytophaga canimorsus infections in The Netherlands: a nationwide survey. Clin Microbiol Infect. missing data preclude us from drawing conclusions regard- 2011;17:312–5. ing this matter. 2010.03195.x The capsular serovars less frequently isolated in hu- 5. Hästbacka J, Hynninen M, Kolho E. Capnocytophaga canimorsus man infections, such as E, I, and J, were mainly found in bacteremia: clinical features and outcomes from a Helsinki ICU cohort. Acta Anaesthesiol Scand. 2016;60:1437–43. patients who had been bitten, which could suggest that these serovars are less virulent than serovars A–D, perhaps 6. Renzi F, Ittig SJ, Sadovskaya I, Hess E, Lauber F, Dol M, et al. requiring a deeper inoculation to provoke an infection. Un- Evidence for a LOS and a capsular polysaccharide in fortunately, the information on dog exposure was missing Capnocytophaga canimorsus. Sci Rep. 2016;6:38914. for the patient infected with the serovar K strain. 7. Hess E, Renzi F, Koudad D, Dol M, Cornelis GR. Identification of Two patients included in this study were reportedly virulent Capnocytophaga canimorsus isolates by capsular typing. bitten on the same day by the same dog. The isolates from J Clin Microbiol. 2017;55:1902–14. these 2 patients (H44 and H46) were both typed as capsu- JCM.00249-17 8. Owren PA. Thrombotest a new method for controlling lar serovar C, suggesting that the same strain infected both anticoagulant therapy. Lancet. 1959;274:754–8. patients. The 2 patients had a severe form of the infection, 10.1016/S0140-6736(59)90857-8 requiring treatment in an intensive care unit. This observa- 9. Tissari P, Zumla A, Tarkka E, Mero S, Savolainen L, Vaara M, et al. tion of 2 patients being infected by the same dog has not Accurate and rapid identification of bacterial species from positive blood cultures with a DNA-based microarray platform: an been reported previously and gives an indication of the epi- observational study. Lancet. 2010;375:224–30. demiology of disease. 10.1016/S0140-6736(09)61569-5 The observation of so few cases of C. canimorsus in- 10. Renzi F, Dol M, Raymackers A, Manfredi P, Cornelis GR. Only a fection is indeed striking, considering that up to 74% of subset of C. canimorsus strains is dangerous for humans. Emerg Microbes Infect. 2016;5:e29. dogs carry C. canimorsus bacteria (14). We hypothesize 2200 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 C. canimorsus Serovar and Disease Severity 11. Mally M, Shin H, Paroz C, Landmann R, Cornelis GR. Pathog. 2011;7:e1002118. Capnocytophaga canimorsus: a human pathogen feeding at ppat.1002118 the surface of epithelial cells and phagocytes. PLoS Pathog. 14. Suzuki M, Kimura M, Imaoka K, Yamada A. Prevalence of 2008;4:e1000164. Capnocytophaga canimorsus and Capnocytophaga cynodegmi in 12. Manfredi P, Renzi F, Mally M, Sauteur L, Schmaler M, Moes dogs and cats determined by using a newly established species- S, et al. The genome and surface proteome of Capnocytophaga specific PCR. Vet Microbiol. 2010;144:172–6. canimorsus reveal a key role of glycan foraging systems in host 10.1016/j.vetmic.2010.01.001 glycoproteins deglycosylation. Mol Microbiol. 2011;81:1050–60. Address for correspondence: Guy R. Cornelis, Unité de Recherche 13. Renzi F, Manfredi P, Mally M, Moes S, Jenö P, Cornelis GR. en Biologie des Microorganismes, Université de Namur, 61 rue de The N-glycan glycoprotein deglycosylation complex (Gpd) from Bruxelles, 5000 Namur, Belgium; email: Capnocytophaga canimorsus deglycosylates human IgG. PLoS etymologia Capnocytophaga canimorsus [kapʺno-si-tofʹǝ-gǝ kanʺǝ-morʹsǝs] Ronnie Henry rom the Greek kapnos (“smoke”) for its dependence on carbon dioxide, which is a large F component of smoke, Capnocytophaga canimorsus (Latin canis, “dog,” and morsus, “bite”) are gram-negative, facultatively anaerobic, rod-shaped bacteria that are part of the normal oral microbiota of dogs and cats. The genus was proposed to distinguish these bacteria from Cytophaga spp. (Greek kytos, “cell,” and phagein, “eat”), which also exhibit gliding motility. C. canimorsus was previously known as CDC group DF-2 (dysgonic fermenter type 2) and was first isolated from a Paul de Vos, Cats man who had ex- Fighting in a Larder 1630–1640. Oil on perienced multiple canvas. Museo Nacional del Prado. https://www. dog bites and devel- coleccion/galeria-on- oped septicemia and line/galeria-on-line/ meningitis. C. cani- obra/pelea-de-gatos- en-una-despensa/, morsus remains a Public Domain, https://commons. major cause of sep- ticemia in persons, php?curid=39117357 particularly those who are asplenic or immunocompro- mised, who are bit- ten by dogs or cats. Sources 2. Leadbetter ER, Holt SC, Socransky SS. 1. Brenner DJ, Hollis DG, Fanning GR, Weaver RE. Capnocytophaga: new genus of gram-negative gliding Capnocytophaga canimorsus sp. nov. (formerly bacteria. I. General characteristics, taxonomic CDC group DF-2), a cause of septicemia follow- considerations and significance. Arch Microbiol. ing dog bite, and C. cynodegmi sp. nov., a cause of 1979;122:9–16. localized wound infection following dog bite. J Clin Microbiol. 1989;27:231–5. Address for correspondence: Ronnie Henry, Centers for Disease Control and Prevention, 1600 Clifton Rd NE, Mailstop E28, Atlanta, GA 30329-4027, USA; email: DOI: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2201 RESEARCH Crimean-Congo Hemorrhagic Fever Virus, Mongolia, 2013–2014 Matthew A. Voorhees, Susana L. Padilla, Dulamjav Jamsransuren, Jeffrey W. Koehler, Korey L. Delp, Dolgorkhand Adiyadorj, Uyanga Baasandagwa, Battsetseg Jigjav, Scott P. Olschner, Timothy D. Minogue, Randal J. Schoepp During 2013–2014, we collected 1,926 serum samples from CCHFV is maintained in the environment in a cycle humans and 4,583 ticks (Hyalomma asiaticum or Derma- between tick vectors, primarily Hyalomma ticks, and wild centor nuttalli) in select regions of Mongolia to determine the and domesticated animals. Birds generally appear to be re- risk for Crimean-Congo hemorrhagic fever virus (CCHFV) fractory to CCHFV infection, although some avian species infection among humans in this country. Testing of human demonstrate infectivity (5). Widespread dispersion is most serum samples by ELISA demonstrated an overall CCHFV likely mediated by infected ticks carried by migrating birds antibody prevalence of 1.4%; Bayankhongor Province had or moving livestock (1,6–8). Viremia in mammals is tran- the highest prevalence, 2.63%. We pooled and analyzed sient, but ticks remain infected throughout their lives (9). tick specimens by real-time reverse transcription PCR; 1 The human species is considered a dead-end host because CCHFV-positive H. asiaticum tick pool from Ömnögovi was virus transmission from infected humans to uninfected identified. In phylogenetic analyses, the virus’s partial small feeding ticks has not been reported. Humans can become segment clustered with CCHFV isolates from Central Asia, and the complete medium segment grouped with CCHFV infected when fed on by infected ticks, exposed to infec- isolates from Africa, Asia, and the Middle East. This study tious bodily fluids during animal slaughter, or caring for confirms CCHFV endemicity in Mongolia and provides in- infected patients (1,10). Infection can result in an unappar- formation on risk for CCHFV infection. Further research is ent or mild disease of fever, chills, nausea, vomiting, head- needed to better define the risk for CCHFV disease to im- ache, and body pains. In severe cases, hemorrhagic disease prove risk mitigation, diagnostics, and surveillance. is characterized by petechial rashes of the skin, bleeding from mucosal membranes and muscles, organ failure, and cerebral hemorrhage. Mortality rates have been estimated rimean-Congo hemorrhagic fever (CCHF) is a tick- to be 5%–30% but have reached as high as 80% in some Cborne disease of humans that is prevalent over a wide outbreaks (10). One report describes CCHF survivors as geographic area, spanning from western China to southern having lasting effects, including prolonged vision and Asia, from the Middle East to southeastern Europe, and over memory problems (11). most of Africa (1–3). The causative agent is Crimean-Congo CCHFV prevalence in Mongolia is generally un- hemorrhagic fever virus (CCHFV), a member of the Bun- known. No published studies provide evidence of human yaviridae family and Orthonairovirus genus. The enveloped infection or of virus circulating in tick vectors. A previous virus has a negative-sense RNA genome composed of 3 am- study in the Mongolia aimag (i.e., province) of Arkhan- bisense segments: small (S), medium (M), and large (L). The gai demonstrated serologic evidence of virus infection in S segment encodes the nucleocapsid protein and the L seg- 1.8% of wildlife species tested ( 12), and a large study in- ment the viral polymerase. The M segment encodes a glyco- volving >2,000 samples collected from sheep demonstrat- protein precursor that is processed into 2 glycoproteins by ed a CCHFV seroprevalence of 27.1% among sheep from the host cell. Genetic drift and segment reassortment of the multiple aimags across Mongolia (13). Furthermore, nu- CCHFV genome results in a high degree of phylogeographic merous CCHFV strains have been identified in Xinjiang diversity, with 5–7 designated genotypes (2,4). Province, a region of China bordering Mongolia (14–16). Author affiliations: United States Army Medical Research Institute of Serologic evidence of CCHFV infections in wildlife and Infectious Diseases, Fort Detrick, Maryland, USA (M.A. Voorhees, livestock suggests the virus circulates in Mongolia and S.L. Padilla, J.W. Koehler, K.L. Delp, S.P. Olschner, T.D. Minogue, that CCHFV infections are possible for persons exposed R.J. Schoepp); Ministry of Health National Center for Zoonotic to biting ticks or infectious tissues or fluids. In this study, Diseases, Ulaanbaatar, Mongolia (D. Jamsransuren, D. Adiyadorj, we sought to better characterize the risk for CCHFV in- U. Baasandagwa, B. Jigjav) fection among the population of Mongolia by determining antibody prevalence in humans associated with livestock DOI: 2202 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Crimean-Congo Hemorrhagic Fever Virus, Mongolia and by surveying ticks collected from the environment Table 1. Prevalence of CCHFV IgG among healthy persons as determined by ELISA, by province and district, Mongolia, and livestock for evidence of CCHFV infection. 2013–2014* No. positive/tested (%) Materials and Methods Province, district 2013 2014 Total Khovd 0/229 (0) NA 0/229 (0) Üyench 0/79 (0) 0/79 (0) Human Samples Bulgan 0/72 (0) 0/72 (0) During 2013–2014, we collected serum samples (n = 1,926) Altai 0/78 (0) 0/78 (0) from persons in regions corresponding to the estimated Govi-Altai 3/314 (0.96) 1/21 (4.76) 4/335 (1.19) Bugat 2/84 (2.38) 2/84 (2.38) geographic distribution of Hyalomma asiaticum ticks (17); Altai 1/84 (1.19) 1/84 (1.19) we collected most samples in 2013. We collected samples Tsogt 0/74 (0) 0/74 (0) from Dornod Aimag in eastern Mongolia; Dornogovi Erdene 0/72 (0) 0/72 (0) Aimag, Dundgovi Aimag, Ömnögovi Aimag, Bayankhon- Taishir 0/5 (0) 0/5 (0) Yesönbulag 1/16 (6.25) 1/16 (6.25) gor Aimag, and Govi-Altai Aimag in southern Mongolia Bayankhongor 4/146 (2.74) 1/44 (2.27) 5/190 (2.63) (the Gobi Desert landscapes); and Khovd Aimag in west- Bayan-Öndör 2/74 (2.70) 0/1 (0) 2/75 (2.67) ern Mongolia (Table 1; Figure 1). We collected blood from Shinejinst 2/72 (2.78) 2/72 (2.78) Jinst 1/43 (2.33) 1/43 (2.33) nomadic herders and local residents of both sexes from the Ömnögovi 7/525 (1.33) 1/40 (2.50) 8/565 (1.42) selected regions by using standard blood sampling tech- Gurvan-tes 3/79 (3.80) 0/10 (0) 3/89 (3.37) niques. We centrifuged and stored serum samples at –20°C Noyon 1/71 (1.41) 1/10 (10.00) 2/81 (2.47) Bayandalai 0/74 (0) 0/10 (0) 0/84 (0) until transporting them to the National Center for Zoonotic Khürmen 1/75 (1.33) 0/10 (0) 1/85 (1.18) Diseases (Ulaanbaatar, Mongolia), where samples were Nomgon 1/75 (1.33) 1/75 (1.33) stored at −70°C. We collected all human samples in ac- Bayan-Ovoo 1/75 (1.33) 1/75 (1.33) cordance with Mongolia Ministry of Health–approved pro- Khanbogd 0/76 (0) 0/76 (0) Dornogovi 6/410 (1.46) 4/72 (5.56) 10/482 (2.07) tocols and tested all human samples at the US Army Medi- Khatanbulag 2/93 (2.15) 2/17 (11.76) 4/110 (3.63) cal Research Institute of Infectious Diseases (USAMRIID) Khövsgöl 3/91 (3.30) 1/14 (7.14) 4/105 (3.81) under approved human use protocol FY11–14. Ulaanbadrakh 0/60 (0) 1/16 (6.25) 1/76 (1.31) Erdene 0/80 (0) 0/14 (0) 0/94 (0) Zamyn-Üüd 1/86 (1.16) 0/11 (0) 1/97 (1.03) IgG ELISA Dundgovi NA 0/76 (0) 0/76 (0) We detected human CCHFV IgG by sandwich ELISA. We Khuld 0/33 (0) 0/33 (0) Luus 0/39 (0) 0/39 (0) coated Immulon 2 HB plates (ThermoFisher Scientific, Saintsagaan 0/4 (0) 0/4 (0) Waltham, MA, USA) with anti-CCHFV hyperimmune Dornod NA 0/49 (0) 0/49 (0) mouse ascitic fluid (1:1,000) in phosphate-buffered saline Bayan-Uul 0/14 (0) 0/14 (0) and 0.01% thimerosal. After incubating plates overnight Bayandun 0/15 (0) 0/15 (0) Dashbalbar 0/20 (0) 0/20 (0) at 4°C and washing 3 times with wash buffer (phosphate- Total 20/1,624 (1.23) 7/302 (2.32) 27/1,926 (1.40) buffered saline with 0.1% tween 20 and 0.01% thimero - *CCHFV, Crimean-Congo hemorrhagic fever virus; NA, none available. sal), we added 100 µL of inactivated CCHFV IbAr10200 subtracting the average OD value of the Vero cell super- supernatant (IgG capture antigen) or noninfected Vero E6 natant control wells from the average OD value of the IgG cell supernatant (negative control) diluted (1:20) in milk capture antigen wells. The positive-negative cutoff of each buffer (wash buffer plus 5% skim milk) to each well. We assay was the mean OD value plus 3 SDs of the 4 negative incubated plates for 60 min at 37°C, washed 3 times with control samples (typically 0.2). wash buffer, and added serum samples diluted 1:100 in milk buffer to test wells and negative control wells. We Plaque Reduction Neutralization Test incubated plates for 60 min at 37°C, washed 3 times with For the plaque reduction neutralization test (PRNT), we wash buffer, and added 100 µL of the detector antibody diluted all test samples 1:10 in minimum essential media (horseradish peroxidase–labeled anti-human IgG F ; Accu- (MEM) with 5% heat-inactivated Hyclone Fetal Bovine rate Chemical and Scientific, Westbury, NY, USA) diluted Serum (FBS) (GE Healthcare Life Sciences, Pittsburgh, 1:8,000 in milk buffer to all wells. We incubated plates for PA, USA) and then placed samples in a 56°C water bath 60 min at 37°C, washed 3 times with wash buffer, and de- for 30 min. For test and positive control serum samples, we veloped using 100 µL of substrate 2, 2’-azino-di-(3-ethyl- 2-fold serially diluted to 1:320 in MEM with 5% FBS. For benzothiazoline-6-sulfonate) with a 30-min incubation at negative control serum samples, we 2-fold serially diluted 37°C. We read plates using a TECAN Infinite M200 Pro to 1:40 in the same buffer. We diluted CCHFV IbAr10200 (Tecan Group Ltd., Männedorf, Switzerland) plate reader in MEM with 5% FBS to a concentration of 1,000 plaque- at 405 nm and determined the optical density (OD) for each forming units/mL. well. We calculated the adjusted OD for each sample by Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2203 RESEARCH Figure 1. Geographic distribution of CCHFV-positive serum samples and tick and CCHFV-negative serum samples, Mongolia, 2013– 2014. CCHFV, Crimean-Congo hemorrhagic fever virus. We mixed positive control, negative control, and test transferred ticks to −70°C storage until shipment to US- serum samples with virus 1:1 in microtiter tubes, covered AMRIID for processing. We collected 4,583 ticks for pro- tubes with parafilm, and then placed them at 37°C for 1 cessing, of which 1,772 were H. asiaticum and 2,811 were h. We incubated 200 µL of serum-virus mixture in dupli- Dermacentor nuttalli. cate wells of 6-well plates with SW-13 cells grown to a We sorted field-collected unfed specimens by sex and minimum of 85% confluence. We incubated plates at 37°C species and placed ticks in pairs into single polycarbonate with 5% CO for 1 h with gentle rocking every 15 min. vials containing prepared media and 3/8-inch steel balls. After incubation, we added 2 mL of 0.5% agar overlay (2 × We homogenized ticks using the 1600 MiniG Automated basal medium Eagle with Earle salts with 10% FBS, 1% Tissue Homogenizer and Cell Lyser (SPEX SamplePrep, penicillin-streptomycin, 1% L-glutamine, and 1% nones- Metuchen, NJ, USA), combined portions of homogenized sential amino acids) to each well and incubated at 37°C for subpools into larger pools (maximum 8 ticks/pool), and ≈36 h. Then, we added a second 0.5% agar overlay contain- treated them with TRIzol LS (ThermoFisher Scientific) ing 5% neutral red to each well, incubated at 37°C for 4–5 according to the manufacturer’s instructions. We purified h, and counted plaques. We reported the reciprocal of the total nucleic acid using the KingFisher Flex Purification highest serum dilutions reducing 50% (PRNT ) and 80% System (ThermoFisher Scientific) and the MagMax 96 (PRNT ) of the plaque assay dose as the titer and calcu- for MicroArrays Total RNA Isolation Kit (ThermoFisher lated the probit titer. Scientific) according to the manufacturer’s recommenda - tions. We stored all tick pools and processed nucleic acids Tick Collections and Processing at −70°C until needed for testing. We collected ticks from livestock and questing ticks from the environment in the same regions of Mongolia where Real-Time Reverse Transcription PCR we collected human serum samples. During the spring and We tested total nucleic acid extracted from tick pools for early summer, we collected adult, unfed ticks by the drag the CCHFV S segment using a real-time reverse transcrip- cloth method (dragging a cotton cloth on a dowel) and by tion PCR (RT-PCR) assay optimized for CCHFV sequence removing ticks from livestock with forceps. We placed diversity (18,19). We performed this assay in triplicate in each collection in a centrifuge tube with a moist cotton. wells of 384-well plates using the SuperScript III One-Step In the field, we stored tick collections in portable thermal RT-PCR Kit (Invitrogen, Carlsbad, CA, USA) with 2.5 µL coolers at 4°C. We sent ticks to the National Center for of sample in 10-µL reactions run on the LightCycler 480 Zoonotic Diseases, where samples were stored at –20°C (Roche Applied Science, Penzberg, Germany). Cycling until species identification was completed. Afterward, we conditions were 50°C for 15 min; 95°C for 5 min; and 45 2204 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Crimean-Congo Hemorrhagic Fever Virus, Mongolia cycles of 94°C for 1 s, 55°C for 20 s, and 68°C for 5 s. reaction. Cycling conditions were 96°C for 1 min and 25 Fluorescence was measured after each extension step. We cycles of 96°C for 10 s, 50°C for 5 s, and 60°C for 4 min. considered a sample negative if the quantification cycle We sequenced libraries using the 3500xL Genetic Analyzer was >40 cycles. (Applied Biosystems). We attempted to amplify CCHFV segments L, M, Nested RT-PCR and S from tick pool 159A using segment-specific primers To confirm real-time RT-PCR results, we performed nested modified for Nextera-based sequencing (Table 2) (21,22); RT-PCR targeting the CCHFV S segment as previously de- however, only the M segment amplified. In brief, we used scribed (20). In brief, we used 15 µL of extracted tick ho- 5 µL of extracted nucleic acid from tick homogenate to re- mogenate and SuperScript III One-Step RT-PCR System verse transcribe the genome and amplify the cDNA of the with Platinum Taq High Fidelity DNA Polymerase (Invi- M segment using the SuperScript III One-Step RT-PCR trogen) in a 50-µL reaction with outer primers CCHF-F2 System with Platinum Taq High Fidelity DNA Polymerase (5′-TGGACACCTTCACAAACTC-3′), CCHF-F2C (5′-TG- (Invitrogen). Cycling conditions were 52.5°C for 30 min; GATACTTTCACAAACTC-3′), and CCHF-R3 (5′-GA- 94°C for 2 min; 40 cycles of 94°C for 15 s, 50.5°C for 30 s, CAAATTCCCTGCACCA-3′). Cycling conditions were and 68°C for 1 min/kb; and 68°C for 5 min. We ran the am- 42°C for 30 min; 94°C for 5 min; 5 cycles of 94°C for 30 s, plified M segment through an agarose gel and purified the 37°C for 30 s, and 72°C for 30 s; 72°C for 2 min; 30 cycles of band using the QIAquick Gel Extraction Kit (QIAGEN). 94°C for 30 s, 52°C for 30 s, 72°C for 30 s; and a final cycle We generated next-generation sequencing libraries of 72°C for 5 min. We assessed for the presence of amplicons using the Nextera XT DNA Library Preparation Kit (Illu- (536 bp) by performing electrophoresis on 2% agarose gels. mina, San Diego, CA, USA) according to the manufactur- We performed nested PCR with 1 µL of amplicon as er’s instructions and sequenced using the 500-cycle MiSeq template in a 50-µL reaction with inner primers CCHF- Reagent Kit v2 (Illumina). We analyzed sequencing reads F3 (5′-GAATGTGCATGGGTTAGCTC-3′), CCHF-F3C with the CLC Genomics Workbench (QIAGEN), filtered (5′-GAGTGTGCCTGGGTTAGCTC-3′), CCHF-R2a and trimmed them for quality, and assembled the complete (5′-GACATCACAATTTCACCAGG-3′), and CCHF-R2b M segment de novo. We generated the final consensus se- (5′-GACATTACAATTTCGCCAGG-3′). Cycling condi- quence by remapping the trimmed reads to the de novo tions were 94°C for 5 min; 30 cycles of 94°C for 30 s, 52°C consensus sequence as previously described (22). for 30 s, 72°C for 30 s; and 72°C for 5 min. We assessed We identified the full-length M segment sequence for the presence of amplicons (260 bp) by performing elec- and partial S segment sequence from tick pool 159A by trophoresis on 2% agarose gels. BLAST analysis ( We aligned the segments from tick pool 159A with the CCHFV Sequencing and Analysis available full-length M and near–full-length S segments in We performed a second RT-PCR with S segment outer GenBank and generated a neighbor-joining phylogenetic primers with the only tick pool that was positive by real- tree (Jukes-Cantor model) with 1,000 bootstrap replicates. time and nested RT-PCR (159A). We then used 5 µL of the reaction for Sanger sequencing. We ran the amplicon (536 Virus Isolation bp, near 5′ end of S segment) through a 2% agarose gel, cut To propagate virus in the RT-PCR–positive tick pool, we the band out, and purified the amplicon using the QIAquick thawed 60 µL of tick homogenates and mixed with 6 mL Gel Extraction Kit (QIAGEN, Desseldorf, Germany). We of freshly prepared MEM with 10% FBS, 1% nonessential created sequencing libraries using the BigDye Terminator amino acids, 1% L-glutamine, 1% penicillin-streptomycin, v3.1 Sequencing Kit (Applied Biosystems, Foster City, and 0.4% fungizone. We inoculated confluent monolayers CA, USA) with 10 µL of the purified amplicon in a 20-µL of SW-13 cells in 2 T75 flasks with 2.5 mL of the virus Table 2. Primer sequences used for Nextera-based sequencing of S, M, and L segments of CCHFV from tick pool 159A, Mongolia, 2013–2014* Primer name Sequence, 5′  3′† CCHFV-M F-next TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGtctcaaagaaatacttgc CCHFV-M R-next GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGtctcaaagatatagtggc CCHFV-S F-next TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGtctcaagaaacacgtgccgc CCHFV-S R-next GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGtctcaaagatatcgttgccgc CCHF-L 1F-next TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGtctcaaagatatcaatcccccc CCHF-L 1R-next GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGttggcactatctttcatttga CCHF-L 2F-next TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGgaagagctatatgacataaggc CCHF-L 2R-next GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGtctcaaagaaatcgttccccccac *CCHF, Crimean-Congo hemorrhagic fever; CCHFV, Crimean-Congo hemorrhagic fever virus; L, large; M, medium; S, small. †The virus-specific sequences are lowercase and the Nextera-specific sequences are capitalized. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2205 RESEARCH mixture and incubated at 37°C for 1 h; then, we added an CCHFV isolate having the highest percent identity at 91%, additional 30 mL of MEM. We incubated flasks at 37°C with IbAr10200 having a percent identity of 81%. Phyloge- with 5% CO for 1 wk or until cytopathic effects (CPE) netic analysis indicated that the partial S segment sequence were evident. To test for unapparent infections (i.e., infec- clusters within the Asia 2 lineage (Figure 2), which con- tions not resulting in CPE), each inoculated culture was tains virus strains from China and Russia (23). CCHFV M tested by real-time RT-PCR as previously described. segments in general do not cluster well by geographic lo- cation on phylogenetic analyses; the M segment sequence Results tick 159A/Mongolia clustered with CCHFV strains from Asia, Africa, and the Middle East (Figure 3). CCHFV Serologic Testing in Humans To assess the overall risk for CCHFV infection in Mongo- Virus Isolation lia, we tested 1,926 serum samples collected from nomadic Attempts to isolate CCHFV from putative PCR-positive herders and local residents in select regions of the country tick pools in SW-13 culture cells were made from multiple by ELISA. In total, 27 samples tested positive for CCH- tick homogenate subpools (N = 18), containing 2 ticks per FV IgG; the overall prevalence rate was 1.4% (Table 1). subpool. None of the inoculated cultures showed evidence Bayankhongor Aimag had the highest prevalence (2.63%); of virus growth by visible CPE or genomic signature by other aimags with CCHFV IgG–positive populations had real-time RT-PCR in either the first or second passages. prevalences of 2.07% (Dornogovi), 1.42% (Ömnögovi), and 1.19% (Govi-Altai). The remaining 3 aimags tested Discussion (Khovd, Dundgovi, and Dornod) showed no evidence of To better determine the risk for CCHFV infection among CCHFV IgG. humans in Mongolia, we conducted an assessment that in- We evaluated 12 positive and 7 randomly selected cluded a CCHFV IgG serosurvey and screening for CCH- negative samples for their ability to neutralize CCHFV FV RNA in ticks collected in parallel with the serosurvey. strain IbAr10200 (Table 3). In total, 8 (67%) of 12 ELI- Overall, this study provides evidence that CCHFV is cir- SA-positive samples and 2 (29%) of 7 ELISA-negative culating in H. asiaticum ticks and humans in Mongolia, samples neutralized CCHFV with PRNT probit titers; 5 provides sequence data on a virus circulating in the region, samples with PRNT probit titers (3 ELISA-positive and 2 and extends the geographic distribution of CCHFV north ELISA-negative) had PRNT probit titers. Neutralization to Mongolia. ability did not correlate with ELISA OD values. Previous studies conducted in Mongolia revealed se- rologic evidence of CCHFV in wildlife species and live- stock. Chumikhin et al. found serologic evidence for CCH- Molecular Testing for CCHFV in Ticks FV infection in small mammals, such as the Tolai hare, We collected questing ticks and ticks from livestock and Mongolian pika, and long-tailed ground squirrel (12,13). identified and pooled them on the basis of tick species, sex, Livestock, such as cattle, sheep, goats, and camels, are am- and geographic location; then, we screened tick pools by plifying hosts of the virus, and CCHFV infections in live- real-time RT-PCR using a broad CCHFV assay (19). Of stock correlate with the risk for infection of humans (24). the 893 tick pools tested, 1 pool (159A) tested positive by Morikawa investigated CCHFV IgG seroprevalence in real-time RT-PCR and was confirmed CCHFV positive sheep in the southern aimags of Mongolia and found preva- by nested RT-PCR; results for 6 other pools were indeter- lences of 7%–28% (13). Evidence of CCHFV in wildlife minate. Tick pool 159A, which contained 7 H. asiaticum and livestock in Mongolia suggests the possibility of virus ticks, was collected in the soum (i.e., district) Nomgon in circulation and human disease, but the actual occurrence Ömnögovi Aimag. was unclear. We found an overall CCHFV IgG seroprevalence of Characterization of CCHFV-Positive Pool 159A 1.4% in Mongolia; Bayankhongor Aimag had the highest To genetically characterize the CCHFV detected in tick rate, 2.63%. Each soum tested within Bayankhongor had pool 159A, we amplified and sequenced the full-length IgG prevalences >2%, suggesting a persistent source of M segment and partial S segment. BLAST analysis of the CCHFV infection is likely present in these soums. Bayank- nucleocapsid coding region located within the partial S hongor is located in the southwestern part of the country segment identified K128_76 (GenBank no. KX013455.1) and has an economy based on livestock, meat, and wool from Kazakhstan as the CCHFV isolate having the great- production. The greater interaction with animals among est percent identity at 98%, and IbAr10200, the prototypi - persons of this aimag could partially explain the higher an- cal CCHFV isolate, had a percent identity of 87%. BLAST tibody prevalence in the population. Dornogovi Aimag had analysis of the M segment identified SPU 97/85 (GenBank the next highest antibody prevalence (2.07%). Dornogovi accession no. DQ211633.1) from South Africa as the 2206 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Crimean-Congo Hemorrhagic Fever Virus, Mongolia Figure 2. Phylogenetic characterization of partial small (S) segment sequence of Crimean-Congo hemorrhagic fever virus (CCHFV) isolate from tick pool 159A, Mongolia, 2013–2014. Near full–length CCHFV S segments from GenBank were aligned with the S segment sequence from tick pool 159A and a phylogenetic tree was generated. A) Genetic clusters are displayed as previously described (23). B) Detailed view of phylogenetic tree of Asia 2 lineage. S segment of the CCHFV isolate from this study (tick 159A/Mongolia; bold) clusters in the Asia 2 lineage. Scale bars indicate nucleotide substitutions per site. is in the southeastern part of Mongolia and lies in the east- samples tested from the other 3 aimags, Khovd, Dundgovi, ern part of the Gobi Desert. Raising livestock previously and Dornod. Dornod Aimag is outside the Hyalomma tick was the main economic driver of this aimag, but this activ- range, and thus, we did not expect to find evidence of CCH- ity has been replaced by mineral exploration. Soums Kha- FV infection in this area. tanbulag and Khövsgöl within Dornogovi had some of the Antibody neutralization tests were used to confirm highest CCHFV antibody prevalences of those sampled. the IgG ELISA data because antibody-mediated virus Ömnögovi Aimag had an antibody prevalence of neutralization suggests specificity to the tested virus. We 1.42%, despite having the greatest number of samples test- found that 67% of the ELISA-positive samples tested ed and being the only aimag having a soum with a CCH- neutralized CCHFV IbAr10200. Because we used an iso- FV-positive tick pool sample. Ömnögovi, the largest aimag late from West Africa in our PRNT, our results are prob- in Mongolia, is located in the southern part of the country ably an underestimate of the true percentage of samples and southern part of the Gobi Desert. Ömnögovi is rich in specific for CCHFV. Nucleic acid differences could ac - mineral deposits and depends heavily on mining, and agri- count for the reduced neutralization. On the other hand, culture is of minor importance. The lowest antibody prev- ELISA-positive, PRNT-negative samples could indicate alence, 1.19%, occurred in the Govi-Altai Aimag, where exposure to related viruses, such as Dugbe virus. Two of the Gobi Desert, steppe grasslands, and mountains of west- 7 randomly selected ELISA-negative samples neutralized ern Mongolia converge. No CCHFV IgG was found in CCHFV to some extent, which could have been the result Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2207 RESEARCH Figure 3. Phylogenetic characterization of full-length medium (M) segment sequence of Crimean-Congo hemorrhagic fever virus (CCHFV) isolate from tick pool 159A, Mongolia, 2013–2014. Full-length and near full–length CCHFV M segments from GenBank were aligned with the M segment sequence of tick pool 159A and a phylogenetic tree was generated. A) M segment lineages weakly cluster by geographic location. B) Detailed view of phylogenetic tree cluster including the isolate from this study (tick 159A/Mongolia; bold). Tick 159A/Mongolia groups with CCHFV isolates from Asia, the Middle East, and Africa. Scale bars indicate nucleotide substitutions per site. of nonspecific binding or low IgG concentrations below to better characterize the virus and the disease severity it the level of detection of our ELISA. causes. Taken together, these data suggest CCHFV is likely As further supporting evidence suggesting CCHFV endemic in Mongolia. Further investigations are needed to circulation in Mongolia, we identified 1 tick pool positive better define the risk for human infection and the genomic for CCHFV genetic material, which enabled us to sequence diversity of CCHFV in the region to improve risk mitiga- the S and M segments of the isolate and perform phyloge- tion, diagnostics, and surveillance. netic analyses. The S segment, which generally clusters on the basis of geographic location in phylogenetic analyses, Acknowledgments was similar to other sequenced CCHFV isolates found in The authors thank all the men and women of Mongolia whose countries of Central and East Asia, including Kazakhstan, contributions directly or indirectly made this work possible. We Uzbekistan, Tajikistan, and China, corresponding to the give a special acknowledgment to Shigeru Morikawa for his Asia 2 CCHFV lineage or lineage IV (21,23). The M seg- contributions to the CCHFV nested RT-PCR. Thanks to Cindy ment, which has a higher mutation rate than the S segment, Rossi and Tamara Clements for their expert technical advice and had the greatest homology to isolates from South Africa, ELISA assistance. Thanks to Lucas Bagley for assistance with the Middle East, and Asia, corresponding to the M segment sorting, identifying, and homogenizing ticks. lineage IV (21,23). Laboratory work was funded in part by the Global Emerging Overall, our study demonstrated the presence of CCH- Infections Surveillance Section of the Armed Forces Health FV in Mongolia, suggesting human infections are occurring Surveillance Branch research plans (ProMIS P00123-16-RD in the area. Our findings are not surprising, considering that and P0017-17-RD) through USAMRIID and the World Bank others have shown by serology and tick testing the pres- project Capability Building for Emerging Infectious Disease ence of CCHFV in the region (15,25). Because of Mon- Preparedness and Response, 2012–2013. golia’s size and diversity, additional studies are required 2208 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Crimean-Congo Hemorrhagic Fever Virus, Mongolia 12. Chumikhin S, Potshild E, Nyamadava P, Peshetnikov I, Disclaimer: Opinions, interpretations, conclusions, and Tkachenko E, Dzagurova T. Serological and virological tests used recommendations are those of the authors and are not necessarily for the detection of arboviruses in wild mammals and birds of endorsed by the US Army. Mongolia. Med Parazitol (Mosk). 1987:10–2. 13. Morikawa S. Serological surveillance of Crimean-Congo haemorrhagic fever infection among sheep in Mongolia. Tokyo: About the Author National Institute of Infectious Diseases; 2013. Mr. Voorhees is a biological sciences laboratory technician 14. Guo R, Shen S, Zhang Y, Shi J, Su Z, Liu D, et al. A new strain of Crimean-Congo hemorrhagic fever virus isolated from Xinjiang, working in the Diagnostic Systems Division at USAMRIID, Fort China. Virol Sin. 2017;32:80–8. Detrick, Maryland. His research interests are in the development s12250-016-3936-9 of immunologic assays and evaluation of platforms to detect and 15. Sun S, Dai X, Aishan M, Wang X, Meng W, Feng C, et al. identify viruses requiring Biosafety Level 3 and 4 containment Epidemiology and phylogenetic analysis of Crimean-Congo hemorrhagic fever viruses in Xinjiang, China. J Clin Microbiol. and their transition to austere settings for disease surveillance 2009;47:2536–43. and diagnostics. 16. Yu-Chen Y, Ling-Xiong K, Ling L, Yu-Qin Z, Feng L, Bao-Jian C, et al. Characteristics of Crimean-Congo hemorrhagic fever virus (Xinjiang strain) in China. Am J Trop Med Hyg. 1985;34:1179–82. References 1. Bente DA, Forrester NL, Watts DM, McAuley AJ, Whitehouse CA, 17. Papa A, Weber F, Hewson R, Weidmann M, Koksal I, Bray M. Crimean-Congo hemorrhagic fever: history, epidemiology, Korukluoglu G, et al. Meeting report: First International pathogenesis, clinical syndrome and genetic diversity. Antiviral Conference on Crimean-Congo Hemorrhagic Fever. Antiviral Res. 2013;100:159–89. Res. 2015; 120:57–65. 07.006 05.005 2. Lukashev AN, Klimentov AS, Smirnova SE, Dzagurova TK, 18. Garrison AR, Alakbarova S, Kulesh DA, Shezmukhamedova D, Drexler JF, Gmyl AP. Phylogeography of Crimean Congo Khodjaev S, Endy TP, et al. Development of a TaqMan minor hemorrhagic fever virus. PLoS One. 2016;11:e0166744. groove binding protein assay for the detection and quantification of Crimean-Congo hemorrhagic fever virus. Am J Trop Med Hyg. 3. Messina JP, Pigott DM, Golding N, Duda KA, Brownstein JS, 2007;77:514–20. Weiss DJ, et al. The global distribution of Crimean-Congo 19. Koehler JW, Delp KL, Hall AT, Olschner SP, Kearney BJ, hemorrhagic fever. Trans R Soc Trop Med Hyg. 2015;109:503–13. Garrison AR, et al. Sequence optimized real-time reverse transcription polymerase chain reaction assay for detection of 4. Hewson R, Gmyl A, Gmyl L, Smirnova SE, Karganova G, Crimean-Congo hemorrhagic fever virus. Am J Trop Med Hyg. Jamil B, et al. Evidence of segment reassortment in Crimean- 2018;98:211–5. Congo haemorrhagic fever virus. J Gen Virol. 2004;85:3059–70. 20. Morikawa S. Laboratory diagnosis of Crimean-Congo hemorrhagic fever (CCHF). Tokyo: National Institute of Infectious Diseases; 5. Mostafavi E, Chinikar S, Moradi M, Bayat N, Meshkat M, Fard MK, 2012. p. 1–15. et al. A case report of Crimean Congo hemorrhagic fever in 21. Deyde VM, Khristova ML, Rollin PE, Ksiazek TG, Nichol ST. ostriches in Iran. Open Virol J. 2013;7:81–3. Crimean-Congo hemorrhagic fever virus genomics and global 10.2174/1874357901307010081 diversity. J Virol. 2006;80:8834–42. 6. Lindeborg M, Barboutis C, Ehrenborg C, Fransson T, Jaenson TG, JVI.00752-06 Lindgren PE, et al. Migratory birds, ticks, and Crimean-Congo 22. Koehler JW, Delp KL, Kearney BJ, Conrad TA, Schoepp RJ, hemorrhagic fever virus. Emerg Infect Dis. 2012;18:2095–7. Garrison AR, et al. Draft genome sequences of eight Crimean- Congo hemorrhagic fever virus strains. Genome Announc. 7. Shepherd AJ, Swanepoel R, Leman PA, Shepherd SP. Field and 2017;5:e00240-17. laboratory investigation of Crimean-Congo haemorrhagic fever 23. Anagnostou V, Papa A. Evolution of Crimean-Congo hemorrhagic virus (Nairovirus, family Bunyaviridae) infection in birds. fever virus. Infect Genet Evol. 2009;9:948–54. Trans R Soc Trop Med Hyg. 1987;81:1004–7. 10.1016/j.meegid.2009.06.018 10.1016/0035-9203(87)90379-8 24. Spengler JR, Bergeron É, Rollin PE. Seroepidemiological studies 8. Mild M, Simon M, Albert J, Mirazimi A. Towards an understanding of Crimean-Congo hemorrhagic fever virus in domestic and wild of the migration of Crimean-Congo hemorrhagic fever virus. J Gen animals. PLoS Negl Trop Dis. 2016;10:e0004210. Virol. 2010;91:199–207. 10.1371/journal.pntd.0004210 9. Papa A, Tsergouli K, Tsioka K, Mirazimi A. Crimean-Congo 25. Xia H, Li P, Yang J, Pan L, Zhao J, Wang Z, et al. hemorrhagic fever: tick-host-virus interactions. Front Cell Infect Epidemiological survey of Crimean-Congo hemorrhagic fever Microbiol. 2017;7:213. virus in Yunnan, China, 2008. Int J Infect Dis. 2011;15:e459–63. 10. Ergönül O. Crimean-Congo haemorrhagic fever. Lancet Infect Dis. 2006;6:203–14. 11. Swanepoel R, Gill DE, Shepherd AJ, Leman PA, Mynhardt JH, Address for correspondence: Randal J. Schoepp, USAMRIID, Diagnostic Harvey S. The clinical pathology of Crimean-Congo hemorrhagic Systems Division, 1425 Porter St, Fort Detrick, MD 21702, USA; email: fever. Rev Infect Dis. 1989;11(Suppl 4):S794–800. 10.1093/clinids/11.Supplement_4.S794 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2209 RESEARCH Novel Type of Chronic Wasting Disease Detected in Moose (Alces alces), Norway Laura Pirisinu, Linh Tran, Barbara Chiappini, Ilaria Vanni, Michele A. Di Bari, Gabriele Vaccari, Turid Vikøren, Knut Ivar Madslien, Jørn Våge, Terry Spraker, Gordon Mitchell, Aru Balachandran, Thierry Baron, Cristina Casalone, Christer M. Rolandsen, Knut H. Røed, Umberto Agrimi, Romolo Nonno, Sylvie L. Benestad Chronic wasting disease (CWD) persists in cervid popula- misfolding of the normal host-encoded cellular prion protein C Sc tions of North America and in 2016 was detected for the first (PrP ) into an abnormal disease-associated isoform (PrP ). Sc time in Europe in a wild reindeer in Norway. We report the PrP is considered to be the main or exclusive component detection of CWD in 3 moose (Alces alces) in Norway, iden- of prions, the transmissible agents for TSEs (1). TSEs might tified through a large-scale surveillance program. The cases have a genetic, infectious, or sporadic origin. Classical scra- occurred in 13–14-year-old female moose, and we detected pie and CWD can be highly contagious, spreading directly Sc an abnormal form of prion protein (PrP ) in the brain but not among animals or through environmental contamination. in lymphoid tissues. Immunohistochemistry revealed that Since its first description in Colorado in 1967, CWD the moose shared the same neuropathologic phenotype, Sc has been detected in new geographic areas and with in- characterized by mostly intraneuronal deposition of PrP . creasing prevalence in captive and free-ranging cervids. This pattern differed from that observed in reindeer and Currently, the disease has been diagnosed in 25 states in the has not been previously reported in CWD-infected cervids. Sc Moreover, Western blot revealed a PrP type distinguish- United States and in 2 Canada provinces (2,3), along with able from previous CWD cases and from known ruminant cases in South Korea associated with importation of in- prion diseases in Europe, with the possible exception of fected cervids from Canada (4). High disease prevalence in sheep CH1641. These findings suggest that these cases in some areas represents a challenge for preservation of wild moose represent a novel type of CWD. cervids and mitigation of human exposure to CWD-related Sc prions (PrP ). The high prevalence might be a plausible explanation for local deer population decline (5,6). ransmissible spongiform encephalopathies (TSEs), Species naturally affected by CWD include white-tailed Tor prion diseases, are fatal and transmissible neuro- deer (Odocoileus virginianus), mule deer (O. hemionus), degenerative diseases that include scrapie in sheep and moose (Alces alces), elk or wapiti (Cervus canadensis), and goats, bovine spongiform encephalopathy (BSE) in cattle, red deer (C. elaphus). In 2016, CWD was reported for the chronic wasting disease (CWD) in cervids, and Creutzfeldt- first time in Europe in wild reindeer ( Rangifer tarandus) (7), Jakob disease in humans. TSEs are characterized by the a species never previously found to be naturally infected. Author affiliations: Istituto Superiore di Sanità, Rome, Italy The biochemical analysis and immunohistochemical (IHC) (L. Pirisinu, B. Chiappini, I. Vanni, M.A. Di Bari, G. Vaccari, Sc distribution of PrP from Norway reindeer revealed a pat- U. Agrimi, R. Nonno); Norwegian Veterinary Institute, Oslo, tern indistinguishable from North America isolates (7). Norway (L. Tran, T. Vikøren, K.I. Madslien, J. Våge, S.L. We report 3 cases of CWD detected in moose in Nor- Benestad); Colorado State University, Fort Collins, Colorado, USA way, characterized by biochemical and IHC features clear- (T. Spraker); Canadian Food Inspection Agency, Ottawa, ly different from CWD cases previously described in North Ontario, Canada (G. Mitchell, A. Balachandran); Anses Lyon America and Norway. Our findings suggest the involve- Unité “Maladies Neuro-Dégénératives,” Lyon, France (T. Baron); ment of a different type of CWD prion. Istituto Zooprofilattico Sperimentale del Piemonte Liguria e Valle d’Aosta, Torino, Italy (C. Casalone); Norwegian Institute for Nature Materials and Methods Research, Trondheim, Norway (C.M. Rolandsen); Norwegian University of Life Sciences, Faculty of Veterinary Science, Oslo Animals and Tissues (K.H. Røed) The 3 moose were found in Trøndelag County in cen- tral Norway. The first case, moose no. 1 (ID P138), was DOI: 2210 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Chronic Wasting Disease in Moose, Norway emaciated and demonstrated abnormal behavior, showing AmpliTaq Gold 360 PCR Buffer (Life Technologies, reduced fear of humans. The second case, moose no. 2 Carlsbad, CA, USA), 2.5 mmol/L MgCl , 1X 360 GC (ID P153), was found dead in a river. Necropsy revealed Enhancer (Life Technologies), 200 µmol/L dNTPs, 0.25 normal body condition and pregnancy with twins; trauma µmol/L of forward (5′-GCTGACACCCTCTTTATTTT - was the cause of death. The third case, moose no. 3 (ID GCAG-3′) and reverse (5′-GATTAAGAAGATAAT- CD11399), was observed showing abnormal behaviors, GAAAACAGGAAG -3′) primers ( 11), and 0.5 µL Am- including reduced fear of humans. Necropsy revealed a pliTaq Gold 360 (Life Technologies), according to the poor body condition and a severe dislocation of the left following amplification protocol: 5 min at 96°C; 30 s at hip joint, which might have influenced the animal’s be - 96°C; 15 s at 57°C; 90 s at 72°C for 40 cycles; and 4 min havior. All 3 were older female moose (13, 14, and 13 at 72°C. Amplicons were purified with the Illustra Exo - years old, based on counts of cementum annuli in the root ProStar 1-Step clean-up kit (GE Healthcare Life Sciences, of the first incisor [8 ]). Little Chalfont, UK), sequenced using the Big Dye Ter- Samples included in this study are described in the minator Cycle Sequencing Kit v1.1 (Life Technologies), Table. We performed the primary diagnostic test (TeSeE and purified with the Big Dye XTerminator Purification ELISA; Bio-Rad Laboratories, Inc., Hercules, CA, USA) Kit (Life Technologies), and detected by using an ABI Sc res for detection of protease-resistant core of PrP (PrP ) on PRISM 3130 apparatus (Life Technologies). the 3 moose and 1 reindeer from Norway (7) in the medulla oblongata at the level of the obex. After the initial positive Anti–Prion Protein Monoclonal Antibodies test results, the remaining brain tissues were divided; one Several antibodies with different epitopes (sheep prion half was fixed in 10% neutral buffered formalin, and the protein [PrP] numbering) were used for discriminatory other half was frozen. In addition, lymph nodes (Ln) from Western blot (WB) and IHC. SAF84 (aa 167–173) was moose no. 1 (retropharyngeal, submandibular, and jejunal obtained from Bertin Pharma (Montigny-le-Bretonneux, Ln), moose no. 3 (retropharyngeal, parotid, prescapular, France), L42 (aa 148–153) from R-Biopharm (Darmstadt, and submandibular Ln, and tonsils), and the reindeer (2 tra- Germany), 9A2 (aa 102–104) and 12B2 (aa 93–97) from cheobronchial Ln) were equally divided and formalin fixed Wageningen Bioveterinary Research (Lelystad, Nether- or frozen. lands), and F99/97.6 (aa 220–225) from VMRD, Inc. (Pull- man, WA, USA). Genotyping of Moose PRNP Immunohistochemistry DNA was extracted from 100 mg of brain tissue by us- Brain, Ln, and tonsil tissues were formalin fixed for >48 h ing a DNeasy Blood and Tissue Kit (QIAGEN, Hilden, and processed by standard histopathologic techniques. We Germany), according to the manufacturer’s instructions. Sc used IHC to visualize the distribution of PrP as previous- The PRNP coding sequence was amplified in a 50 µL fi - ly described (7). We applied a commercially available kit nal volume by using 5 µL of extracted DNA, eluate 1X Table. Summary characteristics of samples used in a study characterizing chronic wasting disease in moose (Alces alces), Norway* Animal or species Geographic origin Age Pathologic phenotype PrP genotype† Moose no. 1 Norway 13 y Atypical CWD KK109MM209 Moose no. 2 Norway 14 y Atypical CWD KK MM 109 209 Moose no. 3 Norway 13 y Atypical CWD KK MM 109 209 Reindeer Norway 3–4 y CWD VV GG SS VV 2 129 138 169 Elk or wapiti Canada 6 y CWD MM Elk or wapiti Canada Adult CWD MM Elk or wapiti Canada 6 y CWD MM White-tailed deer Canada 4 y CWD GG Moose Canada Adult CWD KK II ‡ 109 209 Sheep Italy 2.5 y Classical scrapie AA RR QQ 136 154 171 Sheep Italy Adult Classical scrapie AA RR QQ 136 154 171 Sheep Italy 4.5 y Atypical/Nor98 scrapie AA RH QQ 136 154 171 Sheep Experimental sample NA Scrapie CH1641§ AA HH QQ 136 154 171 Sheep France NA Scrapie CH1641-like¶ AA RR QQ 136 154 171 Cattle Italy 6 y C-BSE NA Cattle Italy 15 y L-BSE NA Cattle France 11 y H-BSE NA *All samples were from naturally infected animals, with the exception of 1 sheep experimentally inoculated with CH1641. BSE, bovine spongiform encephalopathy; C-BSE, classical BSE; CWD, chronic wasting disease; H-BSE, H-type atypical BSE; L-BSE, L-type atypical BSE; NA, not available; PrP, prion protein. †The PrP genotype reported is expressing the amino acid variants present at the major polymorphic codons described for each species. ‡The Canada isolate was homozygous for lysine at codon 109 and for isoleucine at codon 209 (KK109II209) (GenBank accession no. MH230114). §Reference (9). ¶Reference (10). Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2211 RESEARCH (EnVisionTM + System HRP [(AEC)] DAKO, Glostrup, respectively located ≈300 and ≈450 km northeast of Nord - Denmark) by using the monoclonal antibodies (mAbs) fjella, where CWD in reindeer was detected in 2016. Nor- 12B2, 9A2, L42, SAF 84, or F99/97.6 for 30 min at 37°C way is populated by several species of wild cervids with and counterstained with hematoxylin. In each run, tissues varying degrees of overlapping range. Seasonal migrations from CWD-negative moose and reindeer were added as are common and distances might exceed 150 km (13–15). negative controls. However, studies tracking global positioning satellite– collared moose have not documented regular season- res PrP Detection al migrations between Selbu and Lierne municipali- We initially tested the obex samples by using the TeSeE ties, suggesting that these can be considered different SAP ELISA and confirmed positive ELISA results by us- moose subpopulations. Sc ing the TeSeE WB. Both tests were performed as recom- We initially detected PrP in brain samples by using mended by the manufacturer (Bio-Rad). a rapid test and then confirmed by WB (data not shown) and IHC. Sequencing analysis of the entire PrP coding se- res Typing of PrP quence revealed that the 3 moose had the wild type PrP res We performed further characterization of PrP by discrim- genotype, homozygous for lysine at codon 109 and for me- inatory immunoblotting, according to the ISS discrimina- thionine at codon 209 (KK MM ) (GenBank accession 109 209 tory WB method (12). Brain homogenates at 10% (wt/vol) no. MH230115). in 100 mmol/L Tris-HCl (pH 7.4) 2% sarkosyl were incu- Sc bated for 1 h at 37°C with Proteinase K (Sigma-Aldrich, Discriminatory PrP Immunohistochemistry Show St. Louis, Missouri, USA) to a final concentration of 200 Differences between Reindeer and Moose Sc µg/mL. Protease treatment was stopped with 3 mmol/L The distribution of PrP staining was examined by IHC and PMSF (Sigma-Aldrich). Aliquots of samples were added compared in the tissues of the 3 moose and the reindeer by with an equal volume of isopropanol/butanol (1:1 vol/vol) using 5 different antibodies (Figure 1). No staining was ob - and centrifuged at 20,000 × g for 10 min. The pellets were served in CWD-negative reindeer and moose independently Sc resuspended in denaturing sample buffer (NuPAGE LDS of the antibody used. The distribution of PrP in the reindeer Sample Buffer; Life Technologies) and heated for 10 min was identical for each of the 5 antibodies and did not differ Sc at 90°C. from the description of PrP distribution in North America We loaded each sample onto 12% bis-Tris polyacryl - cervids (16–18). The labeling was most consistent within amide gels (Invitrogen) for electrophoresis with subse- the gray matter of the medulla oblongata, particularly in the quent WB on polyvinylidene fluoride membranes using the dorsal motor of the vagus nerve (7). The thalamic and brain Trans-Blot Turbo Transfer System (Bio-Rad) according to stem regions of the brain were most affected, with a minimal Sc the manufacturer’s instructions. The blots were processed amount of PrP identified dorsal to the corpus callosum. Sc with anti-PrP mAbs by using the SNAP i.d. 2.0 system PrP labeling in the moose brains (Figure 1, panels (Millipore, Burlington, MA, USA) according to the man- F–J) was clearly different from that of the reindeer (Figure ufacturer’s instructions. After incubation with horserad- 1, panels A–E). In the moose, after staining with F99/97.6 Sc ish peroxidase–conjugated anti–mouse immunoglobulin and L42, PrP was almost exclusively observed as intra- (Pierce Biotechnology, Waltham, MA, USA) at 1:20,000, neuronal aggregates, although intraastrocytic type (multi- the PrP bands were detected by using enhanced chemilu- ple small granules scattered in the cytoplasm of astrocyte- minescent substrate (SuperSignal Femto; Pierce Biotech- resembling cells) and intramicroglial type (1 single or a nology) and ChemiDoc imaging system (Bio-Rad). The few large granules in close proximity to microglia-like chemiluminescence signal was quantified by using Image nuclei) were also observed in the cerebral cortices and ol- Lab 5.2.1 (Bio-Rad). factory bulb (online Technical Appendix Figure 1, panel We performed deglycosylation by adding 18 μL of 0.2 A, Sc mmol/L sodium phosphate buffer (pH 7.4) containing 0.8% Techapp1.pdf). The degree of PrP staining was more Nonidet P40 (Roche) and 2 μL (80 U/ml) di N-Glycosidase intensive and appeared more widespread in the neuropil F (Roche) to 5 μL of proteinase K–digested and denatur- using SAF84. ated samples. We then incubated the mixtures for 3 h at At the level of the obex, we found stained neurons 37°C with gentle shaking. in all nuclei, whereas the dorsal motor of the vagus nerve was not remarkably stained, as observed in reindeer. The intensity of labeling varied among the 3 moose; no. 2 dis- Results played sparse labeling, no. 3 widespread and abundant CWD was diagnosed in 2 moose in May 2016 in labeling, and no. 1 intermediate labeling intensity. We ob- Norway’s Selbu municipality and in 1 moose in Octo- Sc served PrP in all parts of the brain investigated except the ber 2017 in Lierne municipality. Selbu and Lierne are 2212 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Chronic Wasting Disease in Moose, Norway Figure 1. Immunohistochemical detection of disease-associated prion protein in brain sections at the level of the obex in cervids with chronic wasting disease, Norway. A–E) Reindeer; F–J) moose. mAbs used were 12B2 (A, F), 9A2 (B, G), L42 (C, H), SAF 84 (D, I), and F99/97.6 (E, J). Staining obtained in the reindeer tissues is similar regardless of mAbs used (A–E). Conversely, for moose tissues, the staining was primarily observed intraneuronally with L42, SAF84, and F99/97.6 (H–J) but was not observed using the more N-terminal mAbs 12B2 and 9A2 (F, G). Scale bars indicate 40 µm. mAbs, monoclonal antibodies. cerebella of moose nos. 1 and 2. A diffuse or discrete deglycosylation (Figure 3). Moose samples showed a main punctate staining was observed in the granular layer of C-terminal fragment of ≈17 kDa, detected with SAF84, the cerebellum of moose no. 3, with stronger staining in L42, and 9A2, and an additional glycosylated C-terminal some Golgi neurons (online Technical Appendix Figure fragment of ≈13 kDa (CTF13) detected only with SAF84. 1, panel B). In all 3 moose, the cortical regions showed The N terminal 12B2 epitope was mainly lost, although a res laminar staining of neurons in all the cell layers, especially small amount of PrP was still detectable in moose no. 1 in fusiform-shaped neurons. The neurons of the olfactory (Figure 3) and no. 3 (online Technical Appendix Figure 2) tubercle from all 3 also stained strongly, and some glia- with this antibody. associated staining could be observed. In contrast to the reindeer, the downstream flexible tail mAbs 12B2 and 9A2 did not stain in the moose (Figure Sc 1, panels F and G), suggesting that the moose PrP was truncated by endogenous proteases further upstream in the Sc N terminus than was reindeer PrP . Contrary to previous Sc findings in reindeer, PrP was not detected in the Ln from moose no. 1 or in the Ln and tonsils from moose no. 3 (lym- phoid tissues were not available in moose no. 2) by either IHC or ELISA. Sc PrP from Norway Moose Compared with Other CWD Isolates from Canada and Norway Sc We compared the PrP features in moose from Norway res Figure 2. Western blot analysis of PrP in brains of chronic with those of other CWD isolates from Norway and Can- wasting disease–affected cervids from Norway and Canada. res ada by discriminatory WB, which enabled comparison A) Western blot analysis PrP in brains of moose and reindeer res of PrP by epitope mapping with different antibodies. from Norway. Membrane was probed with L42 monoclonal res antibodies. Molecular weights (kDa) are indicated on the right. Norway moose PrP had a lower apparent molecular res Tissue equivalent loaded per lane was 1 mg. B) Western blot weight (MW) than PrP from Norway reindeer (Fig- res analysis of PrP from moose isolates from Norway (lanes 6–7) ure 2, panel A) or from Canada isolates (Figure 2, panel res compared with PrP from chronic wasting disease–affected elk B). This lower MW was explained by the occurrence of or wapiti (lanes 1–3), white-tailed deer (lane 4), and moose (lane Sc more C-terminal cleavage of PrP by protease K, as con- 5) from Canada. Membranes were probed with L42 (upper) and 12B2 (lower) monoclonal antibodies. A scrapie sheep sample firmed by the partial loss of the 12B2 epitope (Figure 2, from Italy was added as control (lane 8). Molecular weights (kDa) panel B). are indicated on the right of each blot. Tissue equivalents loaded Given the unusual pattern observed in moose isolates per lane were 1 mg for Canadian isolates, 2 mg for Norwegian res from Norway, we further investigated their biochemical isolates, and 0.15 mg for scrapie sheep control. PrP , protease- characteristics with additional mAbs and by enzymatic resistant core of abnormal form of prion protein. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2213 RESEARCH In moose nos. 2 and 3, an additional glycosylated C-ter- minal fragment of ≈16 kDa (CTF16) was detected by SAF84 and L42 mAbs (Figure 3; online Technical Appendix Figure 2). We cannot exclude that a small amount of CTF16 was res also present in moose no. 1, given that a weak PrP frag- ment of ≈16 kDa was detectable upon deglycosylation and long exposure of blots (Figure 3; online Technical Appendix Figure 3). Moose nos. 1 and 3 also had a nonglycosylated in- ternal fragment of ≈10 kDa, cleaved at both N and C termini Sc of PrP , which was recognized by using mAbs 9A2 (Figure res 3). Moreover, the analysis of PrP from different neuroana - tomic regions showed that the slight differences observed among the 3 moose were not dependent on the area analyzed (online Technical Appendix Figure 2). Sc Comparison of the PrP Features of the Norway Moose with Sheep and Cattle Prion Strains from Europe Comparison with ovine and bovine prions was performed to res determine the N terminal cleavage of the main PrP frag- res ment by analyzing the different PrP fragments in each sample, the MW of these fragments, and the L42/12B2 an- tibody ratio (Figure 4; online Technical Appendix Table). Among ovine prions, classical scrapie and atypical/Nor98 were easily discriminated from moose isolates (Figure 5). res res Classical scrapie PrP had a higher MW than moose PrP , Figure 4. Bar graph of antibody-signal ratios (y-axis) showing discrimination of the ovine, bovine, moose, and reindeer samples res Figure 3. Characterization of PrP fragments from moose (x-axis) analyzed in a study characterizing chronic wasting disease (Alces alces) in Europe by epitope mapping. Mapping with mAbs in moose (Alces alces), Norway. Numbers indicate sample type: 1, res spanning the whole prion protein enabled the analysis of PrP scrapie; 2, CH1641; 3, CH1641-like; 4, classical bovine spongiform in moose samples before (PNGase F–) and after (PNGase F+) encephalopathy (BSE); 5, H-type atypical BSE; 6, L-type atypical deglycosylation, based on presence or absence of the epitopes and BSE; 7, moose no. 1; 8, moose no. 2; 9, moose no. 3; 10, reindeer. apparent molecular weight. Lanes 1, moose no. 1; lanes 2, moose The antibody ratio is the L42/12B2 ratio of the chemiluminescence no. 3; lane M, protein standards; lane 3, sheep scrapie sample. signal relative to the L42/12B2 ratio of the control scrapie loaded Solid arrowheads indicate C-terminal fragment of ≈13 kDa fragment in each blot. Bars represent median values of >3 independent (present in both samples and detected with SAF84 mAbs). Open determinations; error bars represent the range of observed values. arrowheads indicate C-terminal fragment of ≈16 kDa fragment in Bars start at y = 2, which is the cutoff value of the antibody ratio for moose no. 2 with SAF84 and L42 mAbs. Asterisk indicates the the discrimination of low molecular weight samples (i.e., suspected internal fragment detected in moose no. 1 with 9A2 mAbs. Molecular bovine spongiform encephalopathy cases) from scrapie, according weights are indicated on the left. In the blots on the right, protein to discriminatory Western blot. Numbers indicate sample type: 1, standards are shown in lane M (10, 15, 20, 25, and 37 kDa). The scrapie; 2, CH1641; 3, CH1641-like; 4, classical bovine spongiform mAbs used are indicated on the right. mAbs, monoclonal antibodies; encephalopathy (BSE); 5, H-type atypical BSE; 6, L-type atypical res PrP , protease-resistant core of abnormal form of prion protein. BSE; 7, moose no. 1; 8, moose no. 2; 9, moose no. 3; 10, reindeer. 2214 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Chronic Wasting Disease in Moose, Norway as confirmed by the preservation of 12B2 epitope. As previ - res ously observed (19), Nor98 PrP was cleaved at both the N and C termini, and the characteristic 11–12 kDa band was detected by L42, 9A2, and 12B2 mAbs (Figure 5). In contrast, CH1641 samples showed molecular features partially over- lapping with the moose (Figure 5). CH1641 samples showed res a PrP of ≈17 kDa and were accompanied by an additional C-terminal fragment of 13–14 kDa detected by using SAF84 res mAbs (20). However, CTF16 and the internal PrP fragment of 10 kDa could not be detected in CH1641 samples. Sc Moose PrP did not overlap with any type of bovine Sc res PrP . The lack of the 12B2 epitope in moose PrP was similar to C-type and atypical L-type BSE, but the 2 bovine prions had neither CTF13, CTF16, nor the internal frag- ment (Figure 6). H-type atypical BSE showed the CTF13 res and the internal fragment similar to moose PrP , but the res main PrP fragment showed a higher MW and preserved the 12B2 epitope (Figure 6). The ratio of reactivity obtained with L42 and 12B2 an- tibodies reflected the N terminal cleavage of the main frag- Sc ment of PrP , enabling confirmation that the differences res observed in MW of PrP actually depend on different N terminal proteinase K cleavages, irrespective of the host species (Figure 4). Values >2 are indicative of BSE-like cleavage, whereas values <1 indicate a better preservation of 12B2 epitope compared with scrapie. In this respect, the res behavior of moose PrP was BSE-like (ratio >2). Howev- er, moose no. 1 had a ratio lower than moose nos. 2 and 3. The CH1641-like field sample was similar to moose no. 1 in this respect, whereas CH1641 was similar to moose nos. res 2 and 3. Finally, the value <1 observed for PrP in H-type atypical BSE, CWD in reindeer, and CWD isolates from Canada reflected their higher MW compared with classical scrapie (online Technical Appendix Table). Discussion Although CWD has been detected in several captive and res Figure 5. Comparison of protease-resistant PrP from moose free-ranging cervid species from a large geographic area (Alces alces) with chronic wasting disease and from sheep with scrapie, Europe. Representative blots show epitope mapping in North America, <10 cases in moose have been reported res analysis of PrP (lane 4, CH1641; lane 5, moose no. 1; lane 6, (21–23). We report 3 naturally occurring cases of prion dis- moose no. 2) in comparison with different ovine transmissible ease in moose in Norway that showed molecular and IHC spongiform encephalopathy isolates (lane 1, atypical/Nor98; lane 2, phenotypes differing from those previously described for classical scrapie; and lane 3, CH1641). A chronic wasting disease classical CWD in North America, as well as in reindeer in isolate from Canada was loaded as control (lane 7). The antibodies used are indicated on the left. Protein standards are shown in lane Norway. The phenomenon of strain variation is well known res M (10, 15, 20, 25, 37, and 50 kDa). The small amount of PrP with in prion diseases and is often associated with phenotype intact 12B2 epitope in moose no.1 had a molecular weight higher variation in natural hosts, as observed in bovines with clas- than that observed with more C-terminal monoclonal antibodies sical, H-type, or L-type BSE, and in sheep with classical or (18.7 +0.3 kDa measured with 12B2 vs. 17.2 +0.1 kDa measured atypical/Nor98 scrapie. Identification of a new CWD phe- with L42). Even if the increase of the apparent molecular weight might be a known behavior when proteinase K cleavage occurs notype in 3 moose in Norway can be suggestive of a new near the epitope, we noted that, in the case of moose no. 1, the CWD strain. Although the existence of CWD strain varia- res 12B2-positive PrP had a molecular weight higher than scrapie tion in North America has been inferred from transmission (18.1 +0.1 kDa measured with 12B2) and CH1641-like sample studies (24–26), this phenomenon has not been directly as- res (18.1 +0.4 kDa when detected with 12B2). PrP , protease-resistant sociated with phenotypic variations in natural hosts so far. core of abnormal form of prion protein. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2215 RESEARCH Sc The phenotype variant found in moose from Norway features different from Norway moose PrP and were in- could be hypothetically attributed to host species fac- distinguishable from other cervids with classical CWD. Sc Sc tors. To address this issue, we directly compared PrP This finding suggests that the variant PrP type observed characteristics in the Norway moose with those in a in Norway moose could not simply reflect a host species Canada moose with CWD. In agreement with the avail- factor. Notably, in both natural and experimental condi- Sc able evidence, we found that the Canada moose PrP had tions, CWD-affected moose in North America have been reported to display disease features indistinguishable Sc from CWD in other cervids and had detectable PrP in lymphoid tissues (21,27). Species-specific amino acid polymorphisms in the cer- vid PrP are associated with CWD susceptibility, incubation time, and pathology (28–30). In transmission experiments, atypical features were reported in elk or wapiti and mule deer with genotypes associated with a relative resistance to disease, extension of the incubation period, or both (31,32). Moose PrP is polymorphic at codon 109 (K/Q) and 209 (M/I), combined in 3 alleles: K M (observed in Europe 109 209 and North America), Q M (observed in Europe), and 109 209 K I (observed in North America) (33,34). The 3 moose 109 209 with CWD from Norway had the KK MM genotype, 109 209 whereas the moose case from Canada used for compari- son had the KK II genotype. Thus, we cannot exclude 109 209 that the differences observed between Norway and Canada moose in our study are dependent on differences in PrP genotype. However, a classical CWD phenotype has been reported in naturally (21) and experimentally infected (27) moose with the KK MM genotype, suggesting that a 109 209 difference at PrP codon 209 is probably not the cause of the variant phenotype observed in moose in Norway. All of these findings suggest that neither the species nor the indi- vidual PrP genotypes are likely to have caused the variant phenotypes observed and imply that this variant phenotype could represent a novel CWD strain. CWD is known to be a highly contagious disease in North America; however, data relating to the disease in moose are sparse and insufficient to understand the epi- demiology and the implications of CWD in this species. The apparent low CWD prevalence reported for moose in North America compared with other cervid species might be attributable to the individual social behavior of moose and the minimal habitat overlap between moose and other cervids in areas with CWD. Additionally, surveillance pro- gram design, disease variability, and host genetics might influence the prevalence of the disease. Based on the epi- Figure 6. Comparison of protease-resistant core of abnormal zootic dynamics in North America, CWD plausibly could form of prion protein from moose (Alces alces) in Europe with have become established in reindeer in Norway more than chronic wasting disease and from cattle with BSE. Representative a decade ago (35). In this scenario, the disease in moose blots show epitope mapping analysis of protease-resistant core could possibly be linked to the disease observed in reindeer, of abnormal form of prion protein in moose (lane 5, moose no. 1; lane 6, moose no. 2) in comparison with different BSE isolates with strain mutation or phenotype shift putatively caused (lane 2, classical BSE; lane 3, H-type BSE; and lane 4, L-type by interspecific transmission. However, a main cause of BSE). A sheep scrapie isolate was loaded as control (lane 1). strain mutation after interspecies transmission (i.e., PrP The antibodies are indicated on the left. Protein standards are amino acid differences between the donor and host species) shown in lane M (10, 15, 20, 25, 37, and 50 kDa). BSE, bovine is not relevant in this case because reindeer and moose spongiform encephalopathy. 2216 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Chronic Wasting Disease in Moose, Norway share the same PrP primary sequence. An alternative hy- biologic association between the atypical CWD in moose pothesis could be that moose have a prion disease that is and small ruminant CH1641. Transmission studies in sev- independent of the reindeer epidemic, being either specific eral rodent models are under way and will help to clarify to the Norwegian moose or acquired by species other than whether the different phenotype observed (designated Nor- the reindeer. 16CWD) could reflect the presence of a new cervid prion The 3 moose were 13, 14, and 13 years of age. Al- strain in moose from Norway. though moose can reach ages beyond 20 years, we con- sider these moose as old because female moose >10–12 Acknowledgments years of age start to show signs of senescence and de- The authors thank the Department of Pathology at the clining survival and reproduction rates (36,37). The old Norwegian Veterinary Institute for their excellent work with age of the moose, the absence of lymphoid tissue involve- TSE testing and sample collection, Liv Midthjell for her ment, and the low disease prevalence observed so far (3 of skillful PRNP genotyping, Magne Haugum for the necropsy 10,531 moose tested) could suggest that CWD in moose of 2 of the moose, and Jacques Grassi for kindly providing the is less contagious than classical CWD or could represent Norwegian Veterinary Institute with SAF84 mAbs. a spontaneous TSE. The finding that the affected moose This work was supported in part by a grant from the Norwegian were from the same geographic area does not seem to Ministry of Agriculture. support a spontaneous origin of the disease; however, the actual evidence for geographic clustering could have been biased by oversampling in Trøndelag County, where the About the Author first positive moose was detected. Lack of detailed data on Dr. Pirisinu is a researcher at the Istituto Superiore di Sanità the ages of the moose tested so far in different geographic in Rome, Italy. Her primary research interests include the areas prevents any definitive conclusion. Still, the recent prion strain characterization and zoonotic potential of animal detection of a positive moose in Finland, several hundred prion diseases. kilometers from Trøndelag County, might indicate that the disease is not restricted to Norway (38). The ongoing References intensive surveillance in Norway and several European 1. Prusiner SB. Molecular biology of prion diseases. Science. Union countries with large moose populations will help 1991;252:1515–22. to better clarify the actual geographic distribution and 2. Benestad SL, Telling GC. Chronic wasting disease: an evolving prevalence and will be critical for understanding the con- prion disease of cervids. Handb Clin Neurol. 2018;153:135–51. tagious or spontaneous nature of the disease. 3. EFSA Panel on Biological Hazards. Scientific opinion on chronic The 3 moose analyzed shared a distinctive IHC pat- wasting disease (CWD) in cervids. EFSA J. 2017;15:4667. tern, mainly characterized by intraneuronal accumulation 4. Kim TY, Shon HJ, Joo YS, Mun UK, Kang KS, Lee YS. Sc Sc of PrP , and common PrP features, such as the proteinase Additional cases of Chronic Wasting Disease in imported deer in Korea. J Vet Med Sci. 2005;67:753–9. K N-terminal cleavage and the presence of an additional jvms.67.753 CTF13 fragment. However, we also observed unexpected 5. Miller MW, Swanson HM, Wolfe LL, Quartarone FG, Huwer SL, differences among the 3 moose. By WB, the CTF16 frag- Southwick CH, et al. Lions and prions and deer demise. PLoS One. ment was observed in moose nos. 2 and 3 but not in moose 2008;3:e4019. 6. Edmunds DR, Kauffman MJ, Schumaker BA, Lindzey FG, no. 1, whereas the nonglycosylated internal fragment of Cook WE, Kreeger TJ, et al. Chronic wasting disease drives 10 kDa was evident in moose nos. 1 and 3 but could not population decline of white-tailed deer. PLoS One. 2016; be detected in moose no. 2. Furthermore, we also showed 11:e0161127. that these differences did not depend on the brain area in - 7. Benestad SL, Mitchell G, Simmons M, Ytrehus B, Vikøren T. First case of chronic wasting disease in Europe in a Norwegian vestigated. We cannot rule out that these slight differences free-ranging reindeer. Vet Res (Faisalabad). 2016;47:88. might depend on technical issues rather than represent Sc actual PrP variations. The outcome of the ongoing bio- 8. Rolandsen CM, Solberg EJ, Heim M, Holmstrøm F, Solem MI, assay experiments will help to clarify the meaning of the Sæther B-E. Accuracy and repeatability of moose (Alces alces) age as estimated from dental cement layers. Eur J Wildl Res. observed variations. 2008;54:6–14. Sc By comparing the moose PrP features with other ani- 9. Pirisinu L, Migliore S, Di Bari MA, Esposito E, Baron T, mal TSEs circulating in Europe, we found no evidence of D’Agostino C, et al. Molecular discrimination of sheep bovine similarities with bovine and ovine prions. Minimal simi- spongiform encephalopathy from scrapie. Emerg Infect Dis. 2011;17:695–8. larities were observed with CH1641 samples; however, 10. Nicot S, Baron TG. Strain-specific proteolytic processing CH1641 cases have not yet been detected in Norway. Bio- of the prion protein in prion diseases of ruminants transmitted assay in a large spectrum of rodent models will assist in in ovine transgenic mice. J Gen Virol. 2010;91:570–4. determining whether these molecular similarities imply Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2217 RESEARCH 11. Kaluz S, Kaluzova M, Flint APF. Sequencing analysis of prion wasting disease to hamsters and transgenic mice: evidence for genes from red deer and camel. Gene. 1997;199:283–6. strains. J Virol. 2007;81:4305–14. JVI.02474-06 12. Community Reference Laboratory of the European Union. TSE 25. Angers RC, Kang HE, Napier D, Browning S, Seward T, strain characterization in small ruminants—a technical handbook Mathiason C, et al. Prion strain mutation determined by prion for national reference laboratories in the EU. Version 8. December protein conformational compatibility and primary structure. 2016 [cited 2016 Dec 12]. Science. 2010;328:1154–8. documents/tse-oie-rl-handbook.pdf science.1187107 13. Rolandsen CM, Solberg EJ, Saether B-E, Moorter BV, Herfindal I, 26. Perrott MR, Sigurdson CJ, Mason GL, Hoover EA. Evidence for Bjørneraas K. On fitness and partial migration in a large distinct chronic wasting disease (CWD) strains in experimental herbivore—migratory moose have higher reproductive performance CWD in ferrets. J Gen Virol. 2012;93:212–21. than residents. Oikos. 2017;126:547–55. 10.1099/vir.0.035006-0 oik.02996 27. Kreeger TJ, Montgomery DL, Jewell JE, Schultz W, Williams ES. 14. Bunnefeld N, Börger L, van Moorter B, Rolandsen CM, Dettki H, Oral transmission of chronic wasting disease in captive Solberg EJ, et al. A model-driven approach to quantify migration Shira’s moose. J Wildl Dis. 2006;42:640–5. patterns: individual, regional and yearly differences. J Anim Ecol. 10.7589/0090-3558-42.3.640 2011; 80:466–76. 28. Jewell JE, Conner MM, Wolfe LL, Miller MW, Williams ES. 01776.x Low frequency of PrP genotype 225SF among free-ranging 15. Meisingset EL, Loe LE, Brekkum Ø, Bischof R, Rivrud IM, mule deer (Odocoileus hemionus) with chronic wasting dis- Lande US, et al. Spatial mismatch between management units and ease. J Gen Virol. 2005;86:2127–34. movement ecology of a partially migratory ungulate. J Appl Ecol. vir.0.81077-0 2018;55:745–53. 29. Johnson CJ, Herbst A, Duque-Velasquez C, Vanderloo JP, 16. Spraker TR, Miller MW, Williams ES, Getzy DM, Adrian WJ, Bochsler P, Chappell R, et al. Prion protein polymorphisms affect Schoonveld GG, et al. Spongiform encephalopathy in free-ranging chronic wasting disease progression. PLoS One. 2011;6:e17450. mule deer (Odocoileus hemionus), white-tailed deer (Odocoileus virginianus) and Rocky Mountain elk (Cervus elaphus nelsoni) 30. Robinson SJ, Samuel MD, O’Rourke KI, Johnson CJ. The role of in northcentral Colorado. J Wildl Dis. 1997;33:1–6. genetics in chronic wasting disease of North American cervids. Prion. 2012;6:153–62. 17. Williams ES, Young S. Neuropathology of chronic wasting disease 31. O’Rourke KI, Spraker TR, Zhuang D, Greenlee JJ, Gidlewski TE, of mule deer (Odocoileus hemionus) and elk (Cervus elaphus Hamir AN. Elk with a long incubation prion disease phenotype nelsoni). Vet Pathol. 1993;30:36–45. have a unique PrPd profile. Neuroreport. 2007;18:1935–8. 030098589303000105 18. Spraker TR, Zink RR, Cummings BA, Wild MA, Miller MW, 32. Wolfe LL, Fox KA, Miller MW. “Atypical” chronic wasting O’Rourke KI. Comparison of histological lesions and disease in PRNP genotype 225FF mule deer. J Wildl Dis. immunohistochemical staining of proteinase-resistant prion protein 2014;50:660–5. in a naturally occurring spongiform encephalopathy of free-ranging 33. Huson HJ, Happ GM. Polymorphisms of the prion protein gene mule deer (Odocoileus hemionus) with those of chronic wasting (PRNP ) in Alaskan moose (Alces alces gigas). Anim Genet. 2006; disease of captive mule deer. Vet Pathol. 2002;39:110–9. 37:425–6. 34. Wik L, Mikko S, Klingeborn M, Stéen M, Simonsson M, Linné T. 19. Pirisinu L, Nonno R, Esposito E, Benestad SL, Gambetti P, Polymorphisms and variants in the prion protein sequence of Agrimi U, et al. Small ruminant nor98 prions share biochemical European moose (Alces alces), reindeer (Rangifer tarandus), roe features with human gerstmann-sträussler-scheinker disease and deer (Capreolus capreolus) and fallow deer (Dama dama) variably protease-sensitive prionopathy. PLoS One. 2013;8:e66405. in Scandinavia. Prion. 2012;6:256–60. pri.19641 20. Baron T, Bencsik A, Vulin J, Biacabe AG, Morignat E, Verchere J, 35. EFSA Panel on Biological Hazards. Scientific opinion on chronic et al. A C-terminal protease-resistant prion fragment distinguishes wasting disease (II). EFSA J. 2018;16:5132. ovine “CH1641-like” scrapie from bovine classical and L-Type 36. Ericsson G, Wallin K. Age-specific moose ( Alces alces) mortality BSE in ovine transgenic mice. PLoS Pathog. 2008;4:e1000137. in a predator-free environment: evidence for senescence in females. Ecoscience. 2001;8:157–63. 21. Baeten LA, Powers BE, Jewell JE, Spraker TR, Miller MW. 01.11682641 A natural case of chronic wasting disease in a free-ranging 37. Ericsson G, Wallin K, Ball JP, Broberg M. Age-related reproductive moose (Alces alces shirasi). J Wildl Dis. 2007;43:309–14. effort and senescence in free-ranging moose, Alces alces. Ecology. 2001;82:1613. 22. Moreno JA, Telling GC. Molecular mechanisms of chronic wasting 38. Finnish Food Safety Authority (Evira). Moose found dead disease prion propagation. Cold Spring Harb Perspect Med. in forest with chronic wasting disease [cited 2018 Mar 8]. 2018;8:a024448. 23. Haley NJ, Hoover EA. Chronic wasting disease of cervids: current dead-in-forest-with-chronic-wasting-disease knowledge and future perspectives. Annu Rev Anim Biosci. 2015;3:305–25. Address for correspondence: Sylvie L. Benestad, Norwegian Veterinary 022114-111001 Institute, Pathology, PO Box 750 Sentrum, 0106 Oslo, Norway; email: 24. Raymond GJ, Raymond LD, Meade-White KD, Hughson AG, Favara C, Gardner D, et al. Transmission and adaptation of chronic 2218 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Genomic Characterization of β-Glucuronidase–Positive Escherichia coli O157:H7 Producing Stx2a Yoshitoshi Ogura, Kazuko Seto, Yo Morimoto, Keiji Nakamura, Mitsuhiko P. Sato, Yasuhiro Gotoh, Takehiko Itoh, Atsushi Toyoda, Makoto Ohnishi, Tetsuya Hayashi Among Shiga toxin (Stx)–producing Escherichia coli circulating STEC O157:H7 strains, frequently causes large (STEC) O157:H7 strains, those producing Stx2a cause outbreaks of severe enteric infections, including diarrhea, more severe diseases. Atypical STEC O157:H7 strains hemorrhagic colitis, and hemolytic uremic syndrome. Stxs showing a β-glucuronidase–positive phenotype (GP STEC are classified as Stx1 or Stx2; Stx1 currently includes 3 sub - O157:H7) have rarely been isolated from humans, mostly types (stx1a, stx1c, and stx1d) and Stx2 includes 7 subtypes from persons with asymptomatic or mild infections; Stx2a- (stx2a to stx2g) (2). The stx genes are encoded by lambda- producing strains have not been reported. We isolated, from like phages and have been acquired by STEC strains through a patient with bloody diarrhea, a GP STEC O157:H7 strain phage transduction (3). Typical STEC O157:H7 produces (PV15-279) that produces Stx2a in addition to Stx1a and Stx1a, Stx2a, and Stx2c, either alone or in combination. Stx2c. Genomic comparison with other STEC O157 strains Stx2a-producing strains cause more severe infections than revealed that PV15-279 recently emerged from the stx1a/ do Stx1a-producing strains (4). In addition, the levels of stx2c-positive GP STEC O157:H7 clone circulating in Ja- pan. Major virulence genes are shared between typical Stx2 production among the STEC O157:H7 strains carrying (β-glucuronidase–negative) and GP STEC O157:H7 strains, stx2c, but not stx2a, are typically very low (5–7). and the Stx2-producing ability of PV15-279 is comparable According to the stepwise evolution model (8,9), to that of typical STEC O157:H7 strains; therefore, PV15- STEC O157:H7 evolved from ancestral enteropathogen- 279 presents a virulence potential similar to that of typical ic E. coli (EPEC) O55:H7 (sorbitol-fermenting [SF] and STEC O157:H7. This study reveals the importance of GP β-glucuronidase–positive [GP]; clonal complex [CC] A1) O157:H7 as a source of highly pathogenic STEC clones. through sequential acquisitions of virulence factors and phe- notypic traits along with a serotype change (Figure 1). Two higa toxin (Stx)–producing Escherichia coli (STEC) phenotypic variants of STEC O157 are known: SF/GP STEC Swith the serotype O157:H7 is characterized by the pos- O157:H– (nonmotile) (CC A4), known as the German clone session of stx gene(s), the locus of enterocyte effacement (SF STEC O157:H–), and NSF/GP STEC O157:H7 (CC A5) (LEE) –encoded type 3 secretion system (T3SS), and a large (GP STEC O157:H7). Both variants are postulated to have virulence plasmid (pO157) that encodes enterohemolysin emerged from a hypothetical intermediate, CC A3; typical and other virulence factors, such as catalase-peroxidase STEC O157:H7 (CC A6) has further emerged from CC A5. KatP, a type II secretion system, and the protease EspP SF STEC O157:H– strains are usually stx2a positive (1). Non–sorbitol-fermenting (NSF) and β-glucuronidase– (10); like typical STEC O157:H7, they have caused many negative (GN) STEC O157:H7 (hereafter referred to as outbreaks and sporadic cases of hemolytic uremic syn- typical STEC O157:H7), the major clone among currently drome in Germany and other countries in Europe. There- fore, this clone is generally thought to be highly pathogenic Author affiliations: Kyushu University, Fukuoka, Japan (Y. Ogura, (11–13). In contrast, although GP STEC O157:H7 strains K. Nakamura, M.P. Sato, Y. Gotoh, T. Hayashi); Osaka Institute have been reported to carry both stx1 and stx2 or only stx2 of Public Health, Osaka, Japan (K. Seto); Hokkaido Institute of (14–16), strains producing the Stx2a subtype have not been Public Health, Hokkaido, Japan (Y. Morimoto); Tokyo Institute of described. Although GP STEC O157:H7 was first isolat - Technology, Tokyo, Japan (T. Itoh); National Institute of Genetics, ed from a patient with bloody diarrhea in 1994 (14), this Shizuoka, Japan (A. Toyoda); National Institute of Infectious variant has rarely been isolated from humans. Moreover, Diseases, Tokyo (M. Ohnishi) human isolates obtained to date have generally been isolat- DOI: ed from patients with asymptomatic or mild infections (16). Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2219 RESEARCH Figure 1. Schematic illustrating model of the stepwise evolution of STEC O157. The proposed stepwise evolution model of STEC O157 was schematically illustrated according to previous reports (8,9). Clonal complexes (CCs) A1 to A6 are indicated, along with phenotypic changes, antigen shifts, and acquisitions of Stx phages and pO157. Squares indicate contemporary circulating STEC O157 clones. EPEC, enteropathogenic E. coli; GP, β-glucuronidase–positive; NSF, non–sorbitol-fermenting; SF, sorbitol-fermenting; STEC, Shiga toxin–producing E. coli. The genomic information available for GP STEC O157:H7 (; >99% identity and is also limited to strain G5101 (9,17) and 4 strains in pub- >99% coverage) for comparisons with previously report- lic databases. Thus, the virulence potential of GP STEC ed reference sequences (2). O157:H7 remains to be fully elucidated. In this study, we isolated a GP STEC O157:H7 strain Genome Sequencing, Assembly, and Annotation that produces Stx2a, in addition to Stx1a and Stx2c, from We determined the complete genome sequence of PV15- a patient with bloody diarrhea in Japan and determined its 279 using PacBio reads obtained with an RS II system complete genome sequence. To reveal the genomic fea- (PacBio, Menlo Park, CA, USA). We assembled the reads tures of the Stx2a-producing strain, we determined the draft with Canu version 1.5 (27) and circularized them using genome sequences of an additional 13 GP STEC O157:H7 Circlator (28). We obtained Illumina paired-end reads isolates, which are all Stx2a negative but Stx1a/Stx2c (or (300 bp × 2) with a MiSeq sequencer (Illumina, San Di- Stx2c) positive, and performed fine phylogenetic and ge - ego, CA, USA) and mapped them to the assembly using nomic comparisons of these GP STEC O157:H7 strains the Burrows-Wheeler aligner (29) for sequence-error cor- with typical STEC O157:H7 and SF STEC O157:H– rection with Pilon (30). We completed further corrections strains. We also analyzed the Stx production levels of GP to the sequences corresponding to Stx phage regions using STEC O157:H7 strains. MiSeq reads obtained by sequencing long-range PCR prod- ucts covering each Stx phage. We performed annotations Methods with the DDBJ Fast Annotation and Submission Tool (31), followed by manual curation using the IMC-GE software Bacterial Strains (In Silico Biology, Kanagawa, Japan). The strains used in this study are listed in online Tech- We obtained the draft genome sequences of 13 GP nical Appendix Table 1 ( STEC O157:H7 strains by assembling Illumina paired-end article/24/12/18-0404-Techapp1.pdf). The GP STEC reads using Platanus (32). The Illumina reads for strain O157:H7 strain PV15-279 was isolated from an adult pa- LB473017, for which the only read data available were tient in Japan who was hospitalized in 2015 with severe from public databases, were also assembled with Platanus. symptoms, including bloody diarrhea. The other 13 GP The genome sequences of PV15-279 and the 13 GP STEC O157:H7 strains sequenced in this study were isolat- STEC O157:H7 strains have been deposited in DDBJ/ ed in Japan during 1988–2013 from humans with or with- EMBL/GenBank under the Bioproject accession numbers out symptoms. We obtained genome sequence information PRJDB6584 and PRJDB6498, respectively. The accession for 3 E. coli O55:H7 (18–20), 2 SF STEC O157:H–, 14 numbers of each sample, including the reference data, are typical STEC O157:H7 (21–26), and 2 GP STEC O157:H7 listed in online Technical Appendix Table 1. strains (17) from public databases. Single-Nucleotide Polymorphism Detection and Subtyping of stx Genes Phylogenetic Analysis We performed in silico subtyping of stx1 and stx2 We performed single-nucleotide polymorphism (SNP) de- in all strains analyzed in this study. We used blastn tection and phylogenetic analyses as described previously 2220 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 β-Glucuronidase–Positive E. coli Producing Stx2a (33). The genome sequences to be examined were aligned Darmstadt, Germany) coated with capture antibodies that with the phage- and insertion sequence (IS)–masked chro- recognize both Stx1 and Stx2. We conjugated monoclo- mosome sequence of the STEC O157:H7 strain Sakai (22) nal antibodies against Stx1 and Stx2 (LSBio, Seattle, WA, using MUMmer (34) to identify conserved regions (cutoff USA) with horseradish peroxidase using the Peroxidase thresholds >98% sequence identity and >1,000-bp align- Labeling Kit–NH (Dojindo, Kumamoto, Japan) and em- ment length) in each strain and the SNP sites located there- ployed them as detection antibodies. We used reagents sup- in. We then determined the core genome sequence that was plied in the RIDASCREEN Verotoxin kit for detection. Fi- conserved in all strains examined. Only SNPs located in nally, we measured the absorbance at 450 nm (A ) using the core genome were subjected to further analysis. After Tecan Infinite 200 PRO (Tecan, Männedorf, Switzerland). reconstructing the genome sequences of each strain using the SNPs and removing recombinogenic SNP sites using Results Gubbins (35), we constructed a maximum-likelihood phy- logenetic tree using RAxML (36) with the general time- Isolation and Genome Sequencing of reversible plus gamma model of nucleotide substitution GP O157:H7 Strains and 500 bootstrap replicates. The tree was displayed and In PV15-279, we detected stx2a in addition to stx1a and annotated using iTOL (37). stx2c (online Technical Appendix Table 1). Determination of the complete genome sequence of PV15-279 revealed Repertoire Analysis of Genes Encoding T3SS Effectors that the genome consisted of a 5,598,151-bp chromosome and Plasmid-Encoded Virulence Factors and a 94,391-bp plasmid. The draft genome sequences of 13 We analyzed the conservation of genes encoding T3SS ef- additional GP STEC O157:H7 strains were also determined fectors and plasmid-encoded virulence factors with blastn for comparison (online Technical Appendix Table 1). These (>90% identity and >30% coverage). All intact effector strains were all stx1a/stx2c-positive and stx2a-negative. The genes and plasmid virulence genes identified in strains uidA gene, which encodes β-glucuronidase, was intact in all Sakai and PV15-279 were clustered using CD-HIT (38) at the GP STEC O157:H7 strains sequenced in this study, as >90% identity and >30% alignment coverage of the lon - well as in 2 GP STEC O157:H7 strains whose draft genome ger sequences, and representative sequences of each cluster sequences were publicly available (online Technical Appen- were used to create a database for blastn analysis. dix Table 1). In typical STEC O157:H7, uidA has been inac- tivated by a frameshift mutation (41). Stx Phage Sequencing We determined the complete sequences of the Stx1a and Comparisons of General Genomic Features and Stx2c phages from strain 980938 (from an asymptomatic Mobile Genetic Elements carrier) and the Stx2c phage from strain 981447 (from a Although the chromosome of PV15-279 was ≈100 kb patient with bloody diarrhea) as described previously (39). larger than that of the typical GP STEC O157:H7 strain We constructed fosmid libraries of the 2 strains using a Sakai (Table and online Technical Appendix Figure 1), the CopyControl fosmid library production kit (Epicenter Bio- chromosome backbone was highly conserved; 97.2% of technologies, Madison, WI, USA). We screened the stx1- Table. General genomic features of the -glucuronidase–positive or stx2-containing clones using PCR and sequenced them STEC O157:H7 strain PV15-279 from Japan and the typical by the shotgun sequencing strategy using an ABI3730 se- STEC O157:H7 strain Sakai* quencer (Applied Biosystems, Foster City, CA, USA). Feature PV15-279 Sakai Chromosome Length, bp 5,598,152 5,498,578 Stx Production Assay CDSs (pseudogenes) 5,295 (107) 5,202 (122) We inoculated bacterial cells into 40 mL of CAYE broth rRNA operons 7 7 (Merck, Darmstadt, Germany) and grew them to mid-log tRNAs 108 104 Prophages 22 18 phase at 37°C with shaking. We then added mitomycin C Integrative elements 5 6 (MMC) to the culture at a final concentration of 500 ng/ IS elements 80 65 mL. After MMC addition, we collected 100 µL of the cul- Plasmid pO157 Length, bp 94,391 92,722 ture every hour and immediately subjected it to sonication CDSs (pseudogenes) 95 (0) 90 (8) using a Bioruptor (Cosmo Bio, Tokyo, Japan). We obtained IS elements 11 8 the soluble fractions of each cell lysate via centrifugation Plasmid pOSAK1 Length, bp NA 3,306 at 14,000 × g for 10 min at 4°C. We determined Stx1 and CDSs (pseudogenes) NA 3 (0) Stx2 concentrations in each cell lysate using a previously Total genome size, kb 5,692,969 5,591,300 described sandwich ELISA (40). We captured Stx using *CDS, coding sequence; IS, insertion sequence; NA, not available; STEC, Shiga toxin–producing Escherichia coli. RIDASCREEN Verotoxin microtiter plates (R-Biopharm, Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2221 RESEARCH the coding sequences (CDSs) identified in Sakai were con - in tandem (online Technical Appendix Table 2, Figure 1, served in PV15-279. Among the various types of mobile Figure 2, panel A). genetic elements (MGEs), integrative elements (defined as The repertoires and total copy numbers of IS elements elements that contain an integrase gene but no other MGE- also exhibited notable differences (online Technical Ap- related genes) were well conserved and showed only minor pendix Table 3). Although 24 types of IS elements were variations. Among the 5 integrative elements identified in identified in Sakai and 18 types were identified in PV15- Sakai, 1 small element (SpLE2) was missing in PV15-279, 279, only 14 were shared by the 2 strains. A greater number and multiple small structural variations were observed in of total copies of IS elements was detected in PV15-279 3 elements (online Technical Appendix Table 2, Figure (91, compared with 73 in Sakai), primarily because of the 2, panel B). In contrast, the prophage contents exhibited proliferation of IS1203 (also annotated as IS629) in PV15- notable differences. PV15-279 contained 22 prophages, 279. The virulence plasmid pO157 also showed notable whereas Sakai contained 18 prophages. This difference is differences, as described in the next section. attributable mainly to the difference in chromosome sizes between the 2 strains. Although most of the integration Phylogenetic Characterization of GP sites used by Sakai prophages (13/16) are also used by STEC O157:H7 Strains PV15-279 prophages (online Technical Appendix Table 2, To determine the precise phylogenetic position of PV15-279 Figure 2, panel A), in most instances prophages that had in STEC O157, we performed high-resolution phylogenetic integrated into the same site showed notable variations be- analysis using the genome sequences of various typical and tween the 2 strains, suggesting the existence of different atypical STEC O157 strains (Figure 2). The results clearly evolutionary histories. Furthermore, PV15-279 carried as indicate that PV15-279 formed a distinct cluster with all oth- many as 4 Stx phages: 1 Stx1a phage, 2 Stx2a phages, and 1 er GP STEC O157:H7 strains, including 2 US isolates, and is Stx2c phage (online Technical Appendix Figure 1). In con- a strain that recently emerged from the stx1a/stx2c-positive trast, Sakai carries 1 Stx1a phage and 1 Stx2a phage (see clone circulating in Japan. The phylogenetic relationship of subsequent sections for a comparison of these phages with this cluster with other STEC O157 lineages was concordant Stx phages from various STEC O157:H7 strains). Surpris- with the stepwise evolution model (Figure 1). ingly, 1 of the 2 Stx2a phages (PV15p10) was integrated In our phylogenetic tree, the GP STEC O157:H7 together with 3 prophages at the ydfJ locus in PV15-279 strains were relatively clonal compared with the typical Figure 2. Whole-genome sequence-based phylogenetic analysis and repertoires of T3SS effectors and plasmid-encoding virulence factors from the study of Stx-producing E. coli O157:H7. The genome sequences of all the strains used in this study were aligned with the complete chromosome sequence of Sakai, and the single-nucleotide polymorphisms located in the 4,074,209-bp backbone sequence that were conserved in all the test strains were identified. After removing the recombinogenic single-nucleotide polymorphisms sites, we performed a concatenated alignment of 5,803 informative sites to generate the maximum-likelihood phylogeny. The conservation of T3SS effectors and plasmid-encoding virulence factors is shown in the tree. Colored boxes indicate the presence, and open boxes the absence, of each gene. GP, β-glucuronidase–positive; LEE, locus of enterocyte effacement; SF, sorbitol-fermenting. Scale bars indicate nucleotide substitutions per site. 2222 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 β-Glucuronidase–Positive E. coli Producing Stx2a STEC O157:H7 strains, although these typical STEC Stx Phages of GP STEC O157:H7 O157:H7 strains were selected to represent their cur - As shown in Figure 3, we performed fine genomic com- rently known phylogenetic diversity (26) (Figure 2). De- parisons of the Stx phages from PV15-279 with the Stx1a, spite the close phylogenetic relationship between the GP Stx2a, and Stx2c phages from other STEC O157:H7 strains STEC O157:H7 strains, only PV15-279 was stx2a-pos- used in the phylogenetic analysis shown in Figure 2 (phag- itive, indicating that PV15-279 acquired stx2a very re- es were included only when complete sequences were cently. No stx genes were detected in the publicly avail- available). The Stx1a and Stx2c phages of the GP STEC able sequence data for the GP STEC O157:H7 strain O157:H7 strain 980938 and the Stx2c phage of the GP G5101, although this strain was previously reported to STEC O157:H7 strain 981447 were sequenced individu- contain stx1 and stx2 (14). ally and included in the analysis. Stx1a phages were integrated in sbcB in the 2 GP STEC Distribution of T3SS Effectors in the O157:H7 strains but were integrated in yehV or argW in GP O157:H7 Lineage typical STEC O157:H7 strains. According to the dot plot In addition to 5 LEE-encoded T3SS effectors, 18 fami - analysis, the 2 Stx1a phages from GP STEC O157:H7 lies of effectors are encoded at non-LEE genomic loci were nearly identical (Figure 3). The phages from typical (non-LEE effectors) in Sakai ( 42). In PV15-279, we STEC O157:H7 strains were classified into 3 groups based identified most of the effector families found in Sakai on sequence similarity. The Stx1a phages from GP STEC (online Technical Appendix Table 4). The major dif- O157:H7 showed various levels of sequence similarity to ferences were the absence of nleF in PV15-279 and the each group. The highest similarity was observed for the 3 presence of ospB and ospG in PV15-279, both of which phages integrated in yehV, but a clear difference was ob- are absent in Sakai. After subgrouping the 4 effector served in the early region. families (espM, espN, espO, and nleG) into 2–5 sub- The 2 Stx2a phages from PV15-279 that integrated in groups based on their sequence diversity, we analyzed yecE and ydfJ were nearly identical (only 2 SNPs), indicat- the effector repertoires of the STEC O157 isolates used ing that these phages were recently duplicated in this strain. in the phylogenetic analysis (Figure 2). This analysis With 1 exception (the phage from strain 155), the Stx2a revealed that the effectors identified in PV15-279 were phages from the typical STEC O157:H7 strains analyzed in mostly conserved in other GP STEC O157:H7 strains, this study were integrated in either wrbA or argW, and all although some variations were detected. shared a similar genomic structure, although considerable variations were observed in the early region, as reported Virulence Plasmid pO157 of GP O157:H7 Lineage previously (7). The Stx2a phages from PV15-279 exhibited The pO157 plasmid of PV15-279 was nearly identical to a genetic structure similar to that of strain 155, which was the plasmid from the GP STEC O157:H7 strain G5101 also integrated in yecE. However, those phages showed (43), with several small variations that were apparently high sequence similarity only in limited regions. generated by mechanisms involving IS (online Techni- The Stx2c phages from GP STEC O157:H7 strains cal Appendix Figure 3). The plasmid-encoded virulence (all integrated in yehV) were nearly identical, showing only genes identified in PV15-279 were well conserved in oth - minor variations, such as IS integration. The Stx2c phages er GP STEC O157:H7 strains, except for 3 strains from from typical STEC O157:H7 strains, which were all inte- which pO157 has apparently been deleted (Figure 2). It grated in sbcB, also exhibited very similar genomic struc- is unknown whether this deletion occurred before or after tures and sequences. Although notable sequence similarity strain isolation. was observed between the Stx2c phages from GP STEC As previously reported (43), the pO157 plasmids from O157:H7 and typical STEC O157:H7, the early regions GP STEC O157:H7 showed high similarity to those from were very different. typical STEC O157:H7 and SF STEC O157:H–. Thus, a pO157-like plasmid was likely acquired by the common Stx Production by GP STEC O157:H7 Strains ancestor of the 3 STEC O157 lineages. A notable difference Stx1 production is partially dependent on phage induction, between the pO157 plasmids of the 3 STEC O157 lineages whereas Stx2 production is strongly dependent on phage was the distribution of katP and espP. These genes were induction (44–46). We compared Stx production levels detected only in typical STEC O157:H7 strains, suggesting between GP STEC O157:H7 and typical STEC O157:H7 that these genes may have been specifically acquired by the using MMC as a phage induction agent; we analyzed typical STEC O157:H7 lineage. Although the roles of katP Stx1 and Stx2 production by representative strains of the and espP in STEC virulence in humans have not yet been 2 lineages (PV15-279, 981447, and 980938 for GP STEC elucidated, at least the SF STEC O157:H– strain causes se- O157:H7; Sakai and EDL933 for typical STEC O157:H7). vere infections even without these genes. MMC exhibited different levels of effectiveness in phage Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2223 RESEARCH Figure 3. Genome comparisons of Stx phages from the study of Stx-producing E. coli O157:H7. The results of the comparison of the genome structure (left) and dot-plot sequence comparisons (right) of the Stx phages are shown. Sequence identities are indicated by different colors. In the dot-plot matrices, phages integrated in the same integration sites are highlighted by gray shading and colored frames. GP, β-glucuronidase–positive; IS, insertion sequence. 2224 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 β-Glucuronidase–Positive E. coli Producing Stx2a induction in each strain, and clear cell lysis (a clear reduc- PV15-279 and other GP STEC O157:H7 strains, although tion in OD values) was observed only in EDL933 and some variations were detected (Figure 2). Moreover, the PV15-279 (Figure 4, panel A). ability of PV15-279 to produce Stx2a was comparable to Consistent with a previous report (47), the Stx1 con- that of typical STEC O157:H7 (Figure 4). These findings centration was not notably elevated by phage induction in suggest that PV15-279 presents a virulence potential simi- any of the strains, and no clear difference was observed be - lar to that of typical STEC O157:H7. In fact, PV15-279 tween the GP STEC O157:H7 and typical STEC O157:H7 was isolated from a patient who had severe enteric symp- strains (Figure 4, panel B). In contrast, levels of Stx2 pro- toms, including bloody diarrhea. However, further investi- duction were highly variable, as previously shown for typi- gations are necessary to determine how the variations in the cal STEC O157:H7 strains (7). Although Stx2 production virulence gene repertoire detected in the comparison with was poorly induced in 2 GP STEC O157:H7 strains carry- other STEC O157 lineages affect the potential virulence of ing stx2c alone, it was strongly induced in the stx2a-positive GP STEC O157:H7. The repertoires of the minor virulence PV15-279 and typical STEC O157:H7 strains (Figure 4, genes of GP STEC O157:H7 must also be investigated. panel C). Therefore, stx2a was strongly induced in PV15- We also obtained several noteworthy findings con- 279, as in typical STEC O157:H7, but stx2c was poorly cerning the evolution of GP O157:H7 in this study. For induced in PV15-279, as in the other GP STEC O157:H7 example, fine genomic comparisons of Stx phages revealed strains. The level of Stx2 production by PV15-279 was that Stx phages differ notably between GP STEC O157:H7 comparable to that of typical STEC O157:H7 strains and and typical STEC O157:H7 (Figure 3). Although many similar to that of the Sakai strain. of these variations may have been generated through ex- tensive recombination with various resident or incoming Discussion phages (48), the acquisition of the Stx2a phage by the GP In this study, we isolated a GP STEC O157:H7 strain STEC O157:H7 strain PV15-279 was apparently an inde- (PV15-279) that produces Stx2a in addition to Stx1a and pendent genetic event. The clear differences in the genetic Stx2c. The whole-genome sequence-based phylogenetic structure, sequence, and integration sites of Stx2c phages analysis, which included additional GP STEC O157:H7 between GP and typical STEC O157:H7 may suggest that strains and representative strains belonging to other E. Stx2c phages were also acquired independently by the 2 coli O55/O157 lineages, revealed that PV15-279 recently lineages after they separated. In contrast, the similar genet- emerged by the acquisition of Stx2a phage from the stx1a/ ic structures and sequences of the pO157 plasmids from GP stx2c-positive GP STEC O157:H7 clone circulating in Ja- and typical STEC O157:H7 and SF STEC O157:H– sug- pan (Figure 2). Most of the major virulence genes identified gest that a pO157-like plasmid might have been acquired in typical STEC O157:H7, such as T3SS effector genes and by their common ancestor. Although more extensive stud- plasmid-encoded virulence genes, were well conserved in ies are needed to obtain a complete understanding of these Figure 4. Lysis curves and levels of Stx produced by STEC O157:H7 strains after MMC treatment in study of Stx-producing E. coli. The lysis curves (A) and levels of Stx1 (B) and Stx2 (C) production by 5 STEC O157:H7 strains after the MMC treatment are shown. After the addition of MMC, the OD of each strain was measured every hour for 8 hours, and 100 µL of the culture was collected at each time point. Cell lysates were prepared, and the Stx1 and Stx2 concentrations in the soluble fractions were analyzed using a sandwich ELISA. All experiments were performed 3 times, and the averages and SEs of the Stx1 and Stx2 concentrations in each sample are plotted. Strains Sakai and EDL933 belong to typical STEC O157:H7, and the other 3 strains belong to GP STEC O157:H7. GP, β-glucuronidase–positive; MMC, mitomycin C; OD, optical density; STEC, Stx-producing E. coli. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2225 RESEARCH 6. Kawano K, Okada M, Haga T, Maeda K, Goto Y. Relationship issues, only a limited number of SF STEC O157:H– ge- between pathogenicity for humans and stx genotype in Shiga nome sequences and no complete sequences of the Stx toxin-producing Escherichia coli serotype O157. Eur J Clin phages of this lineage are currently available. The avail- Microbiol Infect Dis. 2008;27:227–32. able genome sequence information of GP STEC O157:H7 s10096-007-0420-3 7. Ogura Y, Mondal SI, Islam MR, Mako T, Arisawa K, Katsura K, is also highly biased toward Japanese isolates. et al. The Shiga toxin 2 production level in enterohemorrhagic In conclusion, we isolated a Stx2a-producing GP Escherichia coli O157:H7 is correlated with the subtypes of STEC O157:H7 strain that emerged from the stx1a/stx2c- toxin-encoding phage. Sci Rep. 2015;5:16663. positive GP STEC O157:H7 clone circulating in Japan; the 10.1038/srep16663 8. Feng PC, Monday SR, Lacher DW, Allison L, Siitonen A, Keys C, virulence potential of this isolate is similar to that of typi- et al. Genetic diversity among clonal lineages within Escherichia cal STEC O157:H7. Researchers should pay more atten- coli O157:H7 stepwise evolutionary model. Emerg Infect Dis. tion to this less commonly reported STEC O157 lineage, 2007;13:1701–6. particularly the spread of this Stx2a-producing GP STEC 9. Wick LM, Qi W, Lacher DW, Whittam TS. Evolution of genomic content in the stepwise emergence of Escherichia coli O157:H7. O157:H7 clone and the emergence of additional stx2a- J Bacteriol. 2005;187:1783–91. positive clones. Larger-scale genomic analyses including JB.187.5.1783-1791.2005 more GP STEC O157:H7 strains from various geographic 10. Karch H, Bielaszewska M. Sorbitol-fermenting Shiga toxin- regions and more SF STEC O157:H– strains are required to producing Escherichia coli O157:H(–) strains: epidemiology, phenotypic and molecular characteristics, and microbiological obtain a better understanding of the evolution and genomic diagnosis. J Clin Microbiol. 2001;39:2043–9. diversity of GP O157:H7. 10.1128/JCM.39.6.2043-2049.2001 11. Ammon A, Petersen LR, Karch H. A large outbreak of hemolytic Acknowledgments uremic syndrome caused by an unusual sorbitol-fermenting strain of Escherichia coli O157:H-. J Infect Dis. 1999;179:1274–7. We thank A. Yoshida, Y. Inoue, M. Shimbara, N. Kanemaru, N. Kawano, N. Sakamoto, H. Iguchi, M. Horiguchi, 12. Karch H, Böhm H, Schmidt H, Gunzer F, Aleksic S, Heesemann J. M. Kumagai, and Y. Morita for providing technical assistance. Clonal structure and pathogenicity of Shiga-like toxin-producing, sorbitol-fermenting Escherichia coli O157:H-. J Clin Microbiol. This work was funded by the Japan Society for the Promotion of 1993;31:1200–5. Science (KAKENHI) under grant nos. 16K15278 and 13. Rosser T, Dransfield T, Allison L, Hanson M, Holden N, Evans J, 17H04077 to Y.O. and 16H05190 to T.H. and by the Japan et al. Pathogenic potential of emergent sorbitol-fermenting Escherichia coli O157:NM. Infect Immun. 2008;76:5598–607. Agency for Medical Research and Development under grant nos. JP17efk0108127h0001 to Y.O. and JP17fk0108308j0003 to T.H. 14. Hayes PS, Blom K, Feng P, Lewis J, Strockbine NA, Swaminathan B. Isolation and characterization of a beta-D- glucuronidase-producing strain of Escherichia coli serotype About the Author O157:H7 in the United States. J Clin Microbiol. 1995;33:3347–8. Dr. Ogura is an associate professor in the department of 15. Nagano H, Hirochi T, Fujita K, Wakamori Y, Takeshi K, Yano S. Phenotypic and genotypic characterization of beta-D-glucuronidase- bacteriology at the Faculty of Medical Sciences, Kyushu positive Shiga toxin-producing Escherichia coli O157:H7 isolates University, Kyushu, Japan. His primary research interests from deer. J Med Microbiol. 2004;53:1037–43. include bacterial genomics and pathogenicity. 10.1099/jmm.0.05381-0 16. Nagano H, Okui T, Fujiwara O, Uchiyama Y, Tamate N, Kumada H, et al. Clonal structure of Shiga toxin (Stx)-producing and beta- References D-glucuronidase-positive Escherichia coli O157:H7 strains isolated 1. Pennington H. Escherichia coli O157. Lancet. 2010;376:1428–35. from outbreaks and sporadic cases in Hokkaido, Japan. J Med Microbiol. 2002;51:405–16. 2. Scheutz F, Teel LD, Beutin L, Piérard D, Buvens G, Karch H, et al. 51-5-405 Multicenter evaluation of a sequence-based protocol for subtyping 17. Rump LV, Strain EA, Cao G, Allard MW, Fischer M, Brown EW, et Shiga toxins and standardizing Stx nomenclature. J Clin Microbiol. al. Draft genome sequences of six Escherichia coli isolates 2012;50:2951–63. from the stepwise model of emergence of Escherichia coli 3. O’Brien AD, Newland JW, Miller SF, Holmes RK, Smith HW, O157:H7. J Bacteriol. 2011;193:2058–9. Formal SB. Shiga-like toxin-converting phages from Escherichia 10.1128/JB.00118-11 coli strains that cause hemorrhagic colitis or infantile diarrhea. 18. Hazen TH, Sahl JW, Redman JC, Morris CR, Daugherty SC, Science. 1984;226:694–6. Chibucos MC, et al. Draft genome sequences of the diarrheagenic 4. Boerlin P, McEwen SA, Boerlin-Petzold F, Wilson JB, Johnson RP, Escherichia coli collection. J Bacteriol. 2012;194:3026–7. Gyles CL. Associations between virulence factors of Shiga toxin- producing Escherichia coli and disease in humans. J Clin 19. Schutz K, Cowley LA, Shaaban S, Carroll A, McNamara E, Microbiol. 1999;37:497–503. Gally DL, et al. Evolutionary context of non-sorbitol-fermenting 5. de Sablet T, Bertin Y, Vareille M, Girardeau JP, Garrivier A, Shiga toxin-producing Escherichia coli O55:H7. Emerg Infect Dis. Gobert AP, et al. Differential expression of stx2 variants in Shiga 2017;23:1966–73. toxin-producing Escherichia coli belonging to seropathotypes A 20. Zhou Z, Li X, Liu B, Beutin L, Xu J, Ren Y, et al. Derivation of and C. Microbiology. 2008;154:176–86. Escherichia coli O157:H7 from its O55:H7 precursor. PLoS One. mic.0.2007/009704-0 2010;5:e8700. 2226 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 β-Glucuronidase–Positive E. coli Producing Stx2a 21. Eppinger M, Mammel MK, Leclerc JE, Ravel J, Cebula TA. 36. Stamatakis A. RAxML-VI-HPC: maximum likelihood-based Genomic anatomy of Escherichia coli O157:H7 outbreaks. Proc phylogenetic analyses with thousands of taxa and mixed models. Natl Acad Sci U S A. 2011;108:20142–7. Bioinformatics. 2006;22:2688–90. pnas.1107176108 bioinformatics/btl446 22. Hayashi T, Makino K, Ohnishi M, Kurokawa K, Ishii K, 37. Letunic I, Bork P. Interactive tree of life (iTOL) v3: an online tool Yokoyama K, et al. Complete genome sequence of for the display and annotation of phylogenetic and other trees. enterohemorrhagic Escherichia coli O157:H7 and genomic Nucleic Acids Res. 2016;44(W1):W242–5. comparison with a laboratory strain K-12. DNA Res. 2001;8:11–22. 10.1093/nar/gkw290 38. Li W, Godzik A. Cd-hit: a fast program for clustering and 23. Katani R, Cote R, Raygoza Garay JA, Li L, Arthur TM, DebRoy C, comparing large sets of protein or nucleotide sequences. et al. Complete genome sequence of SS52, a strain of Escherichia Bioinformatics. 2006;22:1658–9. coli O157:H7 recovered from supershedder cattle. Genome An- bioinformatics/btl158 nounc. 2015;3:e01569–14. 39. Ogura Y, Abe H, Katsura K, Kurokawa K, Asadulghani M, genomeA.01569-14 Iguchi A, et al. Systematic identification and sequence analysis of 24. Kulasekara BR, Jacobs M, Zhou Y, Wu Z, Sims E, the genomic islands of the enteropathogenic Escherichia coli strain Saenphimmachak C, et al. Analysis of the genome of the B171–8 by the combined use of whole-genome PCR scanning and Escherichia coli O157:H7 2006 spinach-associated outbreak isolate fosmid mapping. J Bacteriol. 2008;190:6948–60. indicates candidate genes that may enhance virulence. Infect 10.1128/JB.00625-08 Immun. 2009;77:3713–21. 40. Ishijima N, Lee KI, Kuwahara T, Nakayama-Imaohji H, Yoneda S, 25. Latif H, Li HJ, Charusanti P, Palsson BO, Aziz RKA. A gapless, Iguchi A, et al. Identification of a new virulent clade in unambiguous genome squence of the enterohemorrhagic enterohemorrhagic Escherichia coli O26:H11/H- sequence type 29. Escherichia coli O157:H7 strain EDL933. Genome Announc. Sci Rep. 2017;7:43136. 2014;2:e00821–14. 41. Monday SR, Whittam TS, Feng PC. Genetic and evolutionary 26. Shaaban S, Cowley LA, McAteer SP, Jenkins C, Dallman TJ, analysis of mutations in the gusA gene that cause the absence of Bono JL, et al. Evolution of a zoonotic pathogen: investigating beta-glucuronidase activity in Escherichia coli O157:H7. J Infect prophage diversity in enterohaemorrhagic Escherichia coli O157 Dis. 2001;184:918–21. by long-read sequencing. Microb Genom. 2016;2:e000096. 42. Tobe T, Beatson SA, Taniguchi H, Abe H, Bailey CM, Fivian A, 27. Koren S, Walenz BP, Berlin K, Miller JR, Bergman NH, et al. An extensive repertoire of type III secretion effectors in Phillippy AM. Canu: scalable and accurate long-read assembly Escherichia coli O157 and the role of lambdoid phages in their via adaptive k-mer weighting and repeat separation. Genome Res. dissemination. Proc Natl Acad Sci U S A. 2006;103:14941–6. 2017;27:722–36. 28. Hunt M, Silva ND, Otto TD, Parkhill J, Keane JA, Harris SR. 43. Rump LV, Meng J, Strain EA, Cao G, Allard MW, Gonzalez- Circlator: automated circularization of genome assemblies Escalona N. Complete DNA sequence analysis of enterohemorrhagic using long sequencing reads. Genome Biol. 2015;16:294. Escherichia coli plasmid pO157_2 in β-glucuronidase-positive E. coli O157:H7 reveals a novel evolutionary path. J Bacteriol. 29. Li H, Durbin R. Fast and accurate short read alignment with 2012;194:3457–63. Burrows-Wheeler transform. Bioinformatics. 2009;25:1754–60. 44. Calderwood SB, Mekalanos JJ. Iron regulation of Shiga-like toxin expression in Escherichia coli is mediated by the fur locus. 30. Walker BJ, Abeel T, Shea T, Priest M, Abouelliel A, Sakthikumar S, J Bacteriol. 1987;169:4759–64. et al. Pilon: an integrated tool for comprehensive microbial jb.169.10.4759-4764.1987 variant detection and genome assembly improvement. PLoS One. 45. Tyler JS, Mills MJ, Friedman DI. The operator and early promoter 2014;9:e112963. region of the Shiga toxin type 2-encoding bacteriophage 933W 31. Tanizawa Y, Fujisawa T, Kaminuma E, Nakamura Y, Arita M. and control of toxin expression. J Bacteriol. 2004;186:7670–9. DFAST and DAGA: web-based integrated genome annotation tools and resources. Biosci Microbiota Food Health. 2016;35:173–84. 46. Wagner PL, Livny J, Neely MN, Acheson DW, Friedman DI, Waldor MK. Bacteriophage control of Shiga toxin 1 production 32. Kajitani R, Toshimoto K, Noguchi H, Toyoda A, Ogura Y, and release by Escherichia coli. Mol Microbiol. 2002;44:957–70. Okuno M, et al. Efficient de novo assembly of highly heterozygous genomes from whole-genome shotgun short reads. Genome Res. 47. Shimizu T , Ohta Y, Noda M. Shiga toxin 2 is specifically released 2014;24:1384–95. from bacterial cells by two different mechanisms. Infect Immun. 33. Ogura Y, Gotoh Y, Itoh T, Sato MP, Seto K, Yoshino S, et al. 2009;77:2813–23. Population structure of Escherichia coli O26: H11 with recent and 48. Asadulghani M, Ogura Y, Ooka T, Itoh T, Sawaguchi A, repeated stx2 acquisition in multiple lineages. Microb Genom. Iguchi A, et al. The defective prophage pool of Escherichia coli 2017;3:e000141. O157: prophage-prophage interactions potentiate horizontal 34. Kurtz S, Phillippy A, Delcher AL, Smoot M, Shumway M, transfer of virulence determinants. PLoS Pathog. 2009;5:e1000408. Antonescu C, et al. Versatile and open software for comparing large genomes. Genome Biol. 2004;5:R12. gb-2004-5-2-r12 Address for correspondence: Yoshitoshi Ogura, Kyushu University, 35. Croucher NJ, Page AJ, Connor TR, Delaney AJ, Keane JA, Department of Bacteriology, Faculty of Medicine Sciences, 3-1-1 Bentley SD, et al. Rapid phylogenetic analysis of large samples of Maidashi Higashi-ku, Fukuoka 812-8582, Japan; email: recombinant bacterial whole genome sequences using Gubbins. Nucleic Acids Res. 2015;43:e15. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2227 RESEARCH Survey of Ebola Viruses in Frugivorous and Insectivorous Bats in Guinea, Cameroon, and the Democratic Republic of the Congo, 2015–2017 1 1 1 Helene M. De Nys, Placide Mbala Kingebeni, Alpha K. Keita, Christelle Butel, Guillaume Thaurignac, Christian-Julian Villabona-Arenas, Thomas Lemarcis, Mare Geraerts, Nicole Vidal, Amandine Esteban, Mathieu Bourgarel, François Roger, Fabian Leendertz, Ramadan Diallo, Simon-Pierre Ndimbo-Kumugo, Justus Nsio-Mbeta, Nikki Tagg, Lamine Koivogui, Abdoulaye Toure, Eric Delaporte, Steve Ahuka-Mundeke, 2 2 Jean-Jacques Muyembe Tamfum, Eitel Mpoudi-Ngole, Ahidjo Ayouba, Martine Peeters To clarify the role of bats in the ecology of Ebola viruses, 27 insectivorous species; 2–37 (0.05%–0.92%) bats were we assessed the prevalence of Ebola virus antibodies in seropositive for Zaire and 0–30 (0%–0.75%) bats for Sudan a large-scale sample of bats collected during 2015–2017 Ebola viruses. We observed Ebola virus antibodies in 1 in- from countries in Africa that have had previous Ebola out- sectivorous bat genus and 6 frugivorous bat species. Cer- breaks (Guinea, the Democratic Republic of the Congo) or tain bat species widespread across Africa had serologic evi- are at high risk for outbreaks (Cameroon). We analyzed dence of Zaire and Sudan Ebola viruses. No viral RNA was 4,022 blood samples of bats from >12 frugivorous and detected in the subset of samples tested (n = 665). Ongoing surveillance of bats and other potential animal reservoirs Author affiliations: TransVIHMI of Institut de Recherche pour le are required to predict and prepare for future outbreaks. Développement, Institut National de la Santé et de la Recherche Médicale and University of Montpellier, Montpellier, France (H.M. De Nys, P. Mbala Kingebeni, A.K. Keita, C. Butel, ince the first outbreak of Ebola virus disease (EVD) in G. Thaurignac, C.-J. Villabona-Arenas, T. Lemarcis, M. Geraerts, S1976 in the northern part of the Democratic Republic of N. Vidal, A. Esteban, A. Toure, E. Delaporte, A. Ayouba, M. Peeters); the Congo (DRC), 26 recognized outbreaks have occurred National Institute of Biomedical Research, Kinshasa, Democratic in humans across Africa; fatality rates of outbreaks have Republic of the Congo (P. Mbala Kingebeni, S.-P. Ndimbo-Kumugo, been 25%–90% (1–4). Each EVD outbreak most likely re- S. Ahuka-Mundeke, J.-J. Muyembe Tamfum); Cliniques sulted from independent zoonotic events. Universitaires de Kinshasa, Kinshasa (P. Mbala Kingebeni, Bats are believed to play a role in the ecology of Eb - S. Ahuka-Mundeke, J.-J. Muyembe Tamfum); Centre de Recherche ola viruses as a reservoir species (5). Bats might infect et de Formation en Infectiologie de Guinée, Conakry, Guinea humans directly or via intermediate amplifying hosts, like (A.K. Keita, A. Toure); ASTRE of Centre de coopération internationale nonhuman primates or duikers (6,7). Bats might serve as a en recherche agronomique pour le développement, Institut national source of infection in certain areas where bats are hunted de la Recherche Agronomique and Univerisity of Montpellier, and eaten as bushmeat, but infection could also occur af- Montpellier (M. Bourgarel, F. Roger); Robert Koch-Institute, Berlin, ter consumption of fruits contaminated with saliva, urine, Germany (F. Leendertz); Ministère de l’Elevage et des Productions or feces from Ebola virus–infected bats (8,9). Ebola virus Animales, Conakry (R. Diallo); Direction de Lutte contre la Maladie, emergence through exposure to bats was suspected for at Kinshasa (J. Nsio-Mbeta); Royal Zoological Society of Antwerp, least 2 outbreaks: Luebo (the DRC) in 2007 and West Af- Antwerp, Belgium (N. Tagg); Université de Conakry, Conakry rica in 2013 (10,11). (L. Koivogui); Institut National de Sante Publique, Conakry (A. Toure); Relatively few data are available to support the role Institut de Recherches Médicales et d’Études des Plantes Médicinales, of bats in the ecology of Ebola viruses. During the EVD Yaoundé, Cameroon (E. Mpoudi-Ngole); Cameroon Institut de outbreaks of 2003 in Gabon and the Congo, Zaire Ebola Recherche pout le Développement, Yaoundé (E. Mpoudi-Ngole) These first authors contributed equally to this article. DOI: 2 These senior authors contributed equally to this article. 2228 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Ebola Virus in Frugivorous and Insectivorous Bats virus RNA and antibodies were detected in live-caught filter paper (GE Healthcare, Feasterville-Trevose, PA, specimens from 3 fruit bat species (Epomops franqueti, USA). We air-dried and preserved samples individually in Hypsignathus monstrosus, Myonycteris torquata); virus plastic bags containing silica desiccant and stored them in sequences were found in the livers or spleens of a few bats hermetic boxes; 2–3 weeks later, we transferred dried blood (6). In subsequent studies in Gabon, the Congo, Ghana, and spots to -20°C until needed for analysis. Data recorded in Zambia, antibodies were detected in additional frugivorous the field included information on capture site (global posi- bat species (Eidolon helvum, Epomophorus gambianus, tioning system coordinates, ecologic environment), capture Rousettus aegyptiacus, Micropteropus pusillus) and 1 in- method, morphology (body measurements, weight, color), sectivorous species (Mops condylurus) (12–16). The am- sex, age class (adult, juvenile), and species (identified visu- plification and sequencing of viral RNA of other filoviruses ally). We collected negative control samples (n = 145) from in bats, such as Marburg virus in bats from Africa (17–20), a captive-born insectivorous bat species (103 Carollia per- Lloviu virus in bats from Europe (21), and new filoviruses spicillata bats) hosted at the Parc Zoologique de Montpel- in bats from China (22), has provided additional evidence lier (Montpellier, France) and 2 frugivorous bat species (19 for a possible role of bats in Ebola virus ecology. Pteropus giganteus bats, 23 R. aegyptiacus bats) hosted at In general, EVD outbreaks have been limited in terms Wilhelma Zoo and Botanical Garden (Stuttgart, Germany). of their geographic spread and chains of human-to-human We collected and preserved samples the same way we did transmission (1). However, during the 2013–2016 out- for free-ranging bats. break, virus spread to the urban areas of 3 countries, infect- ing ≈30,000 persons in Guinea, Sierra Leone, and Liberia, Screening for Ebola Virus Antibodies and ≈11,000 deaths were recorded (23). This outbreak il- We tested dried blood spots with a Luminex-based sero- lustrated the potential for epidemic spread from a single logic assay adapted for bats (28) (online Technical Appen- zoonotic transmission, with severe public health and so- dix, cioeconomic impact (24). Additional studies are urgently Techapp1.pdf). The assay included recombinant Ebola needed to identify the animal reservoir, predict EVD out- virus proteins glycoprotein, nucleoprotein, or viral protein break risks, and improve our capacity to control epidemics. 40 for different lineages: Zaire, Sudan, Bundibugyo, and In previous modeling studies, areas were defined as at Reston. We reconstituted plasma from dried blood spots as risk for EVD outbreaks on the basis of data collected from previously described (28) and incubated 100 μL of sample a limited number of wildlife bat species from a few geo- (final plasma dilution 1:2,000) with 50 µL of recombinant graphic regions (5,25). Also, a wide variety of serologic protein–coated beads (2 µg protein/1.25 × 10 beads) in assays and interpretation criteria have been used, making 96-well flat-bottom filter plates (Millipore, Tullagreen, comparison of results challenging (12–16,26,27). For this Ireland) on a plate shaker at 300 rpm for 16 h at 4°C in study, we performed a large serosurvey with a highly spe- the dark. After washing, we added 0.1 μg/mL of goat anti- cific and sensitive high-throughput assay to assess Ebola bat biotin–labeled IgG (Euromedex, Souffelweyersheim, virus prevalence in bats from Africa (28). We studied bats France) per well and incubated for 30 min at 300 rpm. Af- from Guinea and the DRC, countries with previous EVD ter another round of washing, we added 50 µL of 4 µg/mL outbreaks, and Cameroon, a country considered at high risk streptavidin-R-phycoerythrin (Fisher Scientific, Illkirch, for future EVD outbreaks (5,25). France) per well and incubated for 10 min at 300 rpm. Re- actions were read with BioPlex-200 (BioRad, Marnes-la- Materials and Methods Coquette, France). We expressed results as median fluo- rescence intensity (MFI) per 100 beads. We included 3 Study Sites and Sample Collection samples on every plate to validate interassay repeatability. During November 2015–August 2017, we collected sam- ples from free-ranging frugivorous and insectivorous bats Determination of Cutoffs in Guinea, Cameroon, and the DRC. We captured bats at In the absence of positive control samples, we used 4 dif- night using ground mist nets or harp traps in roosting and ferent statistical methods to determine the MFI cutoff value foraging sites. We set up ground mist nets (12 × 3.2 m) for each antigen (29,30) (online Technical Appendix Table of 30-mm and 60-mm mesh sizes at different heights (1–7 1). First, we used a general formula that involved the MFI m) to maximize capture of different species. We opened of the 145 negative control samples, and we assigned the nets or harp traps just before sunset and checked for bats cutoff as mean plus 4 times the SD (mean + 4×SD). Second, every 1–2 hours. Captured bats were released the same we used a change point analysis (31) to identify the value night immediately after sampling. Using bat whole blood at which statistical properties of the underlying probability taken by venipuncture of the propatagial or brachial vein, distribution changed. This value was used to identify outli- we dropped blood samples directly onto Whatman 903 ers and classify them as reactive. We used the R package Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2229 RESEARCH (37,38). For bat species from Cameroon and Guinea, we Table 1. Bat samples collected for Ebola virus serology by study screened for Zaire Ebola virus RNA by seminested reverse site, Guinea, Cameroon, and the Democratic Republic of the Congo, 2015–2017 transcription PCR (RT-PCR) targeting the nucleoprotein Country, site No. samples region of the virus genome. We amplified a 126-bp frag - Democratic Republic of the Congo ment of Zaire Ebola virus using primers NP1F1 (forward, Boma 156 Kimpese 202 5′-CGGACACACAAAAAGAAWGAA-3′) and NP1R-ZR Zongo 472 (reverse, 5′-CTCTATCTTKGTGATRTGGCTCTGA-3′) Subtotal 830 in the first round of PCR and NP1F2 (forward, 5′- TTGT- Cameroon GTGCGARTAACTAYGAGGAAG-3′) plus NP1R-ZR Yaoundé 126 Libellengoi Sud 44 in the second round. For species from the DRC, we per- Mbalmayo 48 formed seminested RT-PCR targeting the viral protein 35 Bipindi 479 region of the genome using the protocol of He et al. with Campo M’an 344 North Dja 295 modifications (41). In the first round, we amplified a 217- Ekom 122 bp fragment with primers VP35-F (5′-ATYATGTATGAT - Djoum 56 CACYTVCCWGG-3′) and VP35-R (52-AGCGRATGTG- Mambele 348 Mbam Djerem 156 GATSACRGGT-32) and, in the second round, a 184-bp Subtotal 2,018 product with primers VP35-R and VP35-in-F (5′-GCTT - Guinea TYCAYCAAYTAGTRCAAG-3′). Conakry 107 Kindia 323 Kankan 378 Molecular Confirmation of Bat Species Koundara 90 We confirmed bat species identification recorded in the field Mamou 147 on a subset of samples by using molecular tests. We am- Gueckedou 49 Macenta 9 plified an ≈800-bp fragment of mitochondrial cytochrome Nzerekore 71 b using primers cytb-L14724 (forward) and cytb-H15506 Subtotal 1,174 (reverse) (11,39,40). We substituted the cytb-L14724 Total 4,022 primer with cytb-L140217 (5′-ATGACCAACATCC- GAAAATCNCAC-3′) to improve PCR performance for changepoint (32) to calculate a single shift in the arithmetic certain species. We purified PCR products through agarose mean with the at-most-1-change method (33). Third, we gel (1%) and directly sequenced on an ABI 3500 sequencer fitted univariate distributions to our data and defined the (Applied Biosystems, Courtaboeuf, France). We performed cutoff as a 0.001 risk for error, as was used in other virus BLAST analyses ( to serology studies (13,34). We reduced the set of candidate identify the most similar bat species. For samples with no or distributions following a bootstrapped skewness-kurtosis low similarity (<97%) hits with species in GenBank, we per - analysis (35). We performed fitting by maximum-likeli - formed phylogenetic analyses with newly obtained sequenc- hood estimation and selected the best-fit distribution on the es and reference sequences for different bat species using basis of the Akaike information criteria with the R library maximum-likelihood methods implemented with PhyML fitdistrplus (36). A negative binomial distribution best-fit ( to determine genus. the data; however, we also used the negative exponential distribution as in Pourrut et al. and Laing et al. (13,34). For Results every antigen, we computed bootstrap values using 10,000 replicates and averaged. We performed analyses with R Bat Species and Sampling version 3.3.2 software ( We We analyzed blood samples from 4,022 wild bats from 21 considered a blood sample reactive if the MFI of the reac- different regions in Cameroon (n = 10), Guinea (n = 8), and tion was above the cutoff. We defined Ebola virus antibody the DRC (n = 3) (Figure 1; Table 1). To increase species positivity as reactivity to glycoprotein and nucleoprotein of diversity, we captured bats in multiple ecologic settings: for - the same lineage, as was done in our previous study (28). ests (49%), open fields (10%), villages (29%), plantations (7%), and urban areas (5%). For 1,470 (36.5%) samples, Nucleic Acid Extraction and PCR Screening for species identification in the field was confirmed by sequence Ebola Virus RNA analysis. At each site, >1 sample was confirmed per sampling We extracted total DNA and RNA from dried blood spots as date, capture method, and morphologic description. For the previously described using Nuclisens (bioMerieux, Marcy- remaining samples, species identification was extrapolated l’Etoile, France) or m2000sp methods (Abbott Molecular by combining molecular and morphologic data, including Inc., Des Plaines, IL, USA), which are known for a high per- photographs whenever available. For some insectivorous bat formance recovering nucleic acids from dried blood spots 2230 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Ebola Virus in Frugivorous and Insectivorous Bats Figure 1. Study sites for bat blood sample collection for Ebola virus serology, Guinea, Cameroon, and the Democratic Republic of the Congo, 2015– 2017. Yellow dots indicate sampling sites for bats in our study, and green dots indicate sampling sites in previously published studies. Dark red shading indicates highest and light yellow lowest risk for Ebola virus spillover events. Study sites are numbered: 1, Koundara; 2, Conakry; 3, Kindia; 4, Mamou; 5, Kankan; 6, Gueckedou; 7, Macenta; 8, Nzerekore; 9, Mbam Djerem; 10, Libellengoi Sud; 11, Yaoundé; 12, Ekom; 13, North Dja; 14, Bipindi; 15, Mbalmayo; 16, Djoum; 17, Mambele; 18, Campo M’an; 19, Boma; 20, Kimpese; 21, Zongo. Countries with reported index Ebola cases and countries without such cases but deemed at risk are indicated. Map of Africa adapted from Pigott et al. (5) ( licenses/by/4.0/) by adding locations of collection sites. families (Miniopteridae, Molossidae, Nycteridae, Rhinolo- by the mean + 4×SD method, 274 (6.8%) for the change- phidae), identification was possible only at the genus level; point method, 175 (4.4%) for the binomial method, and 457 for some Molossidae bats, we could not distinguish between (11.4%) for the exponential method. Blood samples fre - Mops and Chaerephon genera because of the lack of se- quently reacted with glycoprotein antigens; samples reacted quences in GenBank (Table 2). For 87 (2.16%) samples, spe - most with Zaire and Sudan Ebola virus antigens and least cies identification was not possible because incomplete data with Reston (Table 3). Simultaneous reactivity to >1 anti- were recorded in the field, and available biologic materials gen (i.e., glycoprotein, nucleoprotein, viral protein 40) from were insufficient for molecular confirmation. We collected the same virus lineage was rare. Simultaneous reactivity to samples from 1,736 (43.2%) frugivorous bats (family Ptero- the same antigen from different virus lineages was frequent; podidae) of 12 species and 2,199 (54.7%) insectivorous bats 32.3%–76.7% of blood samples were reactive to glycopro - (7 families) of >27 species. The insectivorous bat families tein from >2 Ebola virus species, 18.4%–34.0% to viral pro- sampled, in order of decreasing frequency, were Hipposid- tein 40, and 1.5%–4.4% to nucleoprotein (online Technical eridae (31.9%), Molossidae (13.4%), Miniopteridae (5.8%), Appendix Table 2). When using the criterion simultaneous Rhinolophidae (2.1%), Vespertilionidae (0.8%), Nycteridae presence of antibodies to nucleoprotein and glycoprotein, the (0.5%), and Emballonuridae (0.12%). Overall, 54.7% of antibody positivity for Zaire or Sudan Ebola virus antibodies bats were female and 43.8% were male; for 1.5% (n = 60) of was generally <1% for all bats tested, regardless of cutoff bats, sex was unknown. Most (77.9%) bats were adults, and method, and was lower among insectivorous than frugivo- 9.6% were juveniles; for 12.5% (n = 502) of bats, age could rous bats: 0.05%–0.27% (insectivorous) and 0.06%–1.79% not be determined or was not recorded. (frugivorous) for Zaire Ebola virus versus 0%–0.09% (in- sectivorous) and 0%–1.61% (frugivorous) for Sudan Ebola Bats Antibodies against Different Ebola Virus Antigens virus (Table 3; Figure 2). Three samples were positive for We tested all samples for Ebola virus antibodies. The num- Zaire and Sudan Ebola viruses, but only by less stringent cut- ber of samples reacting with >1 antigen was 734 (18.2%) off methods (i.e., mean + 4 ×SD). Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2231 RESEARCH Table 2. Bat species sampled for Ebola virus serology, Guinea, Cameroon, and the DRC, 2015–2017* Family Species DRC, no. Cameroon, no. Guinea, no. Total, no. Emballonuridae Coleura afra 0 5 0 5 Hipposideridae Hipposideros abae 0 0 37 37 H. beatus 0 4 0 4 H. cyclops 0 14 0 14 H. fuliginosus 0 2 0 2 H. gigas 2 9 2 13 H. jonesi 0 1 12 13 H. ruber/caffer 127 807 237 1,171 Hipposideros sp. 28 0 0 28 Subtotal 157 837 288 1,282 Miniopteridae Miniopterus sp. 205 0 27 232 Molossidae Chaerephon sp. 0 0 44 44 Mops condylurus 0 0 110 110 Mops sp. 0 256 0 256 Mops/Chaerephon sp. 0 8 120 128 Subtotal 0 264 274 538 Nycteridae Nycteris sp. 0 7 15 22 Rhinolophidae Rhinolophus alcyone 0 16 0 16 R. darlingii 3 0 0 3 R. fumigatus 0 0 19 19 R. landeri 0 0 6 6 Rhinolophus sp. 3 38 1 42 Subtotal 6 54 26 86 Vespertilionidae Glauconycteris variegata 0 3 0 3 Kerivoula sp. 0 1 0 1 Myotis bocagii 0 3 0 3 Neoromicia sp. 0 5 0 5 Scotophilus leucogaster 0 0 15 15 S. nigrita 0 0 1 1 S. nux 0 6 0 6 Subtotal 0 18 16 34 Pteropodidae Eidolon helvum 305 158 17 480 Epomophorus gambianus 0 0 191 191 Epomophorus wahlbergi 0 16 0 16 Epomops buettikoferi 0 0 4 4 Epomops franqueti 20 256 0 276 Hypsignathus monstrosus 1 176 8 185 Lissonycteris angolensis 22 30 32 84 Megaloglossus woermanni 1 19 0 20 Micropteropus pusillus 44 2 18 64 Myonycteris torquata 35 21 0 56 Rousettus aegyptiacus 0 131 228 359 Scotonycteris zenkeri 0 1 0 1 Subtotal 428 810 498 1,736 Inderminate species 34 23 30 87 Total 830 2,018 1,174 4,022 *DRC, the Democratic Republic of the Congo. Zaire and Sudan Ebola Virus Reactivity of Different Zaire Ebola virus and 1.1%–4.3% for Sudan Ebola virus Bat Species in H. monstrosus bat samples, and 0.6%–2.5% for Zaire We estimated specific reactivity to Zaire and Sudan Ebo - Ebola virus and 0.8%–1.4% for Sudan Ebola virus in R. la viruses by bat species. We did not include Bundibugyo aegyptiacus bat samples. We observed 2.4% Zaire Ebola and Reston because recombinant nucleoproteins were virus–seropositive samples for Lissonycteris angolensis not available. Among insectivorous bats, only blood bats and 0.5% for Epomophorus sp. bats, but only by less samples from Mops sp. bats (1–6/494) were positive for stringent cutoff methods. One sample from M. pusillus Zaire or Sudan Ebola virus antibodies (Table 4). Among bats was seropositive for Sudan Ebola virus. No samples frugivorous bats, samples from E. helvum, H. monstro- from E. franqueti or M. torquata bats were reactive with sus, and R. aegyptiacus bats had the highest reactivity. any Ebola virus antigens. Samples from the 1 Scotonyc- We observed Zaire and Sudan Ebola virus seropositivity teris zenkeri bat and 20 Megaloglossus woermanni bats in these 3 species with almost all cutoff methods: 0.2%– were seronegative. Overall, Zaire or Sudan Ebola virus 3.3% for Zaire Ebola virus and 1.0%–2.9% for Sudan antibodies were observed in 7 (1 insectivorous and 6 fru- Ebola virus in E. helvum bat samples, 0.5%–1.6% for givorous) bat species. 2232 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Ebola Virus in Frugivorous and Insectivorous Bats Table 3. Blood samples from bats reactive with Ebola virus antigens in Luminex assay, by antigen, bat type, and statistical method used to determine cutoff, Guinea, Cameroon, and the Democratic Republic of the Congo, 2015–2017* Statistical method, no. (%) Estimated Ebola virus species, antigen Bat type Mean + 4SD Change point Binomial Exponential range, % Zaire NP Frugivorous 57 (3.28) 8 (0.46) 24 (1.38) 51 (2.94) 0.46–3.28 NP Insectivorous 15 (0.68) 1 (0.05) 6 (0.27) 15 (0.68) 0.05–0.68 NP Total 72 (1.79) 9 (0.22) 30 (0.75) 66 (1.64) 0.23–1.79 GP-K Frugivorous 365 (21.03) 141 (8.12) 20 (1.15) 113 (6.51) 1.15–21.03 GP-K Insectivorous 73 (3.32) 18 (0.82) 2 (0.09) 12 (0.55) 0.09–3.32 GP-K Total 440 (10.94) 160 (3.98) 22 (0.55) 125 (3.11) 0.55–10.94 GP-M Frugivorous 226 (13.02) 128 (7.37) 16 (0.92) 103 (5.93) 0.92–13.02 GP-M Insectivorous 31 (1.41) 14 (0.64) 2 (0.09) 12 (0.55) 0.09–1.41 GP-M Total 259 (6.44) 143 (3.56) 18 (0.45) 115 (2.86) 0.45–6.44 VP Frugivorous 55 (3.17) 8 (0.46) 24 (1.38) 44 (2.53) 0.46–3.17 VP Insectivorous 19 (0.86) 5 (0.23) 6 (0.27) 14 (0.64) 0.23–0.86 VP Total 75 (1.86) 14 (0.35) 30 (0.75) 59 (1.47) 0.35–1.86 NP + GP Frugivorous 31 (1.79) 31 (1.79) 1 (0.06) 7 (0.40) 0.06–1.79 NP + GP Insectivorous 6 (0.27) 6 (0.27) 1 (0.05) 1 (0.05) 0.05–0.27 NP + GP Total 37 (0.92) 37 (0.92) 2 (0.05) 8 (0.20) 0.05–0.92 Sudan NP Frugivorous 71 (4.09) 15 (0.86) 34 (1.96) 77 (4.44) 0.86–4.44 NP Insectivorous 12 (0.55) 1 (0.05) 5 (0.23) 18 (0.82) 0.05–0.82 NP Total 84 (2.09) 17 (0.42) 39 (0.97) 96 (2.39) 0.42–2.39 GP Frugivorous 459 (26.44) 147 (8.47) 17 (0.98) 121 (6.97) 0.98–26.44 GP Insectivorous 49 (2.23) 6 (0.27) 1 (0.05) 1 (0.05) 0.05–2.23 GP Total 509 (12.66) 154 (3.83) 18 (0.45) 125 (3.11) 0.45–12.66 VP Frugivorous 102 (5.88) 20 (1.15) 28 (1.61) 61 (3.51) 1.15–5.88 VP Insectivorous 19 (0.86) 4 (0.18) 6 (0.27) 18 (0.82) 0.18–0.86 VP Total 121 (3.01) 24 (0.60) 34 (0.85) 80 (1.99) 0.60–3.01 NP + GP Frugivorous 28 (1.61) 28 (1.61) 0 10 (0.58) 0–1.61 NP + GP Insectivorous 2 (0.09) 2 (0.09) 0 0 0–0.09 NP + GP Total 30 (0.75) 30 (0.75) 0 10 (0.25) 0–0.75 Bundibugyo GP Frugivorous 301 (17.34) 59 (3.40) 0 93 (5.36) 0–17.34 GP Insectivorous 58 (2.64) 8 (0.36) 5 (0.23) 13 (0.59) 0.23–2.64 GP Total 361 (8.98) 68 (1.69) 22 (0.55) 107 (2.66) 0.55–8.98 VP Frugivorous 9 (0.52) 7 (0.40) 12 (0.69) 37 (2.13) 0.40–2.14 VP Insectivorous 0 1 (0.05) 8 (0.36) 20 (0.91) 0–0.91 VP Total 9 (0.22) 8 (0.20) 20 (0.5) 57 (1.42) 0.20–1.42 Reston GP Frugivorous 17 (0.98) 28 (1.61) 26 (1.50) 61 (3.51) 0.98–3.51 GP Insectivorous 3 (0.14) 10 (0.45) 10 (0.45) 29 (1.32) 0.14–1.32 GP Total 20 (0.50) 38 (0.94) 36 (0.90) 90 (2.24) 0.50–2.24 *VP refers to viral protein 40 of Ebola virus. Results are presented for frugivorous (n = 1,736), insectivorous (n = 2,199), and total (n = 4,022) bats. GP, glycoprotein; K, Kissoudougou strain; M, Mayinga strain; NP, nucleoprotein; VP, viral protein. Comparison of Zaire Ebola Virus Seroprevalence in Casinycteris, Megaloglossus, Nanonycteris, and Scotonyc- Bats from Africa across Studies teris, but only a limited number of samples (n = 152) have For comparison, we compiled data regarding Zaire Ebola been tested. Overall, blood samples from 8 frugivorous bat virus serology in bats of known species from previous stud- species have been found reactive with Zaire Ebola virus an- ies (n = 4,493) and this study (n = 3,935; 46.7%) (Tables 5, tigens. Blood samples from E. helvum, H. monstrosus, and 6). Data were available for 3,023 insectivorous bats of ≈30 R. aegyptiacus bats from several countries across Africa species from 7 different families; 2,199 (72.7%) were from have been reported to be seropositive. Reactivity has been this study (Table 5). Insectivorous bat samples originated observed with samples from E. gambianus bats in Ghana from Guinea, Cameroon, the DRC, and Gabon. Zaire Ebola (10.8%) and Guinea. Reactivity was observed with large virus reactivity has been observed only in M. condylurus sample sets from E. franqueti bats derived from Gabon bat samples from Gabon and Mops sp. bat samples from and the Congo and a small sample set from Ghana but not Cameroon. Data were available for 5,405 frugivorous bats Guinea, Cameroon, or the DRC. M. pusillus and M. torqua- of 17 species from 12 genera from West (Guinea, Ghana), ta bats tested positive for Zaire Ebola virus antibodies in West Central (Cameroon, Gabon, the Congo, the DRC), studies in which large sample sets were collected. Among and East (Zambia) Africa (Table 6). No Zaire Ebola virus L. angolensis bat samples, only those from Cameroon have reactivity has been seen in blood samples from bat species tested positive for antibodies. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2233 RESEARCH Figure 2. Bat blood samples reactive to Ebola virus antigens, by statistical method used to determine cutoff, Guinea, Cameroon, and the Democratic Republic of the Congo, 2015–2017. Samples from frugivorous bats (n = 1,736) and insectivorous bats (n = 2,199) were tested by Luminex assay with GP, NP, and VP of the Zaire and Sudan lineages; GP and VP of the Bundibugyo lineage; and GP of the Reston lineage. GP, glycoprotein; K, Kissoudougou strain; M, Mayinga strain; NP, nucleoprotein; VP, viral protein 40. RT-PCR Screening for Zaire Ebola Virus RNA insectivorous bats, we observed Zaire and Sudan Ebola vi- We screened 665 samples from the DRC (n = 193), rus antibodies only in Mops sp. bats, an observation that has Cameroon (n = 399), and Guinea (n = 73) by RT-PCR previously been observed (13). We provided information for the presence of Zaire Ebola virus RNA. Of the 294 on insectivorous Miniopterus and Rhinolophus bats and ex- samples originating from bats previously documented tended knowledge on Mops and Hipposideros bats; all 1,200 to carry Zaire Ebola virus RNA (6) (i.e., H. monstrosus Hipposideros samples were seronegative. We confirmed the [132 from Cameroon, 1 from the DRC], M. torquata [20 presence of Zaire Ebola virus antibodies in only 1 of 3 fru- from Cameroon, 25 from the DRC], and E. franqueti [116 givorous species in which Zaire Ebola virus RNA has been from Cameroon]), all were negative for Zaire Ebola virus reported, that is, in H. monstrosus but not E. franqueti or RNA. Of the 371 samples from bat species E. helvum (58 M. torquata bats (6). However, this result might have been from Cameroon, 165 from the DRC, 3 from Guinea), L. influenced by sample size, test used, and interpretation crite - angolensis (8 from Cameroon, 4 from Guinea), M. pu- ria. We confirmed antibodies in E. helvum bats and showed sillus (2 from the DRC, 1 from Guinea), R. aegyptiacus that Zaire Ebola virus antibodies are widespread among this (45 from Cameroon, 40 from Guinea), E. gambianus (25 species across Africa: Ghana and Zambia, and with our data, from Guinea), and Mops sp. (20 from Cameroon), all were also Cameroon, Guinea, and the DRC (13,14,16). We con- negative for Zaire Ebola virus RNA. firmed antibodies in R. aegyptiacus bats from Cameroon and Guinea, in agreement with previous findings in these bats Discussion from the Congo and Gabon (13). For E. gambianus bats To clarify the role of bats in Ebola virus ecology and iden- from Ghana, we also observed Zaire Ebola virus reactivity tify where the virus circulates between outbreaks, we tested of samples from this species in Guinea (15). In contrast with >4,000 bats, almost doubling the total number of samples a previous study, we observed Sudan Ebola virus antibodies tested in all previous studies in Africa (5–7,42). We provided (not Zaire Ebola virus antibodies) in M. pusillus bats (13). data on bats from Cameroon, added to the existing data on We also identified Zaire Ebola virus antibodies in L. ango- bats from Guinea and the DRC, and substantially increased lensis bats from Cameroon, although only when using less the data available on insectivorous bats. We tested samples stringent cutoff calculations. with the same assay, enabling comparison across species and When combining data from previous Zaire Ebola vi- countries. We used different statistical methods to determine rus seroprevalence studies in bats with data from our study, positive sample numbers and expressed the proportion of only 1 insectivorous bat species (Mops sp.) and 8 frugivo- reactive samples as a range on the basis of the different cut - rous bat species (E. helvum, E. gambianus, E. franqueti, H. off values proposed by those methods. As has been done in monstrosus, L. angolensis, M. pusillus, M. torquata, R. ae- studies of human Zaire Ebola virus survivors (28,43), we de- gyptiacus) exhibited Zaire Ebola virus antibodies (13–16). fined Zaire and Sudan Ebola virus positivity as the presence As seen in bat samples from Zambia, we observed in this of antibodies to both nucleoprotein and glycoprotein. As study Sudan Ebola virus antibodies in E. helvum bats from such, we estimated that 2–37 (0.05%–0.92%) bats were se - Guinea, Cameroon, and the DRC, suggesting that Zaire and ropositive for Zaire Ebola virus and 0–30 (0%–0.75%) bats Sudan Ebola viruses co-circulate and could be widespread were seropositive for Sudan Ebola virus (Table 3). Among among this species. However, only 1 other study has tested 2234 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Ebola Virus in Frugivorous and Insectivorous Bats Table 4. Blood samples from bats reactive with both nucleoprotein and glycoprotein of Zaire or Sudan Ebola virus, by statistical method used to determine cutoff, Guinea, Cameroon, and the Democratic Republic of the Congo, 2015–2017 Statistical method Ebola No. virus Mean + 4SD Change-point Binomial Exponential Bat family, genus tested species No. % (95% CI) No. % (95% CI) No. % (95% CI) No. % (95% CI) Hipposideridae Hipposideros sp. 1,282 Zaire 0 0 (0–0.3) 0 0 (0–0.3) 0 0 (0–0.3) 0 0 (0–0.3) 1,282 Sudan 0 0 (0–0.3) 0 0 (0–0.3) 0 0 (0–0.3) 0 0 (0–0.3) Miniopteridae Miniopterus sp. 232 Zaire 0 0 (0–1.6) 0 0 (0–1.6) 0 0 (0–1.6) 0 0 (0–1.6) 232 Sudan 0 0 (0–1.6) 0 0 (0–1.6) 0 0 (0–1.6) 0 0 (0–1.6) Molossidae Chaerephon sp. 44 Zaire 0 0 (0–8.0) 0 0 (0–8.0) 0 0 (0–8.0) 0 0 (0–8.0) 44 Sudan 0 0 (0–8.0) 0 0 (0–8.0) 0 0 (0–8.0) 0 0 (0–8.0) Mops sp. 494 Zaire 6 1.2 (0.6–2.6) 6 1.2 (0.6–2.6) 1 0.2 (0.03–1.1) 1 0.2 (0.03–1.1) 494 Sudan 2 0.4 (0.1–1.5) 2 0.4 (0.1–1.5) 0 0 (0–0.8) 0 0 (0–0.8) Nycteridae Nycteris sp 22 Zaire 0 0 (0–14.9) 0 0 (0–14.9) 0 0 (0–14.9) 0 0 (0–14.9) 22 Sudan 0 0 (0–14.9) 0 0 (0–14.9) 0 0 (0–14.9) 0 0 (0–14.9) Rhinolophidae Rhinolophus sp. 86 Zaire 0 0 (0–4.3) 0 0 (0–4.3) 0 0 (0–4.3) 0 0 (0–4.3) 86 Sudan 0 0 (0–4.3) 0 0 (0–4.3) 0 0 (0–4.3) 0 0 (0–4.3) Vespertilionidae Glauconycteris sp.* 3 Zaire 0 0 0 0 3 Sudan 0 0 0 0 Kerivoula sp.* 1 Zaire 0 0 0 0 1 Sudan 0 0 0 0 Myotis bocagii* 3 Zaire 0 0 0 0 3 Sudan 0 0 0 0 Neoromicia sp.* 5 Zaire 0 0 0 0 5 Sudan 0 0 0 0 Scotophilus sp. 22 Zaire 0 0 (0–14.9) 0 0 (0–14.9) 0 0 (0–14.9) 0 0 (0–14.9) 22 Sudan 0 0 (0–14.9) 0 0 (0–14.9) 0 0 (0–14.9) 0 0 (0–14.9) Pteropodidae Eidolon helvum 480 Zaire 16 3.3 (2.1–5.4) 16 3.3 (2.1–5.4) 1 0.2 (0–1.2) 4 0.8 (0.3–2.1) 480 Sudan 14 2.9 (1.7–4.8) 14 2.9 (1.7–4.8) 0 0 (0–0.8) 5 1.0 (0.4–2.4) Epomophorus sp. 207 Zaire 1 0.5 (0.08–2.7) 1 0.5 (0.08–2.7) 0 0 (0–1.4) 0 0 (0–1.8) 207 Sudan 0 0 (0–1.8) 0 0 (0–1.8) 0 0 (0–1.8) 0 0 (0–1.8) Epomops sp. 280 Zaire 0 0 (0–1.4) 0 0 (0–1.4) 0 0 (0–1.4) 0 0 (0–1.4) 280 Sudan 0 0 (0–1.4) 0 0 (0–1.4) 0 0 (0–1.4) 0 0 (0–1.4) Hypsignathus 185 Zaire 3 1.6 (0.6–4.7) 3 1.6 (0.6–4.7) 0 0 (0–2.0) 1 0.5 (0.05–3.0) monstrosus 185 Sudan 8 4.3 (2.2–8.3) 8 4.3 (2.2–8.3) 3 1.6 (0.6–4.7) 2 1.1(0.3–3.9) Lissonycteris 84 Zaire 2 2.4 (0.7–8.3) 2 2.4 (0.7–8.3) 0 0 (0–4.4) 0 0 (0–4.4) angolensis 84 Sudan 0 0 (0–4.4) 0 0 (0–4.4) 0 0 (0–4.4) 0 0 (0–4.4) Megaloglossus 20 Zaire 0 0 (0–16.1) 0 0 (0–16.1) 0 0 (0–16.1) 0 0 (0–16.1) woermanni 20 Sudan 0 0 (0–16.1) 0 0 (0–16.1) 0 0 (0–16.1) 0 0 (0–16.1) Micropteropus 64 Zaire 0 0 (0–5.7) 0 0 (0–5.7) 0 0 (0–5.7) 0 0 (0–5.7) pusillus 64 Sudan 1 1.6 (0.3–8.3) 1 1.6 (0.3–8.3) 0 0 (0–5.7) 0 0 (0–5.7) Myonycteris 56 Zaire 0 0 (0–6.4) 0 0 (0–6.4) 0 0 (0–6.4) 0 0 (0–6.4) torquata 56 Sudan 0 0 (0–6.4) 0 0 (0–6.4) 0 0 (0–6.4) 0 0 (0–6.4) Rousettus 359 Zaire 9 2.5 (1.3–4.7) 9 2.5 (1.3–4.7) 0 0 (0–1.1) 2 0.6 (0.2–2.0) aegyptiacus 359 Sudan 5 1.4 (0.6–3.2) 5 1.4 (0.6–3.2) 0 0 (0–1.1) 3 0.8 (0.3–2.4) Scotonycteris 1 Zaire 0 0 0 0 zenkeri* 1 Sudan 0 0 0 0 *Percentages were not calculated because the number of samples collected was too low. for Ebola viruses other than Zaire Ebola virus in E. helvum might reflect previous acute infection with viral clear - bats (16). In our study, we also observed Sudan Ebola virus ance. Unlike inoculations with Marburg virus (44–46), antibodies in Mops sp., H. monstrosus, and R. aegyptiacus experimental inoculation of R. aegyptiacus bats with bats in Cameroon. Almost all samples were positive for ei- Zaire Ebola virus leads to antibody development but in- ther Zaire or Sudan Ebola virus but not for both. frequent or absent detection of viral RNA or shedding Despite the presence of Ebola virus antibodies, the (44,47). R. aegyptiacus bats are therefore able to clear role of bats as reservoir species remains unclear because Zaire Ebola virus after a short infectious period with- viral RNA detection is rare. In only 1 study Zaire Ebola vi- out viral shedding and with little or no transmission. No rus RNA was amplified in a few bats (6). Thus, antibodies antibodies or viral RNA were detected in noninoculated Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2235 RESEARCH Table 5. Zaire Ebola virus antibodies in insectivorous bats from our research, Guinea, Cameroon, and the DRC, 2015–2017, and other published studies* Year of study No. No. (%) Total, no. Family Species Country (reference) Test tested positive† positive/tested (%)† Emballonuridae Coleura afra Cameroon 2015–2017‡ Luminex 5 0–0 (0–0) 0/14 (0) Saccolaimus peli DRC 1979–1980 (26) IFA 9 0 (0) Hipposideridae Hipposideros sp. DRC 2015–2017‡ Luminex 157 0–0 (0–0) 0/1,395 (0) Hipposideros sp. Cameroon 2015–2017‡ Luminex 837 0–0 (0–0) Hipposideros sp. DRC 1979–1980 (26) IFA 69 0 (0) Hipposideros sp. Guinea 2015–2017‡ Luminex 288 0–0 (0–0) Hipposideros sp. Guinea 2014 (11) ELISA 44 0 (0) Miniopteridae Miniopterus sp. Guinea 2015–2017‡ Luminex 27 0–0 (0–0) 0/234 (0) Miniopterus sp. DRC 2015–2017‡ Luminex 205 0–0 (0–0) M. minor DRC 1995 (27) ELISA 2 0 (0) Molossidae Chaerephon sp. Guinea 2015–2017‡ Luminex 44 0–0 (0–0) 0/401 (0) C. pumilus Guinea 2014 (11) ELISA 1 0 (0) C. ansorgei DRC 1995 (27) ELISA 120 0 (0) C. major DRC 1979–1980 (26) IFA 26 0 (0) C. pumilus DRC 1995 (27) Elisa 210 0 (0) Mops sp. Guinea 2015–2017‡ Luminex 230 0–0 (0–0) 4–9/705 (0.6–1.3) Mops sp. Cameroon 2015–2017‡ Luminex 264 1–6 (0.4–2.3) Mops sp. DRC 1979–1980 (26) IFA 158 0 (0) Mops sp. DRC 1995 (27) ELISA 28 0 (0) Mops condylurus Gabon 2003–2008 (13) ELISA 24 3 (12.5) M. condylurus Guinea 2014 (11) ELISA 1 0 (0) Myopterus whitleyi DRC 1995 (27) ELISA 2 0 (0) Nycteridae Nycteris sp. Guinea 2015–2017‡ Luminex 15 0–0 (0–0) 0/43 (0) Nycteris sp. Guinea 2014 (11) ELISA 6 0 (0) Nycteris sp. Cameroon 2015–2017‡ Luminex 7 0–0 (0–0) Nycteris sp. DRC 1979–1980 (26) IFA 14 0 (0) Nycteris hispida DRC 1995 (27) ELISA 1 0 (0) Rhinolophidae Rhinolophus sp. Guinea 2015–2017‡ Luminex 26 0–0 (0–0) 0/86 (0) Rhinolophus sp. DRC 2015–2017‡ Luminex 6 0–0 (0–0) Rhinolophus sp. Cameroon 2015–2017‡ Luminex 54 0–0 (0–0) Vespertilionidae Glauconycteris variegata Cameroon 2015–2017‡ Luminex 3 0–0 (0–0) 0/143 (0) Chalinolobus sp. DRC 1979–1980 (26) IFA 15 0 (0) Eptesicus sp. DRC 1979–1980 (26) IFA 22 0 (0) Eptesicus tenuipinnis DRC 1995 (27) ELISA 1 0 (0) Kerivoula sp. Guinea 2014 (11) ELISA 1 0 (0) Kerivoula sp. Cameroon 2015–2017‡ Luminex 1 0–0 (0–0) Myotis bocagii Cameroon 2015–2017‡ Luminex 3 0–0 (0–0) M. bocagii DRC 1995 (27) ELISA 22 0 (0) M. bocagii DRC 1979–1980 (26) IFA 17 0 (0) Neoromicia sp. Cameroon 2015–2017‡ Luminex 5 0–0 (0–0) Pipistrellus nanus DRC 1995 (27) ELISA 2 0 (0) Scotophilus nux Cameroon 2015–2017‡ Luminex 6 0–0 (0–0) Scotophilus leucogaster Guinea 2015–2017‡ Luminex 15 0–0 (0–0) Scotophilus nigrita Guinea 2015–2017‡ Luminex 1 0–0 (0–0) Scotophilus dinganii DRC 1995 (27) ELISA 19 0 (0) Scotophilus sp. DRC 1979–1980 (26) IFA 10 0 (0) Total 4–9/3,023 (0.13–0.30) *DRC, the Democratic Republic of the Congo; IFA, immunofluorescence assay. †For data from cited studies, the number of positive samples reported in the original study is indicated. For our results, we show the range in the number of samples simultaneously reactive with glycoprotein and nucleoprotein of Zaire Ebola virus on the basis of 4 different statistical methods used to determine cutoff values. ‡This study. bats housed with experimentally Zaire Ebola virus– virus were shown to develop antibodies that protect against infected bats (44). Whether this low level of infectiousness reinfection (49). Long-term survival with Zaire Ebola vi- also occurs for other bat species that carry Ebola virus an- rus antibodies has been reported with E. helvum bats from tibodies remains to be determined. Zaire Ebola virus was Ghana but without information on protection (14). Among experimentally inoculated in other bat species (M. condylu- insectivorous bats, the presence of Ebola virus antibodies rus, Chaerephon pumilus, and Epomophorus wahlbergi) in only Mops sp. is striking, suggesting higher exposure or in only 1 study; virus replication was seen in all species, susceptibility compared with other insectivorous bats. and fecal shedding was seen in E. wahlbergi bats (48). R. In conclusion, we demonstrated higher rates of Eb- aegyptiacus bats experimentally infected with Marburg ola virus antibodies in frugivorous than in insectivorous 2236 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Ebola Virus in Frugivorous and Insectivorous Bats Table 6. Zaire Ebola virus antibodies in frugivorous (Pteropodidae family) bats from our research, Guinea, Cameroon, and the DRC, 2015–2017, and published studies* Year of study No. No. (%) Species Country (reference) Test tested positive Total, no. positive/tested (%) Casinycteris ophiodon Guinea 2014 (11) ELISA 1 0 0/20 Casinycteris argynnis Gabon, Congo 2003–2008 (13) ELISA 18 0 C. argynnis DRC 1995 (27) ELISA 1 0 Eidolon helvum† Guinea 2014 (11) ELISA 6 0 21–36/1,551 (1.4–2.3) Guinea 2015–2017‡ Luminex 17 0–3 (0–17.6) Ghana 2008 (14) IFA 262 1 (0.39) Cameroon 2015–2017‡ Luminex 158 1–9 (0.6–5.7) Gabon, Congo 2003–2008 (13) ELISA 49 0 DRC 1979–1980 (26) IFA 6 0 DRC 2015–2017‡ Luminex 305 0–4 (0–1.3) Zambia 2006–2013 (16) ELISA 748 19 (2.55) Epomophorus gambianus Guinea 2015–2017‡ Luminex 191 0–1 (0–0.5) 4–5/244 (1.6–2.0) Ghana 2007 (15) ELISA 37 4 (10.82) Epomophorus wahlbergi Cameroon 2015–2017‡ Luminex 16 0–0 (0–0) Epomops buettikoferi Guinea 2014 (11) ELISA 17 0 47/1,269 (3.7) Guinea 2015–2017‡ Luminex 4 0–0 (0–0) Epomops franqueti Ghana 2007 (15) ELISA 27 3 (11.2) Cameroon 2015–2017‡ Luminex 256 0–0 (0–0) Gabon, Congo 2001–2005 (6) ELISA 117 8 (6.8) Gabon, Congo 2003–2008 (13) ELISA 805 36 (4.5) DRC 2015–2017‡ Luminex 20 0–0 (0–0) DRC 1979–1980 (26) IFA 21 0 DRC 1995 (27) ELISA 2 0 Hypsygnathus monstrosus Guinea 2015–2017‡ Luminex 8 0–0 (0–0) 15–18/347 (4.3–5.2) Guinea 2014 (13) ELISA 1 0 Ghana 2008 (14) IFA 3 0 Ghana 2007 (15) ELISA 16 2 (12.5) Cameroon 2015–2017‡ Luminex 176 0–3 (0–1.7) Gabon, Congo 2001–2005 (6) ELISA 17 4 (23.5) Gabon, Congo 2003–2008 (13) ELISA 125 9 (7.2) DRC 2015–2017‡ Luminex 1 0–0 (0–0) Lissonycteris angolensis Guinea 2014 (11) ELISA 45 0 0–2/129 (0–1.6) Guinea 2015–2017‡ Luminex 32 0–0 (0–0) DRC 2015–2017‡ Luminex 22 0–0 (0–0) Cameroon 2015–2017‡ Luminex 30 0–2 (0–6.7) Megaloglossus azagnyi Guinea 2014 (11) ELISA 3 0 0/110 Megaloglossus woermanni Cameroon 2015–2017‡ Luminex 19 0–0 (0–0) Gabon, Congo 2003–2008 (13) ELISA 49 0 DRC 2015–2017‡ Luminex 1 0–0 (0–0) DRC 1995 (27) ELISA 38 0 Micropteropus pusillus Guinea 2015–2017‡ Luminex 18 0–0 (0–0) 4/339 (1.2) Cameroon 2015–2017‡ Luminex 2 0–0 (0–0) Gabon, Congo 2003–2008 (13) ELISA 197 4 (2.04) DRC 2015–2017‡ Luminex 44 0–0 (0–0) DRC 1995 (27) ELISA 78 0 Myonycteris leptodon Guinea 2014 (11) ELISA 21 0 23–27/708 (3.2–3.8) Myonycteris torquata Cameroon 2015–2017‡ Luminex 21 0–0 (0–0) Gabon, Congo 2001–2005 (6) ELISA 58 4 (6.9) Gabon, Congo 2003–2008 (13) ELISA 573 19 (3.32) DRC 2015–2017‡ Luminex 35 0–0 (0–0) Nanonycteris veldkampii Guinea 2014 (11) ELISA 17 0 0/21 Ghana 2007 (15) ELISA 4 0 Rousettus aegyptiacus Guinea 2015–2017‡ Luminex 228 0–1 (0–0.4) 24–33/666 (3.6–5.0) Cameroon 2015–2017‡ Luminex 131 0–8 (0–6.1) Gabon, Congo 2003–2008 (13) ELISA 307 24 (7.8) Scotonycteris zenkeri Cameroon 2015–2017‡ Luminex 1 0–0 (0–0) 0–0/1 (0–0) Total 138–172/5,405 (2.55–3.18) *DRC, the Democratic Republic of the Congo; IFA, immunofluorescence assay. †For cited studies, the number of positive samples reported in the original study is indicated. For our results, we show the range in the number of samples simultaneously reactive with glycoprotein and nucleoprotein of Zaire Ebola virus on the basis of 4 different statistical methods used to determine cutoff values. ‡This study. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2237 RESEARCH bats. The total number of frugivorous species shown to About the Author be Zaire Ebola virus seropositive has increased to 8, and Drs. Mbala Kingebeni and Keita are researchers from 1 insectivorous bat species (Mops sp.) was confirmed to ASTRE of Centre de coopération internationale en recherche be seropositive. Zaire and Sudan Ebola viruses circulate agronomique pour le développement, Institut national de la in different species across Africa, with potential co-cir - recherche agronomique and University of Montpellier, culation of both viruses in some species. Although we Montpellier, France. Their research interests include have data on >8,000 bats from >40 species, this sample characterization of hosts that harbor zoonotic pathogens. size is small, given the high numbers of bats that con- stitute colonies. This study illustrates the complexity of References tracking the animal reservoir of Ebola viruses, not only 1. Mylne A, Brady OJ, Huang Z, Pigott DM, Golding N, because sampling of wild bats without performing eu- Kraemer MU, et al. A comprehensive database of the geographic spread of past human Ebola outbreaks. Sci Data. 2014;1:140042. thanasia is difficult and time-consuming but also because of the absence of a reference standard for serologic tests. 2. Baize S, Pannetier D, Oestereich L, Rieger T, Koivogui L, To clarify the significance of Ebola virus antibodies, Magassouba N, et al. Emergence of Zaire Ebola virus disease documenting the extent to which viral RNA and shed- in Guinea. N Engl J Med. 2014;371:1418–25. 10.1056/NEJMoa1404505 ding can be detected in species with antibodies is crucial 3. Maganga GD, Kapetshi J, Berthet N, Kebela Ilunga B, for predicting and controlling the risk for new outbreaks. Kabange F, Mbala Kingebeni P, et al. Ebola virus disease in the Efforts must continue not only to sample bats but also Democratic Republic of Congo. N Engl J Med. 2014;371:2083–91. other animals to elucidate where the virus circulates 4. World Health Organization. Ebola outbreak Democratic in wildlife. Republic of the Congo 2017. 2017 Jul 2 [cited 2018 May 3]. Acknowledgments 5. Pigott DM, Millear AI, Earl L, Morozoff C, Han BA, Shearer FM, et al. Updates to the zoonotic niche map of Ebola virus disease in We thank the staffs from the Ministry of Health and Ministry of Africa. eLife. 2016;5:e16412. Environment and the national ethics committees from the DRC, 6. Leroy EM, Kumulungui B, Pourrut X, Rouquet P, Hassanin A, Cameroon, and Guinea for permission to perform this study. We Yaba P, et al. Fruit bats as reservoirs of Ebola virus. Nature. thank all the field staff from the DRC (Guy Midingi and Servet 2005;438:575–6. 7. Leendertz SA, Gogarten JF, Düx A, Calvignac-Spencer S, Kimbonza); Guinea (Souana Goumou, Mamadou Kalif Diallo, Leendertz FH. Assessing the evidence supporting fruit bats as the Pépé Justin Beavogui, Philippe Kolié, Michel Guilavogui); primary reservoirs for Ebola viruses. EcoHealth. 2016;13:18–25. and Cameroon (Innocent Ndong Bass, Aime Mebenga, Joseph Moudindo, Thomas Atemkem) for the collection of bat samples. 8. Kamins AO, Rowcliffe JM, Ntiamoa-Baidu Y, Cunningham AA, Wood JL, Restif O. Characteristics and risk perceptions of We thank the staffs of the National Institute of Biomedical Ghanaians potentially exposed to bat-borne zoonoses through Research (Kinshasa, the DRC), the Kongo Central Provincial bushmeat. EcoHealth. 2015;12:104–20. Government (Matadi, the DRC), and Projet PRESICA; Donald s10393-014-0977-0 Mbohli from Project Grand Singes; and the staff of the Institut 9. Leroy E, Gonzalez JP, Pourrut X. Ebolavirus and other filoviruses. Curr Top Microbiol Immunol. 2007;315:363–87. National de Santé Publique (Conakry, Guinea) for logistical 10.1007/978-3-540-70962-6_15 support in the field. We thank Seny Mane for his involvement 10. Leroy EM, Epelboin A, Mondonge V, Pourrut X, Gonzalez JP, and support in the implementation of this project and Daouda Muyembe-Tamfum JJ, et al. Human Ebola outbreak resulting from Bangoura for his constant support for this project and the direct exposure to fruit bats in Luebo, Democratic Republic of Congo, 2007. Vector Borne Zoonotic Dis. 2009;9:723–8. facilitation of field missions. We thank the veterinary staff from the Parc Zoologique de Montpellier and Wilhelma Zoo and 11. Marí Saéz A, Weiss S, Nowak K, Lapeyre V, Zimmermann F, Botanical Garden for providing control samples. Düx A, et al. Investigating the zoonotic origin of the West African Ebola epidemic. EMBO Mol Med. 2015;7:17–23. This work was supported in part by grants from Institut 10.15252/emmm.201404792 National de la Santé et de la Recherche Médicale, the Ebola 12. Pourrut X, Délicat A, Rollin PE, Ksiazek TG, Gonzalez JP, Leroy EM. Spatial and temporal patterns of Zaire ebolavirus Task Force, REACTing, EBO-SURSY project funded by the antibody prevalence in the possible reservoir bat species. European Union, Institut de Recherche pour le Développement J Infect Dis. 2007;196(Suppl 2):S176–83. (IRD), and Christophe Mérieux Prize 2015 (to J-.J.M.T.). 10.1086/520541 A.K.K. was supported by a fellowship from the IRD and the 13. Pourrut X, Souris M, Towner JS, Rollin PE, Nichol ST, Gonzalez JP, et al. Large serological survey showing cocirculation University of Montpellier (MUSE, ANR-16-IDEX-0006). of Ebola and Marburg viruses in Gabonese bat populations, and a C.-J.V.-A. was supported by a fellowship from IRD, Labex high seroprevalence of both viruses in Rousettus aegyptiacus. BMC EpiGenMed via the National Research Agency, Programme for Infect Dis. 2009;9:159. Future Investment (ANR-10-LABX-12-01), and the University 14. Hayman DT, Emmerich P, Yu M, Wang LF, Suu-Ire R, Fooks AR, et al. Long-term survival of an urban fruit bat seropositive for of Montpellier. 2238 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Ebola Virus in Frugivorous and Insectivorous Bats Ebola and Lagos bat viruses. PLoS One. 2010;5:e11978. pathogens in wildlife: assessing an appropriate cutoff for henipavirus assays in African bats. J Virol Methods. 2013;193:295– 15. Hayman DT, Yu M, Crameri G, Wang LF, Suu-Ire R, Wood JL, 303. et al. Ebola virus antibodies in fruit bats, Ghana, West Africa. 30. Gilbert AT, Fooks AR, Hayman DT, Horton DL, Müller T, Emerg Infect Dis. 2012;18:1207–9. Plowright R, et al. Deciphering serology to understand the ecology eid1807.111654 of infectious diseases in wildlife. EcoHealth. 2013;10:298–313. 16. Ogawa H, Miyamoto H, Nakayama E, Yoshida R, Nakamura I, Sawa H, et al. Seroepidemiological prevalence of multiple species 31. Lardeux F, Torrico G, Aliaga C. Calculation of the ELISA’s of filoviruses in fruit bats ( Eidolon helvum) migrating in Africa. cut-off based on the change-point analysis method for detection J Infect Dis. 2015;212(Suppl 2):S101–8. of Trypanosoma cruzi infection in Bolivian dogs in the infdis/jiv063 absence of controls. Mem Inst Oswaldo Cruz. 2016;111:501–4. 17. Towner JS, Pourrut X, Albariño CG, Nkogue CN, Bird BH, Grard G, et al. Marburg virus infection detected in a common 32. Killick R, Eckley IA. changepoint: an R package for changepoint African bat. PLoS One. 2007;2:e764. analysis. J Stat Softw. 2014;58:1–19. journal.pone.0000764 jss.v058.i03 18. Towner JS, Amman BR, Sealy TK, Carroll SA, Comer JA, 33. Hinkley DV. Inference about the change-point in a sequence of Kemp A, et al. Isolation of genetically diverse Marburg viruses random variables. Biometrika. 1970;57:1–17. from Egyptian fruit bats. PLoS Pathog. 2009;5:e1000536. 10.1093/biomet/57.1.1 34. Laing ED, Mendenhall IH, Linster M, Low DHW, Chen Y, Yan L, 19. Kuzmin IV, Niezgoda M, Franka R, Agwanda B, Markotter W, et al. Serologic evidence of fruit bat exposure to filoviruses, Breiman RF, et al. Marburg virus in fruit bat, Kenya. Emerg Infect Singapore, 2011–2016. Emerg Infect Dis. 2018;24:114–7. Dis. 2010;16:352–4. 20. Swanepoel R, Smit SB, Rollin PE, Formenty P, Leman PA, 35. Cullen AC, Frey HC. Probabilistic techniques in exposure Kemp A, et al.; International Scientific and Technical Committee assessment. New York: Plenum Press; 1999. p. 81–159. for Marburg Hemorrhagic Fever Control in the Democratic 36. Delignette-Muller ML, Dutang C. fitdistrplus: an R package for Republic of the Congo. Studies of reservoir hosts for Marburg fitting distributions. J Stat Softw. 2015;64:1–34. virus. Emerg Infect Dis. 2007;13:1847–51. 10.18637/jss.v064.i04 10.3201/eid1312.071115 37. Monleau M, Montavon C, Laurent C, Segondy M, Montes B, 21. Negredo A, Palacios G, Vázquez-Morón S, González F, Dopazo H, Delaporte E, et al. Evaluation of different RNA extraction Molero F, et al. Discovery of an ebolavirus-like filovirus in europe. methods and storage conditions of dried plasma or blood spots PLoS Pathog. 2011;7:e1002304. for human immunodeficiency virus type 1 RNA quantification journal.ppat.1002304 and PCR amplification for drug resistance testing. J Clin 22. Yang XL, Zhang YZ, Jiang RD, Guo H, Zhang W, Li B, et al. Microbiol. 2009;47:1107–18. Genetically diverse filoviruses in Rousettus and Eonycteris spp. JCM.02255-08 bats, China, 2009 and 2015. Emerg Infect Dis. 2017;23:482–6. 38. Guichet E, Serrano L, Laurent C, Eymard-Duvernay S, Kuaban C, Vidal L, et al. Comparison of different nucleic acid preparation 23. World Health Organization. Ebola situation report - 30 March methods to improve specific HIV-1 RNA isolation for viral load 2016. 2016 [cited 2018 May 3]. testing on dried blood spots. J Virol Methods. 2018;251:75–9. current-situation/ebola-situation-report-30-march-2016 24. Dudas G, Carvalho LM, Bedford T, Tatem AJ, Baele G, Faria NR, 39. Irwin DM, Kocher TD, Wilson AC. Evolution of the cytochrome b et al. Virus genomes reveal factors that spread and sustained the gene of mammals. J Mol Evol. 1991;32:128–44. Ebola epidemic. Nature. 2017;544:309–15. 10.1007/BF02515385 10.1038/nature22040 40. Kocher TD, Thomas WK, Meyer A, Edwards SV, Pääbo S, 25. Pigott DM, Deshpande A, Letourneau I, Morozoff C, Reiner RC Jr , Villablanca FX, et al. Dynamics of mitochondrial DNA evolution Kraemer MUG, et al. Local, national, and regional viral in animals: amplification and sequencing with conserved haemorrhagic fever pandemic potential in Africa: a multistage primers. Proc Natl Acad Sci U S A. 1989;86:6196–200. analysis. Lancet. 2017;390:2662–72. S0140-6736(17)32092-5 41. He B, Feng Y, Zhang H, Xu L, Yang W, Zhang Y, et al. Filovirus 26. Breman JG, Johnson KM, van der Groen G, Robbins CB, RNA in fruit bats, China. Emerg Infect Dis. 2015;21:1675–7. Szczeniowski MV, Ruti K, et al.; Ebola Virus Study Teams. A search for Ebola virus in animals in the Democratic Republic 42. Han BA, Schmidt JP, Alexander LW, Bowden SE, Hayman DT, of the Congo and Cameroon: ecologic, virologic, and serologic Drake JM. Undiscovered bat hosts of filoviruses. PLoS Negl surveys, 1979–1980. J Infect Dis. 1999;179(Suppl 1):S139–47. Trop Dis. 2016;10:e0004815. journal.pntd.0004815 27. Leirs H, Mills JN, Krebs JW, Childs JE, Akaibe D, Woollen N, 43. Rimoin A W, Lu K, Bramble MS, Steffen I, Doshi RH, Hoff NA, et al. Search for the Ebola virus reservoir in Kikwit, Democratic et al. Ebola virus neutralizing antibodies detectable in survivors of Republic of the Congo: reflections on a vertebrate collection. theYambuku, Zaire outbreak 40 years after infection. J Infect Dis. J Infect Dis. 1999;179(Suppl 1):S155–63. 2018;217:223–31. 10.1086/514299 44. Paweska JT, Storm N, Grobbelaar AA, Markotter W, Kemp A, 28. Ayouba A, Touré A, Butel C, Keita AK, Binetruy F, Sow MS, Jansen van Vuren P. Experimental inoculation of Egyptian fruit et al. Development of a sensitive and specific serological assay bats (Rousettus aegyptiacus) with Ebola virus. Viruses. 2016;8:29. based on luminex technology for detection of antibodies to Zaire Ebola virus. J Clin Microbiol. 2017;55:165–76. 45. Amman BR, Jones ME, Sealy TK, Uebelhoer LS, Schuh AJ, 10.1128/JCM.01979-16 Bird BH, et al. Oral shedding of Marburg virus in experimentally 29. Peel AJ, McKinley TJ, Baker KS, Barr JA, Crameri G, Hayman DT, infected Egyptian fruit bats (Rousettus aegyptiacus). J Wildl Dis. et al. Use of cross-reactive serological assays for detecting novel 2015;51:113–24. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2239 RESEARCH 46. Schuh AJ, Amman BR, Jones ME, Sealy TK, Uebelhoer LS, animals with Ebola virus. Emerg Infect Dis. 1996;2:321–5. Spengler JR, et al. Modelling filovirus maintenance in nature by experimental transmission of Marburg virus between 49. Storm N, Jansen Van Vuren P, Markotter W, Paweska JT. Egyptian rousette bats. Nat Commun. 2017;8:14446. Antibody responses to Marburg virus in Egyptian rousette bats and their role in protection against infection. Viruses. 2018;10:73. 47. Jones ME, Schuh AJ, Amman BR, Sealy TK, Zaki SR, Nichol ST, et al. Experimental inoculation of Egyptian rousette bats (Rousettus aegyptiacus) with viruses of the Ebolavirus and Marburgvirus Adress for correspondence: Martine Peeters, TransVIHMI, Institut genera. Viruses. 2015;7:3420–42. de Recherche pour le Développement, 911 Ave Agropolis, BP34394 48. Swanepoel R, Leman PA, Burt FJ, Zachariades NA, Braack LE, Montpellier CEDEX 1, France; e-mail: Ksiazek TG, et al. Experimental inoculation of plants and February 2016 Ebola • Ebola and Its Control in Liberia, 2014–2015 • Dogs and Opossums Positive for Vaccinia Virus during Outbreak Affecting Cattle and Humans, São Paulo • Epidemiology of Epidemic Ebola Virus Disease in State, Brazil Conakry and Surrounding Prefectures, Guinea, 2014–2015 • African Buffalo Movement and Zoonotic Disease Risk across Transfrontier Conservation Areas, Southern Africa • Hospital Preparations for Viral Hemorrhagic Fever Patients and Experience Gained from the Admission of • Anaplasmataceae–Specific PCR for Diagnosis and an Ebola Patient Therapeutic Guidance for Symptomatic Neoehrlichiosis in Immunocompetent Host • Trematode Fluke Procerovum varium as Cause of Ocular Inflammation in Children, South India • Candidatus Coxiella massiliensis Infection • Association between Landscape Factors and Spatial • Ebola Virus Persistence in Semen Ex Vivo Patterns of Plasmodium knowlesi Infections in Sabah, • Ebola Virus RNA Stability in Human Blood and Urine in Malaysia West Africa’s Environmental Conditions • Feasibility of Xpert Ebola Assay in Médecins Sans • Uveitis and Systemic Inflammatory Markers in Frontières Ebola Program, Guinea Convalescent Phase of Ebola Virus Disease • Prognostic Indicators for Ebola Patient Survival • Louseborne Relapsing Fever among East African • Invasive Group A Streptococcus Infection among Refugees, Italy, 2015 Children, Rural Kenya • Mediterranean Fin Whales (Balaenoptera physalus) • Randomized Controlled Trial of Hospital-Based Threatened by Dolphin Morbillivirus Hygiene and Water Treatment Intervention (CHoBI7) to • Blastomyces gilchristii as Cause of Fatal Acute Reduce Cholera Respiratory Distress Syndrome • Sustained Transmission of Pertussis in Vaccinated, • Effectiveness of Meningococcal B Vaccine against 1–5-Year-Old Children in a Preschool, Florida, USA Endemic Hypervirulent Neisseria meningitidis W Strain, • Molecular Characterization of Invasive Streptococcus England dysgalactiae subsp. equisimilis, Japan • Frequency and Distribution of Rickettsiae, Borreliae, • Epidemiology of Serotype 1 Invasive Pneumococcal and Ehrlichiae Detected in Human-Parasitizing Ticks, Disease, South Africa, 2003–2013 Texas, USA • Hemorrhagic Fever with Renal Syndrome, Zibo City, • High Prevalence of Borrelia miyamotoi among Adult China, 2006–2014 Blacklegged Ticks from White-Tailed Deer To revisit the February 2016 issue, go to: 2240 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Rat Hepatitis E Virus as Cause of Persistent Hepatitis after Liver Transplant Siddharth Sridhar, Cyril C.Y. Yip, Shusheng Wu, Jianpiao Cai, Anna Jin-Xia Zhang, Kit-Hang Leung, Tom W.H. Chung, Jasper F.W. Chan, Wan-Mui Chan, Jade L.L. Teng, Rex K.H. Au-Yeung, Vincent C.C. Cheng, Honglin Chen, Susanna K.P. Lau, Patrick C.Y. Woo, Ning-Shao Xia, Chung-Mau Lo, Kwok-Yung Yuen All hepatitis E virus (HEV) variants reported to infect hu- been reported to infect humans belong to Orthohepevirus mans belong to the species Orthohepevirus A (HEV-A). A (HEV-A). Five genotypes within HEV-A (HEV-1–4 and The zoonotic potential of the species Orthohepevirus C -7) cause hepatitis in humans, and 3 genotypes (HEV-3, -4, (HEV-C), which circulates in rats and is highly divergent and -7) can cause chronic hepatitis in immunocompromised from HEV-A, is unknown. We report a liver transplant recipi- patients after foodborne zoonotic transmission (2,6,9,10). ent with hepatitis caused by HEV-C infection. We detected In addition to HEV-A, the Orthohepevirus genus in- HEV-C RNA in multiple clinical samples and HEV-C antigen cludes 3 other species: Orthohepevirus B circulates in chick- in the liver. The complete genome of the HEV-C isolate had ens, Orthohepevirus C (HEV-C) in rats and ferrets, and Or- 93.7% nt similarity to an HEV-C strain from Vietnam. The thohepevirus D in bats. HEV-C, also known as rat hepatitis patient had preexisting HEV antibodies, which were not pro- E virus, shares only 50%–60% nt identity with HEV-A (8). tective against HEV-C infection. Ribavirin was an effective The zoonotic potential of HEV-C is unknown; cases of clini- treatment, resulting in resolution of hepatitis and clearance of HEV-C viremia. Testing for this zoonotic virus should be cal infection have not been reported. The substantial phylo- performed for immunocompromised and immunocompetent genetic divergence between HEV-A and HEV-C, especially patients with unexplained hepatitis because routine hepati- in critical receptor binding domains, forms a theoretical spe- tis E diagnostic tests may miss HEV-C infection. HEV-C is cies barrier (11). Serologic and molecular tests for HEV are also a potential threat to the blood product supply. designed primarily to detect HEV-A, and they might miss HEV-C infections. Therefore, the threat to human health, including blood and organ supply safety, from HEV-C is epatitis E virus (HEV) infects 20 million humans world- unknown. We aimed to prove definitively that HEV-C can Hwide annually (1). HEV-infected persons usually have infect humans and describe the clinical, epidemiologic, ge- self-limiting acute hepatitis. However, persistent hepatitis nomic, and serologic features of this new zoonosis. can occur in HEV-infected immunocompromised patients who acquire infection by eating undercooked pork, rabbit, Materials and Methods deer, camel, or boar meat (2–6). HEV transmission through blood product transfusion also has been described (7). Study Population The diverse Hepeviridae family, which incorporates We conducted this study in Queen Mary Hospital, a 1,700- all HEV variants, includes members whose primary host bed tertiary care hospital in Hong Kong. We assessed 518 species are terrestrial mammals (genus Orthohepevirus) solid-organ transplant recipients (kidney, liver, lung, and and fish (genus Piscihepevirus) (8). The Orthohepevirus heart transplant) who were followed up in Queen Mary genus is classified into 4 species; HEV variants that have Hospital for persistent biochemical hepatitis from January Author affiliations: The University of Hong Kong, Hong Kong, 1, 2014, or date of transplant (whichever date was later) China (S. Sridhar, C.C.Y. Yip, S. Wu, J. Cai, A.J.-X. Zhang, through December 31, 2017. We defined persistent hepatitis K.-H. Leung, T.W.H. Chung, J.F.W. Chan, W.-M. Chan, as elevation of alanine aminotransferase (ALT) >1.5 times J.L.L. Teng, R.K.H. Au-Yeung, V.C.C. Cheng, H. Chen, S.K.P. Lau, the upper limit of the reference level for a continuous period P.C.Y. Woo, C.-M. Lo, K.-Y. Yuen); The University of Hong Kong– of >6 weeks. For patients whose ALT met this definition, Shenzhen Hospital, Shenzhen, China (J.F.W. Chan, C.-M. Lo, we reviewed clinical records, ultrasonogram results, endo- K.-Y. Yuen); Xiamen University, Xiamen, China (N.-S. Xia) scopic retrograde cholangiopancreatography results, and laboratory results to identify the likely cause of hepatitis. We DOI: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2241 RESEARCH considered patients to have hepatitis B virus (HBV), hepati- Serologic Testing tis C virus (HCV), or cytomegalovirus (CMV) reactivation if We conducted HEV antibody screening for patients any of these viruses were detected in blood during the hepati- with unexplained persistent hepatitis using HEV IgM tis episode. In patients with no identifiable cause of hepatitis, and HEV IgG commercial ELISA kits (Wantai, Beijing, HEV IgM ELISA screening was performed, in accordance China) and detected hepatitis B surface antigen (HB- with the usual practice in Queen Mary Hospital. HEV in - sAg) using the ARCHITECT HBsAg chemiluminescent fection was diagnosed if the HEV IgM assay was positive, microparticle immunoassay (Abbott). HAV IgM and and persistent HEV infection was diagnosed if HEV vire- HCV antibodies were tested using VIDAS immunoas- mia in patient plasma lasted for >3 months. PCR sequencing say kits (bioMérieux, Marcy-L’Étoile, France). For in- was performed for speciation of HEV isolate. We obtained vestigation of the HEV-C transmission event, we sub- ethics approval from the Institutional Review Board of the jected patient and donor serum to HEV-A and HEV-C University of Hong Kong/Hospital Authority West Cluster. Western blots using polyclonal antiserum from mice We obtained written informed consent from all patients with inoculated with HEV-C protein and mAbs as controls. persistent HEV infection. ELISAs using recombinant HEV-A and HEV-C protein- coated plates were designed based on the method de- Nucleic Acid Detection for Hepatitis Viruses and scribed by Shimizu et al. with modifications ( 15). We HEV Complete Genome Sequencing set cutoffs and interpreted results to differentiate HEV- We designed 3 in-house–developed reverse transcription A– and HEV-C–specific serologic responses (online PCRs (RT-PCRs) to detect HEV (online Technical Appen- Technical Appendix). dix Table 1, 0937-Techapp1.pdf). Hepatitis A virus (HAV) RNA and Virus Culture CMV DNA detections were performed using in-house We selected cell lines A549 (lung adenocarcinoma), Huh-7 nucleic acid amplification tests. HBV and HCV viral loads (hepatocellular carcinoma), and Caco-2 (colorectal adeno- were quantified using commercial kits (COBAS TaqMan, carcinoma) to investigate whether human cell lines could Roche, Basel, Switzerland; and RealTime HCV, Abbott, support HEV-C growth. Cell lines were chosen if they sup- Chicago, IL, USA, respectively). ported growth of patient-derived HEV isolates or HEV in- We sequenced the PCR product of the pan-Ortho- fectious clones (16–18) (online Technical Appendix). We hepevirus RT-PCR using the RT-PCR primers. Because subjected supernatants and lysates to HEV-C quantitative the RNA-dependent RNA polymerase sequences of pa- RT-PCR (qRT-PCR) and immunostaining. tient HEV isolates clustered with rat HEV-C strains, primers for complete genome amplification were de - Immunohistochemical and Immunofluorescence Staining signed by multiple alignment of rat HEV-C genomes in We conducted immunohistochemical staining of for- GenBank (online Technical Appendix Table 2). We used malin-fixed paraffin-embedded liver tissue sections and these primers for complete genome sequencing of HEV in infected A549 cell culture monolayers using HEV-C patient feces (strain LCK-3110). We constructed phylo- polyclonal serum antibodies and HEV-A mAbs. We per- genetic trees using MEGA6 with the general time revers- formed immunofluorescence staining of permeabilized ible plus gamma model (12). infected cells using HEV-C polyclonal antiserum (online Technical Appendix). Cloning and Purification of Recombinant HEV-A and HEV-C Open Reading Frame 2 Protein Epidemiologic and Environmental Investigation We used specific primers (online Technical Appendix) We retrieved organ and blood donor serum for HEV to amplify the genes encoding the 239 aa immunogenic ELISA, Western blot, and HEV-C qRT-PCR. To survey recombinant peptides of HEV-A (genotype 4) and HEV- density of rat fecal contamination and collect environ- C. Cloning the amplified genes into a bacterial expression mental specimens for HEV-C qRT-PCR, we visited the vector, expression in Escherichia coli, and protein puri- patient’s housing estate on November 22, 2017. Further- fication were performed as previously described ( 13,14). more, from deep freezers we retrieved archived Rattus sp. liver, spleen, rectal swab, and kidney specimens collected during 2012–2017 within a 2.5-km radius around the pa- Antibodies Against HEV-A and HEV-C tient’s residence for preexisting pathogen surveillance Polyclonal antibodies against the HEV-C recombinant pro- programs and subjected them to HEV-C qRT-PCR. The tein were raised in mice (online Technical Appendix). In HEV-C ORF2 fragment of qRT-PCR–positive specimens addition, we used 2 murine monoclonal antibodies (mAbs) was sequenced using additional primers (online Technical against open reading frame (ORF) 2 antigen of HEV-A in Appendix Table 3). this study. 2242 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Rat HEV after Liver Transplant (LFTs) reverted to normal, and he was discharged on post- Results transplant day 11. He was taking mycophenolate mofetil (500 mg 2×/d), tacrolimus (1 mg 2×/d), and prednisolone Hepatitis E Incidence in Transplant Recipient Cohort (5 mg 2×/d) as antirejection prophylaxis. He was also tak- Of 518 patients, 52 (10.2%) had persistent hepatitis (Table ing entecavir (0.5 mg 1×/d) for HBV suppression; serum 1). Five (9.6%) patients with hepatitis tested positive for HBsAg was negative 6 weeks after the transplant. HEV IgM; 4 of these were kidney transplant recipients, Routine phlebotomy on July 12 (day 59 posttransplant) and 1 was a liver transplant recipient. Together with re- revealed mild derangement of ALT to 74 U/L (reference activation of chronic HBV infection, HEV was the third 8–58 U/L). Other LFTs were normal. One week later, there most common cause of viral hepatitis in the local transplant was further derangement of parenchymal liver enzymes: population. Of the 5 patients, plasma HEV-A qRT-PCR ALT was 138 U/L, aspartate aminotransferase was elevat- of 3 renal transplant recipients was positive; another re- ed to 65 U/L (reference 15–38 U/L), γ-glutamyltransferase nal transplant recipient tested negative for HEV RNA. We was 124 U/L (reference 11–62 U/L), and alkaline phos- have previously reported the clinical details of the 3 HEV- phatase was within reference limits at 70 U/L (reference A–infected patients (9). Rat-derived HEV-C infection was 42–110 U/L). Complete blood count showed lymphopenia, diagnosed in the liver transplant recipient, which accounted at 0.88 × 10 cells/L, although total leukocyte count was for 1.9% (1/52) of persistent hepatitis in our cohort. within reference levels. The patient was empirically managed for acute graft Patient History rejection with increased immunosuppression using a 3-day A 56-year-old man underwent deceased-donor liver trans- course of methylprednisolone. Valganciclovir was pre- plant on May 14, 2017, because of hepatocellular carcino- scribed for low-level whole blood CMV viremia of 5.31 ma complicating chronic HBV carriage. He received 1,000 × 10 IU/mL. However, LFTs continued to deteriorate de- mg hydrocortisone and 20 mg basiliximab (anti–interleu- spite clearance of CMV viremia and increased immunosup- kin-2 receptor mAb) as intraoperative antirejection pro- pression. Liver biopsy showed nonspecific mild to moder- phylaxis and 4 units of platelets (derived from 4 separate ate inflammatory infiltrate comprising small lymphocytes blood donors) during the operation. His liver function tests in the portal tracts. There were no viral inclusion bodies, Table 1. Demographic and clinical characteristics of solid organ and immunohistochemical staining for CMV and hepatitis transplant recipients, Queen Mary Hospital, Hong Kong, January B core antigens was negative. Results of testing for HBsAg 1, 2014–December 31, 2017* in serum, HBV DNA in plasma, HCV antibody in serum, Characteristic Result† HAV IgM in serum, and HAV RNA in plasma and feces No. transplant recipients 518 Organ transplanted were all negative. HEV IgM was detected in serum col- Kidney 430 (83.0) lected on August 22 (day 100 posttransplant). Because of Liver 61 (11.7) the serology result and ongoing LFT derangement, persis- Heart 16 (3.1) Lung 10 (1.9) tent HEV infection was suspected. A qRT-PCR targeting Combined kidney and liver 1 (0.2) HEV-A was performed on patient fecal and plasma speci- Median age, y 56 mens; HEV-A RNA was not detected in either specimen. Sex An RT-PCR capable of detecting all species within the F 203 (39.2) M 315 (60.8) Orthohepevirus genus detected amplicons (online Techni- Prevalence of persistent biochemical 52 (10.2) cal Appendix Figure 1) in plasma, feces, and liver tissue. hepatitis Sequencing confirmed that the products clustered with rat Cause of biochemical hepatitis Viral hepatitis‡ HEV-C strains. Reactivation of chronic HBV infection 5 (9.6) Chronic HCV infection 7 (13.5) Viral RNA Kinetics and Effect of Ribavirin Therapy Chronic HEV infection 5 (9.6) CMV reactivation 8 (15.4) The patient’s archived serum, saliva, urine, feces, and Nonviral causes† nonfixed liver tissue samples were retrieved for HEV- Drug toxicity 7 (13.5) C RNA load testing using HEV-C qRT-PCR (Figure 1, Nonalcoholic fatty liver disease 3 (5.8) panel A). Two pretransplant serum samples and 1 serum Liver graft rejection 7 (13.5) Biliary anastomotic stricture 5 (9.6) sample collected on day 17 after transplant did not contain Liver malignancies 2 (3.8) HEV-C RNA. The first specimen with detectable HEV-C Septic cholestasis 2 (3.8) RNA was a serum sample collected 43 days after trans- Recurrent pyogenic cholangitis 1 (1.9) *CMV, cytomegalovirus; HBV, hepatitis B virus; HCV, hepatitis C virus; plant, which contained an RNA load of 9.48 × 10 cop- HEV, hepatitis E virus. ies/mL; this result preceded onset of LFT derangement †All results are no. (%) unless otherwise indicated. ‡All percentages based on no. patients with biochemical hepatitis. by 3 weeks. After heightened immunosuppression in July Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2243 RESEARCH Figure 1. Natural course of HEV-C infection in a 56-year-old man at Queen Mary Hospital, Hong Kong. A) Timeline of major clinical events. All days are post transplant. B) Kinetics of liver function tests, tacrolimus levels (µg/L), and plasma HEV-C RNA load (log copies/mL) with relation to ribavirin therapy. ALT, alanine aminotransferase; HEV-C, Orthohepevirus C; LFT, liver function test. and August, the HEV-C RNA load in blood steadily rose IgG blot but not in the HEV-C blot (lane 8; Figure 2, pan- along with ALT (Figure 1, panel B). Variation in ALT els B, C) confirming specificity, and the other was cross- correlated with the HEV-C RNA viral load by linear re- reactive against HEV-A and HEV-C (lane 9; Figure 2, pan- gression (R = 0.791). HEV-C RNA was also detected in els B, C). Polyclonal serum raised in mice inoculated with feces, saliva, and liver tissue (Figure 1, panel A); feces HEV-C protein reacted in both blots, showing that the se- contained the highest RNA load. rum was cross-reactive (lane 7). Patient serum collected on Immunosuppressant dosages were decreased after con- day 100 after transplant (lane 1) was tested against HEV- firmation of HEV infection. However, ALT and HEV-C A and HEV-C recombinant proteins. The serum specimen RNA loads continued to increase despite reduction of plas- showed reactivity in both Western blots. ma tacrolimus levels by 55% and rebound of lymphocyte Two patient serum samples, 1 obtained 3 months be- count to 2.27 × 10 cells/L. Therefore, oral ribavirin 400 fore transplant and the other obtained on day 100 after mg twice daily was started on September 7. ALT decreased transplant, were tested in IgG ELISAs using HEV-A and within the first week after start of therapy and normalized HEV-C protein-coated plates. The pretransplant serum within 1 month after starting ribavirin (Figure 1, panel B). (Figure 2, panel D) had cross-reactive antibodies against HEV-C RNA loads also decreased to undetectable levels both HEV-A and HEV-C proteins (<2-fold difference in in plasma obtained on February 13, 2018. Ribavirin was titer using OD cutoff of 0.3). However, the posttransplant stopped in April 2018, and HEV-C RNA in serum remained serum (Figure 2, panel E) showed >16-fold rise in HEV- undetectable as of August 21, 2018, confirming sustained A IgG titer and markedly higher reactivity against HEV- virologic response. A than against HEV-C (>4-fold difference in titer using a cutoff OD of 0.3). Serologic Analysis We retrospectively tested all available patient serum and Liver Histologic and Immunohistochemical Analyses plasma samples for HEV IgG and IgM ELISA using the Serial liver biopsies showed progressively worsening he- Wantai ELISA kit. The patient’s serum before transplant patocyte ballooning and degenerative changes (Figure 3, was HEV IgG positive and IgM negative. HEV IgG and panels A, B). Apoptotic hepatocytes were identified in the IgM optical density rose sharply from June 27, when HEV- biopsy obtained on day 98 posttransplant (Figure 3, panel B). C RNA was first detectable in blood, to July 25, when clini- Immunohistochemical staining with the cross-reactive mAb cal hepatitis began (online Technical Appendix Figure 2). showed positive perinuclear cytoplasmic signals (Figure 3, Despite high IgG levels, HEV-C RNA continued to rise panel C), and negative control with bovine serum albumin until ribavirin was started. instead of mAb showed no signals (Figure 3, panel D). To characterize the serologic response, Western blot using purified HEV-A and HEV-C recombinant proteins Genomic Description (Figure 2, panel A) was performed. Two mAbs raised Complete genome sequencing of the patient’s fecal HEV against HEV-A were used: 1 produced a band in HEV-A isolate (LCK-3110) showed that the genome was 6,942 bp 2244 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Rat HEV after Liver Transplant Figure 2. Serologic testing for HEV infection at Queen Mary Hospital, Hong Kong. A) Sodium dodecyl sulfate polyacrylamide gel electrophoresis gel showing purified HEV-A and HEV-C 239-aa recombinant proteins used in Western blot and ELISA. Lane 1, molecular weight marker; lane 2, HEV-A protein; lane 3, HEV-C protein. B–C) IgM and IgG Western blot using HEV-A protein (B) and HEV-C protein (C). Lane 1, patient serum (posttransplant day 100); lanes 2–5, individual platelet donor serum; lane 6, organ donor serum; lane 7, murine polyclonal serum against HEV-C; lane 8, specific monoclonal antibody against HEV-A; lane 9, cross-reactive monoclonal antibody against HEV-A and HEV-C. D, E) HEV-A and HEV-C ELISA IgG titers of patient pretransplant (D) and posttransplant serum (E) using an OD of 0.3 as assay cutoff as described in the online Technical Appendix ( Techapp1.pdf). HEV, hepatitis E virus; HEV-A, Orthohepevirus A; HEV-C, Orthohepevirus C; OD, optical density. long (GenBank accession no. MG813927). Phylogenetic LA-B350) using ClustalX 2.0 ( trees of the nucleotide and amino acid sequences of ORF1, clustal2/). Alignment revealed significant lack of homol - ORF2, and ORF3 of HEV strains showed that LCK-3110 ogy with HEV-C at the 3′ end of either the forward or is most closely related to the Vietnam-105 strain (Figure reverse primer for the assays described by Jothikumar et 4; online Technical Appendix Figure 3, panels A, B), al. and Rolfe et al. (online Technical Appendix Figure 4, sharing 93.7% nt identity. Because no phylogenetic in - panels A, B) (20,21). Our in-house HEV-A qRT-PCR is congruence was found on comparison of trees of the 3 based on the primer/probe design of Jothikumar et al. and genomic segments, recombination was unlikely (Table 2; was unable to detect HEV-C in patient specimens (20). online Technical Appendix). To determine whether com- For the assays described by Mansuy et al. and Colson et monly used RT-PCRs for HEV nucleic acid amplifica - al. (19,22), there was significant lack of matching of probe tion could detect HEV-C, we aligned published primer/ sequence (40%–45% mismatch) to HEV-C genomes (on - probe sequences of HEV RT-PCRs (19–22) with com- line Technical Appendix Figure 4, panels C, D), which plete genome sequences of HEV-A (genotype 1 reference most likely would result in failure to detect any amplified strain) and HEV-C (strains LCK-3110, Vietnam-105, and nucleic acid. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2245 RESEARCH Figure 3. Histologic and immunohistochemical staining of liver tissue from a 56-year- old man at Queen Mary Hospital, Hong Kong. A, B) Liver tissue sections (original magnification ×200) stained with hematoxylin and eosin obtained at day 0 (A), showing normal hepatocyte architecture, and day 98 (B) after transplant showing progressive increase in hepatocyte ballooning and degenerative changes. C, D) Liver tissue section stained with cross-reactive monoclonal antibody (original magnification ×400); arrows show perinuclear antigen staining (C) and negative control with bovine serum albumin (D). Virus Culture room floor tested negative for HEV-C RNA. To expand the We detected HEV-C RNA in supernatants from all 3 cell investigation, we retrieved archived rodent samples collected lines (Figure 5, panel A) inoculated with patient’s feces at from the area around the patient’s housing estate (≈2.5-km ra - steady levels from day 3 to day 7 after inoculation. RNA dius) as part of preexisting pathogen surveillance programs. loads in cell lysates were ≈1 log higher than concomitantly Spleen, kidney, liver, and rectal swab specimens from 27 rats harvested supernatants, suggesting successful viral cell en- were tested by qRT-PCR. The internal organs of 1 street rat try. Immunohistochemical staining (Figure 5, panels B, C) (Rattus norvegicus) collected in 2012 tested positive for HEV- of A549 cell monolayers and immunofluorescence stain- C RNA (strain name SRN-02). The ORF2 aa sequence of this ing of infected Huh-7 and Caco-2 cells (online Technical isolate had 90.9% identity to LCK-3110. Appendix Figure 5) confirmed the presence of cytoplasmic HEV ORF2 antigen when stained with antiserum against Discussion HEV-C. These findings suggested abortive viral replication Discovered in Germany in 2010, rat HEV variants have of HEV-C in human cell lines. been detected in rodent samples in Asia, Europe, and North America (23–26). Because of high divergence from human- Epidemiologic Investigation pathogenic HEV, rat HEV has been classified into a sepa- The first clinical sample with detectable HEV-C RNA was rate species, Orthohepevirus C, within the family Hepeviri- obtained 43 days after transplant. HEV-C was not detected dae (27). The zoonotic potential of HEV-C is controversial. in serum samples obtained before transplant. Serum samples Virus-like protein ELISAs show possible subclinical infec- from the organ donor and all 4 platelet donors tested negative tion among forestry workers in Germany and febrile inpa- by IgM Western blot against HEV-C recombinant protein tients in Vietnam, although interpretation of such studies (Figure 2, panel C, lanes 2–6) and HEV-C qRT-PCR. is difficult because of serologic cross-reactivity between The patient’s house unit was located adjacent to a refuse HEV-A and HEV-C (15,28). Immunocompetent rhesus chute. He had noticed rodent droppings but had never seen macaques do not appear to be susceptible to experimental rats inside his home. A site visit to the housing estate was con- infection with a North America HEV-C isolate (23). ducted on November 22, 2017. Rodent droppings were found In this study, we detected HEV-C RNA in multiple around refuse collection bins on the ground floor and the floor specimens from a transplant recipient. The HEV-C infec- where the patient lived. Twelve rodent fecal specimens, 2 swab tion manifested as persistent hepatitis, as shown by tempo- samples from the drain, and 2 swab samples from the refuse ral correlation between blood HEV-C RNA detection and 2246 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Rat HEV after Liver Transplant Figure 4. Phylogenetic analysis using complete open reading frame 2 nucleotide sequences of LCK-3110 and other hepatitis E virus strains. The tree was constructed using maximum- likelihood method with bootstrap values calculated from 1,000 trees. Only bootstrap values >70% are shown. GenBank accession numbers are provided. Arrow indicates strain LCK-3110; asterisk indicates strain SRN-02 detected in a street rodent in the epidemiologic investigation. Scale bar indicates mucleotide substitutions per site. hepatitis onset, presence of HEV-C RNA in liver tissue, among both immunocompromised and immunocompetent and normalization of liver function tests with viral clear- patients with hepatitis of unknown etiology. ance. These findings prove that HEV-C can infect humans The patient reported here acquired HEV-C infection to cause clinically significant illness and signal a need to despite having HEV IgG. Interpreted in parallel with the reevaluate the importance of HEV-C as a human zoonosis finding by Sanford et al. that inoculating pigs with HEV-C Table 2. Comparison between nucleotide and deduced amino acid sequence identities of HEV strain LCK-3110 and other HEV strains Rat HEV strain LCK-3110 Nucleotides, % Amino acids, % Entire HEV strain (GenBank accession no.) HEV species genome ORF1 ORF2 ORF3 ORF1 ORF2 ORF3 Genotype 1 (NC_001434) HEV-A 57.6 56.4 60.9 55.0 50.0 56.3 31.0 Genotype 2 (M74506) HEV-A 57.3 56.3 60.0 50.4 49.7 56.1 27.6 Genotype 3 (EU723512) HEV-A 56.6 55.4 60.7 51.9 50.3 56.5 30.7 Genotype 4 (AJ272108) HEV-A 56.5 55.4 59.8 55.4 49.7 56.4 31.0 Rabbit HEV (FJ906895) HEV-A 56.0 54.9 59.9 51.6 50.1 56.4 27.6 Wild boar HEV (AB573435) HEV-A 57.3 56.2 60.4 54.8 49.7 56.2 33.6 Wild boar HEV (AB602441) HEV-A 56.8 55.7 59.6 54.5 50.3 56.5 31.9 Camel HEV (KJ496144) HEV-A 55.9 54.9 59.4 53.5 50.5 56.2 32.2 Camel HEV (KX387867) HEV-A 55.6 54.3 59.7 53.7 50.1 55.8 29.6 Rat HEV Vietnam-105 (JX120573) HEV-C 93.7 93.3 95.2 96.4 98.2 98.0 95.1 Rat HEV LA-B350 (KM516906) HEV-C 77.3 76.3 79.7 79.3 88.0 92.1 64.7 Ferret HEV (JN998606) HEV-C 68.7 67.5 71.1 64.2 74.8 78.7 45.9 Bat HEV (JQ001749) HEV-D 53.8 53.8 54.3 44.7 45.7 47.9 18.1 Avian HEV (AY535004) HEV-B 53.5 54.0 53.0 46.4 45.6 43.5 24.8 *HEV, hepatitis E virus; ORF, open reading frame. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2247 RESEARCH ORF2 protein did not protect them from HEV-A infection and low amino acid homology between HEV-A and HEV- C in critical immunogenic domains (29), our data suggest that HEV-A antibodies do not protect against HEV-C in- fection. The patient’s postinfection serum showed signifi- cantly higher reactivity in an HEV-A–specific ELISA than in an HEV-C ELISA; the humoral immune responses of persons with past HEV-A infection to de novo HEV-C in- fection are worthy of further study to identify whether an- amnestic responses are mounted. The patient’s HEV isolate had high nucleotide similar- ity to the HEV-C Vietnam-105 strain. It shared less homol- ogy with the North America LA-B350 strain, especially in the ORF3 domain, which is important for viral egress (30). Interspecies transmission could not be attributed to specific viral mutations. Future studies will need to include differ- ences in zoonotic potential between HEV-C strains from Asia and the Americas. The patient’s immunosuppression possibly enabled the virus to surmount the species barrier, as described previously for avian influenza ( 31,32). HEV-C infections may go undiagnosed because of amplification failure in RT-PCRs, which are designed based on HEV-A sequenc- es (online Technical Appendix Figure 3). The Wantai ELISA, based on HEV-A genotype 1, was able to detect IgM in this patient, but whether the assay is sensitive for HEV-C infection or was detecting only HEV-A–specific antibodies is uncertain. Therefore, we believe that spe- cific RT-PCR is the most reliable method to diagnose HEV-C infections. Our findings are also relevant to blood and organ dona- tion safety. Because of the inability of commonly used RT- PCRs to detect HEV-C, transmission from asymptomati- cally infected immunocompetent donors may occur, even in countries that screen donated blood for HEV. Studies examining frequency of HEV-C contamination in blood products are needed to quantify this threat. The patient lived in a housing estate with evidence of rat infestation in the refuse bins outside his home. We identified HEV-C in street rodents from the area, but the isolate was not closely related to the patient’s isolate. The route of transmission is unclear; we postulate that con- tamination of food by infected rat droppings in the food supply is possible. Other possibilities include reactivation of a subclinical infection in the patient posttransplant or a Figure 5. Isolation of HEV-C from 56-year-old male patient’s donor-derived infection from residual HEV-C in the trans- feces in cell culture, Queen Mary Hospital, Hong Kong. A) planted organ. However, we found no serologic or viro- HEV-C RNA loads in culture S and day-7 CL of A549, Huh-7, and Caco-2 cell lines after inoculation by patient’s filtered logic evidence of HEV-C infection in donor and recipient fecal suspension. Mean of 3 replicates; error bars indicate serum before transplant. An occult infection in the donated SEM. B) Uninfected A549 cell monolayer stained with liver, which reactivated after transplant as described pre- anti–HEV-C polyclonal antiserum. C) Infected A549 cell viously for HEV-A, cannot be completely excluded. De- monolayer stained with anti–HEV-C polyclonal antiserum. tailed studies are needed to ascertain the route of HEV-C Original magnification ×400. CL, cell lysate; HEV, hepatitis E virus; S, supernatant. infection in humans. 2248 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Rat HEV after Liver Transplant Shenzhen, China: a shift toward foodborne transmission of Acknowledgment hepatitis E virus infection. Emerg Microbes Infect. 2017;6:e115. We thank N.S. Xia for his kind gift of murine mAbs against HEV-A. 11. Zhang J, Li SW, Wu T, Zhao Q, Ng MH, Xia NS. Hepatitis E virus: neutralizing sites, diagnosis, and protective immunity. Rev Med This study was supported by donations from Shaw Virol. 2012;22:339–49. Foundation and from Michael Seak-Kan Tong. The study was 12. Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. MEGA6: partially funded by the Consultancy Service for Enhancing Molecular Evolutionary Genetics Analysis version 6.0. Mol Biol Evol. 2013;30:2725–9. Laboratory Surveillance of Emerging Infectious Disease for 13. Chan JF, Yip CC, Tsang JO, Tee KM, Cai JP, Chik KK, et al. Department of Health of the Hong Kong Special Administrative Differential cell line susceptibility to the emerging Zika virus: Region of China, Seed Fund for Basic Research, and Enhanced implications for disease pathogenesis, non-vector-borne human New Staff Start-up Research Grant of the University of transmission and animal reservoirs. Emerg Microbes Infect. 2016;5:e93. Hong Kong. 14. Chan JF, Chan KH, Choi GK, To KK, Tse H, Cai JP, et al. Differential cell line susceptibility to the emerging novel human J.F.W.C. has received travel grants from Pfizer Corporation betacoronavirus 2c EMC/2012: implications for disease Hong Kong and Astellas Pharma Hong Kong Corporation pathogenesis and clinical manifestation. J Infect Dis. 2013; Limited and was an invited speaker for Gilead Sciences Hong 207:1743–52. Kong Limited and Luminex Corporation. 15. Shimizu K, Hamaguchi S, Ngo CC, Li TC, Ando S, Yoshimatsu K, et al. Serological evidence of infection with rodent-borne hepatitis E virus HEV-C1 or antigenically related virus in humans. About the Author J Vet Med Sci. 2016;78:1677–81. jvms.16-0200 Dr. Sridhar is a clinical assistant professor at the Department of 16. Emerson SU, Nguyen HT, Torian U, Burke D, Engle R, Microbiology, The University of Hong Kong. His major research Purcell RH. Release of genotype 1 hepatitis E virus from cultured interests include viral hepatitis and clinical virology. hepatoma and polarized intestinal cells depends on open reading frame 3 protein and requires an intact PXXP motif. J Virol. 2010;84:9059–69. References 17. Emerson SU, Nguyen H, Graff J, Stephany DA, Brockington A, 1. Rein DB, Stevens GA, Theaker J, Wittenborn JS, Wiersma ST. Purcell RH. In vitro replication of hepatitis E virus (HEV) The global burden of hepatitis E virus genotypes 1 and 2 in 2005. genomes and of an HEV replicon expressing green fluorescent Hepatology. 2012;55:988–97. protein. J Virol. 2004;78:4838–46. 2. Lee GH, Tan BH, Teo EC, Lim SG, Dan YY, Wee A, et al. JVI.78.9.4838-4846.2004 Chronic infection with camelid hepatitis E virus in a liver 18. Tanaka T, Takahashi M, Kusano E, Okamoto H. Development and transplant recipient who regularly consumes camel meat and milk. evaluation of an efficient cell-culture system for Hepatitis E virus. Gastroenterology. 2016;150:355–7.e3. J Gen Virol. 2007;88:903–11. j.gastro.2015.10.048 19. Mansuy JM, Peron JM, Abravanel F, Poirson H, Dubois M, 3. Abravanel F, Lhomme S, El Costa H, Schvartz B, Peron JM, Miedouge M, et al. Hepatitis E in the south west of France in Kamar N, et al. Rabbit hepatitis E virus infections in humans, individuals who have never visited an endemic area. J Med Virol. France. Emerg Infect Dis. 2017;23:1191–3. 2004;74:419–24. 10.3201/eid2307.170318 20. Jothikumar N, Cromeans TL, Robertson BH, Meng XJ, Hill VR. 4. Li TC, Chijiwa K, Sera N, Ishibashi T, Etoh Y, Shinohara Y, et al. A broadly reactive one-step real-time RT-PCR assay for rapid Hepatitis E virus transmission from wild boar meat. Emerg Infect and sensitive detection of hepatitis E virus. J Virol Methods. Dis. 2005;11:1958–60. 2006;131:65–71. 5. Tei S, Kitajima N, Takahashi K, Mishiro S. Zoonotic transmission 21. Rolfe KJ, Curran MD, Mangrolia N, Gelson W, Alexander GJ, of hepatitis E virus from deer to human beings. Lancet. 2003; L’estrange M, et al. First case of genotype 4 human hepatitis E 362:371–3. virus infection acquired in India. J Clin Virol. 2010;48:58–61. 6. Kamar N, Selves J, Mansuy JM, Ouezzani L, Péron JM, Guitard J, et al. Hepatitis E virus and chronic hepatitis in organ-transplant 22. Colson P, Coze C, Gallian P, Henry M, De Micco P, Tamalet C. recipients. N Engl J Med. 2008;358:811–7. Transfusion-associated hepatitis E, France. Emerg Infect Dis. 10.1056/NEJMoa0706992 2007;13:648–9. 7. Hewitt PE, Ijaz S, Brailsford SR, Brett R, Dicks S, Haywood B, 23. Purcell RH, Engle RE, Rood MP, Kabrane-Lazizi Y, Nguyen HT, et al. Hepatitis E virus in blood components: a prevalence and Govindarajan S, et al. Hepatitis E virus in rats, Los Angeles, transmission study in southeast England. Lancet. 2014;384: California, USA. Emerg Infect Dis. 2011;17:2216–22. 1766–73. 8. Sridhar S, Teng JLL, Chiu TH, Lau SKP, Woo PCY. Hepatitis E 24. Wang B, Cai CL, Li B, Zhang W, Zhu Y, Chen WH, et al. virus genotypes and evolution: emergence of camel hepatitis E Detection and characterization of three zoonotic viruses in variants. Int J Mol Sci. 2017;18:E869. wild rodents and shrews from Shenzhen city, China. Virol Sin. ijms18040869 2017;32:290–7. 9. Sridhar S, Chan JFW, Yap DYH, Teng JLL, Huang C, Yip CCY, 25. Mulyanto, Depamede SN, Sriasih M, Takahashi M, Nagashima S, et al. Genotype 4 hepatitis E virus is a cause of chronic Jirintai S, et al. Frequent detection and characterization of hepatitis hepatitis in renal transplant recipients in Hong Kong. J Viral E virus variants in wild rats (Rattus rattus) in Indonesia. Arch Hepat. 2018;25:209–13. Virol. 2013;158:87–96. 10. Sridhar S, Lo SK, Xing F, Yang J, Ye H, Chan JF, et al. Clinical 26. Johne R, Plenge-Bönig A, Hess M, Ulrich RG, Reetz J, Schielke A. characteristics and molecular epidemiology of hepatitis E in Detection of a novel hepatitis E-like virus in faeces of wild rats Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2249 RESEARCH using a nested broad-spectrum RT-PCR. J Gen Virol. 2010;91:750– channel required for release of infectious particles. Proc Natl 8. Acad Sci U S A. 2017;114:1147–52. 27. Smith DB, Simmonds P, Jameel S, Emerson SU, Harrison TJ, pnas.1614955114 Meng XJ, et al.; International Committee on Taxonomy of Viruses 31. Cheng VC, Chan JF, Wen X, Wu WL, Que TL, Chen H, et al. Hepeviridae Study Group. Consensus proposals for classification Infection of immunocompromised patients by avian H9N2 of the family Hepeviridae. J Gen Virol. 2014;95:2223–32. influenza A virus. [Erratum in: Proc Natl Acad Sci U S A. 2017]. J Infect. 2011;62:394–9. 28. Dremsek P, Wenzel JJ, Johne R, Ziller M, Hofmann J, Groschup MH, 32. Chen Y, Liang W, Yang S, Wu N, Gao H, Sheng J, et al. Human et al. Seroprevalence study in forestry workers from eastern infections with the emerging avian influenza A H7N9 virus from Germany using novel genotype 3- and rat hepatitis E virus-specific wet market poultry: clinical analysis and characterisation of viral immunoglobulin G ELISAs. Med Microbiol Immunol (Berl). genome. Lancet. 2013;381:1916–25. 2012;201:189–200. S0140-6736(13)60903-4 29. Sanford BJ, Opriessnig T, Kenney SP, Dryman BA, Córdoba L, Address for correspondence: Kwok-Yung Yuen, The University of Hong Meng XJ. Assessment of the cross-protective capability of recombinant capsid proteins derived from pig, rat, and avian Kong, Carol Yu Centre for Infection, State Key Laboratory of Emerging hepatitis E viruses (HEV) against challenge with a genotype 3 HEV Infectious Diseases, Department of Microbiology, Li Ka Shing Faculty in pigs. Vaccine. 2012;30:6249–55. of Medicine, Queen Mary Hospital, 120 Pokfulam Rd, Pokfulam, Hong 30. Ding Q, Heller B, Capuccino JM, Song B, Nimgaonkar I, Kong HK1, China; email: Hrebikova G, et al. Hepatitis E virus ORF3 is a functional ion The Public Health Image Librar y (PHIL) The Public Health Image Library (PHIL), Centers for Disease Control and Prevention, contains thousands of public health- related images, including high-resolution (print quality) photographs, illustrations, and videos. PHIL collections illustrate current events and articles, supply visual content for health promotion brochures, document the effects of disease, and enhance instructional media. PHIL images, accessible to PC and Macintosh users, are in the public domain and available without charge. Visit PHIL at: 2250 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Influences of Community Interventions on Zika Prevention Behaviors of Pregnant Women, Puerto Rico, July 2016–June 2017 Giulia Earle-Richardson, Christine Prue, Khadija Turay, Dana Thomas We assessed how community education efforts influenced Interventions pregnant women’s Zika prevention behaviors during the 2016 Centers for Disease Control and Prevention–Puerto PRDOH Women, Infants, and Children Program Rico Department of Health Zika virus response. Efforts in - Zika Orientation cluded Zika virus training, distribution of Zika prevention kits, During the tracking period, all newly enrolled pregnant a mass media campaign, and free home mosquito spraying. women at 1 of the island’s 92 Women, Infants, and Children We used telephone interview data from pregnant women (WIC) clinics were given a 20–30-minute presentation on participating in Puerto Rico’s Women, Infants, and Children Zika virus infection and prevention. Orientation (individu- Program to test associations between program participa- ally or in small groups) was provided by the nutrition educa- tion and Zika prevention behaviors. Behavior percentages tor or, during the peak of the epidemic, by a Zika educator ranged from 4% (wearing long-sleeved shirt) to 90% (re- moving standing water). Appropriate mosquito repellent use provided by CDC. The primary advantages of this counsel- (28%) and condom use (44%) were common. Receiving a ing approach are interpersonal communication (including Zika prevention kit was significantly associated with larvicide answering questions) and how easily it can be integrated into application (odds ratio [OR] 8.0) and bed net use (OR 3.1), existing trusted programs, such as WIC prenatal visits (5,6). suggesting the kit’s importance for lesser-known behaviors. Offer of free residential spraying was associated with spray - Zika Prevention Kit Distribution ing home for mosquitoes (OR 13.1), indicating that women The Zika prevention kit (ZPK) was a tote bag containing in- supported home spraying when barriers were removed. sect repellent, condoms, a mosquito bed net, larvicide, and printed Zika education materials. Approximately 26,000 n early 2016, in response to the rising number of Zika vi- ZPKs were distributed in Puerto Rico (CDC–Puerto Rico Irus infections in Puerto Rico and the devastating effects Department of Health, unpub. data, April 26, 2017). When- of Zika infection during pregnancy (1), the Puerto Rico ever possible, the ZPK was given to the pregnant woman Department of Health (PRDOH) activated its emergency at the same time as the WIC Zika orientation. Prevention operations center, with support from the US Centers for kits enable healthy behavior by putting needed items in per- Disease Control and Prevention (CDC) (2). Because there sons’ hands but also by providing a visual reminder of the is currently no Zika virus vaccine and no known measures recommended behavior. Similar home infection prevention can prevent prenatal mother-to-child transmission (3), per- kits were used during the Zika response in the US Virgin sonal protection measures and home vector control are the Islands (7) and during the Ebola epidemic in West Africa only feasible protections for most pregnant women. To (8–10) to provide home caregivers with tools to prevent vi- maximize these self-protection behaviors, the response in- rus transmission. Only HIV infection prevention kits have troduced 4 different community Zika prevention behavior been evaluated to date; these preliminary evaluations indi- promotion interventions. Health behavior interventions can cate kit popularity and suggest supportive effects ( 11,12). change behavior by addressing behavioral barriers, by cre- ating or enhancing incentives, and by increasing persons’ Detén el Zika Campaign capabilities and opportunities to perform the behavior (4). The Detén el Zika (“This Is How We Stop Zika”) cam- paign disseminated strategically designed Zika prevention Author affiliation: Centers for Disease Control and Prevention, Preliminary results from this study were presented as a poster Atlanta, Georgia, USA presentation at the International Conference on Emerging DOI: Infectious Diseases, August 29, 2018, Atlanta, Georgia, USA. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2251 RESEARCH Table 1. Demographic characteristics of all women giving birth in 2016 and interview participants, Puerto Rico, July 2016–June 2017* Women who gave birth in 2016, no. (%)† Sample size, Characteristic no. (%) >18 y of age All ages Total sample 1,329 (100) 27,230 (100) 28,257 (100) Age group, y <18‡ 0 0 1,027 (4) 18–22 353 (27) 7,963 (29) 7,963 (28) 23–25 324 (24) 5,436 (20) 5,436 (19) 26–29 319 (24) 5,884 (22) 5,884 (21) >30 333 (25) 7,947 (29) 7,947 (28) Total sample 1,329 (100) 27,230 (100) 28,257 (100) Educational attainment Some high school or less 24 (3) 427 (2) 579 (2) Attended or completed 12th grade 285 (31) 9,105 (34)§ 9,958§ (35) Attended or completed university 545 (60) 15,648 (58) 15,670 (55) Attended or completed graduate program 55 (6) 2031 (8) 2,031 (7) Total sample 909¶ (100) 27,230 (100) 28,257 (100) Participation in WIC program# 1,329 (100) 23,679 (87) 24,671 (87) Geographic region of Puerto Rico Metropolitan San Juan 203 (15) 2,864 (11) 2,955 (10) Metropolitan Bayamon 182 (14) 1,556 (6) 1,597 (6) Nonmetropolitan regions 941 (71) 22,810 (83) 23,705 (84) Total sample 1,327 (100) 27,230 (100) 28,257 (100) *NCHS, National Center for Health Statistics (Centers for Disease Control and Prevention); WIC, Women, Infants, and Children Program (US Department of Agriculture Food and Nutrition Service). †Source: NCHS’s US Territories, 2016 natality public use file ( ‡Because women had to be >18 years of age to participate, the <18 age category is empty for the WIC sample. §In the NCHS data, this group includes 9th–12th grade, not just 12th grade. ¶The educational attainment data in the WIC dataset (n = 909) were incomplete. The data here represent 68% of the total sample of 1,329. #Source: WIC ( messages through television, radio, print, and social me- vector control behavior (15–17). This offer of free residential dia channels (13). The television advertisement included mosquito spraying was discontinued in August 2016 after a montage showing couples or pregnant women and their a CDC evaluation found that mosquito populations in and families performing the following behaviors: using repel - around sprayed homes had not changed, probably as a result lent, using condoms, using bed nets, removing standing of movement of mosquitoes from nearby homes (18). water, and installing screens. Mass media campaigns have the advantage of reaching multiple audiences (including Intervention Implementation Monitoring partners, families, and pregnant women not enrolled in As these interventions were being implemented, the re- WIC) with repeating messages that appeal cognitively and sponse behavioral science team conducted monthly tele- emotionally by showing relatable images of women taking phone interviews of a random sample of 300 pregnant preventive steps and by showing a healthy baby (14). women participating in WIC to provide feedback to the response leadership about intervention exposure and wom- Offer of Free Residential Mosquito Spraying Services en’s Zika prevention behavior. A subset of 150 respondents When pregnant women attended their WIC appointments, were asked about their performance of the following 10 they were also offered a free residential mosquito spray - CDC-recommended behaviors: using mosquito repellent, ing service. Upon consent, WIC provided women’s contact using condoms, abstaining from sex, wearing long-sleeved information to a contracted professional spraying service. shirts, wearing long pants, sleeping under a bed net, remov- Across the island, ≈3,400 homes were sprayed through this ing or covering standing water, applying larvicide (in water program. For this analysis, this intervention is defined as that cannot be removed), putting screens on windows and the offer of free residential spraying services, meaning that doors, and spraying home and yard for mosquitoes. This women who report being offered the free service are clas - assessment continued until June 2017, when PRDOH de- sified as exposed to the intervention, regardless of whether clared the Zika epidemic over (19). During 2016–2017, a they chose to use the service. In this way, we can determine total of 9 monthly (in 2017, bimonthly) interview rounds whether having free residential spraying services available were conducted. Our analysis addresses the following: 1) affected the overall frequency of spraying the home (or yard) the proportion of pregnant respondents reached by the 4 for mosquitoes. interventions and the factors associated with exposure; 2) Although we might intuit that making residential spray- the Zika prevention behaviors that were most widely prac- ing free would increase use, the literature contains inconsis- ticed and that were most strongly associated with exposure tent evidence about whether removing cost barriers increases to interventions; and 3) additional factors associated with 2252 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Zika Prevention and Pregnant Women, Puerto Rico Table 2. Respondents exposure to 4 Zika prevention interventions, by demographic characteristics and calendar month, Puerto Rico, July 2016–June 2017* Received WIC Received Exposed to Detén Offered free Zika orientation ZPK el Zika campaign home spraying Characteristic Sample No Yes No Yes No Yes No Yes Pregnancy trimester at interview 1st 26.8 8.4 91.6 32.9 67.1 52.2 47.8 68.1 31.9 2nd 48.6 8.2 91.8 24.6 75.4 45.9 54.1 71.7 28.3 3rd 24.6 3.7 96.3 16.9 83.1 53.4 46.6 52.8 47.2 Total no. 1,329 95 1,230 324 976 600 616 873 448 p value 0.019 0.000 0.052 0.000 Calendar month of interview Jul 2016 11.2 4.8 95.2 4.8 95.2 62.9 37.1 29.7 70.3 Aug 2016 11.1 8.2 91.8 23.8 76.2 59.2 40.8 29.3 70.7 Sep 2016 10.1 6.0 94.0 31.3 68.7 44.4 55.6 34.6 65.4 Oct 2016 11.3 10.7 89.3 41.3 58.7 31.6 68.4 65.8 34.2 Nov 2016 11.3 8.0 92.0 31.3 68.7 35.0 65.0 70.1 29.9 Dec 2016 11.3 4.0 96.0 30.0 70.0 36.1 63.9 72.7 27.3 Feb 2017 11.3 6.7 93.3 20.7 79.3 55.2 44.8 97.3 2.7 Apr 2017 11.3 5.3 94.7 16.0 84.0 68.1 31.9 94.7 5.3 Jun 2017 11.2 10.7 89.3 22.1 77.9 52.9 47.1 96.6 3.5 Total no. 1,329 203 1,230 324 976 600 616 873 418 p value 0.225 0.000 0.000 0.000 Age group, y 18–22 26.6 6.8 93.2 22.3 77.7 51.0 49.0 66.2 33.8 23–25 24.4 7.4 92.6 23.5 76.5 52.1 47.9 68.5 68.5 26–29 24.0 6.9 93.1 27.0 73.0 44.1 55.9 64.9 35.1 >30 25.1 7.5 92.5 27.2 72.8 49.8 50.2 64.8 35.2 Total no. 1,329 95 1,230 324 976 600 616 873 448 p value 0.981 0.356 0.217 0.723 Educational attainment Some high school or less 2.6 0.0 100.0 21.7 78.3 54.5 45.5 66.7 33.3 Attended or completed 12th grade 31.4 7.4 92.6 22.3 77.7 49.4 50.6 66.1 33.9 Attended or completed university 60.0 6.1 93.9 23.6 76.4 49.9 50.1 62.2 37.8 Attended or completed graduate program 6.1 7.3 92.7 25.6 74.1 36.2 63.8 58.2 41.8 Total no. 909 58 848 207 681 404 419 572 332 p value 0.512 0.934 0.315 0.579 Population in poverty in ZIP code, % quartiles† >55 below poverty 25.0 5.1 94.9 22.8 77.2 49.5 50.5 65.2 34.8 49–54 below poverty 25.3 7.9 92.1 25.6 74.4 43.7 56.3 67.5 32.5 43–48 below poverty 25.1 6.7 93.3 23.1 76.9 51.6 48.4 64.2 35.8 <43 below poverty 24.5 8.8 91.2 29.2 70.8 53.1 46.9 68.1 31.9 Total no. 1,255 89 1,163 309 918 566 579 826 421 p value 0.305 0.234 0.125 0.700 Municipality population >200,000 63.5 6.1 93.9 23.1 76.9 48.6 51.4 66.0 34.0 >100,00–200,000 9.9 10.6 89.4 31.3 68.7 50.4 49.6 61.8 38.2 >50,000–100,000 12.6 7.7 92.3 27.9 72.1 46.8 53.2 63.3 36.7 <50,000 14.0 8.8 91.2 26.3 73.7 56.1 43.9 72.5 27.5 Total no. 1,326 91 1,184 313 937 578 589 839 431 p value 0.213 0.163 0.328 0.187 *All data are percentages unless otherwise indicated. Statistically significant differences (p<0.05 by  test) are shown in boldface. WIC, Women, Infants, and Children Program (US Department of Agriculture Food and Nutrition Service); ZPK, Zika prevention kit. †Source: US Census Bureau American Community Survey, 2016. American Community Survey 5-year estimates, Table S1701 (generated by G.B.E. using American Fact Finder, 2018 Feb 24). Zika prevention behavior that might provide insight into 10,000–12,000 women currently enrolled (and not previously how the interventions influenced behavior. contacted) for interviews. Vital statistics data indicate that 87% of women giving birth in Puerto Rico in 2016 were en- Methods rolled in WIC (Table 1). The calling list was divided among interviewers so that some began with first trimester women, Interview Population and Sampling some with second, and some with third. As part of the Zika Each month during July–December 2016 and every 2 months response, these interviews were determined to be nonresearch during February–June 2017, a random sample of 950 preg- public health practice and were approved by the US Office nant women >18 years of age (317 women per pregnancy of Management and Budget (control no. 0920–1196). Be- trimester) was drawn from the WIC enrollment database of fore asking women for their verbal agreement to participate, Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2253 RESEARCH Definition of Intervention Exposure Respondents were asked if they had received the WIC Zika orientation, the ZPK, or the offer of free home spraying. They were also asked if they had seen communications from the Detén el Zika campaign. Any woman answering affirmatively to any of these questions was defined as ex- posed to the corresponding intervention. Data Analysis Calculation of Zika Prevention Behavior Variables Because the original interview instrument included multiple questions about each Zika prevention behavior without any clear formula for integrating question responses into a single variable (1 per behavior), analysts had to create such a for- mula. For example, some questions asked whether a woman Figure 1. Percentage of pregnant women reporting exposure to 4 Zika prevention interventions, by interview month, Puerto Rico, performed the behavior any time during pregnancy (or during 2016–2017. August 12, 2016: President declares Zika in Puerto the previous day or week) (yes or no), whereas others used Rico a “public health emergency” ( ordinal frequency scales (e.g., never, sometimes, or always). us-health-zika-usa/u-s-declares-a-zika-public-health-emergency- In addition, a Zika prevention behavior could be reported in in-puerto-rico-idUSKCN10N2KA). September 30, 2016: Free response to the question, “What actions have you taken to residential spraying discontinued. Women who report the offer through December are referring to receiving the offer before protect yourself from being infected by the Zika virus?” September. October 28, 2016: First baby born with microcephaly To describe women’s Zika prevention behavior as com- in Puerto Rico ( pletely as possible, analysts created behavior variables that zika-microcephaly-puerto-rico.html). June 5, 2017: Zika epidemic incorporated 2, 3, or more questions. We prioritized time- declared over by Puerto Rico Department of Health (https://www. bound, behavior-specific questions, such as, “How often Department-Health-Declared-2016-Zika). WIC, Women, Infants, did you use mosquito repellent in the past week?” (never, and Children Program (US Department of Agriculture Food and sometimes, or always), over a more general question such Nutrition Service). as, “What actions have you taken to protect yourself from being infected with the Zika virus?” Among the behavior- specific questions, those questions with multilevel response interviewers explained the purpose of the data collection, the options were prioritized over yes or no or dichotomous re- fact that their participation and all responses would be kept sponse questions, given that the greater number of response confidential, and that they could discontinue the interview any options yielded more information. Zika prevention behavior time without any penalty. The 3 groups of callers continued variables were then created with ordinal scales, combining until 300 total interviews were completed. The interview had the most detailed behavior-specific question available for the 2 parts, administered 2 weeks apart. Those women who con- behavior with other questions that might serve to increase sented to complete part 2 were called in the same order as for the number of levels of Zika prevention behavior. Once pre- part 1 until 150 interviews were completed. liminary scales were created, frequencies and plots were re- viewed by behavioral scientists and epidemiologists involved Data Collection with the Zika response to achieve a consensus on the final The interview consisted of questions about Zika knowledge, composition. We have compiled a list of all candidate ques- attitudes, sources of information, exposure to prevention tions and final variables (online Technical Appendix, https:// interventions, and Zika prevention behaviors. Many of the questions involved binary (e.g., yes or no) or scaled (e.g., never, rarely, sometimes, frequently, or often) responses. Statistical Methods Others were questions in which the interviewer did not pro- Analysts calculated frequencies of intervention exposure vide response options to the participant but coded the re- by interview month and demographic characteristics. In sponse according to a checklist. Although Zika infection sta- addition, because the interventions sought to increase Zika tus was not an interview question, if a participant disclosed prevention behavior by increasing a woman’s concern that she was Zika positive, the interview was excluded from about Zika, her confidence in her ability to protect herself, the dataset. This exclusion was made because Zika virus in- and involvement of partners and families in Zika preven- fection confers immunity and therefore an already positive tion, variables representing these constructs were tested for woman would have no reason to take prevention steps. 2254 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Zika Prevention and Pregnant Women, Puerto Rico Table 3. Associations between Zika prevention intervention exposure and interpersonal communications about Zika and personal risk perceptions, Puerto Rico, July 2016–June 2017* Received WIC Exposed to Detén Offered free Zika orientation Received ZPK el Zika campaign home spraying Variable Sample No Yes No Yes No Yes No Yes Family and interpersonal communication Frequency of talking to family and friends about Zika Not at all 10.7 8.4 10.8 9.9 11.1 14.5 7.3 12.7 6.7 Only once or twice 16.2 21.1 15.9 16.7 16.1 17.7 15.3 17.4 13.6 Sometimes 32.7 45.3 31.8 34.9 32.3 33.0 32.5 33.3 31.5 Often 22.0 16.8 22.4 18.5 22.7 19.2 23.5 20.3 25.2 Every day 18.4 8.4 19.2 20.1 17.8 15.7 21.4 16.3 23.0 Total no. 1,329 79 1,230 600 616 324 976 873 448 p value 0.009 0.472 0.000 0.000 Aware of Zika prevention actions of family No 38.3 38.8 38.2 38.2 38.4 46.0 31.0 41.4 30.9 Yes 61.7 61.2 61.8 61.8 61.6 54.0 69.0 58.6 69.1 Total no. 1,168 85 1,081 511 561 314 850 818 343 p value 0.910 0.966 0.000 0.001 Individual risk perception How concerned women feel about Zika Not at all concerned 8.2 7.4 8.3 5.6 9.0 8.8 7.8 8.9 6.7 Slightly concerned 16.4 13.7 16.7 14.8 17.1 17.9 15.6 17.7 13.8 Somewhat concerned 21.1 20.0 21.2 20.7 21.1 21.7 21.1 21.6 20.3 Moderately concerned 27.3 33.7 26.6 30.2 26.4 27.4 26.8 27.5 26.8 Extremely concerned 27.0 25.3 27.3 28.7 26.4 24.2 28.7 24.3 32.4 Total no. 1,328 95 1,229 599 616 324 975 872 448 p value 0.665 0.182 0.435 0.019 How likely women feel they will become infected with Zika Extremely unlikely 10.0 9.7 10.0 8.1 10.8 10.2 9.9 12.0 5.9 Unlikely 37.4 36.6 37.4 37.4 37.4 36.6 38.0 36.8 38.4 Neither likely nor unlikely 30.6 32.3 30.5 31.5 30.2 30.3 31.4 31.0 30.2 Likely 19.4 19.4 19.4 20.9 18.8 19.3 18.8 18.0 22.1 Extremely likely 2.7 2.2 2.7 2.2 2.8 3.6 2.0 2.1 3.4 Total no. 1,306 93 1,209 587 606 321 957 855 443 p value 0.994 0.607 0.549 0.003 Confidence in ability to protect self and baby from Zika Not confident at all 1.1 2.1 1.1 1.9 0.9 1.2 1.3 1.3 0.9 Somewhat unconfident 9.9 16.0 9.4 10.3 9.6 10.1 9.3 10.0 9.5 Not confident or unconfident 22.3 27.7 21.8 20.6 22.3 24.7 20.2 21.1 24.8 Confident 49.5 45.7 49.9 48.9 50.1 49.5 49.8 50.7 47.1 Very confident 17.2 8.5 17.9 18.4 17.1 14.6 19.3 16.8 17.8 Total no. 1,319 94 1,221 596 610 321 969 867 444 p value 0.030 0.634 0.139 0.530 *All data indicate percentages unless otherwise indicated. Statistically significant differences (p<0.05 by  test) are shown in boldface. WIC, Women, Infants, and Children Program (US Department of Agriculture Food and Nutrition Service); ZPK, Zika prevention kit. associations with intervention exposure and Zika preven- Because the WIC orientation reached nearly all respon- tion behaviors. All analyses were conducted with SPSS dents, the naturally occurring control group of unexposed 21.0 (IBM Corp., Armonk, NY, USA). women was very small, causing concerns about small cell Analysts used logistic regression modeling to estimate size in models with many covariates (20). Conversely, a odds ratios (ORs) for the likelihood of performing recom- small exposure group was a concern with the offer of free mended Zika prevention behaviors by exposure to 1 of the residential mosquito spraying. Therefore, these 2 interven- Zika prevention interventions while controlling for the ef- tions were modeled separately from ZPK distribution and fects of age, education, pregnancy trimester, poverty, cal- Detén el Zika, which were modeled together. In addition, endar month of interview, and exposure to other interven- sparsity concerns led us to consolidate the calendar month tions. For these models, Zika prevention behavior variable of interview variable into 1 representing 3-month intervals. responses were collapsed into dichotomous (yes or no) variables, indicating whether a respondent had performed Results the ideal behavior (e.g., always uses a condom) or not. In the case of mosquito repellent use, the 2 top levels, which Participant Characteristics both include the response always, were combined to make Our sample encompassed 1,329 pregnant WIC partici- the top level. pants interviewed during July 2016–June 2017 (Table 1). Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2255 RESEARCH Table 4. Zika personal protection behaviors among pregnant women, by exposure to 4 interventions, Puerto Rico, July 2016–June 2017* Received Exposed to Offered free WIC Zika Received Detén el Zika home Entire orientation ZPK campaign spraying Behavior sample Yes No Yes No Yes No Yes No Mosquito repellent use Always, reported reapplying 28.3 29.1 18.9 29.7 24.5 31.2 25 32.6 26.1 Always, did not report reapplying 23.9 23.5 28.4 24.1 23.5 27.6 21.5 23.9 23.9 Usually or most of the time 25.9 26.4 21.1 25.9 26 23.1 28.4 23.2 27.1 Sometimes 13.0 12.8 14.7 12.2 15.2 11.7 14.5 13.2 13.1 Rarely or seldom 4.6 4.4 7.4 4.7 4.0 3.9 4.7 4.0 4.9 Never 4.2 3.8 9.5 3.4 6.8 2.6 5.8 3.1 4.8 Total no. 1,328 1,229 95 976 323 614 597 448 873 p value 0.018 0.016 <0.001 0.012 Condom use† Always 44.1 45.3 31.6 45.1 42.6 44.2 26.3 42.5 44.8 Sometimes 29.3 29.5 24.1 30.6 25.8 28.7 26.3 28.3 29.9 Never 26.6 25.2 44.3 24.3 31.6 27.2 47.4 29.2 25.3 Total no. 1,047 964 79 768 256 491 464 353 689 p value 0.001 0.130 0.001 0.266 Bed net use Slept under bed net yesterday 14.8 15.4 7.4 17.7 6.8 16.1 13.8 13.8 15.3 Did not use yesterday, reports use generally 4.9 5.2 1.1 5.7 2.5 4.2 4.7 3.1 5.8 Did not use yesterday, does not report use generally 80.3 79.4 91.6 76.5 90.7 79.7 81.5 83 78.8 Total no. 1,329 1,230 95 976 324 616 600 448 873 p value 0.005 <0.001 0.390 0.094 Wearing long pants Wearing now, every day, all day 21.3 21.4 21.1 20.6 23.5 21.2 20.8 20.6 21.5 Wearing now, every day, part of day 19.2 19.5 15.8 18.7 21.0 20.4 18.3 19.7 19.0 Wearing now, does not wear every day 20.0 20 21.1 20.0 19.4 20.5 20 17.7 21.3 Not wearing long pants now 39.4 39.1 42.1 40.7 36.1 40.8 37.9 41.9 38.1 Total no. 1,327 1,228 95 974 324 614 600 446 873 p value 0.549 0.098 0.378 0.402 Sexual abstinence Had no sex during pregnancy 20.2 20.7 15.8 20.3 19.9 31.2 25.0 20.6 19.9 Had sex during pregnancy 79.8 79.3 84.2 79.7 80.1 80.6 78.2 79.4 80.1 Total no. 1,324 1,225 95 973 322 614 597 447 869 p value 0.256 0.855 0.303 0.773 Wearing long-sleeved shirt Wearing now, every day, all day 3.9 3.8 5.3 3.7 4.7 77.7 79.3 4.0 3.8 Wearing now, every day, part of day 6.7 6.7 7.4 7.2 5.6 6.4 7.2 6.9 6.7 Wearing now, does not wear every day 10.6 10.8 7.4 9.9 13.7 11.1 10.4 8.9 11.5 Not wearing long sleeves now 78.7 78.6 79.8 79.3 79.3 79.3 83.5 80.1 78 Total no. 1,325 1,227 94 974 322 614 598 448 869 p value 0.915 0.289 0.464 0.457 *All data indicate percentages unless otherwise indicated. Statistically significant differences (p<0.05 by Mann-Whitney U nonparametric test) are shown in bold. WIC, Women, Infants, and Children Program (US Department of Agriculture Food and Nutrition Service); ZPK, Zika prevention kit. †Among those reporting having had sex during pregnancy. Among eligible women (i.e., >18 years of age, pregnant, association with exposure to all 4 interventions, whereas and not Zika positive), the response rate was 79%. Age calendar month of interview was significantly associated and educational attainment distributions of the sample with 3 interventions (Table 2). No significant associa - were similar to the general population of women giving tions were observed in terms of age, education, poverty, birth in Puerto Rico in 2016 (21), whereas urban resi- or rurality. dence is somewhat higher. Graphed by calendar month of interview (Figure 1), exposure to the WIC Zika orientation remained consistent- Women’s Exposure to 4 Zika Prevention Interventions ly high (89%–96%). ZPK distribution began high (95%), Women reported exposure to the 4 interventions as dropped in October, then rebounded. Detén el Zika cam- follows: WIC Zika orientation (93%), ZPK distribu - paign exposure began much lower (37%), then steadily tion (75%), Detén el Zika campaign (51%), and of - increased through October (68%), dropped off, and rose fer of free residential mosquito spraying (68% for the again in 2017. Exposure to the offer of free residential mos- months it was running and 34% over the entire period). quito spraying started at 70% in July 2016, then dropped Pregnancy trimester was statistically significant for precipitously after September. 2256 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Zika Prevention and Pregnant Women, Puerto Rico Table 5. Zika home protection behaviors among pregnant women, by exposure to 4 interventions, Puerto Rico, July 2016–June 2017* Received Exposed to Offered free WIC Zika Received Detén el Zika home Samples, % orientation ZPK campaign spraying Behavior (no.) Yes No Yes No Yes No Yes No Removing (or covering) standing water* Removed standing water in past week 90.3 (531) 90.5 87.2 91.8 85.5 93.9 85.6 91.3 90.2 Has not in past week; reports action generally 1.2 (7) 1.1 2.6 0.9 2.2 1.3 1.2 1.9 0.8 Has not in past week; does report action generally 8.5 (50) 8.4 10.3 7.3 12.3 4.7 13.2 6.8 9.0 Total no. 588 546 39 438 138 297 243 377 206 p value 0.516 0.032 0.001 0.637 Spraying home (or yard) for mosquitoes Sprayed for mosquitoes (self or service) 43.1 (569) 43.7 33.7 44.4 37.7 42.6 43.2 82.3 22.9 No home spraying 56.9 (752) 56.3 66.3 55.6 62.3 57.4 56.8 17.7 77.1 Total no. 1,321 1,222 95 971 321 615 595 446 873 p value 0.058 0.036 0.835 <0.001 Larvicide application† Has applied larvicide around home (self or family) 31.3 (308) 24.2 10.8 40.5 7.9 30.0 32.9 20.1 37.3 Never applied larvicide around home (self or family) 68.7 (675) 75.8 89.2 59.5 92.1 70.0 67.1 79.9 62.7 Total no. 983 1,229 93 708 253 476 423 334 641 p value 0.002 <0.001 0.364 <0.001 Installing window or door screens Reports putting screens on windows, doors 17.8 (236) 17.4 22.1 17.6 18.5 18.0 18.7 18.1 17.5 Does not report putting screens on windows, doors 82.2 (1,093) 82.6 77.9 82.4 81.5 82.0 81.3 81.9 82.5 Total no. 1,329 1,230 95 976 324 616 600 448 873 p value 0.247 0.715 0.771 0.803 *All data indicate percentages unless otherwise indicated. Statistically significant differences (p<0.05 by Mann-Whitney U nonparametric test) are shown in boldface. WIC, Women, Infants, and Children Program (US Department of Agriculture Food and Nutrition Service); ZPK, Zika prevention kit. †Among those having yards for which they are responsible, and where water was present. Intervention exposure was more often significantly Zika Home Protection Behaviors associated with family or interpersonal communica- We ranked home protection behaviors from the most fre- tion variables than with individual risk variables (Table quent (removing standing water [90%]) to the least (install - 3). The same pattern was observed for associations with ing window or door screens [18%]) (Table 5). Over time, Zika prevention behaviors (data not shown): “frequency removing standing water declined slightly through Septem- of talking to family and friends about Zika” was signifi - ber but then remained at >85%, whereas spraying the home cantly associated with 10 behaviors and “aware of Zika for mosquitoes had a steep decline during August–June prevention actions of family” with 5 behaviors, whereas 2017 (Figure 3). In contrast, larvicide application began all 3 individual risk perception–related variables were as- low (13%) and then increased through June 2017 (40%). sociated with <3 behaviors. Independent Associations between Interventions and Pregnant Women’s Zika Personal Protection Behaviors Zika Prevention Behaviors Frequencies of recommended personal protection behav- In multivariable logistic regression models, we observed iors (i.e., the top level on the ordinal scale) ranged from 4% a strong association between the offer of free residential (wearing long-sleeved shirt) to 44% (condom use) (Table mosquito spraying services and spraying the home for mos- 4). Although just over half of women reported using repel- quitoes (Table 6). We also observed strong associations be- lent always, fewer (28%) reported the top category, “used tween ZPK receipt and larvicide application and between always and reported reapplying it.” Among the interven- ZPK receipt and bed net use. tions, exposure to the WIC Zika orientation showed the greatest exposed versus not exposed frequency differences Discussion for the top behavior levels (Tables 4, 5). For each intervention, exposure patterns corresponded Over the monthly interview cohorts, the top level of with implementation history; WIC orientation exposure condom use rose steadily with a sustained peak at over was consistently high, Detén el Zika campaign exposure 50%, whereas mosquito repellent use rose to 42%, declined, grew over time, ZPK exposure faltered (because of lo- and peaked again in December (Figure 2). Wearing long gistical problems with kit distribution) and then recov- pants had 2 peaks (in October and December) near 30%, ered, and the free offer of home mosquito spraying was then a steep decline in 2017, whereas sexual abstinence widely received during the offer period. These largely stayed near 20%. Bed net use peaked at 23% in September, successful implementations illustrate the benefits of col - then fluctuated. laborating with a trusted local partner like WIC. WIC Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2257 RESEARCH Figure 3. Percentage of women reporting highest levels of 4 Zika home protection behaviors, by interview month, Puerto Rico, July 2016–June 2017. See Figure 1 legend for timeline of events. Figure 2. Percentage of women reporting highest levels of 6 Zika personal protection behaviors, by interview month, Puerto Rico, July 2016–June 2017. See Figure 1 legend for timeline of events. behavior and some of the strongest associations. This find- ing is consistent with a small but growing body of literature demonstrating the effectiveness of distributing items for was able to incorporate Zika orientations into its regular encouraging prevention behavior (11,23,24). Prevention programming, distribute ZPKs effectively, and provide kits containing prevention products for at-risk populations the free offer of home spraying during a WIC visit. WIC should be considered a best practice, particularly in low- also played an important role in developing the Detén el resource settings. Zika messaging. Larvicide use and bed net use were independently Performance of Zika prevention behaviors varied associated with ZPK receipt, and distributing items widely. Nearly all women removed any standing water associated with these 2 largely unfamiliar behaviors that they saw, and about three quarters usually or al- probably increased use because women were then able ways used mosquito repellent, but very few wore long to try them. According to Rogers’ diffusion of innova - sleeves or put up screens. These findings are consistent tions theory (25), the ability to try a new behavior and with the Pregnancy Risk Assessment Monitoring System observe the results enhances the likelihood of adoption. Zika Postpartum Emergency Response (PRAMS-ZPER) Larvicide application might have been further enhanced study of postpartum women in Puerto Rico (22). Despite by what Rogers calls “relative advantage”; that is, the important methodologic differences between PRAMS- intervention might have been popular because it was ZPER and our analysis, reported frequencies were simi- easier to implement than the other 3 recommended home lar for mosquito repellent use, removing standing water, protection behaviors (removing standing water, install- bed net use, and wearing long sleeves. Where frequencies ing screens, and spraying home for mosquitoes). Many diverged (condom use and spraying home for mosqui- of the ZPKs in the early months of tracking were missing toes), WIC sample frequencies were more similar to larvicide tablets; thus, the dramatic increase in larvicide PRAMS-ZPER when limited to women in their third tri- use over the period is not surprising. The finding also mester. In contrast, interview data from US Virgin Islands suggests that the actual association between ZPKs and in late 2016 (7) showed lower frequencies of using repel- larvicide use is stronger than what our results indicate, lent, using condoms, removing standing water, and spray- given that the incomplete kits might have diluted the ob- ing home for mosquitoes. Only data for bed net use were served association. similar to the results of our analysis. Offer of free residential mosquito spraying ser - Overall, the ZPK distribution had the greatest number vices was strongly associated with spraying the home of independent positive associations with Zika prevention 2258 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Zika Prevention and Pregnant Women, Puerto Rico Table 6. Logistic regression models for Zika prevention behaviors performed by pregnant women that were significantly associated with >1 Zika prevention interventions, Puerto Rico, July 2016–June 2017*† Odds ratio (95% CI) Recalled Detén el Received WIC Offered free Behavior Received ZPK Zika campaign Zika orientation residential spraying Personal protection behaviors Bed net use 3.1 (1.9–5.1) 1.2 (0.8–1.7) 2.2 (1.0–4.8) NA Condom use‡ 1.4 (0.9–2.1) 1.4 (1.0–2.0) 2.4 (1.2–4.7) NA Mosquito repellent use 1.5 (1.1–2.0) 1.6 (1.2–2.1) 1.2 (0.8–1.9) NA Sexual abstinence 0.9 (0.6–1.4) 0.9 (0.6–1.3) 1.2 (0.5–2.5) NA Wearing long sleeves 1.9 (0.6–6.2) 2.9 (0.9–8.8) 1.9 (0.2–14.9) NA Wearing long pants 1.1 (0.7–1.7) 1.0 (0.7–1.5) 1.4 (0.6–3.0) NA Home protection behaviors Larvicide application 8.0 (4.8–13.3) 0.8 (0.6–1.1) 2.7 (1.4–5.5) 0.4 (0.3–0.5) Spraying home or yard for mosquitoes 1.5 (1.1–2.3) 1.0 (0.7–1.4) 1.6 (0.9–2.9) 13.1 (8.5–20.3) Removing or covering standing water 2.2 (0.8–5.7) 2.7 (1.1–6.5) 0.5 (0.1–4.4) 1.1 (0.4–2.9) Installing window or door screens 0.8 (0.6–1.2) 0.8 (0.6,1.2) 0.7 (0.4–1.5) 0.9 (0.6–1.5) *Bold indicates significant result. WIC, Women, Infants, and Children Program (US Department of Agriculture Food and Nutrition Service); ZPK, Zika prevention kit. †Models for WIC orientation and offer of free residential spraying were modeled separately, whereas ZPK distribution and Detén el Zika recall were modeled together to measure independent effects. Thus, each Zika prevention behavior had 3 models. To reduce possible bias associated with sparse data, calendar month of interview was consolidated into a 3-level, 3-month variable. All 5 demographic variables and consolidated calendar month of interview were controlled for in each model, except for the following cases: 1) WIC orientation did not include any calendar month of interview variable; or 2) very few respondents did not receive WIC orientation, thus the naturally occurring control group was very small. To not bias the models, no time of interview variable was included in models of WIC orientation. Education was excluded from bed net, larvicide, and repellent use models. Because of the substantial amount of missing data for education, additional testing was performed to determine whether women with missing education data performed the 10 behaviors with significantly higher or lower frequency. Three behaviors (repellent, bed net, and larvicide use) were significantly associated with whether education data were missing, so education was not included in these models. No calendar month or consolidated month variable was used for any of the larvicide use models because of small cell sizes. ‡Among those reporting having had sex during pregnancy. for mosquitoes, enabling women to overcome both cost ing home for mosquitoes), and 3.1 (ZPK exposure and bed net and logistical barriers. Although efficacy concerns led to use) roughly, convert to risk ratios of 5.2, 3.5, and 2.7, respec- discontinuation of the spraying program, the offer had a tively, whereas the more modest ORs of 2.7 (Detén el Zika strong association with spraying behavior, a finding con - campaign exposure and removing standing water and WIC sistent high percentage (81%) of respondents who rated orientation and larvicide application) and 2.4 (WIC orienta- the offer of insecticide spraying to pregnant women as tion and condom use) undergo a smaller adjustment (1.1, 2.2, very important. and 1.7, respectively). Further research is needed to evaluate The Detén el Zika campaign had the greatest inde- these associations more precisely. pendent effect on removing standing water, significant In our exploration of intervention mechanisms, the 2 effects for repellent use, and modest (marginally signifi - interpersonal communication variables showed stronger as- cant) effects for condom use, whereas the WIC orientation sociation with the interventions and to the Zika prevention appeared to have a slightly greater effect on condom use. behaviors than did the individual variables (Zika concern, Although WIC Zika orientation did not yield the same perceived likelihood of infection, and self-confidence). large number of positive associations in regression mod- This finding suggests that the interpersonal factors were els as was observed in the bivariate analyses, its highly more influential on behavior than individual risk percep - successful implementation left it with a very small natu- tions. Interpersonal communication has long been recog- ral control group, which might have limited the utility of nized as an important mediator of the effects of educational modeling for this intervention. campaigns on health-related behavior change (29–33), and As we consider the public health implications of these re- our results confirm this assertion. sults, we should note that in the context of cross-sectional data The main challenge of this analysis was that the data with outcomes that are not rare, ORs do not equate to relative were collected during an emergency response for nonre- risk. Thus, we cannot say that women receiving the free offer search purposes, meaning that much of the analysis design of home mosquito spraying were 13 times more likely to spray had to be created after the fact, particularly the creation their homes. Unfortunately, estimating relative risks from ORs of Zika behavior outcome variables. Further, this analy- is not straightforward. Simple conversion formulas (26) have sis did not use an optimal research design (i.e., there were been shown to be imprecise (27), but such conversions can no pre–post groups or predesignated control groups). The provide at least a rough sense of the extent to which relative resulting imbalances in naturally occurring control groups risk is more modest than odds with nonrare outcomes (28). prevented the use of a single logistic model for all 4 in- For example, the ORs of 8.0 (ZPK exposure and larvicide ap- terventions. However, the use of random sampling from plication), 13.1 (offer of free residential spraying and spray - a frame representing 87% of the island’s pregnant women Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2259 RESEARCH 4. Michie S, Atkins L, West R. The behaviour change wheel. A guide and logistic regression modeling to control confounding by to designing interventions. London: Silverback Publishing; 2014. demographic factors provide a credible first look at pos - 5. Jones KE, Yan Y, Colditz GA, Herrick CJ. Prenatal counseling on sible effects of Zika prevention interventions during an type 2 diabetes risk, exercise, and nutrition affects the likelihood epidemic response. of postpartum diabetes screening after gestational diabetes. J Perinatol. 2018;38:315–23. Among the 4 intervention strategies, ZPK distribu- 6. Whitlock EP, Orleans CT, Pender N, Allan J. Evaluating primary tion appears to have significant independent effects on care behavioral counseling interventions: an evidence-based the greatest number of Zika prevention behaviors. Con- approach. Am J Prev Med. 2002;22:267–84. sistent with the literature, this intervention should be 10.1016/S0749-3797(02)00415-4 7. Prue CE, Roth JN Jr, Garcia-Williams A, Yoos A, Camperlengo L, considered a best practice for behavioral support in in- DeWilde L, et al. Awareness, beliefs, and actions concerning Zika fectious disease outbreaks, particularly in low-resource virus among pregnant women and community members—U.S. settings. Social context factors appeared to be more in- Virgin Islands, November–December 2016. MMWR Morb fluential in Zika prevention behavior than personal risk Mortal Wkly Rep. 2017;66:909–13. mmwr.mm6634a4 assessment and self-efficacy factors, whereas Zika pre - 8. Samaritan’s Purse. Samaritan’s Purse launches bold new initiative vention behaviors that enable women to try out lesser- to combat Ebola [cited 2018 Feb 28]. https://www.samaritanspurse. known behaviors appeared to garner greater acceptance org/our-ministry/samaritans-purse-launches-bold-new-initiative-to- than other behaviors. Areas for future research include combat-ebola-10-07-14-press-release 9. Sifferlin A. This is how Ebola Patients are equipping their homes developing the evidence base for Zika prevention be- [cited 2018 Feb 9]. havior effectiveness and more precise quantification of treat-ebola-patients intervention mechanisms and effects. 10. Meltzer MI, Atkins CY, Santibanez S, Knust B, Petersen BW, Ervin ED, et al. Estimating the future number of cases in the Ebola epidemic—Liberia and Sierra Leone, 2014–2015. MMWR Suppl Acknowledgments 2014;63:1–14. We thank the pregnant women across Puerto Rico who gave 11. Colindres P, Mermin J, Ezati E, Kambabazi S, Buyungo P, their time and experience to participate in the interviews; the Sekabembe L, et al. Utilization of a basic care and prevention Puerto Rico Department of Health Women, Infants, and package by HIV-infected persons in Uganda. AIDS Care. 2008;20:139–45. Children (WIC) Program leaders and staff; Brenda Rivera- 12. Mabude ZA, Beksinska ME, Ramkissoon A, Wood S, Folsom M. Garcia, Carla Espinet Crespo, Tyler Sharp, Steve Waterman, A national survey of home-based care kits for palliative HIV/AIDS Katherine Lyon Daniel, Amy McMillen, John O’Connor, care in South Africa. AIDS Care. 2008;20:931–7. Carmen Perez, Nicki Pesik, Lyle Petersen, Lee Samuel, Eunice 10.1080/09540120701768438 13. International RTI. Detén el Zika/Stop Zika Campaign, a Soto, Laura Youngblood, Jeffrey Zirger, Mahmoud K. Aboukheir, comprehensive education campaign to fight the Zika epidemic Consuelo Abriles, Jorge Carlo, Pollyanna R. Chavez, Alexander in Puerto Rico [cited 2018 Feb 9]. Cruz-Benitez, Gabriela Escutia, Roberta Lugo, Gisela Medina, DeténDetén-el-zika-stop-zika-campaign Brian D. Montalvo-Martínez, Carlos G. Grana-Morales, 14. Wakefield MA, Loken B, Hornik RC. Use of mass media campaigns to change health behaviour. Lancet. 2010;376:1261–71. Rosalyn Plotzker, Clarissa Valdez, Santos Villarán, and Max Zarate-Bermudez. 15. Boene H, González R, Valá A, Rupérez M, Velasco C, Machevo S, et al. Perceptions of malaria in pregnancy and acceptability of preventive interventions among Mozambican pregnant women: About the Author implications for effectiveness of malaria control in pregnancy. Dr. Earle-Richardson is an epidemiologist and behavioral PLoS One. 2014;9:e86038. scientist with CDC’s National Center for Emerging and journal.pone.0086038 Zoonotic Infectious Diseases. Her research interests include 16. Piltch-Loeb R, Abramson DM, Merdjanoff AA. Risk salience of a novel virus: US population risk perception, knowledge, and the relationship between human behavior and health, the role receptivity to public health interventions regarding the Zika virus of behavioral science in emergency response, and how culture prior to local transmission. PLoS One. 2017;12:e0188666. affects health. 17. Jambulingam P, Gunasekaran K, Sahu S, Vijayakumar T. Insecticide treated mosquito nets for malaria control in India- References experience from a tribal area on operational feasibility and 1. Rasmussen SA, Jamieson DJ, Honein MA, Petersen LR. Zika virus uptake. Mem Inst Oswaldo Cruz. 2008;103:165–71. and birth defects—reviewing the evidence for causality. N Engl J Med. 2016;374:1981–7. 18. Adams L, Bello-Pagan M, Lozier M, Ryff KR, Espinet C, 2. Thomas DL, Sharp TM, Torres J, Armstrong PA, Munoz-Jordan J, Torres J, et al. Update: ongoing Zika virus transmission—Puerto Ryff KR, et al. Local transmission of Zika virus—Puerto Rico, Rico, November 1, 2015–July 7, 2016. MMWR Morb Mortal November 23, 2015–January 28, 2016. MMWR Morb Mortal Wkly Wkly Rep. 2016;65:774–9. Rep. 2016;65:154–8. mmwr.mm6530e1 3. Petersen LR, Jamieson DJ, Powers AM, Honein MA. Zika Virus. 19. BusinessWire. The Puerto Rico Department of health has declared that N Engl J Med. 2016;374:1552–63. the 2016 Zika epidemic is over; Zika transmission has substantially NEJMra1602113 decreased in Puerto Rico below epidemic levels [cited 2018 Feb 28]. 2260 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Zika Prevention and Pregnant Women, Puerto Rico 27. McNutt LA, Hafner JP, Xue X. Correcting the odds ratio in Puerto-Rico-Department-Health-Declared-2016-Zika cohort studies of common outcomes. JAMA. 1999;282:529. 20. Greenland S, Mansournia MA, Altman DG. Sparse data bias: a problem hiding in plain sight. BMJ. 2016;352:i1981. 28. McNutt LA, Wu C, Xue X, Hafner JP. Estimating the relative risk in cohort studies and clinical trials of common outcomes. Am J 21. Centers for Disease Control and Prevention National Center for Epidemiol. 2003;157:940–3. Health Statistics. US Territories, 2016 natality public use file 29. Rogers EM. A prospective and retrospective look at the diffusion [cited 2018 Feb 8]. model. J Health Commun. 2004;9(Suppl 1):13–9. vitalstatsonline.htm 22. D’Angelo DV, Salvesen von Essen B, Lamias MJ, Shulman H, 30. Bandura A. Health promotion by social cognitive means. Health Hernandez-Virella WI, Taraporewalla AJ, et al. Measures Educ Behav. 2004;31:143–64. 10.1177/ taken to prevent Zika virus infection during pregnancy 1090198104263660 Puerto Rico, 2016. MMWR Morb Mortal Wkly Rep. 2017;66:574– 31. Jeong M, Bae RE. The effect of campaign-generated interpersonal 8. communication on campaign-targeted health outcomes: a 23. Robinson MN, Tansil KA, Elder RW, Soler RE, Labre MP, meta-analysis. Health Commun. 2018;33:988–1003. Mercer SL, et al. Mass media health communication campaigns combined with health-related product distribution: a community 32. Gainforth HL, Latimer-Cheung AE, Athanasopoulos P, Moore S, guide systematic review. Am J Prev Med. 2014;47:360–71. Ginis KA. The role of interpersonal communication in the process of knowledge mobilization within a community-based 24. Noar SM. A 10-year retrospective of research in health mass organization: a network analysis. Implement Sci. 2014;9:59. media campaigns: where do we go from here? J Health Commun. 2006;11:21–42. 33. Katz E, Lazarsfeld P, Roper E. Personal influence. New York: 25. Rogers EM. Diffusion of innovations. Fifth edition. New Y ork: Routledge; 2005. Free Press; 2003. 26. Zhang J, Yu KF. What’s the relative risk? A method of Address for correspondence: Giulia Earle-Richardson, Centers for correcting the odds ratio in cohort studies of common outcomes. Disease Control and Prevention, 1600 Clifton Rd NE, Mailstop H16-5, JAMA. 1998;280:1690–1. Atlanta, GA 30329-4027, USA; email: jama. 280.19.1690 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2261 RESEARCH Emerging Multidrug-Resistant Hybrid Pathotype Shiga Toxin–Producing Escherichia coli O80 and Related Strains of Clonal Complex 165, Europe Aurélie Cointe, André Birgy, Patricia Mariani-Kurkdjian, Sandrine Liguori, Céline Courroux, Jorge Blanco, Sabine Delannoy, Patrick Fach, Estelle Loukiadis, Philippe Bidet, Stéphane Bonacorsi Enterohemorrhagic Escherichia coli serogroup O80, involved most serious complication, particularly in young children, in hemolytic uremic syndrome associated with extraintestinal is hemolytic uremic syndrome (HUS), defined by a com - infections, has emerged in France. We obtained circularized bination of renal failure, thrombocytopenia, and hemolytic sequences of the O80 strain RDEx444, responsible for hemo- anemia (1). Post-STEC HUS is a major worldwide public lytic uremic syndrome with bacteremia, and noncircularized health concern because it is the primary cause of acute renal sequences of 35 O80 E. coli isolated from humans and ani- failure in children (1). These clinical features result mainly mals in Europe with or without Shiga toxin genes. RDEx444 from the action of the phage-encoded Shiga toxin (Stx), of harbored a mosaic plasmid, pR444_A, combining extrain- which there are 2 types: Stx1, which has 3 subtypes, Stx1a, testinal virulence determinants and a multidrug resistance– 1c, and 1d; and Stx2, which has 7 subtypes, Stx2a–g. In typi- encoding island. All strains belonged to clonal complex 165, cal EHEC, adhesion to the intestinal epithelium is mediated which is distantly related to other major enterohemorrhagic E. by the locus of enterocyte effacement (LEE), a chromosomal coli lineages. All stx-positive strains contained eae-ξ, ehxA, and genes characteristic of pR444_A. Among stx-negative pathogenicity island (PAI), shared with Enteropathogenic strains, 1 produced extended-spectrum β-lactamase, 1 har- E. coli (EPEC) strains, which encodes a type III secretion bored the colistin-resistance gene mcr1, and 2 possessed system (T3SS), an adhesin called intimin, and its receptor genes characteristic of enteropathogenic and pyelonephritis Tir. Intimin, encoded by the eae gene, is a major virulence E. coli. Because O80–clonal complex 165 strains can inte- factor (VF) involved in the intimate attachment of typical grate intestinal and extraintestinal virulence factors in com- EHEC to intestinal epithelium, causing characteristic attach- bination with diverse drug-resistance genes, they constitute ing and effacing lesions. EHEC enterohemolysin ( ehxA) is dangerous and versatile multidrug-resistant pathogens. a pore-forming cytolysin carried by a plasmid involved in EHEC virulence. This plasmid, initially described as part of nterohemorrhagic Escherichia coli (EHEC), a subset of the O157 serogroup (pO157) (2), can carry 2 additional VFs, EShiga toxin–producing E. coli (STEC), are major food- a catalase peroxidase, encoded by katP, and a serine prote- borne pathogens responsible for outbreaks and sporadic cas- ase, encoded by espP, which can cleave human coagulation es of gastrointestinal diseases ranging from simple diarrhea factor V and might be involved in the development of hem- to hemorrhagic colitis, characterized by bloody diarrhea. The orrhagic colitis (3). Serogroup O157 is the predominant STEC serogroup Author affiliations: Hôpital Robert-Debré (AP-HP), Paris, France worldwide, but non-O157 serogroups are increasingly as- (A. Cointe, A. Birgy, P. Mariani-Kurkdjian, S. Liguori, C. Courroux, sociated with post-STEC HUS, and the unusual serogroup P. Bidet, S. Bonacorsi); Université Paris Diderot, Sorbonne Paris O80 is emerging in France and Europe. In 2016, O80 rep- Cité, Paris (A. Cointe, A. Birgy, P. Mariani-Kurkdjian, P. Bidet, resented the second most frequent serogroup isolated in S. Bonacorsi); Universidade de Santiago de Compostela, Lugo, France, after serogroup O26 (4). This phenomenon is no Spain (J. Blanco); ANSES, Plateforme IdentyPath, Maisons-Alfort, longer restricted to France; strains of serotype O80:H2, all France (S. Delannoy, P. Fach); Université de Lyon, CNRS, INRA, belonging to sequence type (ST) 301, have been identified UCBL, VetAgro Sup, Laboratoire d’Écologie Microbienne, in Spain (5), the Netherlands (6), and Switzerland (7). Villeurbanne, France (E. Loukiadis) This serogroup is unique for several reasons. First, it is always associated with multiple determinants of resistance DOI: 2262 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Hybrid Pathotype Shiga Toxin–Producing E. coli O80 (i.e., resistance to aminopenicillin, aminoglycoside, na- lidixic acid, cotrimoxazole, tetracycline, or phenicols), whereas a resistance phenotype is uncommon among EHECs, which are generally fully susceptible to antibiot- ics, except for rare clones, such as the epidemic O104:H4 German clone carrying a bla gene (8). Furthermore, CTX-M-15 unusual extraintestinal infections have recently been de- scribed for this serogroup (9,10), such as bacteremia, whereas EHEC is generally known to be a strictly intes- tinal pathogen. A recent case in the Netherlands illustrates the potential extreme pathogenicity of this serogroup; a 16-month-old boy died from multiorgan failure and exten- sive cerebral thrombotic microangiopathy attributable to an O80 Stx2d-producing E. coli strain (6). Finally, O80 EHEC appears to be a hybrid pathotype that combines intestinal VFs (Shiga toxin [stx], intimin [eae], enterohemolysin [ehxA]) and extraintestinal VFs (aerobactin [iucC]; salmo- chelin [iroN], an iron uptake protein encoded by sitABCD; serum resistance protein [issp]; a putative secretion system I [etsC]; omptin [ompTp]; hemolysin [hlyF], and 2 bacterio- cins [cia and cva]), suggesting the presence of a pS88-like plasmid (11). pS88 is a ColV plasmid, a key determinant of extraintestinal pathogenic E. coli virulence in poultry and humans. This plasmid is involved in neonatal meningitis (11) and could explain the occurrence of extraintestinal dis- semination in these EHEC infections. The recent diffusion in Europe, high potential extraintestinal pathogenicity, and multidrug resistance (MDR) of this hybrid pathotype led us to further characterize these strains, which might represent a major public health concern. Methods We further characterized O80:H2 EHEC by fully sequencing a recent representative strain, called RDEx444. RDEx444 was responsible for a highly severe case of post-STEC HUS, complicated by bacteremia, in an 8-month-old infant in Feb- ruary 2016 in Bourg-en-Bresse, France. Initial symptoms were febrile diarrhea with signs of sepsis. Blood and stool Figure 1. Genetic characterization of RDEx444, a strain of cultures were positive for O80 EHEC, but urine cultures Escherichia coli serotype O80:H2 isolated in France in February remained negative, suggesting gut translocation that led to 2016 and involved in hemolytic uremic syndrome with bacteremia, bacteremia. The appearance of biologic signs of HUS with carrying both intestinal and extraintestinal virulence factors oliguria necessitated transfer to intensive care, several blood associated with (multidrug-resistance determinants (A) and genetic transfusions, and 5 days of peritoneal dialysis. comparison between plasmid pS88 and mosaic plasmid pR444_A (B). A) Four circularized contigs (chromosome of 5,256,050 bp We performed sequencing of RDEx444 by using the and the 3 plasmids pR444_A [176,500 bp], pR444_B [117,090 PacBio single-molecule real-time method with RS II chem- bp], and pR444_C [95,050 bp]) obtained by using PacBio (Pacific istry 2.4.0 (Pacific Biosciences, Menlo Park, CA, USA); Biosciences, Meno Park, CA, USA) sequencing of RDEx444 we used 1 single-molecule real-time cell. De novo as- are schematically represented. The main virulence factors are sembly was performed twice by using the HGAP pipeline presented as colored rectangles. Intestinal virulence factors are indicated in red for the locus of enterocyte effacement genes, blue ( for prophage-encoded Shiga toxin genes, and green for VFs carried Training/wiki/HGAP) and annotation by using the MAGE by pR444_C, a pO157-like plasmid. Bacteriocin genes (mch, cia, platform ( and cma) are indicated in yellow. B) Comparison of the sequences home/index.php) (Figure 1). For this strain, we performed of pR444_A and pS88, the plasmid of strain S88 involved in plasmid typing by using databases (PlasmidFinder 1.3 and neonatal meningitis. LEE, locus of enterocyte effacement. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2263 RESEARCH pMLST 1.4 [12]) available on the Center for Genomic Epi- strains from France and strains from Spain and Slovakia demiology (CGE) website (http://www.genomicepidemiol - have been partially characterized previously (10). The Plasmids were also characterized by S1 nuclease strains from Spain and Slovakia were isolated during pulsed-field gel electrophoresis (PFGE) with Southern hy- 1998–2012 and the strains from France during 2010–2016. bridization and conjugability of the largest confirmed by The dates of isolation of the strains from Switzerland and experiments using rifampin-resistant E. coli J53, as previ- Germany were not available. We performed assembly by ously described (9). using CLC Genomics Workbench software and SPAdes, We also sequenced a representative set of 35 O80 also available on CGE website. The RAST server (http:// strains from various sources and countries using the Nex- was used for genome annotation and the tera kit (Illumina, San Diego, California, USA) to prepare PHASTER web server ( to identify and the library. Sequencing was performed by using a MiSeq annotate prophage sequences within bacterial genomes and reagent kit V3 600 cycles (Illumina). This panel included plasmids. We established phylogeny by single-nucleotide human isolates from France (n = 21), Spain (n = 3), and polymorphism alignments between the contigs generated Switzerland (n = 2), as well as animal and environmental by CLC Genomics of O80 strains and 9 reference EHEC strains isolated in France (n = 3), Spain (n = 1), Slovakia (n strain sequences of major serotypes available in GenBank = 1), and Germany (n = 4) (Figure 2). Some of the human (O157:H7 EDL933, O26:H11 11368, O111:H- 11128, Figure 2. Phylogeny of 36 Escherichia coli serogroup O80 strains isolated from various sources and countries in Europe during 1998– 2016 and their relationship to major enterohemorrhagic E. coli lineages. General phylogenic tree rooted on E. fergusonii (GenBank accession no. NC_011740), showing (A) the position of O80 strains among major enterohemorrhagic E. coli lineages (O157:H7 EDL933 NC_002655.2, O26:H11 11368 NC_013361.1, O111:H- 11128 NC_013364.1, O103:H2 12009 NC_013353.1, O55:H7 2013C-4465 CP015241, O91:NM 2009C-3745 JHGW00000000, O104:H4 LB226692 EO104H4LB.1, O145:H28 2009C-3292 JHHD00000000, and O121:H19 2009C-4750 JHGL00000000) and (B) a focused view of clonal complex 165, including O80 strains. The highlighted strains carry the Shiga toxin genes; the subtype of Stx is indicated by a color code as follows: purple, stx1a; yellow, stx2a; pink, stx2d ; activable brown, stx1a2a; dark green, stx1a2b. The presence of the pS88 like–plasmid is represented by using a plasmid scheme next to the strain number. R to the right of the plasmid indicates that the strain possesses >2 resistance genes that confer resistance to β-lactams, kanamycin, or cotrimoxazole (R* indicates presence of additional extended-spectrum β-lactamase gene; R** indicates presence of mcr-1 gene conferring additional resistance to colistin). The strain origin (country and source of isolation) is represented by flags and human, animal, or water symbols. CC, clonal complex; HPI, high-pathogenicity island (presence of chromosomal locus encoding the siderophore yersiniabactin); SNP, single-nucleotide polymorphism; ST, sequence type. Scale bar indicates nucleotide substitutions per site. 2264 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Hybrid Pathotype Shiga Toxin–Producing E. coli O80 O103:H2 12009, O55:H7 2013C-4465, O91:NM 2009C- producing Stx2e, encoded by the phage φP27 ( 20). The ≈15- 3745, O104:H4 LB226692, O145:H28 2009C-3292, kb region between Stx subunit B and the phage integrase, and O121:H19 2009C-4750 [online Technical Appen- comprising proteins for DNA replication and repression, dix, shares strong homology with the phage φP27. RDEx444 is Techapp1.pdf]) using CSIPhylogeny 1.4 (13). The neigh- otherwise a typical EHEC, given that it harbors a complete bor-joining tree was built by using MEGA 3.1 (https:// LEE and does not possess atypical adhesins, such as autoag- using bootstraps calculated from glutinating adhesion (saa) or aggregative factors (aggA/R). 100 replicates. The maximum-likelihood method yielded RDEx444 carries the rare variant of the intimin gene eae-ξ the same results (online Technical Appendix Figure 1). (5), which has only very rarely been observed among oth- We performed multilocus sequence typing and iden- er EHEC serotypes, but which is shared by all previously tification of acquired antimicrobial resistance genes for all described O80 EHEC and EPEC isolates (5–7,9,10). Two strains by using suitable databases available on the CGE other chromosomal traits of RDEx444 might also confer website (SerotypeFinder 1.1 [14], MLST 1.8 [15], and a selective advantage: 1) a region encoding microcin H47 ResFinder 3.0 [16]). Investigation of the resistome was (mch A, S, X, B, C, D, E, and F) ≈35 kb downstream from completed by using the Resistance Gene Identifier at the the Stx prophage, and 2) a chromosomal mutation in DNA Comprehensive Antibiotic Resistance Database website gyrase (gyrA S83L), which confers quinolone resistance. ( Moreover, 166 Plasmid pR444_C is 95,050 bp long with a G+C con- genes of intestinal and extraintestinal VFs (list available tent of 49.8% and was predicted to contain 114 CDSs. Plas- from authors) were assessed by local BLAST+ 2.2.31 anal- mid pR444_C is comparable to pO157 because it contains ysis ( Complete results are enterohemolysin (ehxA) and serine protease (espP) but described in online Technical Appendix Figure 2. Nucleo- does not possess catalase peroxidase (katP). Moreover, 2 tide sequences of the 36 sequenced O80 strains (complete other colicines, Ia (cia) and M (cma), are also present on nucleotide sequence of RDEx444 and 35 draft nucleotide this plasmid. sequences) have been deposited in Genbank (project no. Plasmid pR444_B is 117,090 bp long with a G+C con- PRJNA449634); accession numbers are available in the on- tent of 46.5% and contains 135 CDSs and, unexpectedly, line Technical Appendix. those for 3 tRNAs (threonine, asparagine, and tyrosine) re- dundant with those still present on the chromosome. This Results plasmid can be considered to be cryptic because it carries RDEx444 can be considered to be representative of hybrid no drug resistance or virulence-associated genes, but sev- pathotype strains because it was responsible for an extrain- eral phage proteins are present. pR444_B shares high ho- testinal infection and showed similar virulence and resis- mology (99% nucleotide identity and 89% coverage) with tance profiles as other strains described elsewhere (9,10). plasmid pECOH89 (21), encoding an extended-spectrum Complete sequencing yielded 2.3 × 10 bp with 165,041 β-lactamase bla identified in an E. coli strain and be- CTX-M-15 reads and an N50 (the length of the smallest contig among longing to the family of phage-like plasmids. Members of the set of the largest contigs that together cover >50% of this family are generally untypeable, nonconjugative, and the assembly) read length of 19,839 bp. cryptic plasmids, because no known virulence or resistance Four circularized contigs were obtained, including genes have been identified. Their function is unknown, but the chromosome of 5,256,050 bp, containing 5,146 cod- they all have strong homology to the Salmonella-specific ing sequences (CDSs), with an overall G+C content of bacteriophage SSU5 (21). 50.7% and 3 plasmids of 176,500 (pR444_A), 117,090 Plasmid pR444_A is 176,500 bp long, with a G+C (pR444_B), and 95,050 bp (pR444_C). The number and content of 51.8%, and carries 2 replicons: FII_A and FIB_1 size of the RDEx444 plasmids were corroborated by S1 (ST [F2:A-:B1]). We identified 202 CDSs. This plasmid nuclease PFGE (data not shown). We compiled a schematic carries virulence-associated genes characteristic of pS88 representation of the genetic content, including major VFs, (described previously) and MDR genes, and thus can be of this strain (Figure 1, panel A). considered to be a mosaic plasmid. We identified an MDR- As expected, RDEx444 belongs to serotype O80:H2 encoding region (48,237 bp) in addition to the plasmid-re- and ST301. Among the 12 prophage regions identified on lated function (56,106 bp) and virulence-associated domain its chromosome, this strain carries a complete Stx-convert- (72,157 bp), closely related to pS88 (99% nucleotide iden- ing bacteriophage of 43.9 kb. RDEx444 carries the stx2d tity and 96% query coverage) (11). This MDR-encoding re- ac- variant, which has been shown to be predictive of se- gion contains genes encoding resistance tetracycline (tetA), tivable vere clinical outcomes and progression to HUS (17). The trimethoprim (dfrA5), sulfonamide (sul2), β-lactam (bla- Stx prophage is integrated into the chromosomal yecE gene, ), kanamycin (aph[3′]-Ia), and streptomycin (strA and TEM-IB a known integration site (18,19), initially reported in a strain strB) and resistance against heavy metals such as mercury Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2265 RESEARCH (mer gene) (Figure 1, panel B). This resistance cassette has elimination of the stx gene. This finding led us to search only been described for pO26-CRL (100% nucleotide for a scar of the Stx-converting bacteriophage at the inser- identity, 98% coverage) from an O26 EHEC but without tion site described in the RDEx444 strain (yecE). We first VFs (22). Co-localization of extraintestinal VFs and MDR searched for the contig carrying the yecE gene. Then, we genes on the same plasmid was confirmed by Southern hy- used the Phaster webserver system to detect phage regions, bridization (data not shown). Moreover, we confirmed that which were finally blasted against RDEx444. We found this plasmid is conjugative, suggested by the presence on scars of ≈31 kb in the human strain from Spain (IH43632- the annotation of an almost complete F-like transfer region, 03a) and scars of ≈17 kb in the porcine strain from Slovakia by successfully conjugating it with rifampin-resistant E. (FV-4476), similar to the RDEx444 Stx-converting bacte- coli J53 (data not shown). Thus, this plasmid can diffuse by riophage (92% and 94% nucleotide identity and 40% and horizontal transfer. 56% coverage, respectively) (data not shown). We found Accordingly, this mosaic plasmid shows high homol- no scars for the other 2 strains, suggesting either complete ogy with plasmid pS88, responsible for extraintestinal prophage excision or insertion at another site. virulence in neonatal meningitis strains within which a re- All strains carrying the stx gene (28/36) also possess sistance cassette has been integrated (Figure 1, panel B). VFs typical of pS88, and all but 1 (CB12623) also carry Furthermore, pR444_A also carries 2 bacteriocin genes genes conferring resistance to >1 of penicillin, tetracycline, (cia and cva), such as pS88, which might confer a selec- kanamycin, or cotrimoxazole. The consistent association tive advantage by promoting intestinal establishment and between extraintestinal VFs typical of pS88 and MDR colonization by killing other E. coli and freeing up their genes, irrespective of the source or country of isolation, ecologic niches (Figure 1, panel B). might suggest the presence of a mosaic plasmid, such as We sequenced a panel of strains consisting of repre- in the RDEx444 strain. Although it lacks pS88 markers, sentative O80 strains from France and all O80 strains from the water strain from Germany (CB13483) is neverthe- outside of France available at the beginning of the study to less multidrug resistant and carries an extended spectrum establish the genetic relationship between O80 strains iso- β-lactamase gene (bla ), as well as the mphA gene, CTX-M-1 lated in several countries in Europe and analyze the diver- which confers resistance to azithromycin, the only antimi- sity of their genetic content. We obtained an average of 233 crobial drug that can be used for intestinal decontamination contigs, with a mean depth of coverage of 58X and a mean of EHEC (24; online Technical Appendix Figure 2). N50 of 72,312. Statistics of each sequenced genome are Among our panel, 3 STEC strains from cattle in France summarized in online Technical Appendix Table 1. Single- clearly belong to this ST301 clonal group. Cattle might nucleotide polymorphism analysis (41,618 sites total) be- thus represent an animal reservoir for these hybrid patho- tween the 36 sequenced O80 strains and 9 EHEC strains type strains, given that the isolates possess exactly the same of other major serotypes (O157:H7, O26:H11, O111:H-, VFs (eae, genes of T3SS, ehxA, stx, and determinants of the O103:H2, O55:H7, O91:NM, O104:H4, O145:H28, pS88-like plasmid) and resistance genes as human EHECs. O121:H19) enabled us to establish a phylogenic tree (Fig- Two clusters (I and II) can be distinguished within the ure 2), which shows 2 main clusters. EHEC O157:H7 ST301 group we describe. The main difference between (EDL933) and O55:H7 (2013C-4465) group together, as these 2 groups concerns the presence of the unknown cryp- expected by their common origin, demonstrated by Feng et tic plasmid (pR444_B) in RDEx444. All strains of cluster al. (23), and are distantly related to the other major EHEC I (n = 25) possess >85% of the genetic determinants of this serotypes, including the O80 strains. However, O80 iso- plasmid of unknown function. BLAST results are depicted lates clearly form a separate group, suggesting that O80 for each strain in online Technical Appendix Table 2. Con- emerged independently from the other serogroups. All of versely, no strain of group II (n = 7), except 1, has this the O80 strains belong to clonal complex 165 (CC165), plasmid. No strains of ST165 or ST189 carry it either. containing ST301, ST165, and ST189 (Figure 2, panel B). Within cluster I, the pS88-like plasmid carries 2 dis- Almost all of the O80 strains (32/36), including tinct gene profiles, showing its plasticity. The 12 Stx2d acti- RDEx444, belong to the ST301 clonal group of serotype EHECs isolated from humans in France (designated as vable O80:H2. All of these isolates have EHEC markers, includ - subcluster Ia in Figure 2) carry the most complete form of ing genes encoded by the LEE, containing the rare variant the plasmid, identical to pS88. The aerobactin iron-uptake of intimin eae-ξ, ehxA, and stx2a or stx2d genes, except system, encoded by iuc genes, and the type I secretion sys- for 4 strains missing the stx gene (CB13483, IH43632–03a, tem, encoded by etsC, are absent from the pS88-like plas- FV-4476, and VTB262) (online Technical Appendix Fig- mid of the subcluster Ib strains. ure 2). However, the presence of the other EHEC mark- The second clonal group, ST165, is formed by 1 strain ers in these 4 strains (eae, ehxA/espP, or katP) suggests (EC_POI) of serotype O80:H19, which is devoid of all VF that they were initially STEC and underwent subsequent and resistance genes. This isolate might reflect the ancestral 2266 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Hybrid Pathotype Shiga Toxin–Producing E. coli O80 origin from which serogroup O80 EHEC strains were de- lower with a combination of azithromycin and ceftriax- rived after the acquisition of diverse VFs. one assays relative to basal secretion, and we proposed Finally, clonal group ST189 consists of the 3 avian this association for the treatment of such infections (10). strains of serotype O80:H26 from Germany. None has Such a troubling plasmid has never been identified in stx or ehxA genes, but all have a complete LEE with human EHEC isolates. The only example of a similar mo- the variant β1 of the intimin gene ( eae). However, none saic plasmid was reported for S. enterica serovar Kentucky, of these strains carries the plasmid bfp gene of typical in which an AmpC β-lactamase gene (bla ) was inte- CMY-2 EPEC (online Technical Appendix Figure 2). One strain grated into a pS88-like plasmid (29). The insertion of an (CB13938) has a region of ≈27 kb at the same insertion MDR-encoding island in a pS88-like plasmid containing site (yecE gene) that shares homology with RDEx444 extraintestinal virulence genes is particularly worrisome. Stx-converting bacteriophage (88% nucleotide identity, Massive and inappropriate use of veterinary antibiotics, 16% coverage), suggesting a potential scar of Stx-con - such as tetracycline, in food-animal production promotes verting phage. Thus, this strain could have been origi- antimicrobial drug resistance among animals, known to be nally an EHEC which secondarily lost its stx gene. No reservoirs for STEC. This practice can select and favor the similar regions were found in the other 2 strains of this spread of such MDR plasmids in human EHECs. Tetracy- group. Otherwise, 1 strain (CB15046) has several VFs, cline still represented 36.5% of the tonnage of veterinary confirming the presence of the pS88-like plasmid with antibiotic use in 2015 in France (30). In our panel, all the MDR genes. The other 2 strains (CB15387 and CB13938) sequenced O80 STEC strains carry the tetA gene, confer- carry the locus encoding the siderophore yersiniabactin ring resistance to tetracycline. Thus, large veterinary use of (fyuA), also called the high-pathogenicity island; these 2 this drug might favor the selection of these hybrid strains strains also have pyelonephritis-associated pili with Pap- and increase their diffusion. GII adhesin (online Technical Appendix Figure 2). These We indicated a potential reservoir of these hybrid pa- strains constitute another type of hybrid pathotype with thotype strains when we identified 3 O80:H2 strains isolated intestinal and extraintestinal VFs. Although none of them from cattle that carry the same VFs and resistance genes as has VFs typical of pS88 plasmids, CB13938 is multidrug- human strains. However, the presence of the CC165 strains resistant and carries the recently described mcr-1 plasmid in chickens suggest that this clonal complex is also adapted gene (25), conferring resistance to colistin. Blast analysis to poultry. An initial description of pS88-like plasmids in performed with the contig containing mcr-1 (10.119 bp) avian pathogenic E. coli strains reinforces this hypothesis shows strong homology (100% coverage, 99% identity) (31). Moreover, the environmental survival of this clone with 2 chromosomal insertion sites previously described in these potential reservoirs might be enhanced because of in strains Mbl323 and Mbl506 (26). the resistance to mercury shared by all but 1 strain (36047), all carriers of the pS88-like plasmid, irrespective of their Discussion origin. Such resistance to heavy metals has been rarely de- We deciphered the molecular characteristics of O80:H2– scribed in EHEC strains (22). CC165 EHEC, an emerging hybrid pathotype diffusing We also detected an O80:H19–CC165 strain devoid throughout Europe. This pathotype is armed to spread by of virulence genes, which might represent the ances- means of a conjugative plasmid combining extraintesti- tral precursor of CC165, and from which these hybrid nal virulence with resistance to nearly all major classes pathotype strains might have been derived. This strain of antibiotics, improved by the presence of several plas- could be used for tracing the genetic history of this clone mid and chromosomally encoded bacteriocins, such as in future studies. colicines I, V, M, and H47. We used the same criteria of Our genetic description of the emerging hybrid patho- MDR as a recent study in England (27) (bla , strA- type E. coli O80:H2, associated with O80-related strains, TEM-1 strB, sul1/sul2/dfrA, and tetA) and showed that 93% reveals the outstanding capacity of O80–CC165 to acquire (26/28) of O80 STEC have this genotypic resistance pro- the combination of virulence genes involved in intestinal file, whereas only 5% of the strains identified within the and extraintestinal pathogenicity and genes conferring O157 and O26 serogroups in the study in England had broad antibiotic resistance, including extended-spectrum such a profile. MDR observed with this hybrid pathotype β-lactamase–encoding genes and those most recently iden- might complicate patient care, and the use of antimicro- tified, such as mcr-1. O80–CC165 strains, which are able bial drugs during EHEC infections is still a subject of to integrate multiple VFs with various consequences, MDR debate (28). However, the occurrence of invasive infec- genes that encompass nearly all classes, and bacteriocins, tions, such as bacteremia during EHEC infections, with represent a serious threat because of their exceptional ver- this clone warrants antimicrobial treatment for such in- satility and should therefore be closely monitored in all fections. In a previous study, the observed Stx rate was countries in Europe. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2267 RESEARCH 9. Mariani-Kurkdjian P, Lemaître C, Bidet P, Perez D, Boggini L, Acknowledgments Kwon T, et al. Haemolytic-uraemic syndrome with bacteraemia We thank Roger Stephan and Lothar Beutin for providing the caused by a new hybrid Escherichia coli pathotype. New Microbes strains from Switzerland and Germany, respectively. New Infect. 2014;2:127–31. 10. Soysal N, Mariani-Kurkdjian P, Smail Y, Liguori S, Gouali M, This work was financed by Fonds d’Etudes et de Recherche du Loukiadis E, et al. Enterohemorrhagic Escherichia coli hybrid Corps Médical, AP-HP. Work in the Laboratorio de Referencia pathotype O80:H2 as a new therapeutic challenge. Emerg Infect de Escherichia coli was financed by grant no. ED431C-2017-57 Dis. 2016;22:1604–12. 11. Peigne C, Bidet P, Mahjoub-Messai F, Plainvert C, Barbe V, from Consellería de Cultura, Educación e Ordenación Médigue C, et al. The plasmid of Escherichia coli strain S88 Universitaria (Xunta de Galicia) and the European Regional (O45:K1:H7) that causes neonatal meningitis is closely related to Development Fund. The funders had no role in study design, avian pathogenic E. coli plasmids and is associated with high-level data collection and interpretation, or the decision to submit the bacteremia in a neonatal rat meningitis model. Infect Immun. 2009;77:2272–84. work for publication. 12. Carattoli A, Zankari E, García-Fernández A, Voldby Larsen M, Lund O, Villa L, et al. In silico detection and typing of plasmids using PlasmidFinder and plasmid multilocus sequence typing. About the Author Antimicrob Agents Chemother. 2014;58:3895–903. Dr. Cointe is a microbiologist working in the Parisian Public Hospital in France. Her research domain is pediatric infections, 13. Kaas RS, Leekitcharoenphon P, Aarestrup FM, Lund O. Solving the problem of comparing whole bacterial genomes particularly the pathogenicity of intestinal Escherichia coli across different sequencing platforms. PLoS One. 2014;9:e104984. infections and the potential interplay between intraintestinal and extraintestinal E. coli. 14. Joensen KG, Tetzschner AMM, Iguchi A, Aarestrup FM, Scheutz F. Rapid and easy in silico serotyping of Escherichia coli isolates by use of whole-genome sequencing data. J Clin Microbiol. References 2015;53:2410–26. 1. Garg AX, Suri RS, Barrowman N, Rehman F, Matsell D, 15. Wirth T, Falush D, Lan R, Colles F, Mensa P, Wieler LH, et al. Rosas-Arellano MP, et al. Long-term renal prognosis of diarrhea- Sex and virulence in Escherichia coli: an evolutionary perspective. associated hemolytic uremic syndrome: a systematic review, Mol Microbiol. 2006;60:1136–51. meta-analysis, and meta-regression. JAMA. 2003;290:1360–70. j.1365-2958.2006.05172.x 16. Zankari E, Hasman H, Cosentino S, Vestergaard M, Rasmussen S, 2. Schmidt H, Karch H, Beutin L. The large-sized plasmids of Lund O, et al. Identification of acquired antimicrobial resistance enterohemorrhagic Escherichia coli O157 strains encode genes. J Antimicrob Chemother. 2012;67:2640–4. hemolysins which are presumably members of the E. coli 10.1093/jac/dks261 alpha-hemolysin family. FEMS Microbiol Lett. 1994;117:189–96. 17. Bielaszewska M, Friedrich AW, Aldick T, Schürk-Bulgrin R, 3. Brunder W, Schmidt H, Karch H. EspP, a novel extracellular serine Karch H. Shiga toxin activatable by intestinal mucus in Escherichia protease of enterohaemorrhagic Escherichia coli O157:H7 cleaves coli isolated from humans: predictor for a severe clinical outcome. human coagulation factor V. Mol Microbiol. 1997;24:767–78. Clin Infect Dis. 2006;43:1160–7. 18. Serra-Moreno R, Jofre J, Muniesa M. Insertion site occupancy 4. Bruyand M. Surveillance du syndrome hémolytique et urémique by stx bacteriophages depends on the locus availability of post-diarrhéique chez les enfants de moins de 15 ans en France the host strain chromosome. J Bacteriol. 2007;189:6645–54. en 2016 [cited 2016 Feb 16]. content/download/138859/500759/version/1/file/Bilan_ 19. Bonanno L, Loukiadis E, Mariani-Kurkdjian P, Oswald E, Garnier L, SHU_2016.pdf Michel V, et al. Diversity of Shiga toxin-producing Escherichia coli 5. Blanco M, Blanco JE, Mora A, Dahbi G, Alonso MP, González EA, (STEC) O26:H11 strains examined via stx subtypes and insertion et al. Serotypes, virulence genes, and intimin types of Shiga toxin sites of Stx and EspK bacteriophages. Appl Environ Microbiol. (verotoxin)-producing Escherichia coli isolates from cattle in Spain 2015;81:3712–21. and identification of a new intimin variant gene ( eae-ξ). J Clin 20. Recktenwald J, Schmidt H. The nucleotide sequence of Shiga toxin Microbiol. 2004;42:645–51. (Stx) 2e-encoding phage phiP27 is not related to other Stx phage JCM.42.2.645-651.2004 genomes, but the modular genetic structure is conserved. Infect 6. Wijnsma KL, Schijvens AM, Rossen JWA, Kooistra-Smid AMDM, Immun. 2002;70:1896–908. Schreuder MF, van de Kar NCAJ. Unusual severe case of IAI.70.4.1896-1908.2002 hemolytic uremic syndrome due to Shiga toxin 2d-producing 21. Falgenhauer L, Yao Y, Fritzenwanker M, Schmiedel J, Imirzalioglu C, E. coli O80:H2. Pediatr Nephrol. 2017;32:1263–8. Chakraborty T. Complete genome sequence of phage-like plasmid 10.1007/s00467-017-3642-3 pECOH89, encoding CTX-M-15. Genome Announc. 2014; 7. Fierz L, Cernela N, Hauser E, Nüesch-Inderbinen M, Stephan R. 2:e00356–14. Characteristics of Shigatoxin-producing Escherichia coli strains 22. Venturini C, Hassan KA, Roy Chowdhury P, Paulsen IT, Walker MJ, isolated during 2010–2014 from human infections in Switzerland. Djordjevic SP. Sequences of two related multiple antibiotic Front Microbiol. 2017;8:1471. resistance virulence plasmids sharing a unique IS26-related fmicb.2017.01471 molecular signature isolated from different Escherichia coli 8. Bielaszewska M, Mellmann A, Zhang W, Köck R, Fruth A, pathotypes from different hosts. PLoS One. 2013;8:e78862. Bauwens A, et al. Characterisation of the Escherichia coli strain associated with an outbreak of haemolytic uraemic syndrome 23. Feng P, Lampel KA, Karch H, Whittam TS. Genotypic and in Germany, 2011: a microbiological study. Lancet Infect Dis. phenotypic changes in the emergence of Escherichia coli O157:H7. 2011;11:671–6. J Infect Dis. 1998;177:1750–3. 2268 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Hybrid Pathotype Shiga Toxin–Producing E. coli O80 24. Jost C, Bidet P, Carrère T, Mariani-Kurkdjian P, Bonacorsi S. infection, antibiotics, and risk of developing hemolytic uremic Susceptibility of enterohaemorrhagic Escherichia coli to syndrome: a meta-analysis. Clin Infect Dis. 2016;62:1251–8. azithromycin in France and analysis of resistance mechanisms. J Antimicrob Chemother. 2016;71:1183–7. 29. Fricke WF, McDermott PF, Mammel MK, Zhao S, Johnson TJ, 10.1093/jac/dkv477 Rasko DA, et al. Antimicrobial resistance-conferring plasmids with 25. Liu Y-Y, Wang Y, Walsh TR, Yi L-X, Zhang R, Spencer J, et al. similarity to virulence plasmids from avian pathogenic Escherichia Emergence of plasmid-mediated colistin resistance mechanism coli strains in Salmonella enterica serovar Kentucky isolates MCR-1 in animals and human beings in China: a microbiological from poultry. Appl Environ Microbiol. 2009;75:5963–71. and molecular biological study. Lancet Infect Dis. 2016;16:161–8. 30. Agence Nationale de Sécurité Sanitaire de l’Alimentation, de 26. Donà V, Bernasconi OJ, Pires J, Collaud A, Overesch G, l’Environnement et du Travail (ANSES). Suivi des ventes de Ramette A, et al. Heterogeneous genetic location of mcr-1 in médicaments vétérinaires contenant des antibiotiques en France colistin-resistant Escherichia coli isolates from humans and retail en 2015 [cited 2016 Feb 16]. chicken meat in Switzerland: emergence of mcr-1-carrying IncK2 ANMV-Ra-Antibiotiques2015.pdf plasmids. Antimicrob Agents Chemother. 2017;61:e01245–17. 31. Johnson TJ, Siek KE, Johnson SJ, Nolan LK. DNA sequence of a ColV plasmid and prevalence of selected plasmid-encoded 27. Day M, Doumith M, Jenkins C, Dallman TJ, Hopkins KL, Elson R, virulence genes among avian Escherichia coli strains. et al. Antimicrobial resistance in Shiga toxin-producing J Bacteriol. 2006;188:745–58. Escherichia coli serogroups O157 and O26 isolated from human JB.188.2.745-758.2006 cases of diarrhoeal disease in England, 2015. J Antimicrob Chemother. 2017;72:145–52. Address for correspondence: Stéphane Bonacorsi, Service de 28. Freedman SB, Xie J, Neufeld MS, Hamilton WL, Hartling L, Microbiologie, CHU Robert Debré, 48 boulevard Sérurier, 75019 Paris, Tarr PI; Alberta Provincial Pediatric Enteric Infection Team France; email: (APPETITE). Shiga toxin-producing Escherichia coli April 2018 Antimicrobial Resistance • Seroprevalence of Chikungunya Virus in 2 Urban Areas of • Genomic Surveillance of 4CMenB Vaccine Antigenic Variants Brazil 1 Year aer Emer ft gence among Disease-Causing Neisseria meningitidis Isolates, United Kingdom, 2010–2016 • Two Infants with Presumed Congenital Zika Syndrome, Brownsville, Texas, USA, 2016–2017 • Evolution of Sequence Type 4821 Clonal Complex Meningococcal Strains in China from Prequinolone to • Reemergence of Intravenous Drug Use as Risk Factor for Quinolone Era, 1972–2013 Candidemia, Massachusetts, USA • Avirulent Bacillus anthracis Strain with Molecular Assay Targets • Rickettsial Illnesses as Important Causes of Febrile Illness in as Surrogate for Irradiation-Inactivated Virulent Spores Chittagong, Bangladesh • Phenotypic and Genotypic Characterization of • Influence of Population Immunosuppression and Past Enterobacteriaceae Producing Oxacillinase-48–Like Vaccination on Smallpox Reemergence Carbapenemases, United States • Emerging Coxsackievirus A6 Causing Hand, Foot, and Mouth • Bacterial Infections in Neonates, Madagascar, 2012–2014 Disease, Vietnam • Carbapenem-Nonsusceptible Acinetobacter baumannii, • Influenza A(H7N9) Virus Antibody Responses in Survivors 1 8 US Metropolitan Areas, 2012–2015 Year aer In ft fection, China, 2017 • Artemisinin-Resistant Plasmodium falciparum with High • Cooperative Recognition of Internationally Disseminated Survival Rates, Uganda, 2014–2016 Ceftriaxone-Resistant Neisseria gonorrhoeae Strain To revisit the April 2018 issue, go to: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2269 RESEARCH Comparison of 2016–17 and Previous Epizootics of Highly Pathogenic Avian Influenza H5 Guangdong Lineage in Europe 1 1 Pablo Alarcon, Adam Brouwer, Divya Venkatesh, Daisy Duncan, Chrysostomos I. Dovas, George Georgiades, Isabella Monne, Alice Fusaro, Adam Dan, Krzysztof Śmietanka, Vassilios Ragias, Andrew C. Breed, Taxiarchis Chassalevris, Gabriela Goujgoulova, Charlotte Kristiane Hjulsager, Eoin Ryan, Azucena Sánchez, Eric Niqueux, Niina Tammiranta, Siamak Zohari, David A. Stroud, Vladimir Savić, Nicola S. Lewis, Ian H. Brown We analyzed the highly pathogenic avian influenza (HPAI) seasonality, and outbreak duration, making it difficult to H5 epizootic of 2016–17 in Europe by epidemiologic and predict future HPAI epizootics. However, we know that in genetic characteristics and compared it with 2 previous epi- 2005–06 and 2016–17 the initial peak of wild bird detec- zootics caused by the same H5 Guangdong lineage. The tions preceded the peak of poultry outbreaks within Europe. 2016–17 epizootic was the largest in Europe by number of Phylogenetic analysis of 2016–17 viruses indicates 2 main countries and farms affected and greatest diversity of wild pathways into Europe. Our findings highlight the need for birds infected. We observed significant differences among global surveillance of viral changes to inform disease pre- the 3 epizootics regarding region affected, epidemic curve, paredness, detection, and control. Author affiliations: Royal Veterinary College, London, UK ighly pathogenic avian influenza (HPAI) is a zoonotic (P. Alarcon); Animal and Plant Health Agency, Addlestone, UK Hnotifiable disease that can cause high mortality rates (P. Alarcon, A. Brouwer, D. Duncan, A.C. Breed, N.S. Lewis, in most domestic poultry and in some wild bird species. I.H. Brown); University of Cambridge, Cambridge, U K Since 2003, HPAI H5 viruses have been circulating in poul- (D. Venkatesh); Aristotle University of Thessaloniki, Thessaloniki, try in many countries (1). Periodically these poultry HPAI Greece (C.I. Dovas, T. Chassalevris); Ministry of Rural viruses have been reintroduced into the wild migratory bird Development and Food, Thessaloniki (G. Georgiades, V. Ragias); population, representing a key risk pathway for its subse- Istituto Zooprofilattico Sperimentale delle Venezie, Padova, Italy quent global spread (1–3). However, the effect of HPAI (I. Monne, A. Fusaro); Veterinary Diagnostic Institute, Budapest, infection in both wild and domestic birds is variable and Hungary (A. Dan); National Veterinary Research Institute, Pulawy, often strain-specific. Wild birds, particularly of the orders Poland (K. Śmietanka); Department of Agriculture and Water Anseriformes and Charadriiformes, are natural hosts of low Resources, Canberra, Australian Capital Territory, Australia pathogenicity avian influenza ( 4). (A.C. Breed); University of Queensland, Brisbane, Queensland, A passive surveillance system of testing wild birds Australia (A.C. Breed); NDRVMI, Sofia, Bulgaria (G. Goujgoulova); found dead or sick for avian influenza has been in place in Technical University of Denmark, Lyngby, Denmark European Union (EU) member states since 2005 (Commis- (C.K. Hjulsager); Central Veterinary Research Laboratory, sion Decision 2005/94/EC, replaced with 2010/367/EU), Celbridge, Ireland (E. Ryan); Central Veterinary Laboratory, with the objective of timely detection of HPAI subtype Madrid, Spain (A. Sánchez); French Agency for Food, H5N1. Laboratory confirmation of HPAI infection follow- Environmental and Occupational Health & Safety, Ploufragan, ing the development of clinical signs (passive surveillance) France (E. Niqueux); Finnish Food Safety Authority Evira, is the primary method of poultry surveillance in the EU Helsinki, Finland (N. Tammiranta); National Veterinary Institute member states, complemented by a serologic active sur- and World Organisation for Animal Health Collaborating Center, veillance program (5). Uppsala, Sweden (S. Zohari); Joint Nature Conservation During epidemiologic year 2005–06 (epidemiologic Committee, Peterborough, UK (D.A. Stroud); Croatian Veterinary years run from October to September of the next year), Institute, Zagreb, Croatia (V. Savić) These authors are joint first authors. DOI: 2270 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Highly Pathogenic Avian Influenza H5 in Europe HPAI H5N1 clade 2.2 virus of the Guangdong H5 lineage from GISAID ( We performed spread to a number of countries in Europe, infecting poul- phylogenetic analyses on HA sequence data from each try and wild bird populations (3). In 2014–15, another vi- epizootic separately. We used IQ-TREE version 1.5.5 rus of the same lineage, HPAI H5N8 clade, was in- software (15) to infer maximum-likelihood trees with troduced into Europe and associated with variable disease approximate likelihood ratio test (1,000 replicates) and severity, including subclinical infection in wild birds and bootstrap (100 replicates) support values for branches. domestic waterfowl (6). This H5N8 virus showed unprec- We down-sampled each dataset using Cluster Database edented intercontinental spread to the United States and at High Identity with Tolerance to remove sequences with Canada and was associated with both wild bird infection >99.9% sequence identity ( 16). We performed root-to- and, subsequent to local genetic reassortment, large HPAI tip regression analyses using Tempest version 1.5 on the H5N2 outbreaks in poultry (7). downsampled datasets (17). Then, we inferred Bayesian In October 2016, a novel HPAI H5 clade virus phylogenetic trees from each downsampled dataset using of the Guangdong lineage was detected in Hungary and was BEAST version 1.8.4 to determine the mean substitution subsequently reported in other countries in Europe, infect- rate and TMRCA (time to most recent common ancestor) ing many poultry farms and causing both large-scale and (18). We annotated the final trees using FigTree version sporadic deaths in wild bird populations. The hemagglu- 1.4.3 ( Details tinin (HA) gene of this virus was considered phylogeneti- of criteria and priors used in the analyses are provided in cally distinct from the previous 2014 clade viruses the online Technical Appendix. and was nominally suffixed by A (the 2016 clade) or B (the 2014 clade (8) but this subclade definition requires verifi - Results cation by the World Health Organization H5 nomenclature group. We describe the epidemiology and genetic charac- Epizootic Size teristics of the 3 major wild-bird mediated epizootics in In 2016–17, a total of 1,108 poultry outbreaks were reported Europe associated with the Guangdong HPAI H5 lineage. in 21 countries in Europe. Extensive farm-to-farm spread, predominantly in ducks, seemed apparent in France, which Methods had >400 farms affected, and Hungary, with >200 farms in- fected (19). Conversely, in 2005–06, a total of 230 poultry Epidemiologic Data and Analyses outbreaks occurred in 6 countries, mostly located in Roma- We collected data from the 3 major HPAI H5 epizootics nia (86%) and Hungary (13%). In 2014–15, only 13 poultry in Europe: HPAI H5N1 in epidemiologic year 2005–06 outbreaks were reported in 5 countries. The estimated num- (2); HPAI H5N8 in 2014–15; and HPAI H5 in 2016–17. ber of poultry culled was 8 times higher in 2016–17 than in For 2016–17, we collected data through July 31, 2017. We 2005–06 (Table 1). obtained epidemiologic data from the Animal Disease No- The number of wild bird detections was substantially tification System and the Directorate-General for Health different between epizootics: 1,559 incidents in 27 coun- and Food Safety, managed by the European Commission, tries in 2016–17, 487 in 18 countries in 2005–06, and only and from country notifications sent to the EU Reference 5 in 3 countries in 2014–15. Almost half of the wild bird Laboratory for avian influenza (Animal and Plant Health incidents reported in all 3 epizootics were in Germany. Agency, Weybridge, UK). We conducted analyses to describe each epizootic, Wild Birds Species and Mass Mortality Events examined the geographic and temporal spread (epidemic A total of 49 different wild bird species were reported in - curves), and assessed differences in clinical illness and fected with HPAI H5 virus of the Guangdong lineage in death rates. For spatial analysis, we grouped countries into 2016–17, 28 in 2005–06, and 6 in 2014–15 (Table 2,3). 4 regions (North, South-West, South-East, and Central Eu- Swans (Cygnus spp.), particularly mute swans (Cygnus rope) on the basis of the broad migration patterns of the olor), were the most frequent species infected in 2005– major migratory water bird species affected by HPAI (on- 06 (41% of all wild birds) and 2016–17 (20% of all wild line Technical Appendix Figure 1, birds). Ducks were the second most common type of wild EID/article/24/12/17-1860-Techapp1.pdf) (9–14). A full birds infected. In 2005–06 and 2016–17, tufted duck (Ay- description of the methods used is presented in the online thya fuligula) was the most frequent duck species detected Technical Appendix. positive (5% of all wild birds). In 2005–06, a total of 28 (6%) mass mortality events (>5 birds dead in 1 location) Viruses’ Sequence Data and Phylogenetic Analyses were reported, whereas 112 (7%) mass mortality events We obtained virus HA gene sequence data from coun- were reported in 2016–17; none were reported in 2014–15 tries’ submissions to the EU Reference Laboratory and (online Technical Appendix Figure 2). The number of wild Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2271 RESEARCH birds found dead by incident was significantly different be- Ducks, geese, turkeys, and broiler chickens on average tween epizootics (p<0.001 by Mann-Whitney U test). had higher illness rates in 2005–06 than in the other epizoot- ics (Figure 1). In 2016–17, average mortality rate was lowest Type of Poultry Farm and Clinical Manifestations in ducks (7%) and turkeys (6%); few farms (<5%) reported The types of poultry infected in each epizootic are shown a >25% mortality rate. In contrast, 32% of affected broiler in Table 4. In 2016–17, a large proportion of infected farms and 27% of affected layer farms reported mortality farms (40%) kept ducks. In 2005–06, many affected back- rates >25%. In 2005–06, more than half of broiler farms re - yard flocks in Romania (176/230, 77%) had <100 birds, ported mortality rates >25%. When comparing overall esti - whereas 70% (9/13) of poultry farms infected in 2014–15 mates, we found the observed poultry illness and death rates had >10,000 birds and >60% in 2016–17 had >1,000 birds to be substantially higher in 2005–06 than in 2016–17. (difference in flock size distribution, p<0.001 by Kruskal- Wallis test). When we excluded Romania from the com- Temporal Spread parison of flock size, there was no statistical difference in We determined the epidemiologic curves of the 3 epizootics flock size between 2005–06 and 2016–17 (online Technical (Figure 2, panels A–C). In 2016–17, H5 was first detected Appendix Figure 3). in Europe in a mute swan in Hungary; the first outbreak in Table 1. Highly pathogenic avian influenza outbreaks by country in 3 epizootics in Europe* H5N1 2005–06 epizootic H5N8 2014–15 epizootic H5N8 2016–17 epizootic No. No. No. No. No. No. wild No. No. wild captive No. No. wild captive No. poultry birds poultry poultry birds birds poultry poultry birds birds poultry Country infected infected culled† infected infected infected culled† infected infected infected culled† France 1 21 11,700 – – – – 485 51 3 1,529,361 Hungary 29 12 251,948 1 – – 22,000 238 86 5 2,678,191 Germany 1 220 14,300 5 2 1 58,964 89 738 15 1,150,631 Bulgaria – 4 – – – – – 71 13 2 511,832 Poland – 29 – – – – – 65 66 – 1,167,282 Romania 197 17 755,372‡ – – – – 45 93 2 2,222 Czech – 14 – – – – – 38 39 – 79,308 Republic Italy – 19 – 1 – – 31,985 16 6 – 357,049 Spain – 1 – – – – –– 10 2 – 28,330 Croatia § § § – – – – 9 12 – 1,546 United – 1 – 1 – – 6,178 12 23 – 102,849 Kingdom Netherlands – – – 5 1 – 245,600 8 48 10 202,004 Slovakia – 2 – – – – – 8 58 3 351 Greece – 25 – – – – – 5 8 – 28,275 Serbia § § § – – – – 4 20 – 289 Sweden 1 13 692 – 2 – – 4 30 2 203,053 Austria – 46 – – – – 2 55 1 1,258 Ukraine § § § – – – – 2 3 1 10,288 Bosnia and § § § – – – – 1 1 1 148 Herzegovina Denmark 1 26 102 – – – – 1 49 1 69 FYROM § § § – – – – 1 1 – 438 Belgium – – – – – – – 2 3 13 4,047 Finland – – – – – – – – 15 2 – Ireland – – – – – – – – 10 – – Lithuania – – – – – – – – 5 – – Portugal – – – – – – – – 1 – – Slovenia – 28 – – – – – – 41 – – Switzerland – 9 – – – – – – 87 – – Luxembourg – – – – – – – – – 4 – Totals 230 487 1,034,114 13 5 1 364,727 1,116 1,565 64 8,058,831 Total infected 717 19 2,745 *Table includes all reported HPAI H5N8 outbreaks through July 31, 2017. It excludes the new wave of secondary H5N8 outbreaks observed in Italy from the beginning of July 2017 through September 2017, which has different drivers and kinetics with maintenance in the poultry (primarily turkey) population rather than through wild bird introduction. FYROM, the former Yugoslav Republic of Macedonia; HPAI, highly pathogenic avian influenza. †It is uncertain if for some outbreaks only the number of poultry in one farm building or if the poultry population in the area of the farm were reported. This estimate should be used as an approximation and indicator of impact. ‡One observation contained 600,000 birds, representing the overall population of backyard flocks affected in Romania. This number is an approximation. §These countries did not submit data to the Animal Disease Notification System in 2005–06; however, there is other evidence of H5N1 incursion in the period. 2272 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Highly Pathogenic Avian Influenza H5 in Europe poultry was detected 11 days later in a turkey farm, also in 2005–06 had significantly higher values (p<0.001 by Hungary. We observed 3 major epidemic peaks on the inci- 2-sample Kolmogorov-Smirnov test) than 2014–15. Tem- dence of poultry outbreaks (Figure 2, panel D): on day 54 poral median of the poultry epizootic was substantially (14.9 outbreaks/wk), following large farm-to-farm spread different between epizootics (mean/median distance for in Hungary; day 79 (12.1 outbreaks/wk) caused by farm- 2005–06, 189/223 days; for 2014–15, 33.5/26; for 2016– to-farm transmission in France and Bulgaria; and on day 17, 92/90 days). Seasonal analysis of poultry outbreaks 121 (16.9 outbreaks/wk), caused by the large farm-to-farm indicates significant differences (p<0.001 by Pearson χ spread in France and Poland. test) between epizootics; >50% of poultry outbreaks oc- In 2005–06 and 2016–17, a peak in wild bird inci- curred in May in 2005–06, in November in 2014–15, and dents preceded the peak in poultry outbreaks (Figure 2, in December–February in 2016–17 (Figure 2, panel E). panel A, C). Statistical analysis of the distribution of the epidemic curves indicates that the 2016–17 outbreak had Spatial Spread significantly higher incidence values (p<0.001 by 2-sample We mapped a temporal-spatial analysis of the 3 epizootics Kolmogorov-Smirnov test) than the other 2 epizootics; (Figures 3–5). The data shown in Figure 5, panel B, suggest Table 2. Wild bird species of the orders Podicipediformes, Anseriformes, and Charadriiformes, reported by event in 3 highly pathogenic avian influenza epizootics in Europe No. (%) events H5N1 2005–06 H5N8 2014–15 H5N8 2016–17 Species group Species epizootic epizootic epizootic Rails Eurasian coot (Fulica atra) 5 (1) 8 (0.5) Crested coot (Fulica cristata) 1 (0.1) Purple swamphen (Porphyrio porphyrio) 4 (1) Common moorhen (Gallinula chloropus) 1 (0.2) 2 (0.1) Total 10 (2) 11 (1) Swans Unspecified. 197 (38) 2 (22) 262 (16) Mute swan (Cygnus olor) 92 (18) 344 (20) Whooper swan (Cygnus cygnus) 2 (0.4) 80 (5) Total 291 (56) 2 (22) 683 (41) Ducks Unspecified 57 (11) 143 (9) Northern pintail (Anas acuta) 2 (0.4) Eurasian wigeon (Anas penelope) 1 (11) 21 (1) Mallard (Anas platyrhynchos) 4 (1) 1 (11) 43 (3) Common pochard (Aythya farina) 4 (1) 8 (0.5) Red-crested pochard (Netta rufina) 2 (0.1) Common goldeneye (Bucephala clangula) 1 (0.1) Greater scaup (Aythya marila) 2 (0.4) Common merganser (Mergus merganser) 5 (1) Tufted duck (Aythya fuligula) 18 (3) 82 (5) Eurasian teal (Anas crecca) 1 (11) 3 (0.2) Smew (Mergus albellus) 1 (0.2) Shelduck (Tadorna tadorna) 2 (0.1) Common eider (Somateria mollissima) 2 (0.1) Total 93 (18) 3 (33) 307 (18) Geese Unspecified 30 (6) 94 (6) Canada goose (Branta canadensis) 5 (0.3 Barnacle goose (Branta leucopsis) 1 (0.2) Greater white-fronted goose (Anser albifrons) 9 (1) Lesser white-fronted goose (Anser erythropus) 2 (0.4) 4 (0.2) Greylag goose (Anser anser) 1 (0.2) 21 (1) Red-breasted goose (Branta ruficollis) 1 (0.2) Bean goose (Anser fabalis) 1 (0.1) Pink-footed goose (Anser brachyrhynchus) 1 (0.1) Total 35 (7) 134 (8) Gulls Unspecified 9 (2) 89 (5) Great black-backed gull (Larus marinus) 11 (1) Herring gull (Larus argentatus) 1 (0.2) 28 (2) Black-headed gull (Larus ridibundus) 1 (0.2) 1 (11) 23 (1) Lesser black-backed gull (Larus fuscus) 1 (0.1) Common gull (Larus canus) 2 (0.1) Total 11 (2) 1 (11) 154 (9) Waders Green sandpiper (Tringa ochropus) 1 (0.1) Eurasian curlew (Numenius arquata) 1 (0.1) Total 2 (0.1) Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2273 RESEARCH that, in the first 2 months of the 2016–17 epizootic, 2 different Poultry detections by region were significantly different viral incursions may have occurred: one spreading through for the 3 epizootics (p<0.001 by Pearson χ test), whereas Hungary, Croatia, Switzerland, and southern Germany, and wild bird detections by region were only significantly dif- another spreading in northern Europe (Poland, Denmark, ferent (p<0.001 by Pearson χ test) between 2005–06 and northern Germany, Sweden, and the Netherlands). The 2016–17. 2005–06 epizootic indicated a similar progression pattern, initiating in Romania and spreading up to northern Europe Phylogenetic Analysis and down to southeastern Europe (Figure 3). Genetic analysis of the HA gene for the 2014–15 and 2016– Comparison by region of Europe according to wild 17 epizootics shows the involvement of H5 clade bird migratory patterns indicates poultry outbreaks were in all cases where data were available (Figure 6). Patterns mostly observed in the South-East and South-West re- found in maximum-likelihood trees are largely in agreement gions in 2005–06 and 2016–17 but in the North in 2014–15 with the Bayesian analysis; however, a greater proportion (online Technical Appendix Figure 4). Most wild bird de- of the clades remain unresolved in the maximum-likelihood tections were reported in the North and Central regions. trees (Figure 6; online Technical Appendix Figure 8). The Table 3. Wild bird species of orders other than Podicipediformes, Anseriformes, and Charadriiformes reported by event in 3 highly pathogenic avian influenza epizootics in Europe No. (%) events H5N1 2005–06 H5N8 2014–15 H5N8 2016–17 Species group Species epizootic epizootic epizootic Birds of prey Unspecified 30 (6) Buzzard 1 (0.2) 6 (0.4) Eagle 1 (0.1) Falcon 1 (0.2) 3 (0.2) Hawk 1 (0.2) 3 (0.2) Owl. 2 (0.4) 4 (0.2) Barn owl (Tyto alba) 1 (0.2) Peregrine falcon (Falco peregrinus) 1 (0.2) 8 (0.5) White-tailed eagle (Haliaeetus albicilla) 24 (1) Common buzzard (Buteo buteo) 7 (1) 70 (4) Rough-legged buzzard (Buteo lagopus) 1 (0.2) Eurasian eagle-owl (Bubo bubo) 2 (0.4) 1 (0.1) Eurasian sparrowhawk (Accipiter nisus) 1 (0.1) Common kestrel (Falco tinnunculus) 2 (0.1) Northern goshawk (Accipiter gentilis) 1 (0.1) Total 47 (9) 124 (7) Crows Unspecified 1 (0.2) Eurasian magpie (Pica pica) 1 (0.2) 4 (0.3) Hooded crow (Corvus cornix) 3 (0.2) Rook (Corvus frugilegus) 2 (0.1) Carrion crow (Corvus corone) 1 (0.1) Common raven (Corvus corax) 1 (0.1) Total 2 (0.4) 11 (1) Grebes Great crested grebe (Podiceps cristatus) 7 (1) 12 (1) Little grebe (Tachybaptus ruficollis) 1 (0.2) 4 (0.2) Total 8 (2) 16 (1) Thrushes Blackbird (Turdus merula) 1 (0.1 Song thrush (Turdus philomelos) 2 (0.1) Total 3 (0.2 Pigeons, doves Wood pigeon (Columba palumbus) 2 (0.1) Collared dove (Streptopelia decaocto) 1 (0.2) 1 (0.1) Rock dove (Coumbia livia) 1 (11) Total 1 (0.2) 1 (11) 3 (0.2) Herons Unspecified 2 (0.4) 16 (1) Grey heron (Ardea cinerea) 4 (1) 48 (3) Total 6 (1) 64 (4) Storks Unspecified 2 (0.4) White stork (Ciconia ciconia) 3 (0.2) Total 2 (0.4) 3 (0.2) Pelicans Unspecified. (Pelcanus spp.) 2 (0.1) Terns Common tern (Sterna hirundo) 2 (0.1) Cormorants Great cormorant (Phalacrocorax carbo) 6 (1) 17 (1) Other Unspecified 9 (2) 2 (22) 140 (8) 2274 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Highly Pathogenic Avian Influenza H5 in Europe Table 4. Types of poultry on infected farms in 3 highly pathogenic avian influenza epizootics in Europe* H5N1 2005–06 epizootic H5N8 2014–15 epizootic H5N8 2016–17 epizootic No. with only No. with only No. with only Type of poultry No. (%) farms 1 species No. (%) farms 1 species No. (%) farms 1 species Ducks 3 (23) 0 495 (44) 433 Geese 113 (10) 81 Ducks and geese 29 (13) 0 Turkey 5 (2) 1 3 (23) 0 91 (8) 82 Broilers 23 (10) 17 4 (31) 0 93 (8) 48 Laying hens 47 (4) 29 Pigeons 9 (1) 1 Guinea fowl 10 (1) 1 Peacocks 2 (0) 0 Pheasants 8 (1) 5 Quail 2 (0) 1 Ostrich 1 (0) 0 Backyard† 176 (77) NA Unknown 2 360(32) NA Total infected farms 230 13 1,116 *NA, not available. †Backyard represents those households that keep few birds, normally layer hens, for their own consumption. The category was used only in the 2005–06 epizootic. 2016–17 viruses form a distinct clade and can be clearly dif- March 2015–August 2016 [0.9 posterior probability] and ferentiated from the clade viruses present in Europe November 2014–October 2015 [0.82 posterior probability]). in 2014–15. In agreement with the geospatial results, analy- This finding potentially indicates 2 major incursion path - sis of the HA gene of the viruses from the 2016–17 epizootic ways via wild birds. shows that most originate from a common progenitor (time We also found smaller clusters and singleton se- to most recent common ancestor estimated May 2014–Au- quences including sequences from European viruses; vi- gust 2015) (online Technical Appendix Figure 8). However, ruses from 2014–15 form 1 subclade, estimated to have these viruses differ in their evolutionary pathway thereafter, emerged in January–February 2014 (Figure 6; online evolving in 2 co-circulating subclades without clear geo- Technical Appendix Figure 8). The 2005–06 data show graphic restriction (time to most recent common ancestor viruses in several subclades, but the branching pattern in Figure 1. Morbidity (A) and mortality (B) rates as percentages of populations reported in infected poultry farms during 3 highly pathogenic avian influenza epizootics in Europe, 2005–06, 2014–15, and 2016–17. Years given are epidemiologic years (October through September of the next year). Diamonds with error bars indicate means and 95% CIs. Asterisks indicate farms with unique poultry species used for analysis; dagger indicates large majority of data from backyard farms reported in Romania. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2275 RESEARCH Figure 2. Epidemic curve of 3 HPAI H5 virus epizootics in Europe: A) 2005–06 H5N1; B) 2014–15 H5N8; C) 2016–17 H5N8. Years given are epidemiologic years (October through September of the next year). Dashed lines indicate number of countries reporting an HPAI infection since the beginning of the epizootic; vertical line in panel C indicates data collected through July 31, 2017. D) Weekly average number of poultry outbreaks for each epizootic. Horizontal lines indicate the day at which half of the poultry outbreaks have occurred (diamonds); error bars indicate 1 SD. E) Number of poultry outbreaks for each month for the 3 epizootics. HPAI, highly pathogenic avian influenza. this dataset is generally less distinct and many sequences 2005–06 outbreak. The nucleotide diversity for each epizo- remain unresolved. otic (online Technical Appendix Figure 9) shows that per- BEAST analyses ( site diversity (average pairwise nucleotide differences in a beast/) also revealed that the 2014–15 epizootic viruses population) is lowest in the 2005–06 epizootic (0.0038), show the highest mean substitution rate (measured per site consistent with the lower substitution rate inferred from per year), followed by 2016–17 and then by the 2005–06 BEAST. The 2014–15 epizootic has the highest diversity epizootic, which is significantly lower (one-way analysis (0.0086); the rate for 2016–17, calculated from viruses col- of variance p<0.001) (online Technical Appendix Figure lected through June 2017, is 0.0063. 6). These data are in agreement with the results of the root- to-tip regression analysis (online Technical Appendix Fig- Discussion ure 7), which show a much steeper slope for the 2014–15 The 2016–17 epizootic of HPAI H5 clade vi- epizootic compared with the others. However, the spread ruses in Europe has 5 times more outbreaks in poultry of the data is high for the 2016–17 epizootic, where the than observed in the H5 clade 2.2 epizootic in 2005–06 SD of rates is an order of magnitude higher than that for and 80 times more than in the H5 clade epizo- the 2014–15 epizootic and 2 orders greater than for the otic in 2014–15. This study highlights the unprecedented 2276 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Highly Pathogenic Avian Influenza H5 in Europe Figure 3. Geographic and temporal spread of the 2005–06 HPAI H5N1 epizootic. A) Location of each incident reported. Blue shading indicates countries where cases were reported. B) Month of first report of an HPAI H5N1 incident. Years given are epidemiologic years (October through September of the next year). HPAI, highly pathogenic avian influenza. magnitude of the 2016–17 HPAI H5 epizootic in Europe, 2016 (20,21). Despite reduced passive surveillance efforts in in terms of size (both number of poultry outbreaks and recent years, more virus detections were made in wild birds wild bird incidents), geographic spread, speed of inci- in calendar year 2016 compared with 2006, indicating a like- dents/outbreaks, and diversity of wild bird species re- ly increase in viral burden within bird populations in Europe, ported infected. As a result, the economic impact is many leading to an increased risk for incursion into poultry. Al- times higher for 2016–17, which resulted in an >8-fold though we found a lower rate of substitution and diversity in increase in poultry that died or were culled. 2016–17 compared with 2014–15, the viruses in the 2016–17 A greater passive surveillance effort to detect influenza epizootic might be more efficient in capacity to adapt and virus in wild birds was reported in the EU in 2006 than in infect avian hosts. Different rates and diversity between Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2277 RESEARCH Figure 4. Geographic and temporal spread of the 2014–15 HPAI H5N8 epizootic. A) Location of each incident reported. Blue shading indicates countries where cases were reported. B) Month of first report of an HPAI H5N8 incident. Years given are epidemiologic years (October through September of the next year). HPAI, highly pathogenic avian influenza. 2005–06 and the 2 more recent epizootics may be caused farmed duck sector in 2016–17, possibly because of rapid by overall differences in the H5 lineages (clade 2.2 versus attenuation of viral symptoms. Hence, on several farms, which could influence viral spread. The greater ge - with clinically healthy birds, we detected HPAI infections netic distances we observed in viruses detected in the 2014– through active epidemiologic tracings and not on the ba- 15 epidemic could also be due to lower sensitivity of surveil- sis of clinical signs, as reported in data from some mem- lance for this virus compared with the other 2 epidemics due ber states. The results may also indicate that infection and to an apparently lower mortality rate in wild birds. transmission between domestic ducks is relatively easy for Extensive secondary spread is the most probable ex- these viruses. The type of husbandry practices and frequent planation for the large number of outbreaks reported in the movement of birds, coupled with poor biosecurity and lack 2278 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Highly Pathogenic Avian Influenza H5 in Europe Figure 5. Geographic and temporal spread of the 2016–17 HPAI H5N8 epizootic. A) Location of each incident reported. Blue shading indicates countries where cases were reported. B) Month of first report of an HPAI H5N8 incident. Years given are epidemiologic years (October through September of the next year). HPAI, highly pathogenic avian influenza. of robust hygiene practices, may also make the spread of In light of these results, we recommend a review of the the viruses between farms easier (22). target species for avian influenza surveillance ( 5) to im- Swans and ducks were the predominant hosts in- prove sensitivity of surveillance. Clarifying the precise fected in 2005–06 and 2016–17. Of interest, although origins of the current epizootic viruses from reported wild mallards (Anas platyrhynchos) are the most frequently bird mortality data is problematic, because these data do tested in EU passive surveillance (4), tufted ducks (Ay- not allow distinction between migratory carrier species thya fuligula) were the most commonly identified species and resident sentinel species. Many of the reported spe- of duck with HPAI in 2005–06 and 2016–17. In addition, cies are not migratory (e.g., mute swan or little grebes) the 2016–17 epizootic demonstrated a much expanded and so might play a role as regional amplifiers of viruses wild bird host range compared with previous outbreaks. but not in long-distance spread (23). Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2279 RESEARCH Figure 6. Maximum-likelihood tree from viral sequences of the 2016–17 highly pathogenic avian influenza H5 epizootic in Europe. Circles represent node support values, filled according to approximate likelihood ratio test values 0–100. Light gray boxes indicate distinct clades with support >50 with isolates from Europe; dark gray boxes indicate clades with <50 or unresolved. Scale bar indicates nucleotide substitutions per site. An expanded figure showing trees for all 3 epizootic years is available online ( 2280 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Highly Pathogenic Avian Influenza H5 in Europe Epidemic curves for the 3 epizootics were significantly demonstrate movements from breeding areas from Sibe- different. The incidence values in order of magnitude were ria both southwest toward the Black and Aegean Seas and 2016–17 > 2005–06 > 2014–15. In the period of the re- ultimately the coastal wetlands of the eastern Mediterra- view, the mean temporal distances to the midpoint in the nean, and further north and west through the Baltic Sea to poultry epizootic were different; 2014–15 was relatively coastal and other wetlands of the southern North Sea and short, consistent with the incursion into the poultry sector northwestern countries (11–14). These represent migratory and potentially lower virus infectivity present in the wild tendencies only; several studies have shown the high-level bird reservoir, whereas in 2005–06 and in 2016–17, epi- complexity of these movements and their variation due to demic curves show a clear peak of detection of wild bird both short-term weather patterns and longer-term climate incidence preceding the peak of poultry incidences, which change (27,28). The fact that these corridors were appar- demonstrates the importance of wild bird surveillance. ent in 2 temporally distant epizootics suggests the need for For the 2016–17 epizootic, the epidemic curve shows a further research to focus surveillance in these areas. long extended tail with small sporadic peaks relating to lo- This study presents many limitations (online Techni- calized but limited detection and spread in both poultry and cal Appendix). Differences in the implementation of pas- wild birds (Figure 2, panel C). These data might suggest sive wild bird surveillance between countries, which are greater infection pressure from migratory birds in 2016–17, implied in the EU avian influenza annual report for 2016 leading to higher risks for incursion, greater environmen- (20), suggest that sensitivity of wild bird surveillance var- tal contamination, and exposure of local indigenous wild ies across countries (29), which could affect the distribu - bird populations and poultry. The observed spatiotemporal tion of cases we observed. The true probability of detect- relationships between poultry incursions and wild bird de- ing HPAI is dependent on many factors that may influence tections represent a complex dynamic. Exploration of the both the frequency of wild bird deaths and the likelihood epidemic curves by country in 2016–17 shows important of identification and sampling of wild bird carcasses in dif- differences that relate to the type of poultry production in- ferent regions and countries. Public awareness, the current fected (online Technical Appendix Figure 7). For example, avian influenza status of the country area, media coverage, we detected infections in Hungary relatively early in the prevailing climatic conditions, available food sources, and epizootic; their rapid peak and decline may reflect exten- removal by predators may affect wild bird mortality, detec - sive infection within the major duck-producing regions and tion rates, or both (30). Furthermore, the efficacy of pas- less susceptible populations through infection and depopu- sive surveillance is difficult to measure because capturing lation. In contrast, infection in Germany and Poland was the expended effort depends on observation and testing of more consistent and may reflect a more continuous expo- deceased birds. On the other hand, surveillance has high sure and incursion risk into a variety of poultry sectors. sensitivity in farmed poultry, mainly because of higher vir- The viruses showed close genetic similarity to viruses ulence and much closer observation of these populations. contemporaneously circulating in Central and Southeast Despite apparent heavy infection pressure in wild birds Asia. The lower genetic diversity observed in 2016–17 in 2016–17, the virus was not detected early in the epizo- was accompanied by reassortment of all gene segments, as otic in areas in eastern Europe, such as the Danube Delta, shown in previous studies (8,24,25). The high reassortment with high density of early migratory waterfowl. There were observed in the 2016–17 epizootic also resulted in novel significant incursions in poultry in northern Europe, par- NA reassortants such as the H5N6 and H5N5 viruses. The ticularly Germany and Poland, and these areas also report- H5N6 viruses circulating in Europe were a reassortant of ed the greatest number of infected wild birds. This finding HPAI H5N8 and classical European LPAI present in wild may reflect the implementation of enhanced surveillance birds (data not shown). We can clearly differentiate the ge- in wild bird populations rather than true increased risk. netic characteristics of this strain from viruses known to be Southwestern Europe had relatively few wild bird detec- circulating in poultry and wild birds in the Far East with tions compared to the number of poultry outbreaks, per- occasional spillover to humans. haps because of the establishment of the virus in the duck Epidemiologic results suggest 2 broad corridors of vi- production sector in southwestern France, not as a result of rus incursion in 2005–06 and 2016–17, through northern increased introductions from wild birds (31). and central Europe with subsequent spread, later corrobo- The extent of the 2016–17 H5 epizootic indicates an rated through phylogenetic analyses of the HA gene of the urgent need to reappraise the effectiveness of surveillance viruses from the 2016–17 epizootic. This dual incursion strategies in both wild and domestic birds and to monitor probably relates broadly to known postbreeding move- key populations for emergence of viral variants. The dif- ments of northern duck species, which breed widely across ferences we observed in the 3 epizootics illustrate the dif- northern Eurasia (11,13,26). These movements occur on ficulty of predicting HPAI epizootics. However, the tem - a broad front, but ringing recoveries and other analyses poral peak of wild bird detections preceding the peak of Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2281 RESEARCH 2018 Sep 28]. poultry outbreaks at the EU level highlighted the utility of control-measures_ai_surv-rslt_pltry-wld-brds_2015.pdf surveillance in wild birds, as observed in other studies (29). 5. European Union Occupational Health and Safety Information The spatial corridors of HPAI we identified may provide Service. Commission decision 2010/367/EU of 25 June 2010 on the basis for an increase in targeted surveillance to improve the implementation by Member States of surveillance programmes for avian influenza in poultry and wild birds. Official Journal of the system sensitivity. Although the H5N8, H5N5, and H5N6 European Union. 2010;166:22–32. European-reassortant viruses have not been shown to infect 6. European Food Safety Authority. Highly pathogenic avian humans and remain avian influenza–like strains with no influenza A subtype H5N8. EFSA Journal. 2014;12:3941–32. evidence of key mammalian adaptation markers (27), their 7. Lee, DH, Torchetti MK, Winker K, Ip HS, Song CS, Swayne DE, Intercontinental spread of Asian-origin H5N8 to genetic volatility represents a potential threat that requires North America through Beringia by migratory birds. J Virol. continuous monitoring and surveillance of virus incidence 2015;89:6521–4. and genetics to continue to protect public safety. 8. Lee DH, Sharshov K, Swayne DE, Kurskaya O, Sobolev I, Kabilov M, et al. Novel reassortant clade avian influenza A(H5N8) virus in wild aquatic birds, Russia, 2016. Emerg Infect Acknowledgments Dis. 2017;23:359–60. The following laboratories supplied virus sequence data used 9. Boere G, Galbraith C, Stroud D; Scottish Natural Heritage. The in our analysis: National Food Chain Safety Office, Veterinary flyway concept: what it is and what it isn’t. In: Waterbirds around Diagnostic Directorate, Laboratory for Molecular Biology, the world: a global overview of the conservation, management, and research of the world’s waterbird flyways, Edinburgh: The Hungary; the Croatian Veterinary Institute, Croatia; the National Stationery Office; 2006. Veterinary Research Institute, Poland; Wageningen UR, the 10. Delany S, Scott D, Dodman T, Stroud DA. An atlas of wader Netherlands; the National Veterinary Institute, Denmark; the populations in Africa and western Eurasia. Wageningen (the Friedrich-Loeffler Institute, Germany; National Veterinary Netherlands): Wetlands International; 2009. 11. Scott DA, Rose PM. Atlas of Anatidae populations in Africa and Institute, Sweden; Wageningen University and Research Centre, western Eurasia. Publication No. 41. Wageningen (the Netherlands): the Netherlands. Wetlands International; 1996. 12. Wernham CV, Toms M, Marchant JH, Clark J, Siriwardena G, The work at Animal and Plant Health Agency was jointly Baillie S, editors. The migration atlas: movements of the birds of funded by the European Commission and the Department for Britain and Ireland. London: T. & A.D. Poyser Ltd; 2002. Environment, Food and Rural Affairs, London, through the EU 13. Viksne J, Švažas S, Czajkowski A, Janaus M, Mischenko A, reference laboratory. This work was funded in part by National Kozulin A, et al. Atlas of duck populations in eastern Europe. Vilnius (Lithuania): Oiseaux Migrateurs du Palearctique Institute of Allergy and Infectious Diseases (NIAID)–funded Occidental; 2010. Centers of Excellence in Influenza Research and Surveillance 14. Veen J, Delany S. An atlas of movements of southwest Siberian (contract HHSN272201400008C) and a US Defense Threat waterbirds. Wageningen (the Netherlands): Wetlands International; Reduction Agency Broad Agency Announcement award 2005. 15. Nguyen LT, Schmidt HA, von Haeseler A, Minh BQ. IQ-TREE: (FRBAA09-6-2-0114). a fast and effective stochastic algorithm for estimating maximum- likelihood phylogenies. Mol Biol Evol. 2015;32:268–74. About the Author 16. Fu L, Niu B, Zhu Z, Wu S, Li W. CD-HIT: accelerated for Dr. Alarcon is a lecturer in animal health economics at the clustering the next-generation sequencing data. Bioinformatics. Royal Veterinary College, London. During this study, he was 2012;28:3150–2. 17. Rambaut A, Carvalho LM. Exploring the temporal structure of a veterinary epidemiologist at the Animal and Plant Health heterochronous sequences using TempEst (formerly Path-O-Gen). Agency, United Kingdom, where his role and research focused Virus Evolution. 2016;2:vew007. on the analysis of avian influenza surveillance data in Europe. 18. Drummond AJ, Suchard MA, Xie D, Rambaut A. Bayesian phylogenetics with BEAUti and the BEAST 1.7. Mol Biol Evol. 2012;29:1969–73. References 19. Scoizec A, Niqueux E, Thomas R, Daniel P, Schmitz A, 1. Sims LD, Brown IH. Multi-continental panzootic of H5 highly Le Bouquin S. Airborne detection of H5N8 highly pathogenic avian pathogenic avian influenza (1996–2015). In: Swayne DE, editor. influenza virus genome in poultry farms, France. Front Vet Sci. Animal influenza, 2nd ed. New York: Wiley & Sons; 2016. 2018;5:15. 2. Global Consortium for H5N8 and Related Influenza V iruses. Role 20. European Union Reference Laboratory For Avian Influenza. Annual for migratory wild birds in the global spread of avian influenza report on surveillance for avian influenza in poultry and wild birds H5N8. Science. 2016;354:213–7. in member states of the European Union in 2016. 2017 [cited 2018 aaf8852 Apr 27]. 3. Hesterberg U, Harris K, Stroud D, Guberti V, Busani L, Pittman M, ad_control-measures_ai_surv-rslt_pltry-wld-brds_2016.pdf. et al. Avian influenza surveillance in wild birds in the European 21. Hesterberg U, Harris K, Cook A, Brown I. Annual report of the Union in 2006. Influenza Other Respi Viruses. 2009;3:1–14. EU avian influenza surveillance in wild birds 2006. Community Reference Laboratory for Avian Influenza and Newcastle Disease, 4. European Union Reference Laboratory For Avian Influenza. Annual European Commission 2007 [cited 2018 Apr 27]. https://ec.europa. report on surveillance for avian influenza in poultry and in wild eu/food/sites/food/files/animals/docs/ad_control-measures_ai_ birds in member states of the European Union in 2015. 2016 [cited surv-rslt_wld-brds_2006.pdf. 2282 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Highly Pathogenic Avian Influenza H5 in Europe 22. Guinat C, Nicolas G, Vergne T, Bronner A, Durand B, Courcoul A, wintering distributions of three common waterbird species. et al. Spatio-temporal patterns of highly pathogenic avian influenza Glob Change Biol. 2013;19:2071–81. virus subtype H5N8 spread, France, 2016 to 2017. Euro Surveill. gcb.12200 2018; 23. 28. Pavón-Jordán D, Fox AD, Clausen P, Dagys M, Deceuninck B, 23.26.1700791 Devos K, et al. Climate-driven changes in winter abundance of 23. Hill NJ, Takekawa JY, Ackerman JT, Hobson KA, Herring G, a migratory waterbird in relation to EU protected areas. Divers Cardona CJ, et al. Migration strategy affects avian influenza Distrib. 2015;21:571–82. dynamics in mallards (Anas platyrhynchos). Mol Ecol. 2012;21:5986– 29. Breed AC, Harris K, Hesterberg U, Gould G, Londt BZ, 99. 2012.05735.x Brown IH, et al. Surveillance for avian influenza in wild birds in 24. Pohlmann A, Starick E, Harder T, Grund C, Höper D, Globig A, the European Union in 2007. Avian Dis. 2010;54(Suppl):399–404. et al. Outbreaks among wild birds and domestic poultry caused by reassorted influenza A(H5N8) clade viruses, Germany, 30. Breed AC, Irvine RM, Duncan D, Rae D, Snow L, Cook AJ, 2016. Emerg Infect Dis. 2017;23:633–6. et al. An evaluation of wild bird avian influenza surveillance in eid2304.161949 Great Britain. Avian Dis. 2012;56(Suppl):986–91. 25. Fusaro A, Monne I, Mulatti P, Zecchin B, Bonfanti L, Ormelli S, 10.1637/10166-040912-Reg.1 et al. Genetic diversity of highly pathogenic avian influenza 31. Bahl J, Pham TT, Hill NJ, Hussein IT, Ma EJ, Easterday BC, et al. A(H5N8/H5N5) viruses in Italy, 2016–17. Emerg Infect Dis. Ecosystem interactions underlie the spread of avian influenza A 2017;23:1543–7. viruses with pandemic potential. PLoS Pathog. 2016;12:e1005620. 26. Atkinson PW, Robinson RA, Clark JA, Miyar T, Downie IS, du Feu CR, et al. Migratory movements of waterfowl: a web-based mapping tool. EURING report to the EU Commission. 2007. Address for correspondence: Adam Brouwer, Animal and Plant Health Agency, Woodham Lane Addlestone, Surrey, KT15 3NB, UK; email: 27. Lehikoinen A, Jaatinen K, Vähätalo AV, Clausen P, Crowe O, Deceuninck B, et al. Rapid climate-driven shifts in @CDC_EIDjournal Follow the EID journal on Twitter and EID get the most current information journal from Emerging Infectious Diseases. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2283 RESEARCH CTX-M-65 Extended-Spectrum β-Lactamase–Producing Salmonella enterica Serotype Infantis, United States Allison C. Brown, Jessica C. Chen, Louise K. Francois Watkins, Davina Campbell, Jason P. Folster, Heather Tate, Jamie Wasilenko, Christine Van Tubbergen, Cindy R. Friedman Extended-spectrum β-lactamases (ESBLs) confer resis- Salmonellosis usually causes a self-limited gastro- tance to clinically important third-generation cephalospo- enteritis; however, current guidelines recommend that rins, which are often used to treat invasive salmonellosis. antimicrobial therapy be considered for groups of persons In the United States, ESBLs are rarely found in Salmonel- at increased risk for invasive infection. For those patients, la. However, in 2014, the US Food and Drug Administra- treatment with ceftriaxone, ciprofloxacin, trimethoprim/ tion found bla ESBL-producing Salmonella enterica CTX-M-65 sulfamethoxazole, or amoxicillin is recommended (3). serotype Infantis in retail chicken meat. The isolate had Extended-spectrum β-lactamases (ESBLs) confer resis - a rare pulsed-field gel eletrophoresis pattern. To clarify tance to most third-generation cephalosporins and peni- the sources and potential effects on human health, we cillins, including ampicillin. examined isolates with this pattern obtained from hu - Some Enterobacteriaceae produce CTX-M ESBLs, man surveillance and associated metadata. Using broth which are encoded by bla genes that were discovered microdilution for antimicrobial susceptibility testing and CTX-M whole-genome sequencing, we characterized the iso- in 1989 (4). Since then, their prevalence has increased dra- lates. Of 34 isolates, 29 carried the bla gene with <9 matically (5) and they have been isolated worldwide, pri- CTX-M-65 additional resistance genes on 1 plasmid. Of 19 patients marily from Escherichia coli (6). Identification of ESBLs, with travel information available, 12 (63%) reported recent including the CTX-M types, in Salmonella in the United travel to South America. Genetically, isolates from travel- States is relatively rare (7,8). ers, nontravelers, and retail chicken meat were similar. In the United States, the National Antimicrobial Re- Expanded surveillance is needed to determine domestic sistance Monitoring System (NARMS) is a surveillance sources and potentially prevent spread of this ESBL- system that tracks changes in the antimicrobial suscepti- containing plasmid. bility of certain enteric bacteria isolated from ill persons, retail meats, and food animals. In July 2015, the US Food leading cause of bacterial foodborne disease in the and Drug Administration (FDA) notified the Centers for A United States is nontyphoidal Salmonella (1). Sal- Disease Control and Prevention (CDC) of a CTX-M-65– monella enterica serotype Infantis (hereafter called Sal- producing Salmonella Infantis strain isolated from retail monella Infantis) is one of the most common Salmonella chicken meat in December 2014 (9). The isolate had a rare serotypes in the United States and 1 of 3 serotypes for pulsed-field gel electrophoresis pattern, JFXX01.0787 which incidence has substantially increased (by 60%) (pattern 787). To clarify the sources and potential effects in the past 10 years (2). Salmonella Infantis has been of this strain on human health, we analyzed data from sev- identified in a variety of foods, animals, and environ - eral CDC surveillance systems to describe the prevalence, mental settings. epidemiology, antimicrobial drug resistance, and molecu- lar phylogenetics of Salmonella Infantis pattern 787 iso- Author affiliations: Centers for Disease Control and Prevention, lates from humans. Atlanta, Georgia, USA (A.C. Brown, J.C. Chen, L.K. Francois Watkins, D. Campbell, J.P. Folster, C. Van Tubbergen, C.R. Friedman); Food and Drug Administration, Laurel, Maryland, 1 Preliminary results from this analysis were presented at the USA (H. Tate); US Department of Agriculture, Athens, Georgia, International Salmonella and Salmonellosis Symposium, June 6–8, 2016, St. Malo, France, and at the International Association USA (J. Wasilenko) for Food Protection Annual Meeting, July 9–12, 2017, Tampa, DOI: Florida, USA. 2284 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 CTX-M-65 Salmonella Infantis, United States chloramphenicol, ciprofloxacin, nalidixic acid, and tetra- Methods cycline (10). These agents were categorized into 9 classes defined by the Clinical and Laboratory Standards Institute Background Rates guidelines (; where available, CLSI inter- To determine the expected demographics, rates of hos- pretive criteria were used to define resistance. Transforma- pitalization, and international travel among patients with tion studies to confirm plasmid-associated genes were con- Salmonella Infantis infections compared with patients with ducted by using plasmid purification and electroporation as infections caused by other common nontyphoidal Salmo- previously described (11). nella serotypes, we analyzed data collected through the Whole-genome sequencing was performed according to Foodborne Disease Active Surveillance Network (Food- the PulseNet protocol for the Illumina MiSeq (Illumina, La Net; during Jolla, CA, USA) (12). Closed PacBio genomes were gener- 2012–2015. Begun in 1996, FoodNet has conducted ac- ated as part of a previous study and used as references where tive, population-based surveillance for culture-confirmed indicated (8). Genome assemblies for short-read data were cases of infection caused by 9 pathogens transmitted com- generated de novo and analyzed for acquired antimicrobial monly through food, including Salmonella. FoodNet is drug–resistance determinants and plasmid replicons by using a collaboration of CDC, 10 state health departments, the ResFinder and PlasmidFinder (13,14). To confirm the ab- US Department of Agriculture Food Safety and Inspection sence of certain genes, we performed read mapping in CLC Service (USDA-FSIS), and the FDA. The FoodNet surveil- Genomics Workbench version 8.5 (https://www.qiagenbio - lance area includes 15% of the US population; these data To are used to estimate the burden of US foodborne illnesses, identify mutational resistance, we extracted the gyrA and hospitalizations, and deaths (1). We defined other common parC genes from genome assemblies by using a perl script nontyphoidal Salmonella as the top 20 S. enterica serotypes ( - (excluding Infantis) isolated from humans: Typhimurium, To identify mutations in the quinolone resis- Enteritidis, Newport, Heidelberg, Javiana, Saintpaul, Mon- tance–determining regions of these genes, we aligned gene tevideo, Agona, Oranienburg, Muenchen, Thompson, Ha- sequences in CLC Workbench. To assess isolate relatedness, dar, Braenderup, Derby, I 4,[5],12:i:-, Paratyphi B var. we generated high-quality single-nucleotide polymorphism L(+) tartrate+, Blockley, Anatum, Mississippi, and Pana- (hqSNP) phylogenies. In brief, isolates were aligned to the ma. These 20 serotypes represented 69% of nontyphoidal closed chromosomal sequence of 2014AM-3028 (omitting Salmonella isolates reported to FoodNet in 2015. the plasmid contig) by using Lyve-SET-v1.1.4f (https:// Genome alignments were Case Finding processed by using Gubbins (https://sanger-pathogens. We looked for pattern 787 isolates reported to the National to omit areas of recombination, unin- Molecular Subtyping Network for Foodborne Disease Sur- formative sites were removed, and the resulting SNP align- veillance (PulseNet; ment was used to calculate pairwise differences and gener - html) through 2017. The PulseNet database contains pulsed- ate hqSNP trees by using scripts bundled with Lyve-SET field gel electrophoresis patterns from state and local public (15,16). We performed a phylogeographic analysis (a type health laboratories and food regulatory agencies. Only the of molecular clock analysis) by using Bayesian Evolutionary first isolate from each patient was included in case counts. Analysis Sampling Trees (BEAST; CDC also requested patient data and clinical isolates ploscompbiol/article?id=10.1371/journal.pcbi.1003537). from state and local public health departments for any case To sample more diverse Salmonella Infantis isolates, with pattern 787 reported through October 2015. Patient we obtained sequence data on isolates from sources other data included age, sex, date of symptom onset, hospital- than CDC NARMS assigned to the same National Cen- ization, and recent international travel (defined as travel ter for Biotechnology Information (NCBI) SNP cluster outside of the United States in the 7 days before symptom PDS000003955.192 as our study isolates on the NCBI Patho- onset). Isolate data included specimen collection date(s) gen Detection page (17). These additional genomes were and source. from isolates recovered from hospitalized patients, meat, and environmental samples from Peru during 2010–2014 Isolate Characterization and collected by the Center for Food Safety and Applied We used the NARMS standard broth microdilution pro- Nutrition at FDA and the US Naval Medical Research Unit tocol (Sensititer; Thermo Fisher Scientific, Oakwood Vil- (18), and isolates from chicken samples obtained at slaugh- lage, OH, USA) to determine the MICs for 14 antimicrobial ter by USDA-FSIS. These isolates were used to produce a agents: gentamicin, streptomycin, ampicillin, amoxicillin/ time-measured maximum clade credibility tree to estimate clavulanic acid, ceftiofur, ceftriaxone, cefoxitin, azithro- the dates that study isolates in addition to all isolates (study mycin, sulfasoxazole, trimethoprim/sulfamethoxazole, Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2285 RESEARCH isolates and isolates from Peru and cecal isolates from US- DA-FSIS) shared most recent common ancestors (MRCAs), and the geographic location of these MRCAs (18). The date of the MRCA was estimated by reporting the height of the most ancestral node of the maximum clade credibility tree and the 95% highest posterior density (HPD) intervals for these estimates. More details on genetic analyses are avail- able in the online Technical Appendix (https://wwwnc.cdc. gov/EID/article/24/12/18-0500-Techapp1.pdf). Statistical Analyses We used the χ test (or Fisher exact test when cell counts were ≤5) for statistical comparisons. All denominators ex- Figure 1. Total of 312 Salmonella enterica serotype Infantis clude persons for whom data were missing. We considered isolates from humans with pattern JFXX01.0787 as a p<0.05 to be significant. All p values were 2-tailed. We percentage of all Salmonella Infantis isolates by year, United used SAS 9.3 or 9.4 (SAS Institute, Cary, NC, USA) to States, 2012–2017. Source: PulseNet ( conduct our analyses. pulsenet/index.html). Results infected with CTX-M-65 Infantis with patients infected The first pattern 787 Salmonella Infantis isolate from a hu- with all strains of Salmonella Infantis and with patients man in the United States was collected from a patient in infected with common Salmonella serotypes other than In- June 2012. By the end of 2017, PulseNet contained 312 fantis. The median age of patients with CTX-M-65 Infantis Salmonella Infantis pattern 787 isolates from persons liv- was 25 years (interquartile range [IQR] 15–50 years), and ing in 43 states; Washington, DC; and Puerto Rico. The 69% were female (Table 3). The median age of patients number of cases detected each year increased from 5 in with any strain of Salmonella Infantis was 37 years (IQR 2012 to 174 in 2017 and represented 8.4% of all Salmo- 12–58 years), and 56% were female. The median age of nella Infantis isolated that year (Figure 1). patients infected with other common Salmonella serotypes State health departments submitted 34 pattern 787 was 29 years (IQR 6–54 years), and 52% were female. isolates from humans to CDC; 29 (85%) had resistance Of 18 patients infected with CTX-M-65 Infantis for phenotypes consistent with ESBL-conferred resistance to whom outcome data were available, 8 (44%) were hospital- ceftriaxone, ceftiofur, and ampicillin (Tables 1, 2). All 29 ized, compared with a hospitalization rate of 29% among isolates with ESBL-resistant phenotypes had the bla patients in both comparison groups (those with Infantis and CTX- gene. In addition to ESBL-conferred resistance, these those with other Salmonella serotypes; Table 3). Patients M-65 isolates were also resistant to chloramphenicol, sulfisoxa- with CTX-M-65 Infantis infection were more likely (17%) zole, tetracycline, nalidixic acid, and trimethoprim/sulfa- to have the organism isolated from urine compared with methoxazole and intermediately susceptible to ciprofloxa- all patients with Salmonella Infantis (9%, p = 0.14) and cin and gentamicin. Resistance to these drugs was plasmid patients with other common serotypes (5%, p<0.01). mediated and transferable to E. coli, except for nalidixic For 20 CTX-M-65 Infantis–infected patients with in- acid and ciprofloxacin, for which resistance was caused formation about symptom onset date and isolation date(s), by a chromosomal mutation in gyrA (D87Y). Transferable Salmonella Infantis was recovered from 12 who provided resistance resulted from a large IncFIB-like plasmid con- fecal samples >2 weeks after their reported date of symp- taining the resistance genes aph(3′)-Ic, aph (4)-Ia, aadA1, tom onset. A total of 8 patients were still reporting symp- aac (3)-IVa, bla , fosA, floR, sul1, tetA, and dfrA14. toms and had positive cultures >30 days after symptom CTX-M-65 These observations are consistent with the organization of onset, including 3 patients who sought care and had mul- these genes on a single IncF1B-like plasmid closed by us- tiple isolates recovered from 50 days to 8 months after ing long read sequencing, which was completed as part of symptom onset. a previous study by Tate et al., who recently published a Twelve (63%) patients with CTX-M-65 Infantis in- further detailed description of this plasmid and a compara- fections reported international travel in the 7 days before tive genomics analysis of bla –positive IncFIB-like symptom onset. All reported travel to South America, 10 CTX-M-65 plasmids from Salmonella Infantis in the United States (9). to Peru, and 2 to Ecuador (Table 3). Travel was less com- We focused our analysis on the 29 patients with Salmo- mon among all Salmonella Infantis–infected patients (7%, nella Infantis isolates containing the bla gene (here- p<0.01) and those infected with other common serotypes CTX-M-65 after called CTX-M-65 Infantis). We compared patients (8%, p<0.01; Table 3). 2286 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 CTX-M-65 Salmonella Infantis, United States A maximum-likelihood hqSNP phylogeny of the 34 evolved from a common ancestor. Included in clade A was isolates from humans and that from retail chicken meat the isolate from retail chicken meat collected in 2014. revealed that isolates from patients with travel-associated Of the 5 CTX-M-65-negative isolates, 2 grouped with- infections formed a well-supported clade (clade A) with in clade A and contained a multidrug-resistance plasmid isolates from patients with infections not associated with that was lacking the bla gene. An additional 3 CTX- CTX-M-65 travel (Figure 2). Isolates in clade A differed by only 2–47 M-65–negative isolates lacked the IncFIB-like plasmid pairwise hqSNPs, suggesting that these isolates recently replicon, lacked resistance determinants, and differed by Table 1. Characteristics of patients and 30 Salmonella enterica serotype Infantis pattern JFXX01.0787 isolates associated with CTX- M-65, United States* Accession Specimen Recent Prolonged no.† Year source Resistance profile Plasmids gyrA parC Hospitalized travel infection‡ SRR2485281 2012 Urine ACSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt Yes Peru Gen(I) SRR2485278 2012 Blood ACSuTCxTioFoxCip(I)NalCot IncFIB-like D87Y Wt No data Peru Yes Gen(I) SRR4019593 2013 Feces ASuTCxTioCip(I)NalCotGen IncFIB-like D87Y Wt No data No data SRR4019592 2013 Feces ACSuTCxTioCip(I)NalCotGen IncFIB-like D87Y Wt No data No data SRR2485284 2013 Feces ACSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt No data No data Yes Gen(I) SRR4025935 2013 Feces ACSuTCxTioCip(I)Nal IncFIB-like D87Y Wt No Peru SRR4025936 2013 Feces ACSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt No data Peru Yes SRR3178069 2013 Feces ACSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt No Peru Yes SRR4019589 2014 Feces ACSSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt No Ecuador Gen SRR2407791 2014 Retail ACTCxTioCip(I)NalGen(I) IncFIB-like D87Y Wt NA NA NA chicken meat SRR2353201 2014 Feces ACSSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt No data No data Gen(I) SRR4025938 2014 Feces ACSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt No data No data Gen(I) SRR2485287 2014 Feces ACSSuTAug(I)CxTioFox IncFIB-like D87Y Wt No data No data Yes Cip(I)NalCotGen(I) SRR4019601 2014 Feces ASSuTCxTioCip(I)NalCotGen IncFIB-like D87Y Wt No data No data Yes SRR4019595 2014 Feces ACSSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt No No travel SRR2485288 2014 Feces ACSSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt No data No travel Gen(I) SRR2485282 2014 Feces ACSSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt No Peru Yes Gen(I) SRR2485286 2014 Feces ASSuTCxTioCip(I)NalCotGen IncFIB-like D87Y Wt Yes Peru SRR3178071 2014 Feces ACSSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt Yes Peru SRR4019602 2015 Feces ACSSuTCxTioFox(I)Cip(I)Nal IncFIB-like D87Y Wt No Ecuador Cot SRR4019594 2015 Feces ACSSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt No data No data SRR3185043 2015 Urine ACSSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt No No data Yes SRR4019591 2015 Feces; ACSSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt Yes No data Yes urine SRR4019588 2015 Feces ACSuTCxTioCip(I)NalCot IncFIB-like, D87Y Wt Yes No travel Yes Gen(I) colE SRR4019587 2015 Feces ACSSuTCxTioCip(I)Nal IncFIB-like D87Y Wt No No travel Gen(I) SRR3184311 2015 Feces ACSSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt Yes No travel Yes Gen(I) SRR4019590 2015 Feces ACSuTCxTioCip(I)NalCot pXuzhou21, D87Y Wt Yes No travel IncFIB-like SRR4019603 2015 Urine ACSSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt No No travel SRR3178070 2015 Urine ACSSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt Yes Peru SRR4019600 2015 Feces ACSuTCxTioCip(I)NalCot IncFIB-like D87Y Wt No Peru Yes *A, ampicillin; Aug, amoxicillin-clavulanic acid; C, chloramphenicol; Cip, ciprofloxacin; Cot, trimethoprim/sulfamethoxazole; Cx, ceftriaxone; Cx, ceftriaxone; Fox, cefoxitin; Gen, gentamicin; (I), intermediate susceptibility; NA, not applicable; Nal, nalidixic acid; S, streptomycin; Su, sulfisoxazole; T, tetracycline; Tio, ceftiofur. †Accession numbers are the sequence read archive run identification numbers from the National Center for Biotechnology Information. ‡Documented illness >2 wk after symptom onset. Blank cells indicate no evidence of prolonged infection (either no additional data were available or additional data indicated no clear evidence of prolonged infection). Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2287 RESEARCH Table 2. Characteristics of patients and 5 Salmonella enterica serotype Infantis pattern JFXX01.0787 isolates without CTX-M-65, United States* Accession Specimen no.† Year source Resistance profile Plasmids gyrA parC Hospitalized Recent travel SRR2485280 2012 Feces No resistance detected No replicons None Wt No No travel detected SRR4025941 2012 Feces CSuTCip(I)NalCotGen(I) IncFIB-like D87Y Wt No Peru, Bolivia, Ecuador, Chile SRR4019596 2014 Urine CSSuTFox(I)Cip(I)NalCot IncFIB-like D87Y Wt No No data SRR3185042 2014 Feces No resistance detected No replicons None Wt Yes No travel detected SRR2485289 2015 Feces No resistance detected No replicons None Wt No No data detected *A, ampicillin; Aug, amoxicillin-clavulanic acid; C, chloramphenicol; Cip, ciprofloxacin; Cot, trimethoprim/sulfamethoxazole; Cx, ceftriaxone; Fox, cefoxitin; Gen, gentamicin; (I), intermediate susceptibility; Nal, nalidixic acid; S, streptomycin; Su, sulfisoxazole; T, tetracycline; Tio, ceftiofur. †Accession numbers are the sequence read archive run identification numbers from the National Center for Biotechnology Information. 96–273 pairwise hqSNPs from isolates in clade A. Absence returned from South America, whereas subsequent infec- of the bla gene in these 5 isolates was confirmed by tions were acquired domestically. CTX-M-65 read-mapping procedures and supported by the observed The demographic and clinical characteristics among phenotypic susceptibility to β-lactam antimicrobial drugs. patients with CTX-M-65 Infantis infections differed from A phylogeographic analysis of the 32 isolates in the those of patients infected with all strains of Salmonella In- main clade and additional genomes from NCBI generated fantis or other common Salmonella serotypes. CTX-M-65 a time-measured phylogenetic tree of isolates from the Infantis–infected patients were younger and more frequent- United States and Peru (online Technical Appendix Figure ly female, and rates of hospitalization were 50% higher for 1). These isolates last shared a common ancestor around these patients than for those in the other 2 groups. Among 2006 (95% HPD interval 2003–2008). This analysis also patients with CTX-M-65 Infantis infections, 63% reported suggests that the MRCA of these isolates existed in Peru recent travel to South America, predominantly to Peru, with a probability of 98.7%. The probability of the tree be- compared with <10% in the comparison groups reporting ing rooted in the United States was ≈1.3%. Clinical iso- travel. This finding is consistent with those of other studies lates from humans in the United States sequenced as part of that found foreign travel to be a risk factor for CTX-M– our study last shared a common ancestor sometime around type ESBLs (19–21). Studies have shown that CTX-M-65 2009 (95% HPD interval 2008–2009), before the first isola- has recently emerged in commensal E. coli in Bolivia (22) tion of pattern 787 in the United States in 2012. and in Salmonella Infantis in Ecuador (23), 2 countries to which patients in our study also traveled. Discussion Our phylogeographic analysis, which included many A new strain of Salmonella Infantis, which has pattern 787 isolates from Peru, suggested that isolates from patients in and frequently carries a multidrug-resistant plasmid with a the United States last shared a common ancestor that exist- CTX-M-65 ESBL, has emerged in the United States. This ed in Peru in 2009. This finding makes sense, given the high strain possesses clinically important resistance associated proportion of patients in our study who reported travel to with higher hospitalization rates. Using epidemiologic and Peru and the known circulation of CTX-M-65–positive En- phylogenetic methods, we demonstrated that the earliest terobacteriaceae in South America. In our study, patients cases of CTX-M-65 Infantis infection were among travelers with CTX-M-65 Infantis infection and a history of travel Table 3. Comparison of demographic and clinical characteristics of patients with CTX-M-65–positive isolates with patients infected with all Salmonella enterica serotype Infantis strains or other common Salmonella serotypes, FoodNet, United States, 2012–2015* CTX-M-65-positive Salmonella Infantis, pattern JFXX01.0787, n = 29* All Salmonella Infantis, n = 723† Other Salmonella, n = 21,285‡ Characteristic Proportion§ % p value Proportion§ % p value Proportion§ % p value Female sex 20/29 69 Referent 403/722 56 0.16 11,067/21,258 52 0.07 Urine isolation 5/29 17 Referent 66/723 9 0.14 1,076/21,214 5 <0.01 Hospitalization 8/18 44 Referent 199/695 29 0.14 5,951/20,619 29 0.14 International 12/19 63 Referent 51/611 8 <0.01 1,431/16,496 9 <0.01 travel *Patient age range 0–82 y; median age (interquartile range) 25 (15–50) y. †Patient age range 0–95 y; median age (interquartile range) 37 (12–58) y. ‡Patient age range 0–101 y; median age (interquartile range) 29 (6–54) y. Other Salmonella defined as the top 20 Salmonella enterica serotypes (excluding Infantis): Typhimurium, Enteritidis, Newport, Heidelberg, Javiana, Saintpaul, Montevideo, Agona, Oranienburg, Muenchen, Thompson, Hadar, Braenderup, Derby, I 4,[5],12:i:-, Paratyphi B var. L(+) tartrate+, Blockley, Anatum, Mississippi, Panama. §Unknowns were excluded from the denominator. 2288 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 CTX-M-65 Salmonella Infantis, United States were reported every year from 2012 through 2015; patients Poultry consumption may be a source of CTX-M-65 In- with no history of travel were first reported in 2014. fantis infection in the United States and abroad. In our study, One caveat is that genetic information for CTX-M In- more than one third of patients with CTX-M-65 Infantis in- fantis from countries in South America other than Peru was fections did not report recent international travel and thus not available; therefore, we could not distinguish the role were exposed via a domestic source. CTX-M genes have that other countries in South America may have played in been linked to poultry, particularly broiler chickens (24–26). the spread of this Salmonella Infantis strain to the United A study of 14 chicken farms in Henan Province, China, con- States. We cannot definitively determine how patients with ducted during 2007–2008, was the first to describe detection this strain of Salmonella Infantis who did not report travel of CTX-M-65–producing E. coli in chickens (27). Recently, to South America became infected; however, our analysis CTX-M-65 Infantis was found to be transmitted from broiler does show a close genetic relationship between clinical iso- chickens and chicken meat to humans in Italy (28). lates from these patients and isolates collected by USDA- The results of our initial hqSNP analysis demonstrated FSIS from chickens in the United States. that the original isolate collected from retail chicken in Figure 2. High-quality single-nucleotide polymorphism–based phylogenetic tree of clinical and retail meat isolates of Salmonella enterica serotype Infantis with pulsed-field gel electrophoresis pattern JFXX01.0787 collected in the United States and submitted to the National Antimicrobial Resistance Monitoring System for whole-genome sequencing. Tree tips are labeled with National Center for Biotechnology Information accession numbers (sequence read archive run identification numbers); shading indicates patients’ international travel history for clinical isolates (black, recent international travel; gray, no recent international travel; white, travel data missing). Black circles indicate isolates that are missing the bla gene. Isolates in top clade differed by 2–47 high-quality single-nucleotide polymorphisms. Numbers CTX-M-65 displayed on nodes are bootstrap support values, an indication of the reliability of the tree. Only bootstrap values >80 are displayed. More information on patient and isolate characteristics are provided in Tables 1 and 2. Scale bar indicates nucleotide substitutions per site. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2289 RESEARCH 2014 was genetically related to isolates from humans. In widespread dissemination in the United States. Enhanced addition, our data show a clear partition between the dates surveillance and additional studies in humans and food ani- of specimen collection from the first travel-associated in - mals may help pinpoint the sources of infection for imple- fections detected in the United States (2012) and the dates mentation of prevention and control measures. Meanwhile, of the first domestically acquired infections detected (2014) travelers and healthcare providers should be aware of the (online Technical Appendix Figure 2). Sampling in poul- risks and implications of infection with this strain, includ- try production plants and sequence data from USDA-FSIS ing the potential for antimicrobial treatment failure. demonstrate that this strain was present in domestic food processing plants during the latter part of the study period Acknowledgments (2014–2015) (29). We cannot, however, determine pre- We thank Heather Carleton and Lee Katz for helpful cisely when or how this strain was introduced into poultry discussion and guidance on some analyses, Beth Tolar for stock in the United States because enhanced poultry plant fulfilling repeated data requests, the USDA-FSIS Eastern sampling was not conducted before 2014. International dis- Laboratory Microbiology Characterization Branch for their work tribution of infected breeder stocks, chicken feed, or feed on adding whole-genome sequencing data to NCBI, Lauren additives contaminated with CTX-M-65 Infantis could Ahart for manuscript preparation, and the FoodNet sites and help explain how these broiler-associated infections have public health departments in the following states for their spread globally during the same period. collaboration on this investigation: Arizona, California, CTX-M-65, in comparison with more well-charac- Colorado, Florida, Illinois, Kentucky, Louisiana, Massachusetts, terized CTX-M enzymes, differs by only 2 substitutions Michigan, New York, Oregon, Pennsylvania, Texas, Utah, (A77V, S272R) from CTX-M-14, one of the more com- Virginia, and Wisconsin. monly detected CTX-M variants worldwide (30,31). The About the Author bla gene was transmitted on a large IncFIB-like plas- CTX-M-65 mid containing multiple resistance genes. The presence of Dr. Brown is team lead for AR Capacities and Special Studies in multidrug resistance in CTX-M-65 Infantis isolates in our the Division of Healthcare Quality Promotion, National Center study suggests that a variety of antimicrobial drugs could for Emerging and Zoonotic Infectious Diseases, CDC. Her provide positive selection pressure and thus promote per- research interests include emerging mechanisms of antimicrobial sistence of this strain. These characteristics suggest that the drug resistance in healthcare-associated infections. potential for spread of the gene is high. Research conducted in Bolivia has shown that even in the absence of selective References pressure from antimicrobial drug use, plasmid transfer of 1. Scallan E, Hoekstra RM, Angulo FJ, Tauxe RV, Widdowson MA, CTX-M-65 from E. coli to other pathogens was achieved at Roy SL, et al. Foodborne illness acquired in the United States—major pathogens. Emerg Infect Dis. 2011;17:7–15. high frequency and shown to be stable (22). Additional studies on antimicrobial drug use and man- 2. Marder EP, Griffin PM, Cieslak PR, Dunn J, Hurd S, Jervis R, agement practices in food animals may help us understand et al. Preliminary incidence and trends of infections with pathogens which factors contribute most to the emergence, persis- transmitted commonly through food—Foodborne Diseases Active Surveillance Network, 10 U.S. sites, 2006–2017. MMWR Morb tence, and spread of resistance genes such as bla . CTX-M-65 Mortal Wkly Rep. 2018;66:397–403. New efforts to perform whole-genome sequencing on all mmwr.mm6615a1 Salmonella isolates at public health laboratories nationwide 3. Shane AL, Mody RK, Crump JA, Tarr PI, Steiner TS, Kotloff K, will help determine whether plasmid-mediated bla et al. 2017 Infectious Diseases Society of America clinical practice CTX-M-65 guidelines for the diagnosis and management of infectious diarrhea. has spread to other Salmonella serotypes. One limitation of Clin Infect Dis. 2017;65:1963–73. our study was that we were unable to obtain epidemiologic cix959 data for all 312 cases. The data available from FoodNet en- 4. Bauernfeind A, Casellas JM, Goldberg M, Holley M, Jungwirth R, abled us to compare variables such as patient demograph- Mangold P, et al. A new plasmidic cefotaximase from patients infected with Salmonella typhimurium. Infection. 1992;20:158–63. ics, travel, and hospitalizations; however, we did not have a control population for evaluating other variables of interest 5. Poirel L, Bonnin RA, Nordmann P. Genetic support and diversity to determine potential domestic sources of transmission. of acquired extended-spectrum β-lactamases in Gram-negative The spread of CTX-M-65 is concerning because the rods. Infect Genet Evol. 2012;12:883–93. j.meegid.2012.02.008 presence of ESBLs eliminates 2 recommended treatment 6. Livermore DM, Canton R, Gniadkowski M, Nordmann P, options, ceftriaxone and ampicillin, for the management Rossolini GM, Arlet G, et al. CTX-M: changing the face of of salmonellosis. Given the multidrug-resistant profile of ESBLs in Europe. J Antimicrob Chemother. 2007;59:165–74. CTX-M-65 Infantis, potential for plasmid-mediated trans- 7. Sjölund-Karlsson M, Howie R, Krueger A, Rickert R, Pecic G, mission, increased hospitalization rate, and evidence of Lupoli K, et al. CTX-M-producing non-Typhi Salmonella spp. this strain in domestic poultry, action is needed to prevent 2290 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 CTX-M-65 Salmonella Infantis, United States isolated from humans, United States. Emerg Infect Dis. 21. Arcilla MS, van Hattem JM, Haverkate MR, Bootsma MC, 2011;17:97–9. van Genderen PJ, Goorhuis A, et al. Import and spread of 8. Sjölund-Karlsson M, Howie RL, Blickenstaff K, Boerlin P, Ball T, extended-spectrum β-lactamase-producing Enterobacteriaceae Chalmers G, et al. Occurrence of β-lactamase genes among non- by international travellers (COMBAT study): a prospective, Typhi Salmonella enterica isolated from humans, food animals, and multicentre cohort study. Lancet Infect Dis. 2016. retail meats in the United States and Canada. Microb Drug Resist. 22. Riccobono E, Di Pilato V, Di Maggio T, Revollo C, Bartoloni A, 2013;19:191–7. Pallecchi L, et al. Characterization of IncI1 sequence type 71 9. Tate H, Folster JP, Hsu CH, Chen J, Hoffmann M, Li C, et al. epidemic plasmid lineage responsible for the recent dissemination Comparative analysis of extended-spectrum-beta-lactamase of CTX-M-65 extended-spectrum β-lactamase in the Bolivian CTX-M-65-producing Salmonella enterica serovar Infantis isolates Chaco region. Antimicrob Agents Chemother. 2015;59:5340–7. from humans, food animals, and retail chickens in the United States. Antimicrob Agents Chemother. 2017;61:e00488-17. 23. Cartelle Gestal M, Zurita J, Paz Y Mino A, Ortega-Paredes D, Alcocer I. Characterization of a small outbreak of Salmonella 10. Centers for Disease Control and Prevention. National Antimicrobial enterica serovar Infantis that harbour CTX-M-65 in Ecuador. Resistance Monitoring System for enteric bacteria (NARMS): Braz J Infect Dis. 2016;20:406–7. human isolates surveillance report for 2015 (final report). Atlanta: j.bjid.2016.03.007 The Centers; 2016. p. 22–5. 24. Maciuca IE, Williams NJ, Tuchilus C, Dorneanu O, Guguianu E, 11. Folster JP, Pecic G, McCullough A, Rickert R, Whichard JM. Carp-Carare C, et al. High prevalence of Escherichia coli- Characterization of bla(CMY)-encoding plasmids among producing CTX-M-15 extended-spectrum β-lactamases in poultry Salmonella isolated in the United States in 2007. Foodborne Pathog and human clinical isolates in Romania. Microb Drug Resist. Dis. 2011;8:1289–94. 2015;21:651–62. 12. Centers for Disease Control and Prevention. Laboratory standard 25. Silva KC, Fontes LC, Moreno AM, Astolfi-Ferreira CS, operating procedure for PulseNet Nextera XT Library Prep and Run Ferreira AJ, Lincopan N. Emergence of extended-spectrum-β- Setup for the Illlumina MiSeq. Code PNL32. 2015 [cited 2016 Feb lactamase CTX-M-2-producing Salmonella enterica serovars 1]. Schwarzengrund and Agona in poultry farms. Antimicrob 13. Zankari E, Hasman H, Cosentino S, Vestergaard M, Rasmussen S, Agents Chemother. 2013;57:3458–9. Lund O, et al. Identification of acquired antimicrobial resistance AAC.05992-11 genes. J Antimicrob Chemother. 2012;67:2640–4. 26. Weill FX, Lailler R, Praud K, Kérouanton A, Fabre L, Brisabois A, 10.1093/jac/dks261 et al. Emergence of extended-spectrum-beta-lactamase 14. Carattoli A, Zankari E, García-Fernández A, Voldby Larsen M, (CTX-M-9)-producing multiresistant strains of Salmonella enterica Lund O, Villa L, et al. In silico detection and typing of plasmids serotype Virchow in poultry and humans in France. J Clin using PlasmidFinder and plasmid multilocus sequence typing. Microbiol. 2004;42:5767–73. Antimicrob Agents Chemother. 2014;58:3895–903. JCM.42.12.5767-5773.2004 27. Yuan L, Liu JH, Hu GZ, Pan YS, Liu ZM, Mo J, et al. Molecular 15. Katz LS, Petkau A, Beaulaurier J, Tyler S, Antonova ES, characterization of extended-spectrum β-lactamase-producing Turnsek MA, et al. Evolutionary dynamics of Vibrio cholerae O1 Escherichia coli isolates from chickens in Henan Province, China. following a single-source introduction to Haiti. MBio. 2013; J Med Microbiol. 2009;58:1449–53. 4:e00398-13. jmm.0.012229-0 16. Croucher NJ, Page AJ, Connor TR, Delaney AJ, Keane JA, 28. Franco A, Leekitcharoenphon P, Feltrin F, Alba P, Cordaro G, Bentley SD, et al. Rapid phylogenetic analysis of large samples of Iurescia M, et al. Emergence of a clonal lineage of multidrug- recombinant bacterial whole genome sequences using Gubbins. Nucleic resistant ESBL-producing Salmonella Infantis transmitted from Acids Res. 2015;43:e15. broilers and broiler meat to humans in Italy between 2011 and 17. US National Library of Medicine. Pathogen detection [cited 2018 2014. PLoS One. 2015;10:e0144802. Jan 23]. journal.pone.0144802 18. Gopinath G, Chase H, Gangiredla J, Patel I, Addy N, 29. Food and Drug Administration . NARMS Now [cited 2018 Jan 19]. Beaubrun JJG, et al. Comparative genomics of blaCTX-M-65- resistant clinical strains of Salmonella enterica serovar Infantis Resistance/NationalAntimicrobialResistanceMonitoringSystem/ from Peru and resistant strains from chicken, cattle and humans. ucm416741.htm Poster presented at: International Association for Food Protection 30. Yin J, Cheng J, Sun Z, Ye Y, Gao YF, Li JB, et al. Characterization annual meeting; 2017 Jul 9–12; Tampa, FL, USA. of two plasmid-encoded cefotaximases found in clinical Escherichia 19. Tängdén T, Cars O, Melhus A, Löwdin E. Foreign travel is a major coli isolates: CTX-M-65 and a novel enzyme, CTX-M-87. J Med risk factor for colonization with Escherichia coli producing Microbiol. 2009;58:811–5. CTX-M-type extended-spectrum β-lactamases: a prospective 31. Bush K. Proliferation and significance of clinically relevant study with Swedish volunteers. Antimicrob Agents Chemother. β-lactamases. Ann N Y Acad Sci. 2013;1277:84–90. 2010;54:3564–8. 20. Hopkins KL, Batchelor MJ, Liebana E, Deheer-Graham AP, Threlfall EJ. Characterisation of CTX-M and AmpC genes in Address for correspondence: Allison C. Brown, Centers for Disease human isolates of Escherichia coli identified between 1995 and 2003 Control and Prevention, 1600 Clifton Rd NE, Mailstop C16, Atlanta, GA in England and Wales. Int J Antimicrob Agents. 2006;28:180–92. 30329-4027, USA; email: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2291 RESEARCH Substance Use and Adherence to HIV Preexposure Prophylaxis for Men Who Have Sex with Men Martin Hoenigl, Sonia Jain, David Moore, Deborah Collins, Xiaoying Sun, Peter L. Anderson, Katya Corado, Jill S. Blumenthal, Eric S. Daar, Joel Milam, Michael P. Dubé, Sheldon Morris, for the California Collaborative Treatment Group 595 Team In support of improving patient care, this activity has been planned and implemented by Medscape, LLC and Emerging Infectious Diseases. Medscape, LLC is jointly accredited by the Accreditation Council for Continuing Medical Education (ACCME), the Accreditation Council for Pharmacy Education (ACPE), and the American Nurses Credentialing Center (ANCC), to provide continuing education for the healthcare team. Medscape, LLC designates this Journal-based CME activity for a maximum of 1.00 AMA PRA Category 1 Credit(s)™. Physicians should claim only the credit commensurate with the extent of their participation in the activity. All other clinicians completing this activity will be issued a certificate of participation. To participate in this journal CME activity: (1) review the learning objectives and author disclosures; (2) study the education content; (3) take the post-test with a 75% minimum passing score and complete the evaluation at; and (4) view/print certificate. For CME questions, see page 2401. Release date: November 15, 2018; Expiration date: November 15, 2019 Learning Objectives Upon completion of this activity, participants will be able to: • Assess the efficacy of preexposure prophylaxis (PrEP) in different patient populations • Distinguish rates of substance use and PrEP adherence in the current cohort of predominantly men who have sex with men (MSM) • Analyze the effect of substance use on adherence to PrEP • Evaluate substances associated with a higher risk for sexually transmitted diseases among MSM CME Editor Thomas J. Gryczan, MS, Technical Writer/Editor, Emerging Infectious Diseases. Disclosure: Thomas J. Gryczan, MS, has disclosed no relevant financial relationships. CME Author Charles P. Vega, MD, Clinical Professor, Health Sciences, Department of Family Medicine, University of California, Irvine School of Medicine, Irvine, California. Disclosure: Charles P. Vega, MD, has disclosed the following relevant financial relationships: served as an advisor or consultant for Johnson & Johnson Pharmaceutical Research & Development, L.L.C.; Shire; and Sunovion Pharmaceuticals Inc.; served as a speaker or a member of a speakers bureau for Shire. Authors Disclosures: Sonia Jain, PhD; David Moore, PhD; Deborah Collins, PA-C, MSPAS; Xiaoying Sun, MS; and Joel Milam, PhD, have disclosed no relevant financial relationships. Martin Hoenigl, MD, has disclosed the following relevant financial relationships: served as a speaker or a member of a speakers bureau for Merck & Co., Inc.; received grants for clinical research from Gilead Sciences, Inc. Peter L. Anderson, PharmD, has disclosed the following relevant financial relationships: received grants for clinical research from Gilead Sciences, Inc. Katya Corado, MD, has disclosed the following relevant financial relationships: received grants for clinical research from Gilead Sciences, Inc. Jill Blumenthal, MD, has disclosed the following relevant financial relationships: served as an advisor or consultant for Gilead Sciences, Inc.; received grants for clinical research from Gilead Sciences, Inc. Eric S. Daar, MD, has disclosed the following relevant financial relationships: served as an advisor or consultant for Gilead Sciences, Inc.; Merck & Co., Inc.; ViiV Healthcare; received grants for clinical research from Gilead Sciences, Inc.; Merck & Co., Inc.; ViiV Healthcare. Michael P. Dubé, MD, has disclosed the following relevant financial relationships: served as an advisor or consultant for Gilead Sciences, Inc.; Thera; received grants for clinical research from Bristol-Myers Squibb Company; Gilead Sciences, Inc.; Thera; ViiV Healthcare. Sheldon Morris, MD, has disclosed the following relevant financial relationships: received grants for clinical research from Gilead Sciences, Inc. This article may discuss off-label uses of drugs, mechanical devices, biologics, or diagnostics approved by the Food and Drug Administration for use in the United States. Original data from this study were presented in part at the 2017 Conference on Retroviruses and Opportunistic Infections, February 13–16, 2017, Seattle, Washington, USA; and at the 12th International Conference on HIV Treatment and Prevention Adherence, June 4–6, 2017, Miami, Florida, USA. 2292 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Page 1 of 1 Substance Use and Adherence to HIV Prophylaxis The effectiveness of oral HIV preexposure prophylaxis seroconverison in the treatment arm (9,10). However, in (PrEP) strongly depends on maintaining adherence. We that study, study participants had daily observed dosing in investigated the association between substance use and conjunction with substance use disorder treatment. There- PrEP adherence, as well as incident sexually transmitted in- fore, adherence remains uncertain among substance users fections (STIs) in a high-risk cohort of 394 participants (391 without observed therapy. men who have sex with men and 3 transgender women) who The effectiveness of TDF/FTC for HIV PrEP strongly were enrolled in a PrEP demonstration project. We assessed depends on maintaining adherence (11,12). Although stud- baseline and ongoing substance use over a 48-week period ies have indicated that different strategies might be required for stimulants and nonstimulant substances and for each for PrEP implementation for MSM who use stimulant sub- substance separately. We measured PrEP adherence by stances and alcohol (13), comprehensive/demonstrative using dried blood spots to obtain levels of tenofovir diphos- studies that evaluate adherence among MSM or transgender phate. No differences in these levels were found between substance users and nonsubstance users. Baseline stimu- women using different classes of substances are lacking. lant use was strongly associated (odds ratio 3.4; p<0.001) We hypothesized that, among MSM and transgender with incident STIs during the study. Thus, PrEP adherence women enrolled in a randomized controlled PrEP demon- was not decreased by substance use. Because substance stration trial, substance users would have lower levels of users had increased rates of STIs, indicating higher-risk be- PrEP adherence. The objective of our study was to inves- havior, they might be excellent candidates for PrEP. tigate the association between substance/alcohol use and adherence to PrEP, as well as sexually transmitted infec- tions (STIs) and study completion, in a well-characterized ver the past 2 decades, substance use, in particular high-risk cohort of MSM and transgender women who par- Ouse of stimulants, such as methamphetamine, remains ticipated in the California Collaborative Treatment Group prevalent among men who have sex with men (MSM) and (CCTG) 595 Study. transgender women in the United States (1). Alcohol, stim- ulant use, and injection drug use (IDU) are independently Materials and Methods associated with increased risk behavior and HIV acquisi- CCTG 595 was a randomized controlled trial of individu- tion among MSM and transgender women (1–4). Thus, alized text messaging versus standard care for adherence HIV-uninfected MSM and transgender women with sub- to daily TDF/FTC PrEP ( stance use might represent ideal candidates for preexposure show/NCT01761643) (14). In CCTG 595, PrEP was given prophylaxis (PrEP). in combination with safety monitoring, HIV/STI testing, The efficacy of tenofovir disoproxil fumarate (TDF) and risk reduction counseling. On a daily basis, participants combined with emtricitibine (FTC) for HIV PrEP has been in the intervention arm received a mixture of health pro- documented in several randomized and controlled trials (5– motion and factoid messages at a personally selected time 7). In the iPrEx study, TDF/FTC reduced the risk for HIV consistent with when they planned to take PrEP. The study infection in MSM by 44% vs. placebo, and a 73% lower was conducted during February 2014–February 2016. Pa- risk of HIV infection was reported for persons who had tients were enrolled at 4 medical centers in southern Cali- >90% self-reported adherence (5), and >90% lower risk for fornia (University of California San Diego, University of persons who had adherence defined by tenofovir diphos- Southern California, Harbor–University of California Los phate (TFV-DP) drug levels commensurate with >4 tablets Angeles, and Long Beach Health Department), and partici- per week (8). pants were provided with mobile phones in case they did The Bangkok TDF Study randomized 2,413 persons not have a mobile phone (14). who injected drugs (PWIDs; <5% were MSM) 1:1 to TDF Eligible participants for CCTG 595 were HIV-unin- or placebo, and results showed a 48% reduction in HIV fected MSM and transgender women (age >18 years) con- firmed by a negative result for an antigen/antibody assay Author affiliations: Medical University of Graz, Graz, Austria or antibody assay plus HIV nucleic acid amplification test. (M. Hoenigl); University of California, San Diego, California, USA Participants needed to have a persistent increased risk for (M. Hoenigl, S. Jain, D. Moore, X. Sun, J.S. Blumenthal, HIV acquisition as determined by >1 of the following cri- S. Morris); Department of Health and Human Services, Long teria: 1) >1 HIV-infected sexual partner for >4 weeks; 2) Beach, California, USA (D. Collins); University of Anschutz condomless anal intercourse with >3 male sex partners who Medical Campus, Aurora, Colorado, USA (P.L. Anderson); Harbor were HIV positive or of unknown HIV status during the UCLA Medical Center, Torrance, California, USA (K. Corado, previous 3 months; or 3) condomless anal sex with >1 male E.S. Daar); University of Southern California Keck School of partner and an STI diagnosis during the previous 3 months. Medicine, Los Angeles, California, USA (J. Milam, M.P. Dubé) Participants were required to have acceptable laboratory test values during the previous 30 days; exclusion criteria DOI: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2293 RESEARCH included active hepatitis B. Study visits occurred at base- fmoL/punch of a paper disk containing DBS (e.g., https:// line and at weeks 4, 12, 24, 36, and 48 for the primary out- was defined as an come. Study participants were allowed to continue receiv- average of >4 tablets/week. This value is the unrounded ing the study drug past week 48 until the last participant level corresponding to 700 fmoL/punch level used in the completed his or her week 48 visit. IPREX OLE study, which reported 0 of 28 seroconver- At each visit, we collected data by using a confiden- sions when the TFV-DP level was >700 fmoL/punch (16). tial in-person interview and computer-assisted survey self- We determined intracellular TFV-DP concentrations at the report (e.g., Patient Health Questionnaire 9 [PHQ9]). We week 12 visit and the last on-drug visit on or before the found no significant differences in the primary adherence week 48 visit by using a validated method (15). outcome between the 2 study arms (72.0% in text messag- The primary DBS adherence outcome was a composite ing arm vs. 69.2% in standard of care; p = 0.58), in adequate outcome for being adherent as defined by a DBS TFV-DP adherence at week 12 (91.7% vs. 85.6%; p = 0.07) or week level >719 fmoL/punch (i.e., adequate adherence) at the 48 (83.4% vs. 81.6%; p = 0.77), or in baseline substance week 12 visit and, if continued past week 12, the last study use (p = 0.11) or Drug Abuse Screening Test (DAST10) visit through week 48 (e.g., week 24, 36, or 48). Missing or result (p = 0.30) (14). not completing the visit at week 12 was considered nonad- For this analysis, we included randomized CCTG 595 herence. If week 12 was the last study visit while receiving participants who had completed the baseline substance use drug, then the adherence of the participant was based only questionnaire (n = 394) to examine associations with sub- on that 1 value. The secondary DBS near-perfect adherence stance use over 48 weeks and used dried blood spot (DBS) composite outcome included the same composite outcome intracellular TFV-DP levels as a biologic measure of PrEP for DBS TFV-DP dose associated with taking 7 doses of adherence (15). We assessed substance use during the pre- TDF in the past week (>1,246 fmoL/punch) (15,16). In ad- vious 3 months at baseline and week 4, 12, 24, 36, and 48 dition to the composite outcomes, we also performed cross- visits by using a Substance Use Screening Questionnaire sectional analyses at weeks 12 and 48 on the basis of avail- (SCID). Each substance variable was categorized into no able samples. use, some use (1–4 times), and frequent use (>5 times) As a secondary objective, we assessed whether sub- on the basis of the frequency of use during the previous stance or alcohol use reported at baseline impacted study 3 months. We also analyzed use of combined stimulant completion and incident STIs during the study (i.e., mea- substances (i.e., poppers, methamphetamine, cocaine, ec- sure of sexual risk behavior). STI screening assessments stasy, amphetamine, and other stimulants); nonstimulant at baseline and every 3 months over 12 months included substances (i.e., heroin, other opioids [e.g., hydrocodone syphilis (serum rapid plasma reagin, and if a positive result bitartrate/acetaminophen and oxycontin], sedatives, anti- was obtained, a confirmatory treponemal test), nucleic acid anxiety drugs, hallucinogens, dissociative drugs, and inhal- amplification test of urine, and testing of pharyngeal and ants); and any substances (i.e., stimulant and nonstimulant rectal swab specimens for Chlamydia spp. and gonorrhea substances listed previously, not including alcohol and (Aptima; Hologic, Marlborough, MA, USA). Information marijuana use). about newly diagnosed STIs was communicated to partici- We assessed problematic use at baseline by using the pants, and referrals were made to their providers or a local DAST10 and the Alcohol Use Disorders Identification Test STI clinic for treatment. Incident STI was defined as having (AUDIT). DAST10 score was grouped into 3 categories: no positive results for gonorrhea or infection with Chlamydia or low problems, DAST10 score <3; moderate problems, spp. at any site or a positive rapid plasma reagin result for DAST10 score >3–<6; and substantial or severe problems, syphilis during the study visits after baseline. DAST10 score >6. AUDIT score was grouped into 3 cat- We compared baseline characteristics, DBS adherence egories: <8, >8–<16, and >16. Ongoing substance use was composites, and incident STI during the study and at study defined as >50% of completed study visits (study had 6 completion regarding substance/alcohol use categories by regular visits) with reported use in the SCID. Ongoing sub- using the Fisher exact test for categorical variables and stance use was defined in a hierarchical way. We first de- analysis of variance test. We used the Wilcoxon rank-sum fined frequent ongoing user as reporting frequent substance test to test continuous variables. To assess the association use on the SCID at >50% of visits; if this criterion was not between substance/alcohol use and outcomes (i.e., adher- reached, we looked further at whether the study participant ence, study completion, and incident), we used separate reported any substance use (including some and frequent) logistic regression models adjusted for study arm and other at >50% of visits; if yes, we defined them as some ongoing baseline factors that were associated with outcome. We as- user; otherwise, they were counted as not an ongoing user. sessed model discrimination by using the goodness-of-fit We determined adherence by measuring intracellular Hosmer–Lemeshow statistic. In addition, we used Cox re- TFV-DP levels in DBS. A TFV-DP concentration >719 gression models to study the association between baseline 2294 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Substance Use and Adherence to HIV Prophylaxis Figure 1. Flow chart for selection of patients from randomized controlled trial for study of substance use and adherence to HIV preexposure prophylaxis among men who have sex with men and transgender women, February 2014–February 2016, California, USA. substance use and time to early study termination or time Univariate analyses showed no significant difference to first incident STI diagnosis and reported hazard ratios in the primary or secondary DBS adherence outcomes be- (HRs). We defined time to study termination as the last tween persons with or without ongoing substance/alcohol completed visit of a participant. Participants who did not use (all p values >0.2; Table 2). There was also no signifi- leave the study early were censored at week 48. Participants cant association between baseline substance/alcohol use or who did not reach the event were censored at their last visit between baseline DAST10 and AUDIT scores and adher- before or at week 48. A p value <0.05 was considered sta- ence outcomes (all p values >0.5; Table 2). Similar results tistically significant. No adjustment was made for multiple were confirmed in multivariable logistic regression models comparisons. Statistical analyses were performed by using adjusted for study arm, race, and baseline PHQ9 scores (all R software version 3.3.2 ( p values >0.1). We created boxplots of DBS TFV-DP levels at weeks Results 12 and 48 for those with no, some, and frequent ongo- A total of 394 persons participated in the study and com- ing substance use (alcohol and marijuana excluded), as pleted their baseline substance use questionnaire (Figure 1). well as boxplots of DBS TFV-DP levels at week 48 by Of these participants, any substance use was reported by 288 AUDIT and DAST10 score categories (Figure 2). We (73%) and any alcohol use by 327 (83%). Overall, substance also developed cross-sectional associations of previous use remained relatively stable over the course of the study 3 months substance use with adequate and perfect ad- (e.g., 39% reported frequent substance use at baseline and herence at weeks 12 and 48 (online Technical Appendix 42% at week 48). Some ongoing substance use was reported Table 1, by 37%, and frequent ongoing substance was reported by 0400-Techapp1.pdf). Although at week 48 persons with 38%. Participants with ongoing substance use had higher the highest category of AUDIT scores were significantly levels of depressive symptoms (PHQ9 scores) than those less likely to have adequate adherence (p = 0.03), persons without ongoing substance use. We obtained demographic who had substantial or severe substance use problems data and PHQ9 and DAST10 scores of subgroups with no, according to DAST10 were significantly more likely to some, and frequent ongoing substance use (Table 1). reach near-perfect adherence (p = 0.04). However, when Overall, 89% of participants at week 12 and 83% of we compared DBS TFV-DP as a continuous variable be- participants at week 48 had adequate DBS TFV-DP levels tween the AUDIT and DAST10 groups, differences were (i.e., >719 fmoL/punch); 48% of participants at week 12 not significant (p = 0.847 for AUDIT and p = 0.099 for and 44% of participants at week 48 had estimated near- DAST10; Figure 2). perfect DBS TFV-DP levels (i.e., >1,246 fmoL/punch). Overall, 322/394 (82%) participants completed the A total of 279/394 (71%) study participants reached the study (Table 3). In the Cox regression model adjusting primary DBS adherence composite (i.e., adequate adher- for study arm (Table 4), we found that frequent base- ence), 115/394 (29%) reached the secondary DBS adher- line substance use was significantly associated with ence composite (i.e., near-perfect adherence), and 322/394 study completion (HR for early study termination 0.541; (82%) completed the study (i.e., the week 48 visit) ( 14). p = 0.036) compared with persons who had no substance Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2295 RESEARCH use; 86% of persons who had frequent substance use com- methamphetamine use), but frequent baseline substance pleted the study compared with 81% who had some sub- use tended to be associated with a lower tendency to leave stance use and 76% who had no substance use). We also cal- the study early (5.2%; all p>0.05). culated the same model after replacing substance use with Overall, 152 (39%) of 394 participants were given alcohol use and methamphetamine use. Although baseline a diagnosis of an incident STI during the study. By us- alcohol use was not a strong predictor of study completion, ing univariate analysis (Table 3), we found that incident we found that some methamphetamine use at baseline was STIs occurred more frequently in participants with some associated with a significantly lower likelihood of study and frequent stimulant use at baseline (incident STIs oc- completion compared with no methamphetamine use, but curred in 48% of both groups vs. 21% in persons with frequent methamphetamine use at baseline was not associ- no stimulant use at baseline; p<0.001). This difference ated with study completion (HR for early study termina- was driven by use of poppers (52% incident STIs in per - tion for some methamphetamine use 1.885; p = 0.046; 70% sons with frequent popper use and 50% in persons with study completion for some methamphetamine use vs. 83% some popper use vs. 26% in persons with no popper use; study completion for no or frequent methamphetamine p<0.001). We also found significantly higher rates of in - use). The logistic regression models yielded similar find- cident STIs in those with some methamphetamine use ings (online Technical Appendix Table 2). (58% STI incidence vs. 36% in persons with no metham - In an explorative analysis, we focused on 39 (9.9%) phetamine use and 39% in persons with frequent meth - persons who left the study early (before the week 24 vis- amphetamine use; p = 0.037). No difference was found it). We found that persons who had some baseline meth- for alcohol use. amphetamine use had a greater tendency to leave the By using Cox regression models adjusting for study early (17.5% left the study early vs. 4.3% who had study arm, age, and baseline STI status, we found that frequent methamphetamine use and 9.3% who had no stimulant use was strongly associated with incident STI Table 1. Baseline demographic characteristics by substance use status used for assessing substance use and adherence to HIV preexposure prophylaxis among MSM and transgender women, California, USA* Substance use Characteristic Overall, n = 394 None, n = 102 Some, n = 144 Frequent, n = 148 p value Sex 0.191 M 391 (99) 100 (98) 143 (99) 148 (100) F 3 (1) 2 (2) 1 (1) 0 Median age, y (IQR) 33 (28–41) 33 (29–40) 33 (28–41) 33.5 (28–42) 0.885 Race, n = 386† 0.238 Asian 12 (3) 2 (2) 5 (4) 5 (3) Black 51 (13) 20 (20) 19 (13) 12 (8) White 292 (76) 72 (72) 104 (73) 116 (81) Multiple 24 (6) 6 (6) 10 (7) 8 (6) Other 7 (2) 0 4 (3) 3 (2) Hispanic ethnicity, n = 391† 119 (30) 30 (29) 54 (38) 35 (24) 0.048 English primary language 3,786 (95) 98 (96) 133 (92) 145 (98) 0.066 Education 0.296 High school or less 35 (9) 12 (12) 15 (10) 8 (5) Some college 146 (37) 36 (35) 58 (40) 52 (35) Bachelor’s degree 132 (33) 31 (30) 42 (29) 59 (40) Postgraduate or advanced degree 81 (21) 23 (23) 29 (20) 29 (20) Household income/mo 0.434 <$2,000 84 (21) 27 (26) 31 (22) 26 (18) >$2,000 248 (63) 63 (62) 89 (62) 96 (65) Not known 62 (16) 12 (12) 24 (17) 26 (18) Randomization arm 0.019 Standard of care 196 (50) 60 (59) 75 (52) 61 (41) Text messaging 198 (50) 42 (41) 69 (48) 87 (59) Study site 0.660 Harbor–UCLA 48 (12) 11 (11) 15 (10) 22 (15) Long Beach 46 (12) 15 (15) 17 (12) 14 (9) UCSD 173 (44) 48 (47) 62 (43) 63 (43) USC 127 (32) 28 (27) 50 (35) 49 (33) Median PHQ9 for depression (IQR) 3 (1–7) 2 (0–5) 3.5 (1–7) 5 (2–8) <0.001 Median baseline DAST10 (IQR) 2 (0–3) 0 (0–1) 2 (1–3) 3 (2–4) <0.001 *Values are no. (%) unless otherwise noted. DAST, Drug Abuse Screening Test; IQR, interquartile range; MSM, men who have sex with men; PHQ, Patient Health Questionnaire; UCLA, University of California Los Angeles; UCSD, University of California San Diego; USC, University of Southern California. †Characteristics were not available for all study participants. 2296 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Substance Use and Adherence to HIV Prophylaxis during the study (HR 2.7 for some use, 2.6 for frequent Discussion use; both p<0.001) (Table 4). We also obtained signifi - We investigated the association between substance/alco- cant results when stimulant use was replaced with pop- hol use and adherence to PrEP, as well as study comple- per use (HR 2.3 for some use, 2.5 for frequent use; both tion and incident STIs, in a high-risk cohort of mostly p<0.001) or any substance use (HR 2.1 for some use, MSM who participated in a randomized controlled PrEP p = 0.002; HR 2.0 for frequent use, p = 0.004). When adherence trial. Three main findings are evident. First, we replaced stimulant use with methamphetamine use, substance use was not associated with decreased adher- some methamphetamine use was a significant predictor ence to PrEP, as measured by TFV-DP in DBS. Second, of incident STI (HR 1.9, p = 0.005), but frequent use was baseline frequent substance use was associated with high- not a significant predictor. In contrast, alcohol use was er likelihood of study completion. Third, baseline stimu- not a strong predictor of incident STI. Logistic regres- lant use was strongly associated with higher rates of inci- sion models yielded similar findings (online Technical dent STIs during the study, suggesting greater sexual risk Appendix Table 2). behavior in users of stimulant substances. Taken together, Table 2. Associations of DAST10 and AUDIT results at baseline and ongoing substance/alcohol use with primary and secondary DBS adherence endpoints for MSM and transgender women, California, USA* Primary endpoint Secondary endpoint Characteristic No Yes p value No Yes p value Substance use baseline, n = 394 DAST10 problems 0.80 0.55 No/low 70 (28) 179 (72) 173 (69) 76 (31) Moderate 37 (31) 83 (69) 86 (72) 34 (28) Substantial/severe 8 (32) 17 (68) 20 (80) 5 (20) AUDIT score 0.09 0.84 <8 81 (29) 201 (71) 197 (70) 85 (30) 8–15 25 (26) 70 (74) 69 (73) 26 (27) >15 9 (53) 8 (47) 13 (76) 4 (24) Ongoing substance use, n = 394 Methamphetamine 0.82 0.32 No 97 (29) 238 (71) 240 (72) 95 (28) Some 10 (28) 26 (72) 26 (72) 10 (28) Frequent 8 (35) 15 (65) 13 (57) 10 (43) Heroin 0.79 >0.99 No 113 (29) 275 (71) 274 (71) 114 (29) Some 1 (25) 3 (75) 3 (75) 1 (25) Frequent 1 (50) 1 (50) 2 (100) 0 Poppers 0.54 0.66 No 62 (31) 139 (69) 146 (73) 55 (27) Some 32 (30) 75 (70) 75 (70) 32 (30) Frequent 21 (24) 65 (76) 58 (67) 28 (33) Cocaine 0.48 0.33 No 97 (29) 241 (71) 236 (70) 102 (30) Some 13 (29) 32 (71) 33 (73) 12 (27) Frequent 5 (45) 6 (55) 10 (91) 1 (9) Stimulant substances† 0.37 0.75 No 42 (30) 100 (70) 98 (69) 44 (31) Some 43 (33) 88 (67) 96 (73) 35 (27) Frequent 30 (25) 91 (75) 85 (71) 36 (29) Nonstimulant substances‡ 0.96 0.32 No 67 (30) 158 (70) 166 (74) 59 (26) Some 31 (28) 79 (72) 74 (67) 36 (33) Frequent 17 (29) 42 (71) 39 (66) 20 (34) Any substance‡ 0.34 >0.99 No 31 (30) 71 (70) 72 (71) 30 (29) Some 47 (33) 97 (67) 102 (71) 42 (29) Frequent 37 (25) 111 (75) 105 (71) 43 (29) Alcohol 0.27 0.88 No 23 (38) 38 (62) 42 (69) 19 (31) Some 32 (29) 79 (71) 78 (70) 33 (30) Frequent 60 (27) 162 (73) 159 (72) 63 (28) *Values are no. (%). Primary endpoint value was TFV-DP >719 fmol/punch; secondary endpoint value was TFV-DP >1,246 fmoL/punch. For baseline use data for specific substances and substance classes not shown, all p values were >0.2. AUDIT, Alcohol Use Disorders Identification Test; DAST, Drug Abuse Screening Test; DBS, dried blood spot; MSM, men who have sex with men; TSF-DV, tenofovir diphosphate. †Includes poppers, methamphetamine, cocaine, ecstasy, amphetamine, and other stimulants. ‡Marijuana and alcohol excluded. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2297 RESEARCH Figure 2. Substance use and adherence to HIV preexposure prophylaxis among transgender women and men who have sex with men, California, USA. A, B) Boxplots showing dried blood spot TFV-DP levels at weeks 12 (A) and 48 (B) for persons with no, some, and frequent ongoing substance use. C, D) Boxplots showing dried blood spot TFV-DP levels at week 48 in persons with and without alcohol (C) and substance use (D) problems, according to assessments with AUDIT (C) and DAST (D) (cross-sectional analysis). In each case, dried blood spot TFV-DP levels were compared among the 3 groups by using the analysis of variance test. Circles indicate outliers; horizontal lines within boxes indicate medians, box bottoms and tops indicate 25th and 75th quartiles; and error bars indicate levels within 1.5 times the interquartile range of the lower quartile and upper quartiles. AUDIT, Alcohol Use Disorders Identification Test; DAST, Drug Abuse Screening Test; TFV-DP, tenofovir diphosphate. these findings indicate that substance use should not be face major individual barriers (e.g., HIV-related stigma, used as a reason to withhold PrEP because of concerns substance use) and structural barriers (e.g., economic, about adherence. healthcare) that might reduce linkage and adherence to We and others have shown that substance use in PrEP (13,22–26). general, and methamphetamine and other stimulant use Adherence is probably the major factor affecting PrEP in particular, is a likely cause of increased sexual risk effectiveness in those linked to PrEP, as outlined by a behavior among MSM and therefore a predictor for HIV recently published mathematical model, which showed that acquisition (3,17–21). Thus, HIV-uninfected MSM who increased adherence was the only factor resulting in reduc- use substances should be considered a target popula- tions of the number needed to treat with PrEP to prevent 1 tion for PrEP. However, substance-using MSM often HIV infection (27). Our study indicates that substance use 2298 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Substance Use and Adherence to HIV Prophylaxis was not associated with decreased adherence to PrEP. No- loss to care, the reasons for the increased study dropout of tably, in this study, persons who used methamphetamines these persons remains unknown. did not have worse adherence than persons who did not use Some or frequent stimulant use at baseline was strongly methamphetamines. A previous qualitative study indicated associated with contracting an incident STI during the study. that barriers to PrEP uptake and adherence differ by type This finding was driven mainly by popper use. We also found of substance used. In that study, stimulant drug users were higher rates of incident STIs for persons with some metham- more likely to be concerned that substance use would affect phetamine use when compared with persons who had fre- PrEP adherence, and were less concerned about HIV stig- quent or no methamphetamine use. One speculative explana- ma as a barrier to PrEP uptake compared with alcohol users tion for this association is that many MSM take psychoactive (13). However, in our study, we did not find an association drugs, in particular methamphetamine, and engage in sex at between stimulant use and PrEP adherence. the same time. Also known as chemsex or party and play, Baseline frequent substance use was associated with this practice is associated with condomless anal sex, multiple higher likelihood of study completion, and no associations sex partners, and the transmission of HIV and other STIs were found for alcohol use. Some methamphetamine use (4,28,29). This intermittent methamphetamine use might not was associated with lower likelihood of study completion occur frequently (i.e., not fullfilling frequent methamphet - when compared with frequent or no methamphetamine amine use in an SCID questionnaire) but might be associated use. Although some methamphetamine use might relate with high-risk sexual activities. Together with the finding to MSM who use methamphetamines occasionally, in that those with some methamphetamine use also have lower intermittent binges that are more likely to impart risk for study completion rates, this finding might warrant further Table 3. Associations of substance/alcohol use at baseline with study completion and incident STI among MSM and transgender women, California, USA* Study completion Incident STI Substance use baseline, n = 394 No Yes p value No Yes p value DAST10 problems 0.59 0.043 No/low 42 (17) 207 (83) 161 (65) 88 (35) Moderate 25 (21) 95 (79) 63 (53) 57 (48) Substantial/severe 5 (20) 20 (80) 18 (72) 7 (28) Methamphetamine 0.15 0.037 No 56 (17) 275 (83) 211 (64) 120 (36) Some 12 (30) 28 (70) 17 (43) 23 (57) Frequent 4 (17) 19 (83) 14 (61) 9 (39) Heroin 0.24 0.80 No 67 (18) 310 (82) 230 (61) 115 (39) Some 2 (22) 7 (78) 6 (67) 1 (33) Frequent 3 (38) 5 (63) 6 (75) 2 (25) Poppers 0.25 <0.001 No 41 (21) 150 (79) 142 (74) 49 (26) Some 19 (17) 94 (83) 57 (50) 56 (50) Frequent 12 (13) 78 (87) 43 (48) 47 (52) Cocaine 0.31 0.18 No 58 (17) 276 (83) 211 (63) 123 (37) Some 9 (20) 35 (80) 24 (55) 20 (45) Frequent 5 (31) 11 (69) 7 (44) 9 (56) Stimulant substances† 0.40 <0.001 No 30 (22) 109 (78) 110 (79) 29 (21) Some 23 (18) 106 (82) 67 (52) 62 (48) Frequent 19 (15) 107 (85) 65 (52) 61 (48) Nonstimulant substances‡ 0.95 0.33 No 42 (19) 179 (81) 140 (63) 81 (37) Some 19 (18) 89 (82) 60 (56) 48 (44) Frequent 11 (17) 54 (83) 42 (65) 23 (35) Any substance 0.18 <0.001 No 25 (24) 81 (76) 81 (76) 25 (24) Some 25 (19) 110 (81) 76 (56) 59 (44) Frequent 22 (14) 131 (86) 85 (56) 68 (44) Alcohol 0.13 0.28 No 16 (24) 51 (76) 47 (70) 20 (30) Some 21 (22) 74 (78) 57 (60) 38 (40) Frequent 35 (15) 197 (85) 138 (59) 94 (41) *Values are no. (%). DAST, Drug Abuse Screening Test; DBS, dried blood spot; MSM, men who have sex with men; STI, sexually transmitted infection. †Includes poppers, methamphetamine, cocaine, ecstasy, amphetamine, and other stimulants. ‡Marijuana and alcohol excluded. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2299 RESEARCH Table 4. Cox regression models used for assessing substance use and early study termination and incident STIs among MSM and transgender women, California, USA* Model HR (95% CI) p value Model 1 Intervention arm (receiving individualized texting for adherence to daily TDF/FTC) 1.377 (0.862–2.200) 0.180 Baseline some substance use (any) 0.743 (0.426–1.293) 0.293 Baseline frequent substance use (any) 0.541 (0.304–0.961) 0.036 Model 2 Intervention arm 0.924 (0.671–1.272) 0.626 Age 0.973 (0.955–0.992) 0.005 Baseline some stimulant use 2.690 (1.727–4.190) <0.001 Baseline frequent stimulant use 2.604 (1.665–4.072) <0.001 Positive STI test result at baseline 1.450 (1.031–2.039) 0.033 *Model 1 assessed the effect of baseline substance use on early study termination. Model 2 assessed the association of stimulant use and incident STIs during the study. FTC, emtricitibine; HR, hazard ratio; MSM, men who have sex with men; STI, sexually transmitted infection; TDF, tenofovir disoproxil fumarate. investigations into tailored HIV prevention counseling, as hypothesis. In our study, drug interaction seems unlikely, well as retention counseling, for this group of persons. given the consistency of findings across different drugs As a secondary finding, we found that substantial or of abuse, including alcohol, that have different pharma - severe problems with alcohol use, according to the AU- cologic profiles. DIT questionnaire at week 48, were associated with lower In conclusion, for MSM who participated in a random- likelihood of adequate adherence in cross-sectional analy- ized controlled trial, we found that baseline substance us- sis, although we found no strong association when DBS ers had increased STI rates, indicating higher risk behavior, TFV-DP levels were used as a continuous outcome. Also, but PrEP adherence was not decreased by substance use. we found no strong associations between baseline AUDIT Our findings suggest that substance-using persons are ap- scores and adherence composites. propriately diligent with PrEP adherence and therefore are Limitations of our study include that DBS TFV-DP excellent candidates for PrEP. levels were only measured at 2 time points, and that the composite adherence outcome logistic regression models Acknowledgments did not account for missing follow-up data and time effects. We thank the volunteers for participating in this study; and In addition, we assessed frequency of substance use with study nurses Edward Seefried, Connie Funk, as well as other validated SCID questionnaires that use categories (with study staff, including Marvin Hanashiro, Kelly Walsh, Fang the highest category being >5 times) instead of assessing Wang, Daisy Villafuerte, Luis Mendez, Mario Guerrero, frequency as a continuous outcome. This limitation is ap- Romero Correa, Sadia Shaik, and Ruben Lopez, for their efforts plicable particularly to the assessment of alcohol use, in during this study. which the frequent use category (i.e., >5 times within 3 This study was supported by the California HIV Research months) might not seem appropriate. However, although Program (grants CHRP-MC08-SD-700 and EI-11-SD-005) and our study did not look specifically into the effect of more the National Institutes of Health (grants AI036214, MH113477, frequent substance and alcohol use (e.g., >10 or >20 times AI064086, MH081482, MH062512, DA026306, AI106039, in the previous 3 months), the study included AUDIT and and MH100974). DAST scores that have been accepted as measures of prob- lematic alcohol and substance use. M.H. was principal investigator for this study, designed the Another limitation was that IDU was not assessed analysis, assisted with analyzing data, and drafted the separately. However, when we analyzed heroin use as manuscript; D.M. and S.M. were the principal investigators a proxy for IDU, we found no negative associations be- of the parent randomized controlled trial and assisted with tween heroin use and adherence, although these analyses designing the analysis and drafting the manuscript; S.J. and X.S. were limited by small sample size. A recent dynamic com- analyzed data and assisted with designing the analysis; D.C., partmental cost model suggested oral PrEP for PWIDs as K.C., J.S.B., E.S.D., J.M., and M.P.D. were investigators a potentially cost-effective strategy to control HIV in re - at the participating centers of the randomized controlled trial, gions where IDU is a major driver of the substance use and helped with designing this analysis and drafting the epidemic (30). Future studies are needed to evaluate PrEP manuscript; and P.L.A. analyzed dried blood spots and assisted adherence in PWIDs. Finally, there is a chance that drug in interpretation of data and drafting of the manuscript. All interactions could increase TFV-DP concentrations in authors revised the manuscript critically for intellectual substance abusers, resulting in misclassification of sub - content, approved the final manuscript, and agreed to be stance users as adherent, but no evidence supports this accountable for all aspects of the study. 2300 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Substance Use and Adherence to HIV Prophylaxis pre-exposure prophylaxis implementation. AIDS Patient Care About the Author STDS. 2014;28:622–7. Dr. Hoenigl is a research professor at the University of 13. Oldenburg CE, Mitty JA, Biello KB, Closson EF, Safren SA, California San Diego, San Diego, CA. His primary research Mayer KH, et al. Differences in attitudes about HIV pre-exposure interests are HIV prevention and primary HIV infection. prophylaxis use among stimulant versus alcohol using men who have sex with men. AIDS Behav. 2016;20:1451–60. References 14. Moore DJ, Jain S, Dube MP, Daar E, Sun X, Yung J, et al. 1. Freeman P, Walker BC, Harris DR, Garofalo R, Willard N, Ellen JM; Randomized controlled trial of daily text messages to support Adolescent Trials Network for HIV/AIDS Interventions 016b adherence to PrEP in at-risk for HIV individuals: the TAPIR Study. Team. Methamphetamine use and risk for HIV among young men Clin Infect Dis. 2018;66:1566–72. who have sex with men in 8 US cities. Arch Pediatr Adolesc Med. cix1055 2011;165:736–40. 15. Castillo-Mancilla JR, Zheng JH, Rower JE, Meditz A, Gardner EM, 2. Santos GM, Coffin PO, Das M, Matheson T , DeMicco E, Predhomme J, et al. Tenofovir, emtricitabine, and tenofovir diphos- Raiford JL, et al. Dose-response associations between number phate in dried blood spots for determining recent and cumulative drug and frequency of substance use and high-risk sexual behaviors exposure. AIDS Res Hum Retroviruses. 2013; 29:384–90. among HIV-negative substance-using men who have sex with 16. Grant RM, Anderson PL, McMahan V, Liu A, Amico KR, Mehrotra M, men (SUMSM) in San Francisco. J Acquir Immune Defic et al.; iPrEx Study Team. Uptake of pre-exposure prophylaxis, Syndr. 2013;63:540–4. sexual practices, and HIV incidence in men and transgender QAI.0b013e318293f10b women who have sex with men: a cohort study. Lancet Infect Dis. 3. Hoenigl M, Chaillon A, Moore DJ, Morris SR, Smith DM, 2014;14:820–9. Little SJ. Clear links between starting methamphetamine and 17. Buchbinder SP, Vittinghoff E, Heagerty PJ, Celum CL, increasing sexual risk behavior: a cohort study among men who Seage GR III, Judson FN, et al. Sexual risk, nitrite inhalant use, have sex with men. J Acquir Immune Defic Syndr. 2016;71:551–7. and lack of circumcision associated with HIV seroconversion in men who have sex with men in the United States. J Acquir 4. Hoenigl M, Chaillon A, Morris SR, Little SJ. HIV infection Immune Defic Syndr. 2005;39:82–9. rates and risk behavior among young men undergoing 01.qai.0000134740.41585.f4 community-based testing in San Diego. Sci Rep. 2016;6:25927. 18. Drumright LN, Little SJ, Strathdee SA, Slymen DJ, Araneta MR, Malcarne VL, et al. Unprotected anal intercourse and substance 5. Grant RM, Lama JR, Anderson PL, McMahan V, Liu AY, Vargas L, use among men who have sex with men with recent HIV infection. et al.; iPrEx Study Team. Preexposure chemoprophylaxis for J Acquir Immune Defic Syndr. 2006;43:344–50. HIV prevention in men who have sex with men. N Engl J Med. 10.1097/01.qai.0000230530.02212.86 2010;363:2587–99. 19. Koblin BA, Murrill C, Camacho M, Xu G, Liu KL, Raj-Singh S, 6. Molina JM, Capitant C, Spire B, Pialoux G, Cotte L, Charreau I, et al. Amphetamine use and sexual risk among men who have sex et al.; ANRS IPERGAY Study Group. On-demand preexposure with men: results from the National HIV Behavioral Surveillance prophylaxis in men at high risk for HIV-1 infection. N Engl J Med. study—New York City. Subst Use Misuse. 2007;42:1613–28. 2015;373:2237–46. 7. Baeten JM, Donnell D, Ndase P, Mugo NR, Campbell JD, Wangisi J, 20. Pines HA, Gorbach PM, Weiss RE, Shoptaw S, Landovitz RJ, et al.; Partners PrEP Study Team. Antiretroviral prophylaxis for Javanbakht M, et al. Sexual risk trajectories among MSM in the HIV prevention in heterosexual men and women. N Engl J Med. United States: implications for pre-exposure prophylaxis delivery . 2012;367:399–410. J Acquir Immune Defic Syndr. 2014;65:579–86. 8. Anderson PL, Glidden DV, Liu A, Buchbinder S, Lama JR, 10.1097/QAI.0000000000000101 Guanira JV, et al.; iPrEx Study Team. Emtricitabine–tenofovir 21. Hoenigl M, Anderson CM, Green N, Mehta SR, Smith DM, concentrations and pre-exposure prophylaxis efficacy in men Little SJ. Repeat HIV-testing is associated with an increase in who have sex with men. Sci Transl Med. 2012;4:151ra125. behavioral risk among men who have sex with men: a cohort study. BMC Med. 2015;13:218. 9. Choopanya K, Martin M, Suntharasamai P, Sangkum U, Mock PA, s12916-015-0458-5 Leethochawalit M, et al.; Bangkok Tenofovir Study Group. 22. Mimiaga MJ, Closson EF, Kothary V, Mitty JA. Sexual Antiretroviral prophylaxis for HIV infection in injecting drug users partnerships and considerations for HIV antiretroviral pre-exposure in Bangkok, Thailand (the Bangkok Tenofovir Study): a ran- prophylaxis utilization among high-risk substance using men domised, double-blind, placebo-controlled phase 3 trial. who have sex with men. Arch Sex Behav. 2014;43:99–106. Lancet. 2013;381:2083–90. S0140-6736(13)61127-7 23. Escudero DJ, Kerr T, Wood E, Nguyen P, Lurie MN, Sued O, et al. 10. Martin M, Vanichseni S, Suntharasamai P, Sangkum U, Mock PA, Acceptability of HIV pre-exposure prophylaxis (PREP) among Chaipung B, et al. Factors associated with the uptake of and people who inject drugs (PWID) in a Canadian setting. AIDS adherence to HIV pre-exposure prophylaxis in people who have Behav. 2015;19:752–7. injected drugs: an observational, open-label extension of the 24. Guise A, Albers ER, Strathdee SA. ‘PrEP is not ready for our Bangkok Tenofovir Study. Lancet HIV. 2016. community, and our community is not ready for PrEP’: 11. Liu A, Glidden DV, Anderson PL, Amico KR, McMahan V, pre-exposure prophylaxis for HIV for people who inject drugs and Mehrotra M, et al.; iPrEx Study Team. Patterns and correlates of limits to the HIV prevention response. Addiction. 2017;112:572–8. PrEP drug detection among MSM and transgender women in the Global iPrEx Study. J Acquir Immune Defic Syndr. 2014;67:528– 25. Edelman EJ, Moore BA, Calabrese SK, Berkenblit G, 37. Cunningham C, Patel V, et al. Primary care physicians’ willingness 12. Liu AY, Hessol NA, Vittinghoff E, Amico KR, Kroboth E, Fuchs J, to prescribe HIV pre-exposure prophylaxis for people who inject et al. Medication adherence among men who have sex with men drugs. AIDS Behav. 2017;21:1025–33. at risk for HIV infection in the United States: implications for s10461-016-1612-6 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2301 RESEARCH 26. Marshall BD, Milloy MJ. Improving the effectiveness and delivery 29. Hegazi A, Lee MJ, Whittaker W, Green S, Simms R, Cutts R, of pre-exposure prophylaxis (PrEP) to people who inject drugs. et al. Chemsex and the city: sexualised substance use in gay Addiction. 2016. bisexual and other men who have sex with men attending 27. Jenness SM, Goodreau SM, Rosenberg E, Beylerian EN, sexual health clinics. Int J STD AIDS. 2017;28:362–6. Hoover KW, Smith DK, et al. Impact of the Centers for Disease Control’s HIV preexposure prophylaxis guidelines for men 30. AIDSVu. Rollins School of Public Health, Emory University [cited who have sex with men in the United States. J Infect Dis. 2017 Dec 2]. 2016;214:1800–7. 28. Grov C, Rendina HJ, Breslow AS, Ventuneac A, Adelson S, Address for correspondence: Martin Hoenigl, Antiviral Research Center, Parsons JT. Characteristics of men who have sex with men (MSM) Division of Infectious Diseases, Department of Medicine, University of who attend sex parties: results from a national online sample in the California, 200 W Arbor Dr, #8208, San Diego, CA 92103, USA; email: USA. Sex Transm Infect. 2014;90:26–32. sextrans-2013-051094 EID Podcast: Ending the HIV/AIDS Pandemic Dr. Anthony Fauci, Director of the National Institute of Allergy and Infectious Diseases at the National Institutes of Health, discusses progress made toward ending HIV/AIDS and other pandemic infectious diseases. Visit our website to listen: index.aspx#/media/id/380134 2302 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Highly Pathogenic Clone of Shiga Toxin–Producing Escherichia coli O157:H7, England and Wales 1 1 Lisa Byrne, Timothy J. Dallman, Natalie Adams, Amy F.W. Mikhail, Noel McCarthy, Claire Jenkins We used whole-genome sequencing to investigate the evo- can be further divided into subtypes Stx1a–1d and Stx2a– lutionary context of an emerging highly pathogenic strain 2g; Stx2a is strongly associated with causing severe disease of Shiga toxin–producing Escherichia coli (STEC) O157:H7 (4,5). The STEC O157:H7 population has previously been in England and Wales. A timed phylogeny of sublineage delineated into 3 main lineages (I, I/II, and II) (6), and 7 IIb revealed that the emerging clone evolved from a STEC sublineages (Ia, Ib, Ic, IIa, IIb, IIc, and I/ll) (5). O157:H7 stx-negative ancestor ≈10 years ago after acquisi- In England, the most common STEC serotype is tion of a bacteriophage encoding Shiga toxin (stx) 2a, which O157:H7, which causes an average of 800 cases/year ( 1). in turn had evolved from a stx2c progenitor ≈20 years ago. All STEC O157:H7 isolated at local hospital laborato - Infection with the stx2a clone was a significant risk factor ries from fecal samples from hospitalized patients and all for bloody diarrhea (OR 4.61, 95% CI 2.24–9.48; p<0.001), cases in the community are submitted to the Gastrointes- compared with infection with other strains within sublineage tinal Bacteria Reference Unit (GBRU) at Public Health IIb. Clinical symptoms of cases infected with sublineage IIb stx2c and stx-negative clones were comparable, despite the England for confirmation of identification and typing. loss of stx2c. Our analysis highlighted the highly dynamic From 2000 through 2016, phage type (PT) 8 with the stx nature of STEC O157:H7 Stx-encoding bacteriophages and profile stx1a/stx2c and PT21/28 with the stx profile stx2a revealed the evolutionary history of a highly pathogenic or stx2a/stx2c were detected most frequently in England, clone emerging within sublineage IIb, a sublineage not pre- with PT21/28 the most frequently associated with severe viously associated with severe clinical symptoms. disease (2,7). Since 2015, all isolates submitted to GBRU have been genome sequenced. Whole-genome sequencing higa toxin–producing Escherichia coli (STEC) (WGS) demonstrates unparalleled sensitivity and accu- SO157:H7 cause a wide range of gastrointestinal symp- racy in identifying linked cases (8). Using WGS data dur- toms, including mild gastroenteritis, abdominal pain, vom- ing outbreak investigations has improved the robustness iting, and bloody diarrhea (1). A subset of patients, most of case ascertainment and provided forensic evidence commonly the very old and the very young, go on to devel- for linking human cases to the source of their infection op hemolytic uremic syndrome (HUS) (2). STEC O157:H7 (9,10). Phylogenetic inference can also reveal how strains are zoonotic, and transmission to humans is most common- are related over time and space more accurately than other ly associated with ruminants, especially cattle and sheep. molecular typing methods and may provide insight into Transmission occurs by consumption of contaminated food the evolutionary and epidemiologic context of emerging or water or by direct contact with animals or their environ- pathogenic clones (8,10,11). ment. The infectious dose is low (10–100 organisms), and In 2015, a total of 47 persons were affected by an person-to-person spread can occur in households, nurseries, outbreak in England of foodborne gastrointestinal illness and other institutional settings (1). The STEC pathotype is caused by STEC O157:H7 PT8 stx2a. The outbreak was as- defined by the presence of the genes encoding Shiga toxin sociated with the consumption of contaminated prepacked (Stx) type 1, type 2, or both, located on a bacteriophage salad leaves (11). The outbreak strain continued to cause incorporated into the bacterial genome (3). Stx1 and Stx2 sporadic infection and outbreaks of foodborne disease throughout 2016 and 2017 (11). The aim of our analysis Author affiliations: Public Health England, London, UK (L. Byrne, was to investigate the evolutionary history of this newly T.J. Dallman, N. Adams, A.F.W. Mikhail, C. Jenkins); National emergent strain of STEC O157:H7 PT8 stx2a and assess Institute for Health Research, London (T.J. Dallman, N. Adams, the risk to public health. N. McCarthy); University of Warwick, Coventry, UK (N. McCarthy) These first authors contributed equally to this article. DOI: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2303 RESEARCH reference genome Sakai (GenBank accession no. BA000007) Materials and Methods using Burrows Wheeler Aligner-Maximum Exact Matching (BWA-MEM) (13). We identified single- nucleotide poly - Bacterial Strains morphisms (SNPs) using Genome Analysis Toolkit version All isolates submitted to GBRU for confirmation and typ- 2 (14) in unified genotyper mode and extracted core genome ing from local hospital laboratories in England and Wales positions that had a high-quality SNP (>90% consensus, mini - during July 2015–December 2017 were sequenced for rou- mum depth ×10, GQ >30) in >1 isolate for further analysis. tine surveillance (National Center for Biotechnology In- We performed hierarchical single linkage clustering on the formation Short Read Archive bioproject PRJNA248042). pairwise SNP difference between all strains at various dis - We included an additional 74 clinical isolates of STEC tance thresholds (250, 100, 50, 25, 10, 5, 0). The result of the O157:H7 belonging to sublineage IIb, the lineage con- clustering is a SNP profile, or SNP address, that can be used taining STEC O157:H7 PT8 stx2a, that were submitted to to describe population structure based on clonal groups (15). GBRU between January 2010–June 2015 from previous We performed recombination analysis using Gub- studies (5,8) (online Technical Appendix Table, https:// bins (16) and reconstructed timed phylogenies using BEAST-MCMC version 2.4.7 (17). We computed alter- pdf). We selected these STEC O157:H7 isolates on the native clock models and population priors and assessed basis of stx subtype and phage type diversity to provide their suitability on the basis of Bayes factor tests. The context as a sample of the background population. We highest supported model was a relaxed log-normal clock defined STEC O157:H7 isolates from patients who were rate with a Bayesian skyline population model. We ran hospitalized as a result of their gastrointestinal symptoms all models with a chain length of 1 billion. We recon- or who reported bloody diarrhea as highly pathogenic or structed a maximum clade credibility tree using TreeAn- as having increased pathogenic potential compared with notator version 1.75 (17). isolates from patients who were asymptomatic or reporting We performed Stx subtyping as described by Ashton nonbloody diarrhea. et al. (18). The integration of Stx-encoding prophage into the host genome has been characterized into 6 target genes: Genome Sequence Analysis wrbA, encoding a NAD quinone oxidoreductase; yehV, a For WGS, we extracted DNA from cultures of STEC transcriptional regulator; sbcB, an exonuclease; yecE, a gene O157:H7 for sequencing on the HiSeq 2500 instrument (Il- of unknown function; the tRNA gene argW; and Z2577, lumina, San Diego, California, USA). We mapped qual- which encodes an oxidoreductase (5). We mapped short ity trimmed Illumina reads (12) to the STEC O157:H7 Figure 1. Cases of Shiga toxin–producing Escherichia coli O157:H7 belonging to sublineage IIb, 250 single-nucleotide polymorphism single-linkage cluster 18.%, by stx subtype profile, submitted to the Gastrointestinal Bacterial Reference Unit at Public Health England from England and Wales during June 1, 2010–December 31, 2017. 2304 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Highly Pathogenic Clone of STEC O157 reads from the STEC O157:H7 genomes to intact refer - Evolutionary Timescale and Stx Prophage Insertion in ence sequences of these genes, and aligned them with BWA STEC O157:H7 MEM (13). We defined occupied Stx bacteriophage inser - We reconstructed a timed phylogeny of sublineage IIb tion (SBI) sites as those strains that had disrupted alignments (Figure 2). We calculated the mutation rate of STEC (5). We used Tablet to visualize read pileups (19). O157:H7 within sublineage IIb to be ≈2 mutations per genome per year (95% highest posterior density [HPD] Data Analyses 1.7–2.4). This rate is less than the 2.6 mutations per ge- The National Enhanced Surveillance System for STEC nome per year previously calculated across the complete (NESSS) in England was implemented on January 1, 2009, STEC O157 population (5). Our analysis revealed that the and has been described in detail elsewhere (1). For this study, emerging stx2a clone evolved from a stx-negative recent we extracted data from NESSS for the cases identified as be - ancestor with the acquisition of stx2a ≈10 years ago (95% ing infected with strains that had been sequenced and belong- HPD 9.0 years 12.7 years). Previously, this stx-negative ing to the sublineage IIb cluster of interest (containing the clone had evolved from a stx2c progenitor ≈20 years ago STEC O157:H7 PT8 stx2a, outbreak strain). We excluded (95% HPD 17.6 years–24.6 years) after the loss of stx2c. asymptomatic carriers detected through screening high-risk contacts of symptomatic patients as well as patients who did not return the enhanced surveillance questionnaire (ESQ) to NESSS. Data analyzed included age, gender, and whether the patient reported symptoms of nonbloody diarrhea, bloody diarrhea, and vomiting along with whether cases were hos- pitalized, developed typical HUS, or died. Cases were cat- egorized into children (<16 years of age) or adults, based on a priori knowledge that children are most at risk for both STEC infection and progression to HUS (1). Where clinical symptoms were blank on the ESQ, we coded them as nega - tive responses for these symptoms. We divided cases into 3 groups based on stx subtype: stx2a, stx2c, and stx-negative. We first described patients’ symptoms by stx subtype as well as by age group and sex and also examined the dis- tribution of stx subtype by age and gender. We used Fisher exact tests to compare proportions among different groups. We assessed reporting of bloody diarrhea or hospitalization as a marker of disease severity by stx subtype. We used logistic regression to calculate odds ratios (ORs) to assess bloody diarrhea by stx subtype while adjusting for age (child/adult) and sex. We performed all analyses in Stata 13.0 (StataCorp LLC, College Station, TX, US). Results Sublineage IIb The STEC O157:H7 stx2a clone analyzed in this study was located within sublineage IIb, and belonged to a 250 SNP single-linkage cluster, designated 18%. This cluster com- prised 251 clinical isolates: 138 of STEC O157:H7 stx2a, 77 of stx-negative E. coli O157:H7, and 36 of STEC O157:H7 stx2c (Figure 1; online Technical Appendix Table). Since July 2015, when Public Health England implemented the use of WGS for STEC, the number of cases identified with- Figure 2. Timed phylogeny of Shiga toxin–producing in sublineage IIb has remained stable (≈60/y). However, Escherichia coli O157:H7 sublineage IIb isolates illustrating the number of cases of the stx-negative E. coli O157:H7 the sequential loss of stx2c and subsequent gain of stx2a. Red indicates stx2a; green, stx negative; blue, stx2c. Scale bar clone has declined, whereas the stx2a and stx2c clones are indicates years in the past. increasing (Figure 1). Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2305 RESEARCH Table 1. Clinical features of cases of Shiga toxin–producing Escherichia coli O157:H7 belonging to the IIb 250 SNP single-linkage cluster 18.% by stx subtype clone, England and Wales, July 2015–December 2017* Diarrhea Bloody diarrhea† Vomiting Hospitalization‡ No. No. % Patients No. % Patients No. % Patients No. % Patients stx subtype patients patients (95% CI) patients (95% CI) patients (95% CI) patients (95% CI) stx2c 36 31 86.1 7 22.6 10 32.3 3 8.3 (74.2–98.0) (5.9–33.0) (12.4–43.1) 1.1 to 17.8) stx-negative 77 72 93.5 12 16.7 23 31.9 9 11.7 (87.8–99.1) (7.3–23.9) (19.4–40.3) (4.3–19.0) stx2a 138 127 92.0 69 54.3 38 29.9 35 25.4 (87.4–96.6) (41.5–58.4) (20.0–35.1) (18.0–32.7) Total 251 230 91.6 88 38.3 71 30.9 47 18.7 (88.1–95.1) (29.1–41.0) (22.7–33.9) (13.9–23.5) *SNP, single-nucleotide polymorphism. †Statistically significant difference in reporting by stx subtype using Fisher exact test (p<0.001) ‡Statistically significant difference in reporting by stx subtype using Fisher exact test (p = 0.011) Historically, the majority of strains in sublineage IIb har- (9.8% [16/163], 95% CI 5.2–14.1; p<0.001). Half (50.0%) bored a Stx2c-encoding prophage at sbcB, with the yehV SBI of patients infected with stx2a isolates reported bloody di- site occupied by a truncated non–Stx-encoding prophage (5). arrhea (69/138, 95% CI 41.5–58.4), compared with 15.6% Analysis of the short read data indicated that in the stx-nega- of patients infected with stx-negative isolates (12/77, 95% tive sublineage IIb clone, yehV was disrupted but sbcB was in- CI 7.3–23.9) and 19.4% of those infected with stx2c isolates tact, indicating the loss of the Stx2c-encoding prophage from (7/36, 95% CI 5.9–33.0; p<0.001). No patients were known the SBI site. The more recently emerged sublineage IIb stx2a to experience HUS, and none died. clone had disrupted SBI sites at sbcB and yehV only, indicat- Among the 251 clinical cases, 141 (56.2%, 95% CI ing that a Stx2a-encoding phage had been inserted into sbcB, 50.0%–62.3%) were adults and 136 (54.2%, 95% CI the site left vacant in the stx-negative clone after the loss 48.0%–60.4%) were female. Adult patients were infected of stx2c. with stx2a strains (61.0% [86/141], 95% CI 52.8%–69.1%) more often than children (47.3% [52/110], 95% CI 37.8%– Disease Severity of Clinical Cases within 56.7%; p = 0.030). Conversely, children were more often in- the Sublineage IIb Cluster by stx Subtype fected with stx-negative strains than adults: 41.8% (46/110) Overall, 91.6% patients (230/251, 95% CI 88.1–95.1) had of children (95% CI 32.4%–51.2%) versus 22.0% (31/141) symptoms of diarrhea, and similar percentages were re- of adults (95% CI 15.1%–28.9%; p = 0.001). There was ported regardless of the stx subtype profile of the STEC also variation in stx subtype by sex; proportionately more O157:H7 causing the infection (Table 1). Rates of other female patients were infected with stx2a strains (61.0% symptoms varied; 28.3% of patients (71/251, 95% CI 22.7– [83/136], 95% CI 52.7%–69.3%) than were male patients 33.9) reported vomiting, 35.1% (88/251, 95% CI: 29.1–41.0) (47.8% [55/115], 95% CI 38.6%–57.1%; p = 0.036). Adult experienced bloody diarrhea, and 18.7% (47/251, 95% CI: patients reported bloody diarrhea (46.8% [66/141], 95% CI 13.9–23.5) were hospitalized. Hospitalization occurred more 38.5%–55.1%) more often than children (20.0% [22/110], often for patients reporting bloody diarrhea (35.2% [31/88, 95% CI 12.2%–27.6%; p<0.001), as did female patients 95% CI 25.0–45.4]) than those without bloody diarrhea (40.4% [55/136], 95% CI 32.1%–48.8%) compared with Table 2. Univariable and multivariable logistic regression analysis for reported bloody diarrhea in cases of Shiga toxin–producing Escherichia coli O157:H7 belonging to the IIb, 250 SNP single-linkage cluster 18.% by stx subtype clone, England and Wales, July 2015–December 2017* No. (%) No. (%) patients with patients without Univariable analysis Multivariable analysis† Exposure bloody diarrhea bloody diarrhea OR (95% CI) p value OR (95% CI) p value stx subtype stx negative 12 (16.7) 65 (83.3) Reference Reference stx2c 7 (22.6) 29 (77.4) 1.31 (0.47–3.7) 0.61 1.01 (0.35–2.94) 0.978 stx2a 69 (54.3) 69 (45.7) 5.42 (2.69–10.91) <0.001 4.61 (2.24–9.48) <0.001 Age Child, <16 y 22 (20.0) 88 (80.0) 0.28 (0.16–0.50) <0.001 0.31 (0.17–0.58) <0.001 Adult, >16 y 66 (46.8) 75 (53.2) Reference Reference Sex F 55 (40.4) 81 (59.6) Reference Reference M 33 (28.7) 82 (71.3) 0.59 (0.35–1.01) 0.053 0.81 (0.45–1.47) 0.494 *OR, odds ratio; SNP, single-nucleotide polymorphism. †Adjusted for each exposure variable (stx subtype, sex, age). 2306 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Highly Pathogenic Clone of STEC O157 male patients (28.7% [33/115], 95% CI 20.3%–37.1%), al- severity; specifically, bloody diarrhea linked to higher rates though the difference was not statistically significant (p = of hospitalization. Previous studies have reported evidence 0.05). The proportion of patients hospitalized did not differ of increased pathogenicity of STEC harboring stx2a (4,5). significantly by sex or age group (data not shown). However, these studies report on STEC from a wide range After adjusting for age (adult or child) and sex, the of different serotypes, exhibiting a wide variety of stx sub- odds ratio of experiencing bloody diarrhea was significant- types and are based on relatively small datasets. In this ly higher in those infected with the stx2a clone compared study, we present the analysis of a large dataset focusing with patients infected with the stx-negative clone (Table 2). on a specific clade within a single serotype characterized by The odds of bloody diarrhea were no different for cases in- a limited number of stx subtype combinations, specifically fected with the stx2c clone than for the stx-negative clone. stx2c, stx negative, and stx2a only. This analysis enabled Among the cases analyzed, being a child was protective for us to make direct comparisons between specific stx profiles symptoms of bloody diarrhea. while limiting the influence of other factors in the genome. Strains of Stx-negative E. coli O157:H7 are regarded Discussion as atypical enteropathogenic E. coli (EPEC), defined by the The data described here support previous studies that presence of the intimin gene (eae) and the absence of stx showed the acquisition and loss of the Stx-encoding phage and the E. coli adherence factor (EAF) plasmid (24). EPEC is highly dynamic in STEC O157:H7 ( 5,20). Most com- are a common cause of infantile diarrhea and travelers’ di- monly described is the acquisition of stx1a or stx2a by a arrhea and are known to cause mild diarrhea in adults (25). STEC O157:H7 stx2c progenitor, followed by the subse- In this study, the fact that clinical cases infected with the quent loss of stx2c in strains that acquired stx2a. The in- E. coli O157:H7 stx-negative clone reported a similar fre- volvement of a stx-negative intermediate in this process, as quency of symptoms, including bloody diarrhea and hospi- captured here, has not been previously described. The loss talization, as those infected with STEC O157:H7 stx2c de- of the Stx2c-encoding phage appears to have facilitated the spite the loss of stx was an unexpected finding that requires acquisition of the Stx2a-encoding phage because the latter further investigation. was inserted into the same SBI site, sbcB, left vacant by the A timed phylogenetic reconstruction of the evolution- Stx2c-encoding phage. ary history of a cluster of sublineage IIb charted the recent Using phylogenetic analysis of variation at the whole- emergence of a highly pathogenic clone of STEC O157:H7 genome level, we reconstructed the recent evolutionary stx2a. The symptom of bloody diarrhea, a marker of sever- history of this emerging pathogenic clone within STEC ity and predictor of HUS development (2), was strongly O157:H7 sublineage IIb. We observed the loss of stx2c associated with cases infected with isolates of STEC from the stx2c progenitor that caused a stx-negative clone O157:H7 harboring stx2a compared with those isolates ≈20 years ago, followed by the acquisition of stx2a ≈10 without stx or those with stx2c. Our analysis also illustrated years ago, and later expansion as shown in Figure 1. Previ- the highly dynamic nature of the Stx-encoding phages. In ously, we showed that the historic acquisition of a Stx2a- contrast to the observed excision events of stx2c-encoding encoding bacteriophage by a population of STEC O157:H7 phages in O157:H7, there is evidence to suggest that once PT2 stx2c, belonging to lineage I/II indigenous in the UK a Stx2a-encoding phage is integrated into a population it cattle population, was associated with the first outbreaks of tends to be maintained (5). As such, the emergence of yet childhood HUS in England in the early 1980s (5,7). Sub- another sublineage of STEC O157:H7 acquiring stx2a is of sequently, the increase in the incidence of STEC O157:H7 public health concern. Through this study, we demonstrate PT21/28 during the 1990s was linked to the acquisition of that STEC O157:H7 WGS surveillance data have a role stx2a by an indigenous population of STEC O157:H7 stx2c in monitoring and anticipating emerging threats to public belonging to sublineage Ic, resulting in the highly pathogen- health and in contributing to our understanding of the un- ic contemporary clone STEC PT21/28 stx2a/stx2c (1,2,5,7). derlying pathogenic mechanisms associated with severe This clone has been associated with several outbreaks in gastrointestinal illness. the United Kingdom associated with a high incidence of HUS (10,21–23). Here, we described an E.coli O157:H7 Acknowledgments clone from yet another UK domestic lineage (sublineage We thank Florence Aromona, Lukeki Kaindama, Nalini Purohit, IIb) that has recently acquired the Stx2a-encoding phage and Mike Harte for their contributions to the National Enhanced and is showing evidence of increasing pathogenic potential. Surveillance System for STEC in England, and all public health The analysis of disease severity of clinical cases by practitioners who report to the system. We also thank all the stx subtype of isolates of STEC O157:H7 within the same laboratory staff in the Gastrointestinal Bacteria Reference Unit, sublineage IIb cluster showed a significant association especially Michela Wright, Amy Gentle, Neil Perry, and between the presence of stx2a and markers of disease Dawn Hedges. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2307 RESEARCH 11. Mikhail AFW, Jenkins C, Dallman TJ, Inns T, Martín AIC, Fox The research was funded by the National Institute for Health A, et al. An outbreak of Shiga toxin–producing Escherichia coli Research (NIHR) Health Protection Research Unit (in Gastro- O157:H7 associated with contaminated salad leaves: intestinal Infections at University of Liverpool in partnership epidemiological, genomic, and food trace back investigations. with Public Health England, in collaboration with University of Epidemiol Infect. 2018;146:187–96. 12. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer East Anglia, University of Oxford, and the Quadram Institute. for Illumina sequence data. Bioinformatics. 2014;30:2114–20 The views expressed are those of the authors and not necessarily those of the National Health Service, the NIHR, the Department 10.1093/bioinformatics/btu170 of Health, or Public Health England. 13. Li H, Durbin R. Fast and accurate long-read alignment with Burrows-Wheeler transform. Bioinformatics. 2010;26:589–95. About the Author 14. McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, et al. The Genome Analysis Toolkit: a MapReduce Dr. Byrne is the lead epidemiologist for STEC at the National framework for analyzing next-generation DNA sequencing data. Infection Service, Public Health England, London, UK. Her Genome Res. 2010;20:1297–303. interests include public health surveillance of STEC and gr.107524.110 hemolytic uremic syndrome and the investigation of outbreaks of 15. Dallman TJ, Ashton P, Schafer U, Jironkin A, Painset A, Shaaban S, et al. SnapperDB: A database solution for routine sequencing foodborne gastrointestinal disease. analysis of bacterial isolates. Bioinformatics. 2018;34:3028–29. 16. Croucher NJ, Page AJ, Connor TR, Delaney AJ, Keane JA, References Bentley SD, et al. Rapid phylogenetic analysis of large samples of 1. Byrne L, Jenkins C, Launders N, Elson R, Adak GK. The recombinant bacterial whole genome sequences using Gubbins. epidemiology, microbiology and clinical impact of Shiga toxin– Nucleic Acids Res. 2015;43:e15. gku1196 producing Escherichia coli in England, 2009–2012. Epidemiol Infect. 2015;143:3475–87. 17. Drummond AJ, Suchard MA, Xie D, Rambaut A. Bayesian S0950268815000746 phylogenetics with BEAUti and the BEAST 1.7. Mol Biol Evol. 2. Launders N, Byrne L, Jenkins C, Harker K, Charlett A, Adak GK. 2012;29:1969–73. Disease severity of Shiga toxin–producing E. coli O157 and 18. Ashton PM, Perry N, Ellis R, Petrovska L, Wain J, Grant KA, et al. Insight into Shiga toxin genes encoded by Escherichia coli factors influencing the development of typical haemolytic uraemic syndrome: a retrospective cohort study, 2009–2012. BMJ Open. O157 from whole genome sequencing. PeerJ. 2015;3:e739. 2016;6:e009933. 3. Croxen MA, Law RJ, Scholz R, Keeney KM, Wlodarska M, 19. Milne I, Bayer M, Stephen G, Cardle L, Marshall D. Tablet: Finlay BB. Recent advances in understanding enteric pathogenic visualizing next-generation sequence assemblies and mappings. Methods Mol Biol. 2016;1374:253–68. Escherichia coli. Clin Microbiol Rev. 2013;26:822–80. 978-1-4939-3167-5_14 4. Persson S, Olsen KE, Ethelberg S, Scheutz F. Subtyping method 20. Kyle JL, Cummings CA, Parker CT, Quiñones B, Vatta P, for Escherichia coli shiga toxin (verocytotoxin) 2 variants Newton E, et al. Escherichia coli serotype O55:H7 diversity and correlations to clinical manifestations. J Clin Microbiol. supports parallel acquisition of bacteriophage at Shiga toxin phage insertion sites during evolution of the O157:H7 lineage. J Bacteriol. 2007;45:2020–4. 5. Dallman TJ, Ashton PM, Byrne L, Perry NT, Petrovska L, Ellis R, 2012;194:1885–96. et al. Applying phylogenomics to understand the emergence of 21. Ihekweazu C, Carroll K, Adak B, Smith G, Pritchard GC, Shiga toxin–producing Escherichia coli O157:H7 strains causing Gillespie IA, et al. Large outbreak of verocytotoxin-producing severe human disease in the UK. Microb Genom. 2015;1:e000029. Escherichia coli O157 infection in visitors to a petting farm in South East England, 2009. Epidemiol Infect. 2012;140:1400–13. 6. Zhang Y, Laing C, Steele M, Ziebell K, Johnson R, Benson AK, et al. Genome evolution in major Escherichia coli O157:H7 lineages. BMC 22. Rowell S, King C, Jenkins C, Dallman TJ, Decraene V, Lamden K, Genomics. 2007;8:121. et al. An outbreak of Shiga toxin–producing Escherichia coli 7. Adams NL, Byrne L, Smith GA, Elson R, Harris JP, Salmon R, serogroup O157 linked to a lamb-feeding event. Epidemiol Infect. 2016;144:2494–500. et al. Shiga toxin–producing Escherichia coli O157, England and Wales, 1983–2012. Emerg Infect Dis. 2016;22:590–7. 23. Wilson D, Dolan G, Aird H, Sorrell S, Dallman TJ, Jenkins C, et al. Farm-to-fork investigation of an outbreak of Shiga toxin– 8. Dallman TJ, Byrne L, Ashton PM, Cowley LA, Perry NT, Adak G, producing Escherichia coli O157. Microb Genom. 2018;4. et al. Whole-genome sequencing for national surveillance of 24. Trabulsi LR, Keller R, Gomes TA. Typical and atypical Shiga toxin–producing Escherichia coli O157. Clin Infect Dis. 2015;61:305–12. enteropathogenic Escherichia coli. [Review]. Emerg Infect Dis. 9. Jenkins C, Dallman TJ, Launders N, Willis C, Byrne L, 2002;8:508–13. Jorgensen F, et al. Public health investigation of two outbreaks of 25. Sakkejha H, Byrne L, Lawson AJ, Jenkins C. An update on the Shiga toxin–producing Escherichia coli O157 associated with microbiology and epidemiology of enteropathogenic Escherichia coli in England 2010–2012. J Med Microbiol. 2013;62:1531–4. consumption of watercress. Appl Environ Microbiol. 2015;81:3946–52. 10. Butcher H, Elson R, Chattaway MA, Featherstone CA, Willis C, Jorgensen F, et al. Whole genome sequencing improved case Address for correspondence: Claire Jenkins, Public Health England, ascertainment in an outbreak of Shiga toxin–producing Escherichia Gastrointestinal Bacteria Reference Unit, 61 Colindale Ave, London, coli O157 associated with raw drinking milk. Epidemiol Infect. 2016;144:2812–23. NW9 5HT, UK; email: 2308 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Prevalence of Avian Influenza A(H5) and A(H9) Viruses in Live Bird Markets, Bangladesh 1 1 1 Younjung Kim, Paritosh K. Biswas, Mohammad Giasuddin, Mahmudul Hasan, Rashed Mahmud, Yu-Mei Chang, Steve Essen, Mohammed A. Samad, Nicola S. Lewis, Ian H. Brown, Natalie Moyen, Md. Ahasanul Hoque, Nitish C. Debnath, Dirk U. Pfeiffer, Guillaume Fournié We conducted a cross-sectional study in live bird mar- Live bird markets (LBMs) form the backbone of poul- kets (LBMs) in Dhaka and Chittagong, Bangladesh, to try trade in many countries in Asia. Birds of different types estimate the prevalence of avian influenza A(H5) and and from different geographic areas are introduced daily A(H9) viruses in different types of poultry and environ - into LBMs and might be caged together, promoting local mental areas by using Bayesian hierarchical logistic re- transmission of multiple virus subtypes and generating op- gression models. We detected these viruses in nearly all portunities for reassortment (10–12). Surveys and routine LBMs. Prevalence of A(H5) virus was higher in waterfowl surveillance have described the abundance and diversity of than in chickens, whereas prevalence of A(H9) virus was avian influenza A viruses (AIVs) in LBMs in AIV-endemic higher in chickens than in waterfowl and, among chicken countries, including Bangladesh (1,4,11,13–21). However, types, in industrial broilers than in cross-breeds and in- only the proportion of positive samples is usually reported, digenous breeds. LBMs with >1 wholesaler were more without accounting for the hierarchical data structure, es- frequently contaminated by A(H5) virus than retail-only LBMs. Prevalence of A(H9) virus in poultry and level of pecially the clustering of sampled poultry per LBM. There- environmental contamination were also higher in LBMs fore, a robust assessment of AIV prevalence in LBMs is with >1 wholesaler. We found a high level of circulation lacking, although this knowledge is essential to understand of both avian influenza viruses in surveyed LBMs. Preva - AIV epidemiology and optimize surveillance design. lence was influenced by type of poultry, environmental Multiple poultry species and, for each poultry spe- site, and trading patterns. cies, multiple breeds are offered for sale in LBMs in Ban - gladesh. Desi, Sonali, and broiler are the most commonly traded chicken types. Desi, which means “local” in Ben- ow pathogenicity avian influenza A(H9N2) virus and gali, are indigenous chicken breeds raised in backyard Lhighly pathogenic avian influenza A(H5N1) virus are farms. Sonali is a cross-breed of the Rhode Island Red endemic in poultry populations in Bangladesh (1–4). In ad- cocks and Fayoumi hens. Broilers are industrial white- dition to their adverse effect on poultry production, these feathered breeds. In addition to varying levels of suscep- viruses have resulted in sporadic influenza cases in humans tibility, different poultry types might be raised in different (2,3). Because there is potential for generating novel reas- farming systems and traded through different value chains sortant variants between them or with other virus subtypes, (i.e., the range of activities that businesses perform to their persistent circulation in poultry poses a serious threat deliver products to customers), therefore being exposed to animal and human health globally (5–9). to different pathogens and pathogen loads ( 22,23). How- Author affiliations: The Royal Veterinary College, Hatfield, UK ever, the proportion of AIV-positive samples is gener- (Y. Kim, Y.-M. Chang, N. Moyen, D.U. Pfeiffer, G. Fournié); ally reported as an overall estimate or stratified only by City University of Hong Kong, Hong Kong, China (Y. Kim, poultry species. Likewise, the relative position of indi- D.U. Pfeiffer); Chittagong Veterinary and Animal Sciences vidual LBMs in a regional or national live poultry trad- University, Chittagong, Bangladesh (P.K. Biswas, R. Mahmud, ing network might also affect AIV prevalence; the sources M.A. Hoque); Bangladesh Livestock Research Institute, Dhaka, from which poultry are supplied to traders and the time Bangladesh (M. Giasuddin, M. Hasan, M.A. Samad); Animal they spend in LBMs influence the likelihood of virus in - Health and Plant Agency, Weybridge, UK (S. Essen, N.S. Lewis, troduction and amplification in LBMs and vary depend - I.H. Brown); Food and Agriculture Organization of the United ing on traders being wholesalers or retailers (12,24). Nations, Dhaka (N.C. Debnath) These authors contributed equally to this article. DOI: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2309 RESEARCH However, such information is generally poorly document- from the bottom 50% of eligible LBMs in terms of number ed, or even ignored. of poultry traded. To address these issues, we conducted a cross-section- At the second sampling stage, we randomly selected al survey in the 2 largest cities in Bangladesh, Dhaka and stalls and environmental sites in each LBM independently Chittagong, during February–March 2016. First, we esti- for each type of poultry and environmental site. We cre- mated prevalence of influenza A(H5) and A(H9) viruses ated a list of stalls selling each poultry type for each LBM. in marketed poultry and the LBM environment. We also We then selected stalls from these lists by using a random accounted for the clustering effect at LBM level by using number generator. Likewise, for each type of environmen- Bayesian hierarchical logistic regression models. Second, tal site, we selected sites from a list of sites identified in we assessed the effect of type of poultry and environmental each LBM by using a random number generator. We col- site, and the position of LBMs in the poultry value chain on lected 1 swab specimen from each environmental site. We AIV prevalence. pooled 5 swab specimens collected from the same LBM and site type. Materials and Methods At the third sampling stage, for each poultry type, we randomly selected 5 birds from each of the stalls selected Sample Collection for that type and collected cloacal and oropharyngeal swab An LBM was defined as an open space in which >2 poultry specimens from each of the selected birds. We pooled 5 stalls sell live poultry at least once a week, and only those swab specimens collected from the same stall and poultry selling >400 poultry/day were considered eligible for this type separately for cloacal and oropharyngeal swab speci- study. We aimed to sample 40 LBMs, and from each of mens. We transported samples collected in Chittagong these LBMs, 60 birds and 50 environmental sites (sample on the day of sampling to the Chittagong Veterinary and size calculations in online Technical Appendix 1, https:// Animal Sciences University (Chittagong) and samples col- lected in Dhaka on the day of sampling to the Bangladesh pdf). We used a stratified cluster sampling design. For Livestock Research Institute (Dhaka). Samples were stored poultry, LBMs, stalls within selected LBMs, and birds at −80°C until diagnostic laboratory processing. within selected stalls constituted the primary, secondary, and tertiary sampling units, respectively. For environmen- Sample Screening tal sites, LBMs constituted primary sampling units and We screened pools for AIVs by using a real-time reverse environmental sites within selected LBMs constituted sec- transcription PCR (RT-PCR) and specific primers and ondary sampling units. probes (27,28). We extracted virus RNA by using the Mag- We stratified LBMs by city for Dhaka and Chittagong MAX RNA Isolation Kit (QIAGEN, Hilden, Germany) and and, within each city, by poultry sales into large and small reverse transcribed and amplified virus RNA by using the LBMs, hypothesizing that the risk for AIV infection var- AgPath-ID One-Step RT-PCR (ThermoFisher Scientific, ies between geographic locations and the number of poul- Waltham, MA, USA). We then screened a pool with a cy- try traded. Also, simple random sampling with too small a cle threshold (C ) <40 for the AIV matrix gene for the H5 sample size of LBMs was not likely to capture diversity of and H9 genes. Results were considered positive for the H5 LBM types because the distribution of LBMs as a function subtype if C <38 and positive for the H9 subtype if C <40 t t of their size tended to be right-skewed; the largest LBMs (27,28). A pool was considered positive for AIV if its C were often wholesale markets (24,25). We hypothesized for the AIV matrix gene <38 or if it was positive for any that samples of different origins have different AIV preva- of the H5 and H9 subtypes. A given group of 5 birds was lences and thus stratified birds and environmental sites considered positive if any of its cloacal and oropharyngeal into 5 types of poultry and 10 types of environmental sites pools showed a positive result. commonly found to be contaminated with AIV (26) (online Technical Appendix 2 Tables 1, 2, Bayesian Hierarchical Logistic Regression Models EID/article/24/12/18-0879-Techapp2.pdf). We developed 2-level Bayesian hierarchical logistic regres- At the first sampling stage, we sampled 40 LBMs. sion models to estimate LBM-level, bird-level, and environ- The number of LBMs selected in Dhaka (n = 26) and Chit - mental swab specimen–level prevalence from pooled swab tagong (n = 14) was proportional to the number of LBMs samples, accounting for lower-level (swab specimens) and eligible in each city (n = 80 for Dhaka and n = 36 for Chit - higher-level (LBMs) risk factors. We developed separate tagong). In each city, we further stratified LBMs by size: models for poultry and environmental samples to avoid pa- 50% of the selected LBMs were large, trading the highest rameters related to different sampling units interfering with number of poultry (13 largest LBMs in Dhaka and 7 largest each other (29). We used LBM type (retail or mixed), city LBMs in Chittagong); 50% were small, randomly selected (Chittagong or Dhaka), and size (small or large) as LBM- 2310 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Influenza Viruses in Live Bird Markets, Bangladesh level risk factors. We defined an LBM with only retailers waterfowl, ducks (19.3%) and geese (16.7%) had a similar (i.e., trader selling poultry to end-users only) as retail and prevalence of A(H9) virus–positive pools, but the preva- an LBM with >1 wholesaler (i.e., trader selling poultry to lence of A(H5) virus–positive pools was higher in ducks other traders) as mixed. Each LBM-level risk factor (LBM (36.8%) than in geese (22.2%). For both H5 and H9 sub- type, city, and size) was assessed separately because we types, the prevalence of positive pools was higher for oro- could not include them simultaneously in any given model pharyngeal samples (8.6% for H5 and 31.9% for H9) than due to the small number of sampled LBMs. In models for for cloacal samples (3.6% and 9.9%) in all surveyed poul- poultry samples, we differentiated birds into 1) chicken and try types (online Technical Appendix 2 Table 1). waterfowl or 2) broiler, Desi, Sonali (i.e., chicken types), Approximately 25% of environmental pools were pos- and waterfowl. Waterfowl were not differentiated further itive for A(H9) virus, and the prevalence of positive pools because of the small number of pools collected from ducks was higher in slaughter areas (31.5%), especially knives and geese. In models for environmental samples, we differ - and boards used for slaughter and processing, than stall entiated environmental sites into stall and slaughter areas areas (20.2%). The prevalence of A(H5) virus–positive en- or classified as environmental area without differentiation. vironmental pools was lower (10.8%) and did not vary be- We ran models (online Technical Appendix 1) by using a tween slaughter and stall areas (online Technical Appendix Markov Chain Monte Carlo simulation in JAGS (30) and 2 Table 2). R.3.4.2 (31). Bayesian Model Results Results Convergence was achieved for all models; the Gelman and Rubin statistic was <1.001 and the effective sample size Descriptive Results for Pooled Swab Samples was >10,000 for all parameters. For each AIV subtype, the We collected 477 pairs of cloacal and oropharyngeal pooled best models reasonably predicted the number of positive samples from 2,384 birds, and 400 environmental pooled pools (online Technical Appendix 2 Figure 1). In the best samples from 2,000 environmental sites in 40 LBMs in H5 models (i.e., lowest deviance information criterion), Chittagong and Dhaka. Each pool was composed of 5 swab A(H5) virus prevalence differed according to poultry spe- specimens, except for 1 pair of cloacal and oropharyngeal cies (chicken, waterfowl), but not according to the type pools made from 4 swab specimens collected from geese. of environmental site. In contrast, in the best H9 models, We collected 12 pairs of cloacal and oropharyngeal pooled A(H9) virus prevalence differed according to type of poul - samples from all LBMs, except for 11 pairs from 3 LBMs. try (broiler, Desi, Sonali, waterfowl) and environmental site We sampled chickens in all LBMs (8–12 pairs/LBM), and (slaughter and stall area). For both subtypes, LBM size and waterfowl in 25 LBMs (0–4 pairs/LBM). Broilers account- city did not improve model fit when compared with LBM ed for most samples (32.1%), followed by Desi (26.6%) type. For ease of comparison between the 2 AIV subtypes, and Sonali (25.6%). Ducks accounted for 76% of 75 pool we report LBM-level, bird-level, and environmental swab pairs collected from waterfowl and geese accounted for specimen–level prevalences of A(H5) and A(H9) viruses 24%. We collected 10 environmental pools in each LBM on the basis of the best H9 models with LBM type (Tables (stall areas: 4–8 pools, slaughter areas, 2–6 pools). 1, 2). This reporting did not affect interpretation of results, Of 47.4% (416/877) pools considered positive for and we provide estimates obtained with more parsimonious AIV, 6.5% pools were negative for the AIV matrix gene models (online Technical Appendix 2 Tables 3–6). but positive for any of the H9 and H5 subtypes. The H9 LBM-level A(H5) virus prevalence was lower in re- subtype (63.2% positive pools) was detected more fre- tail LBMs than in mixed LBMs, and the posterior median quently than the H5 subtype (21.6%), and 12.3% of pools estimate was ≈100% for mixed LBMs. However, among were positive for both subtypes and 27.4% of pools were contaminated LBMs, levels of virus detection in birds and negative for both subtypes. Although 80.0% of the LBMs environmental areas did not vary between LBM types, but had >1 A(H5) virus–positive poultry or environmental A(H5) virus prevalence in waterfowl was ≈6 times higher pool, 97.5% had >1 A(H9) virus–positive poultry or en- than in chickens (Figure). The prevalence did not vary be- vironmental pool. We determined the prevalence of pools tween chicken breeds or environmental areas. that were positive for A(H5) and A(H9) viruses according In contrast to that for A(H5) virus, we found that the to sample and LBM type (Table 1). posterior median estimate of the LBM-level A(H9) virus Approximately 33.3% of pools collected from water- prevalence was ≈100% for retail and mixed LBM groups, fowl were positive for A(H5) virus, whereas only 5.5% of but the level of virus detection in birds and environmental those collected from chickens were positive. In contrast, areas was higher for mixed LBMs than for retail LBMs. the prevalence of A(H9) virus–positive pools was higher A(H9) virus prevalence was highest in broilers and lowest in chickens (36.3%) than in waterfowl (18.7%). Among in waterfowl. The prevalence in broilers was 3.8 times as Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2311 RESEARCH Table 1. Prevalence of avian influenza A(H5) and A(H9) viruses in pooled poultry and environmental samples, Chittagong and Dhaka, Bangladesh* A(H5) virus prevalence, % A(H9) virus prevalence, % Observed pool Estimated bird level Observed pool Estimated bird level Sample type No. pools level (95% HDI)† level (95% HDI)† Poultry Retail LBM Broiler 96 5.2 0.9 (0–9.4) 37.5 10.8 (2.3–22.5) Sonali 62 3.2 1.4 (0–13.2) 32.3 6.6 (1.2–14.5) Desi 61 3.4 1.3 (0–12.2) 29.5 6.8 (1.3–14.9) Waterfowl 20 50.0 8.1 (0–46.8) 25.0 2.8 (0.3–7.0) Mixed LBM Broiler 57 1.8 0.9 (0–4.0) 47.4 13.1 (1.2–30.1) Sonali 60 10.0 1.4 (0–5.7) 31.7 8.0 (0.4–19.8) Desi 66 9.1 1.3 (0–5.2) 39.4 8.3 (0.5–20.4) Waterfowl 55 27.3 7.6 (0–24.6) 16.4 3.4 (0.1–9.7) Environmental site Retail LBM Stall area 101 5.9 1.5 (0–10.4) 16.8 3.2 (0.1–9.1) Slaughter area 99 7.1 1.4 (0–10.1) 25.3 6.2 (0.2–16.6) Mixed LBM Stall area 102 15.7 3.1 (0–11.3) 23.5 5.2 (0.1–14.1) Slaughter area 98 14.3 3.0 (0–11.0) 37.8 9.9 (0.4–25.0) *Desi, which means “local” in Bengali, are indigenous chicken breeds raised in backyard farms. Sonali is a cross-breed of the Rhode Island Red cocks and Fayoumi hens. HDI, high-density interval; LBM, live bird market. †Bird and environmental swab specimen–level prevalence in contaminated live bird markets. Median values are reported. high as that in waterfowl and 1.6 times as high as that in in Chittagong (21), which found that 17.5% of LBMs had Desi and Sonali (Figure). The environmental swab speci- >1 environmental sample pool contaminated by A(H5) vi- men–level prevalence was ≈2 times as high for slaughter rus and 12.5% of LBMs had >1 environmental sample pool areas than for stall areas (Figure). contaminated by A(H9) virus. This difference might have been caused by different sampling schemes; in our study, Discussion we collected a larger number of pools per LBM. We detected A(H5) and A(H9) viruses in marketed poul- Bird-level prevalence was also higher than that re- try and environmental sites in nearly all LBMs sampled ported in other AIV-endemic countries, including Ban- in Chittagong and Dhaka. The prevalence of A(H5) virus gladesh (1,4,16,19). However, care must be taken when was higher in waterfowl than in chickens, whereas the comparing these results because studies used different prevalence of A(H9) virus was higher in chickens than study designs and sample screening protocols over differ - waterfowl and also varied among chicken types, being ent periods. Bird-level prevalence for contaminated LBMs more prevalent in broilers than in Desi and Sonali breeds. was much higher than for virologic surveys conducted in Slaughter areas were more frequently contaminated by backyard and commercial farms in Bangladesh (1,4,33,34). A(H9) virus than stall areas. Whereas mixed LBMs were This finding suggests that virus transmission was amplified more frequently contaminated by A(H5) virus than were along the value chain from farms to LBMs. Overcrowding retail LBMs, prevalence of A(H9) virus was higher in and continuous supply of susceptible birds of different spe- mixed LBMs than in retail LBMs for birds and environ- cies and breeds might have created conditions promoting mental areas. the silent transmission of AIVs within these markets (10). AIVs were ubiquitous in surveyed LBMs. The LBM- Our results suggest that birds in LBMs with a mixture level prevalence of A(H5) virus in Bangladesh was higher of wholesalers and retailers were at higher risk for infec- than in other AIV-endemic countries, including Egypt (32) tion than birds in LBMs with primarily retail poultry busi- and Vietnam (16). For both AIV subtypes, LBM-level nesses. Poultry value chains supplying different business prevalence was also higher than in another study conducted types might differ structurally, thereby affecting the risk for Table 2. Prevalence of avian influenza A(H5) and A(H9) viruses in LBMs, Chittagong and Dhaka, Bangladesh* Sample type No. LBMs H5 virus median prevalence, % (95% HDI) H9 virus median prevalence, % (95% HDI) Poultry Retail LBM 20 69.9 (40.2–100.0) 96.4 (85.5–100.0) Mixed LBM 20 92.0 (72.3–100.0) 96.0 (84.0–100.0) Environmental sites Retail LBM 20 76.5 (47.2–100.0) 94.9 (80.5–100.0) Mixed LBM 20 93.2 (75.5–100.0) 96.0 (84.0–100.0) *Prevalence estimates were made by using the best models. HDI, high-density interval; LBM, live bird market. 2312 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Influenza Viruses in Live Bird Markets, Bangladesh Figure. Bird and environmental swab specimen–level avian influenza A(H5) and A(H9) virus prevalence ratios, Bangladesh. Dotted lines indicate H5 subtypes, and solid lines indicate H9 subtypes. Diamonds indicate median values, and horizontal bars indicate 95% high-density interval of a given prevalence ratio. Asterisks (*) indicate reference groups for each comparison. Desi, which means “local” in Bengali, are indigenous chicken breeds raised in backyard farms. Sonali is a cross-breed of the Rhode Island Red cocks and Fayoumi hens. introduced birds being already infected. Wholesalers gen- However, these prevalence patterns might also be caused erally trade a larger number of birds from more diverse geo- by varying levels of genetic susceptibility to AIV infection graphic origins than do retailers (23) and therefore might (35,36). Further investigations are needed to disentangle have increased likelihood of virus introduction into mixed the possible influences of trade-related and genetic factors LBMs. Moreover, because wholesalers might sell birds to on AIV transmission in these chicken types. The higher retailers in the same LBM (23), virus amplification might level of contamination with A(H9) virus in slaughter areas be increased through the presence of wholesalers. than in stall areas suggests that, in the absence of appropri- The higher prevalence of A(H9) virus in broilers than ate biosecurity measures, slaughtering is likely to expose in Sonali and Desi might result from differences in the struc- humans to AIVs by fomite transmission (37). ture of their respective value chains (23). Depending on the Co-circulation of A(H5) and A(H9) viruses arouses chicken type, different value chain actors might be involved concerns over evolution of novel reassortant variants (5– and their trading practices might differ ( 23). The amount 8). Detection of both subtypes in some poultry pools sug- of time chickens spend with traders, the density at which gests that these subtypes co-circulated near each other or chickens are kept in flocks of traders, and the frequency of in the same host during the study period. Although A(H5) contact with chickens from other flocks might vary with viruses have considerable variability in their ability to in- chicken type. The greater number of broilers marketed in fect, cause disease, and be transmitted among waterfowl surveyed LBMs might mean that broilers are more likely (38), waterfowl are generally known to be less suscep- than Desi and Sonali to be sourced from large numbers of tible to highly pathogenic avian influenza A(H5N1) vi - flocks, which are then mixed in densely populated trucks ruses (39). Therefore, waterfowl could harbor this virus during transport to LBMs, promoting AIV transmission. but remain asymptomatic and serve as a potential host Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2313 RESEARCH for genesis of novel AIVs in the presence of other virus Fourth, we collected samples over a short period to subtypes. Also, the high level of A(H9) virus circulation reduce variability that could arise from seasonal varia- among chickens could provide an ideal environment for tions in AIV prevalence. We focused on winter months, virus diversification and selection in the LBM system. which are often reported to be periods of higher risk for The different prevalence patterns in chickens and water - AIV infection (44). Therefore, our estimates only repre- fowl observed suggest that these poultry species should sented AIV prevalence during that period and did not cap- be separated in LBMs and that active surveillance of nov- ture seasonal changes. el reassortant variants should be implemented. Contrary to previous cross-sectional studies, our ap- This study had some limitations. First, our models only proach enabled us to estimate AIV prevalence not only by accounted for clustering of sampled birds at the LBM level, poultry species but also by chicken type and account for the but not at stall level. It is plausible that clustering of sam- type of LBMs in which sampled poultry were marketed. ples at stall level has less influence on AIV infection prob- Despite most AIV surveys and surveillance activities being ability across the study population than clustering at the based on multistage sampling, single-level analytic meth- LBM level because stallholders in a given LBM in Bangla- ods are generally used to analyze their results, while ignor- desh are likely to be supplied by the same traders and trade ing within-market correlation in poultry infection status. between each another (23). However, potential risk factors Accounting for this effect by incorporating LBM-specific at stall level, such as ducks and hygiene level (21), might random effects in a hierarchical model, and enabling mutu - cause heterogeneous levels of AIV infection across stalls. al influence between bird-level, environmental swab speci- Second, our models did not account for the fact men–level, and LBM-level parameters, improved the reli- that sampling units in each stratum were selected with ability of prevalence estimates (29). When applied to other unequal probabilities. Although we selected different settings, this approach needs to be adapted on the basis of numbers of birds for each poultry type to account for an understanding of the variety of poultry value chains. In- variations in poultry populations, birds were still se- formation about LBM locations and about trading practices lected with different probabilities because their popula - and numbers and types of poultry sold within these LBMs tions varied between clusters and strata. This selection is rarely readily available and would need to be collected to might have resulted in larger SEs and thus less precise inform the study design. estimates compared with what could have been obtained In conclusion, LBMs surveyed in Bangladesh were with proportional sample sizes. Moreover, the overall highly contaminated by A(H5) and A(H9) viruses. The lev- prevalence might have been biased toward prevalence in el of virus detection was associated with the type of poultry samples selected with higher probabilities. and environmental area and the presence of wholesalers in Third, our models assumed perfect sensitivity and spec- LBMs. These findings need to be included in the design of ificity of real-time RT-PCR for pooled samples. The assays risk-based surveillance and control interventions aimed at used in this study are considered highly sensitive and specif- reducing AIV prevalence, human exposure, and the risk for ic (27,28), and previous studies did not report any differenc - emergence of novel virus reassortant variants. es in virus detection for pooled and individual samples (40– 42). Furthermore, virus detection in our study was based on Acknowledgments parallel interpretation of cloacal or oropharyngeal sample We thank Eric Brum for his support during study implementation test results (i.e., positive if >1 was positive). However, pools and the participants involved in the study. that were negative for the AIV matrix gene but positive for This study was supported by the BALZAC research program any of the H5 and H9 subtypes indicate that accounting for “Behavioural adaptations in live poultry trading and farming actual test sensitivity and specificity would enable more systems and zoonoses control in Bangladesh” (BB/L018993/1) robust prevalence estimation. Virus isolation might be at- and is 1 of 11 programs supported by the Zoonoses and tempted for RT-PCR–positive pools to assess the viability Emerging Livestock Systems, a joint research initiative between of virus material. However, this testing was not attempted the Biotechnology and Biological Sciences Research Council, in our study. Each pool consisted of swab specimens from the Defence Science and Technology Laboratory, the different birds or environmental sites. Thus, multiple AIV Department for International Development, the Economic and subtypes and virus species, including Newcastle disease vi- Social Sciences Research Council, the Medical Research ruses, could be present in the same pool and interfere with Council, and the Natural Environment Research Council. growth of each virus in chicken eggs (43). Should such stud- ies be replicated, the collection of individual swab speci- mens and their pooling at the laboratory is recommended to About the Author enable analysis of individual swab specimens that formed a Mr. Kim is a doctoral student at the College of Veterinary virus-positive pool. Medicine and Life Sciences, City University of Hong Kong, 2314 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Influenza Viruses in Live Bird Markets, Bangladesh Virus Genes. 2013;47:317–29. Hong Kong, China. His primary research interests are the s11262-013-0954-7 socioeconomic, cultural, and epidemiologic factors that shape 15. Phan MQ, Henry W, Bui CB, Do DH, Hoang NV, Thu NT, et al. zoonotic and animal infectious disease transmission, including Detection of HPAI H5N1 viruses in ducks sampled from live bird avian influenza. markets in Vietnam. Epidemiol Infect. 2013;141:601–11. 16. Nguyen DT, Bryant JE, Davis CT, Nguyen LV, Pham LT, Loth L, References et al. Prevalence and distribution of avian influenza A(H5N1) virus 1. Negovetich NJ, Feeroz MM, Jones-Engel L, Walker D, Alam SM, clade variants in live bird markets of Vietnam, 2011–2013. Hasan K, et al. Live bird markets of Bangladesh: H9N2 viruses and Avian Dis. 2014;58:599–608. the near absence of highly pathogenic H5N1 influenza. PLoS One. 10814-030814-Reg 2011;6:e19311. 17. Huang Y, Zhang H, Li X, Hu S, Cai L, Sun Q, et al. Detection 2. Chakraborty A. Outbreak of mild respiratory disease caused by and genetic characteristics of H9N2 avian influenza viruses H5N1 and H9N2 infections among young children in Dhaka, from live poultry markets in Hunan Province, China. PLoS One. Bangladesh, 2011. Health Science Bulletin. 2011;9:5–12. 2015;10:e0142584. 3. World Health Organization. Cumulative number of confirmed 18. Chen LJ, Lin XD, Guo WP, Tian JH, Wang W, Ying XH, et al. human cases for avian influenza A(H5N1) reported to WHO, Diversity and evolution of avian influenza viruses in live poultry 2003–2017; 2017 [cited 2017 Oct 9]. markets, free-range poultry and wild wetland birds in China. J Gen human_animal_interface/H5N1_cumulative_table_archives/en/ Virol. 2016;97:844–54. 4. Turner JC, Feeroz MM, Hasan MK, Akhtar S, Walker D, Seiler P, 19. Thuy DM, Peacock TP, Bich VTN, Fabrizio T, Hoang DN, et al. Insight into live bird markets of Bangladesh: an overview Tho ND, et al. Prevalence and diversity of H9N2 avian influenza of the dynamics of transmission of H5N1 and H9N2 avian in chickens of Northern Vietnam, 2014. Infect Genet Evol. influenza viruses. Emerg Microbes Infect. 2017;6:e12. 2016;44:530–40. 20. Wang X, Wang Q, Cheng W, Yu Z, Ling F, Mao H, et al. Risk 5. Lin YP, Shaw M, Gregory V, Cameron K, Lim W, Klimov A, et al. factors for avian influenza virus contamination of live poultry Avian-to-human transmission of H9N2 subtype influenza A viruses: markets in Zhejiang, China during the 2015–2016 human relationship between H9N2 and H5N1 human isolates. Proc Natl influenza season. Sci Rep. 2017;7:42722. Acad Sci U S A. 2000;97:9654–8. srep42722 pnas.160270697 21. Sayeed MA, Smallwood C, Imam T, Mahmud R, Hasan RB, 6. Monne I, Yamage M, Dauphin G, Claes F, Ahmed G, Giasuddin M, Hasan M, et al. Assessment of hygienic conditions of live bird et al. Reassortant avian influenza A(H5N1) viruses with H9N2-PB1 markets on avian influenza in Chittagong metro, Bangladesh. gene in poultry, Bangladesh. Emerg Infect Dis. 2013;19:1630–4. Prev Vet Med. 2017;142:7–15. j.prevetmed.2017.04.009 7. Gao R, Cao B, Hu Y, Feng Z, Wang D, Hu W, et al. Human 22. Biswas PK, Biswas D, Ahmed S, Rahman A, Debnath NC. infection with a novel avian-origin influenza A (H7N9) virus. A longitudinal study of the incidence of major endemic and N Engl J Med. 2013;368:1888–97. epidemic diseases affecting semi-scavenging chickens reared NEJMoa1304459 under the Participatory Livestock Development Project areas in 8. Chen H, Yuan H, Gao R, Zhang J, Wang D, Xiong Y, et al. Bangladesh. Avian Pathol. 2005;34:303–12. Clinical and epidemiological characteristics of a fatal case of avian 10.1080/03079450500178972 influenza A H10N8 virus infection: a descriptive study. Lancet. 23. Moyen N, Ahmed G, Gupta S, Tenzin T, Khan R, Khan T, et al. 2014;383:714–21. A large-scale study of a poultry trading network in Bangladesh: (14)60111-2 implications for control and surveillance of avian influenza viruses. 9. Lee DH, Bertran K, Kwon JH, Swayne DE. Evolution, global BMC Vet Res. 2018;14:12. spread, and pathogenicity of highly pathogenic avian influenza 1331-5 H5Nx clade J Vet Sci. 2017;18(S1):269–80. 24. Fournié G, Tripodi A, Nguyen TT, Nguyen VT, Tran TT, Bisson A, et al. Investigating poultry trade patterns to guide avian influenza 10. Webster RG. Wet markets—a continuing source of severe acute surveillance and control: a case study in Vietnam. Sci Rep. respiratory syndrome and influenza? Lancet. 2004;363:234–6. 2016;6:29463. 25. Molia S, Boly IA, Duboz R, Coulibaly B, Guitian J, Grosbois V, 11. Nguyen DC, Uyeki TM, Jadhao S, Maines T, Shaw M, Matsuoka Y, et al. Live bird markets characterization and trading network et al. Isolation and characterization of avian influenza viruses, analysis in Mali: implications for the surveillance and control of including highly pathogenic H5N1, from poultry in live bird avian influenza and Newcastle disease. Acta Trop. 2016;155:77–88. markets in Hanoi, Vietnam, in 2001. J Virol. 2005;79:4201–12. 26. Indriani R, Samaan G, Gultom A, Loth L, Irianti S, Adjid R, et al. 12. Fournié G, Guitian FJ, Mangtani P, Ghani AC. Impact of the Environmental sampling for avian influenza virus A (H5N1) in implementation of rest days in live bird markets on the dynamics live-bird markets, Indonesia. Emerg Infect Dis. 2010;16:1889–95. of H5N1 highly pathogenic avian influenza. J R Soc Interface. 2011;8:1079–89. 27. Monne I, Ormelli S, Salviato A, De Battisti C, Bettini F, 13. Liu M, He S, Walker D, Zhou N, Perez DR, Mo B, et al. The Salomoni A, et al. Development and validation of a one-step influenza virus gene pool in a poultry market in South central real-time PCR assay for simultaneous detection of subtype H5, china. Virology. 2003;305:267–75. H7, and H9 avian influenza viruses. J Clin Microbiol. viro.2002.1762 2008;46:1769–73. 14. Okamatsu M, Nishi T, Nomura N, Yamamoto N, Sakoda Y, 28. Heine HG, Foord AJ, Wang J, Valdeter S, Walker S, Morrissy C, Sakurai K, et al. The genetic and antigenic diversity of avian et al. Detection of highly pathogenic zoonotic influenza virus H5N6 influenza viruses isolated from domestic ducks, muscovy ducks, by reverse-transcriptase quantitative polymerase chain reaction. and chickens in northern and southern Vietnam, 2010–2012. Virol J. 2015;12:18. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2315 RESEARCH 29. Kruschke JK. Doing Bayesian data analysis: a tutorial with R, JAGS, and Stan. 2nd ed. New York: Academic Press; 2014. EID Podcast: 30. Plummer M. JAGS version 3.4.0 user manual, 2013 [cited 2018 Aug 23]. Emerging Infectious jags_user_manual.pdf 31. R Core Team. R: a language and environment for statistical Diseases Cover Art computing. Vienna: R Foundation for Statistical Computing, 2016 Byron Breedlove, managing editor of [cited 2018 Aug 23]. the journal, elaborates on aesthetic 32. Abdelwhab EM, Selim AA, Arafa A, Galal S, Kilany WH, Hassan MK, et al. Circulation of avian influenza H5N1 in live bird considerations and historical factors, markets in Egypt. Avian Dis. 2010;54:911–4. as well as the complexities of obtaining 10.1637/9099-100809-RESNOTE.1 artwork for Emerging Infectious Diseases. 33. Khatun A, Giasuddin M, Islam KM, Khanom S, Samad MA, Islam MR, et al. Surveillance of avian influenza virus type A in semi-scavenging ducks in Bangladesh. BMC Vet Res. 2013;9:196. 34. Sarkar S, Khan SU, Mikolon A, Rahman MZ, Abedin J, Zeidner N, et al. An epidemiological study of avian influenza A (H5) virus in nomadic ducks and their raising practices in northeastern Bangladesh, 2011-2012. Influenza Other Respi Viruses. 2017;11:275–82. 35. Blohm U, Weigend S, Preisinger R, Beer M, Hoffmann D. Immunological competence of different domestic chicken breeds against avian influenza infection. Avian Dis. 2016;60(Suppl): 262–8. 36. Ruiz-Hernandez R, Mwangi W, Peroval M, Sadeyen JR, Ascough S, Balkissoon D, et al. Host genetics determine susceptibility to avian influenza infection and transmission dynamics. Sci Rep. 2016;6:26787. 37. Fournié G, Høg E, Barnett T, Pfeiffer DU, Mangtani P. A systematic review and meta-analysis of practices exposing humans to avian influenza viruses, their prevalence, and rationale. Am J Trop Med Hyg. 2017;97:376–88. 38. Hulse-Post DJ, Sturm-Ramirez KM, Humberd J, Seiler P, Govorkova EA, Krauss S, et al. Role of domestic ducks in the propagation and biological evolution of highly pathogenic H5N1 influenza viruses in Asia. Proc Natl Acad Sci U S A. 2005;102:10682–7. 39. Kim JK, Negovetich NJ, Forrest HL, W ebster RG. Ducks: the “Trojan horses” of H5N1 influenza. Influenza Other Respi Viruses. 2009;3:121–8. 40. Ladman BS, Spackman E, Gelb J Jr. Comparison of pooling 11 or 5 oropharyngeal swabbings for detecting avian influenza virus by real-time reverse transcription-PCR in broiler chickens. Avian Dis. 2012;56:227–9. 41. Arnold ME, Slomka MJ, Coward VJ, Mahmood S, Raleigh PJ, Brown IH. Evaluation of the pooling of swabs for real-time PCR detection of low titre shedding of low pathogenicity avian influenza in turkeys. Epidemiol Infect. 2013;141:1286–97. http://dx.doi.or g/ 10.1017/S0950268812001811 42. Spackman E, Pedersen JC, McKinley ET, Gelb J Jr. Optimal specimen collection and transport methods for the detection of avian influenza virus and Newcastle disease virus. BMC Vet Res. 2013;9:35. 43. Ge S, Zheng D, Zhao Y, Liu H, Liu W, Sun Q, et al. Evaluating viral interference between Influenza virus and Newcastle disease virus using real-time reverse transcription-polymerase chain reaction in chicken eggs. Virol J. 2012;9:128. 44. Park AW, Glass K. Dynamic patterns of avian and human influenza in east and southeast Asia. Lancet Infect Dis. 2007;7:543–8. Visit our website to listen: Address for correspondence: Younjung Kim, City University of Hong Kong, Rm 1B-410, 4/F, Block 1, To Yuen Bldg, 31 To Yuen St, Kowloon, podcasts/player. Hong Kong, China; email: asp?f=8646224 2316 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Human Exposure to Novel Bartonella Species from Contact with Fruit Bats Ying Bai, Modupe O.V. Osinubi, Lynn Osikowicz, Clifton McKee, Neil M. Vora, Maria Rosales Rizzo, Sergio Recuenco, Lora Davis, Mike Niezgoda, Ajoke M. Ehimiyein, Grace S.N. Kia, Akin Oyemakinde, Olufunmilayo Sanni Adeniyi, Yemi H. Gbadegesin, Olugbon A. Saliman, Abiodun Ogunniyi, Albert B. Ogunkoya, Michael Y. Kosoy; Idanre Bat Festival Investigation Team Twice a year in southwestern Nigeria, during a traditional abundant and frequently roost within or in close proximity bat festival, community participants enter designated caves to humans and domestic animals. In Asia and Africa, larger to capture bats, which are then consumed for food or trad- fruit bats (family Pteropodidae) are used as food, for either ed. We investigated the presence of Bartonella species in cultural reasons or subsistence (2). In some cultures, bat Egyptian fruit bats (Rousettus aegyptiacus) and bat flies caves serve as spiritual sanctuaries (3). (Eucampsipoda africana) from these caves and assessed One particular situation that has attracted the attention whether Bartonella infections had occurred in persons from of scientists is a bat festival that takes place biannually in the surrounding communities. Our results indicate that the Idanre Hills area of Nigeria. During the festival, which these bats and flies harbor Bartonella strains, which multi- has occurred for many years, men enter designated caves, locus sequence typing indicated probably represent a novel often without appropriate personal protective equipment, Bartonella species, proposed as Bartonella rousetti. In se- rum from 8 of 204 persons, we detected antibodies to B. to capture bats. Local customs forbid persons from enter- rousetti without cross-reactivity to other Bartonella species. ing the caves outside of these festivities without permission This work suggests that bat-associated Bartonella strains from the community leadership. The captured bats are then might be capable of infecting humans. eaten, used in cultural rituals, or sold as bushmeat (3). The predominant bat species within the caves is the Egyptian fruit bat (Rousettus aegyptiacus); colony sizes can reach ats are natural reservoirs for a variety of pathogens >1,000 (4). Egyptian fruit bats are known reservoirs of zoo- B(1). However, despite the risk to human health, per- notic pathogens including Lagos bat virus, Marburg virus, sons around the world still intentionally handle bats, often and Yersinia pseudotuberculosis (5–8). Given the close without taking appropriate precautions. This lack of precau- human-to-bat contact that occurs during the festival, there tions is particularly evident in the tropics, where bats are is a risk for spillover of batborne pathogens to humans. The genus Bartonella currently includes >30 species Author affiliations: Centers for Disease Control and Prevention, of bacteria (9), many of which have been described only Fort Collins, Colorado, USA (Y. Bai, L. Osikowicz, C. McKee, recently. Various arthropod vectors seem to play an es- M.R. Rizzo, M.Y. Kosoy); Centers for Disease Control and sential role in the maintenance and transmission of most Prevention, Atlanta, Georgia, USA (M.O.V. Osinubi, N.M. Vora, known Bartonella species (9,10). In recent years, recog- L. Davis, M. Niezgoda); Colorado State University, Fort Collins nition of multiple Bartonella species as human pathogens (C. McKee); Universidad Nacional Mayor de San Marcos, Lima, responsible for a wide range of clinical manifestations has Peru (S. Recuenco); Ahmadu Bello University, Zaria, Nigeria grown. Numerous novel strains of Bartonella have been (A.M. Ehimiyein, G.S.N. Kia); African Field Epidemiology Network, discovered in bats of various species around the globe, in- Abuja, Nigeria (A. Oyemakinde); Federal Ministry of Health, cluding the human pathogen Candidatus Bartonella mayo- Abuja (O.S. Adeniyi); Federal Ministry of Science and Technology, timonensis, which was originally detected in aortic valve Ibadan, Nigeria (Y.H. Gbadegesin); Ministry of Agriculture and tissue of a person with endocarditis (11–13). In addition, a Natural Resources, Ilorin, Nigeria (O.A. Saliman); Nigeria Center novel Bartonella genotype found in bats from the country for Disease Control, Abuja (A. Ogunniyi); Centre for Control and of Georgia clustered with genotypes found in human forest Prevention of Zoonoses/Rabies in West Africa International, workers from Poland (14). Ibadan (A.B. Ogunkoya) DOI: Team members are listed at the end of this article. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2317 RESEARCH During 2010 and 2013, we researched the health risk to We purified and sequenced all PCR products of gltA humans participating in the Idanre bat festival. We sampled in both directions by using an ABI 3130 Genetic Analyzer bats and their ectoparasites from the caves and used them to (Applied Biosystems, Foster City, CA, USA). We used the identify a variety of zoonotic pathogens, including Barton- Lasergene software package (DNASTAR, Madison, WI, ella. We recruited human participants from the surrounding USA) to compare the generated gltA sequences with all community and surveyed them (through an orally adminis- available Bartonella species/genotypes. Once the sequenc- tered questionnaire and serologic testing) to understand risk es were identified, we selected 1 representative strain (R- factors and the occurrence of pathogen spillover from bats 191) for further characterization with multilocus sequence to humans. We examined whether bats and ectoparasites of typing on the basis of sequence analysis of 8 molecular these bats within the caves used in the Idanre bat festival markers (ftsZ, gltA, nuoG, ribC, rpoB, ssrA, 16S rRNA, are infected with Bartonella, characterized any Bartonella and internal transcribed spacers [ITS]) (16). For phyloge- species identified in bats or bat flies, and screened human netic analyses, we used the neighbor-joining method by the serum samples for evidence of Bartonella infection. Kimura 2-parameter distance method and bootstrap calcu- lations with 1,000 replicates. Materials and Methods Human subjects work was approved by the Centers for Dis- Bat Ectoparasite Collection and Detection of ease Control and Prevention (CDC) Institutional Review Bartonella DNA Board, the Ahmadu Bello University Human Ethics Board, We collected ectoparasites from the skin and pelage of bats and the National Health Research Ethics Committee of Ni- and stored them in microcentrifuge tubes with 70% etha- geria. All animal procedures were conducted in compliance nol. Ectoparasite species were identified by using available with a field protocol approved by the CDC Animal Institu- morphologic keys (17), and identifications were later con- tional Care and Use Committee. firmed by sequencing of the mitochondrial 16S rRNA and cytochrome oxidase I (COI) genes (18,19). Field Sites, Bat Capture, and Sample Collection Using a Bullet Blender Gold homogenizer (Next Ad- We captured bats by nets in 2 caves in Idanre Hills, Ondo vance, Averill Park, NY, USA), we homogenized whole State, southwestern Nigeria, in September 2010 (n = 106) ectoparasites in Navy Eppendorf bead tubes (Next Ad- and February 2013 (n = 71). We identified all bats by vance) containing 400 µL brain–heart infusion broth (CDC, morphologic characteristics as Egyptian fruit bats (R. ae- Atlanta, GA, USA). We extracted DNA from the homog- gyptiacus). Captured bats were anesthetized by intramus- enates by using the KingFisher Flex Purification System cular injection of ketamine hydrochloride (0.05–0.1 mg/g and the associated MagMAX Pathogen RNA/DNA Kit bat weight) and exsanguinated via cardiac puncture after (both ThermoFisher, Waltham, MA, USA) according to the surface sterilization with 75% alcohol. Serum and blood manufacturer’s protocols. Detection of Bartonella DNA in clots were separated by centrifugation. Clots were stored at ectoparasite samples was performed by nested PCR for –80°C except while still in the field or being shipped, dur- gltA (20) because of low concentrations of DNA and by ing which time they were stored on dry ice. conventional PCR for ITS (21), followed by sequencing and sequence analysis of amplicons. Bat Blood Culture and Characterization of Preparation of Antigen from the Bartonella Strain Bartonella Strain Obtained from Bats We plated bat blood clots on heart infusion agar containing We produced a whole-cell antigen by co-cultivating Vero E6 10% rabbit blood and incubated in an aerobic atmosphere cells with the pure culture (≈10 agar-grown organisms) of with 5% carbon dioxide at 35°C for up to 4 weeks. Bacte- the Bartonella strain (R-191) obtained from Egyptian fruit rial colonies morphologically identified as Bartonella were bats. Both were put into T-150 flasks that contained mini - subcultured to obtain pure cultures. mum essential medium supplemented with 10% fetal calf We prepared crude genomic DNA by heating a heavy serum, 10 mmol HEPES buffer solution, 10 mmol nonessen - suspension of pure culture for 10 minutes at 95°C, followed tial amino acids, and 2 mmol L-glutamine. The flasks were by centrifugation of the lysed cells for 1 minute at 3,000 incubated at 35°C and harvested on postinoculation day 4. rpm. The supernatant was then transferred to a clean centri- At harvest, all but 2 mL of the medium was removed from fuge tube to be used as the template DNA. We first verified the flask, sterile glass beads were added, and the flask was all isolates obtained from the blood clots as Bartonella spp. gently rocked to remove the Vero E6 cell monolayer. Drops by PCR amplification targeting a fragment of the citrate (≈15 µL) of the cell suspension were mounted on each well synthase gene (gltA) (15). Positive (B. doshiae) and nega- of 12-well glass slides, which were then air dried, fixed in tive (deionized water) controls were included to ensure that acetone for 15 minutes, and stored at –70°C until use. the PCR worked properly. 2318 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Human Exposure to Novel Bartonella Species Human Serum Collection and Testing for Antibodies genetic loci (ftsZ, nuoG, ribC, rpoB, ssrA, 16S rRNA, Persons in communities surrounding the caves who gave and ITS) further confirmed the uniqueness of this strain. consent were enrolled in the study 11–15 days after the first Sequencing information for each genetic marker demon- bat festival of 2013 (February 19, 2013); not all of these strated that the Bartonella strain from the Egyptian fruit persons had participated in all bat festival activities. Partici- bats was distant from all other known Bartonella species pants were asked about their contact with bats and their role and genotypes, including those reported from other bats in the festival, and some provided a blood sample (consid- from Africa. Sequences of all genetic loci obtained dur- ered an acute-phase specimen). About 69–78 days later, a ing the analyses were deposited in GenBank (accession follow-up survey was conducted and a second blood sample nos. HM363769, KM387321, HM363779, HM363774, (considered a convalescent-phase specimen) was collected KM382247, HM363784, and KM382255). We compared (the second bat festival of 2013 did not take place between the fragment sequences of each target with those from other collection of the acute- and convalescent-phase samples). Bartonella species/genotypes. The Egyptian fruit bat–asso- Serum and blood clots were separated by centrifugation; ciated Bartonella formed a separate genetic group that was serum was stored at –80°C except while in the field or be - distant from all other Bartonella species with >20% genetic ing shipped, during which time it was stored on dry ice. distance and probably represents a novel Bartonella spe- To screen human serum, we used an indirect immu- cies, according to the definition of La Scola et al. (26). We nofluorescence assay at an initial dilution of 1:32 for IgG proposed that this bacterial species be named Bartonella against the specific Bartonella antigen from the bat-associ- rousetti, to reflect the Egyptian fruit bat (Rousettus aegyp- ated isolate. Antigen-covered wells of the slide were over- tiacus) as the natural host. A phylogenetic tree based on the laid with dilutions of human serum. Separate slides were ITS locus illustrates the relationship of this proposed novel included with positive and negative controls. The positive species to other Bartonella species (Figure). control against the Egyptian fruit bat–associated Barton- ella species was produced in laboratory mice via mouse Identification of Bat Flies and Detection of immunization with heat-inactivated bacterium (ProSci In- Bartonella DNA corporated, Poway, CA, USA). All slides were incubated In 2013, we collected 51 ectoparasites from Egyptian fruit at 35°C for 30 minutes and then washed in phosphate-buff- bats. With the exception of 1 unidentified mite, all arthro - ered saline for 15 minutes. We used anti-human and anti- pods were identified as the bat fly Eucampsipoda africana mouse conjugates (Kirkegaard & Perry Laboratories Inc., Theodor (Diptera: Nycteribiidae). The morphologic identi - Gaithersburg, MD, USA) for human and control serum fication of every bat fly was confirmed by 1 or both mito - samples, respectively. Each human serum sample reactive chondrial markers (16S rRNA or COI). Representative 16S at the initial dilution was further titrated in 2-fold dilutions rRNA (accession nos. MH138030–MH138037) and COI to endpoint; to check for cross-reactivity, we tested the fi- (accession nos. MH151059–MH151066) sequences have nal positive samples (defined as a titer >1:64) for 3 other been deposited in GenBank. Bartonella antigens (B. elizabethae, B. henselae, and B. Of the 50 DNA extracts from bat flies, 21 (42%) pro- quintana) previously reported in Africa (22–24). duced >1 ITS or gltA sequence that was confirmed via BLAST ( as Bar- Results tonella. Positive samples yielded 19 ITS sequences and 18 gltA sequences; 16 samples yielded sequences for both loci Bartonella in Egyptian Fruit Bats and 5 samples yielded only 1 sequence. All but 1 of the 19 We recovered Bartonella isolates from 22 of 177 Egyptian ITS sequences matched closely to the proposed B. rousetti fruit bat blood clots, giving an overall prevalence of 12.4%. (>95% sequence identity) (Figure); the remaining sequence The gltA sequences of all Bartonella strains obtained from was identical to B. tamiae (DQ395180). Of the 18 (66.7%) Egyptian fruit bats were identical or similar (>97% identity) gltA sequences, 12 were close matches for B. rousetti to each other and represented 4 unique variants (GenBank (>98.3% sequence identity); all 12 of these samples also accession nos. HM363764, MH069693–MH069695). A produced ITS sequences matching this strain. variant is defined when it differs by >1 nt from others. To- The remaining 6 gltA sequences were identical to Bar- gether with Bartonella strains obtained from Egyptian fruit tonella sequences detected in a louse (Neohaematopinus bats in Kenya (25), these variants constitute a monophy- sciuri) collected from a dead Eastern gray squirrel (Sciurus letic genogroup that is distant from all other genotypes pre- carolinensis) at a zoo in Greenville, SC, USA (GenBank viously found in other bat species and any other described accession no. EU368000) and an unidentified tick collected Bartonella species. from a sheep in Peru (GenBank accession no. AF415209). Multilocus sequence typing of the type strain (gltA; These sequences were also closely (>99% sequence iden- GenBank accession no. HM363764) with 7 additional tity) related to other sequences from fleas (Ctenocephalides Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2319 RESEARCH Figure. Phylogenetic relationships of Bartonella rousetti (proposed name) obtained from Egyptian fruit bats (Rousettus aegyptiacus) collected in Nigeria, 2010 and 2013, and other Bartonella species and bat-associated Bartonella based on internal transcribed spacer sequences. The neighbor-joining method by the Kimura 2-parameter distance method and bootstrap calculation was conducted with 1,000 replicates for phylogenetic analysis. The internal transcribed spacer sequence obtained from the bat flies was closely clustered with B. rousetti. GenBank accession numbers are provided for the B. rousetti sequence and the comparison sequences. felis and C. canis) collected from dogs in Tunisia (GenBank titers >1:64 (Table). The positive samples were retested for accession nos. KP126468–74), a louse pool (Polyplax spp. 3 other Bartonella species—B. henselae, B. quintana, and and Hoplopleura spp.) collected from rodents in Thailand B. elizabethae, all of which have been reported in Africa (GenBank accession no. KT324560), and an unidentified (22–24); antibodies against these Bartonella species were flea collected from a dog in Peru (GenBank accession no. not detected in any of the samples. Five seropositive partic- GU583843). Of the specimens with this particular gltA se- ipants reported having eaten bats and having either touched quence, 2 yielded no ITS sequence, 3 yielded ITS sequenc- bats or been scratched or bitten by them, although not all es matching B. rousetti, and 1 yielded the single B. tamiae reported having ever participated in the bat festival. Three sequence. Representative ITS (GenBank accession nos. seropositive participants reported never having eaten bats, MH14262–MH142639) and gltA (GenBank accession nos. touched bats, or been scratched or bitten by bats; in addi- MH151067–79) sequences for each novel sequence variant tion, these 3 participants claimed to have never participated were submitted to GenBank. in the bat festival. Of the 8 seropositive participants, only 1 reported having experienced a febrile illness since the bat Human Exposure to B. rousetti festival that had taken place earlier in the year. A total of 305 serum samples from 204 participants were tested for IgG against B. rousetti; 12 samples from differ- Discussion ent persons showed reactivity at an initial dilution of 1:32. We made several observations during this investigation. Further 2-fold titration confirmed that 8 were positive, with First, Egyptian fruit bats carry a unique Bartonella strain 2320 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Human Exposure to Novel Bartonella Species Table. Epidemiologic data for persons with antibodies to Bartonella rousetti detected in study of human exposure to a novel Bartonella species from contact with fruit bats, Nigeria, 2013* Last time touched, Febrile illness Participant Titer in acute- Titer in convalescent- Ever participated scratched, or bitten since first bat age, y/sex phase serum† phase serum† Ever ate bat in bat festival by bat festival of 2013 45/F <1:32 1:64 Yes No 6–12 mo ago No 37/M <1:32 1:64 Yes No >12 mo ago No 25/F <1:32 1:512 Yes Yes <1 mo ago No 30/F <1:32 1:512 No No Never No 21/M 1:64 <1:32 No No Never No 44/F 1:64 <1:32 Yes Yes <1 mo ago No 70/M 1:256 No sample Yes No >12 mo ago Yes 32/F 1:256 No sample No No Never No *Bartonella rousetti is the proposed name for the novel Bartonella species identified in Egyptian fruit bats in Nigeria. †Acute-phase samples collected within 11–15 d after first bat festival of 2013; convalescent-phase samples collected 69–78 d after acute-phase sample collection (the second bat festival of 2013 did not take place between collections of acute- and convalescent-phase samples). that probably represents a new species, for which we pro- Detection of antibodies against B. rousetti in serum pose the name Bartonella rousetti. Second, bat flies, the samples from several study participants indicates their common ectoparasites of bats, carry this same strain of exposure to the bacteria. However, with serologic results, Bartonella. Because this organism was detected by PCR cross-reactivity is a concern. For example, phylogeneti- only, the presence of the DNA does not necessarily indicate cally closely related B. henselae and B. quintana (the caus- that the organism is viable. Last, persons from the commu- ative agents of cat-scratch disease and trench fever, respec- nities surrounding the bat caves were exposed to this partic- tively) exhibit a high level of serologic cross-reactivity ular Bartonella strain, which might cause human infection. (26,38,39). We tested the positive human serum samples Since 2010, several reports have described finding di - for 3 other Bartonella species (B. henselae, B. quintana, verse Bartonella genotypes in bats of many species (25,27– and B. elizabethae) that circulate in Africa, and we did not 30). The relationships between Bartonella genotypes and bat detect any positive results. Given that immunofluorescence species that harbored these bacteria are not always simple. assays have good discriminatory ability for a wide range The same Bartonella species may circulate among different of antigens (40–42), the results lead us to conclude that the bat species, showing no specific relationship between the bats antibodies in these participants were indeed reactive with and the Bartonella species (27). Sometimes, multiple Barton- B. rousetti but not the other Bartonella species tested, al- ella species are associated with bats of only 1 species. For ex- though cross-reactivity with other non-Bartonella agents ample, 6 Bartonella species have been identified in straw-col - cannot be ruled out. ored fruit bats (Eidolon helvum) in Africa (16,29). Our study Our study is not the first attempt to identify antibodies indicates that Egyptian fruit bats carry a specific Bartonella against bat-associated Bartonella in humans. Mannerings strain that has not been identified in bats of other species. et al. (43) conducted a serologic survey of 335 volunteers Similarly, we found that the most prevalent Bartonella from Ghana for antibodies against 6 species of Bartonella, species found in bat flies parasitizing Egyptian fruit bats is including Bartonella strains isolated from straw-colored B. rousetti. The ectoparasite bat flies E. africana are pre- fruit bats. In that study, only 2 serum samples were positive dominantly associated with Egyptian fruit bats (17,31,32). for B. henselae antibodies at low titers, whereas none was Although sequences matching other Bartonella species positive against the bat strains. were identified in the bat flies, these genogroups may be All known species of Bartonella are transmitted between primarily associated with arthropods and not mammals. natural animal hosts by arthropods (29,44). The presence of One sequence from a bat fly was identified as B. tamiae. B. rousetti DNA in E. africana bat flies parasitizing Egyptian The presence of B. tamiae in bat flies from Algeria has been fruit bats suggests that these ectoparasites may act as vectors recorded (33), and the bacterium reportedly has been iden- for the transmission of Bartonella infection among bats, but tified in chigger mites collected from rodents in Thailand it is unclear how bat flies would play a role in transmitting the (34). It is possible that Bartonella species found only in ar- bacterium to humans because bat flies do not commonly bite thropods and not their associated mammal hosts may repre- humans (C. McKee, unpub. data). Instead, human exposure sent facultative symbionts that are uniquely adapted to live may potentially occur via other routes, such as 1) directly by in the arthropod gut or other body system (35,36). The risks bat bites or scratches, which is similar to how humans acquire posed to humans by these primarily arthropod-associated infections with B. henselae through cat scratches (10,45); 2) Bartonella species are still unclear, although B. tamiae is a indirectly by contamination of open wounds with blood or reported human pathogen that may cause febrile illness and other materials (e.g., saliva, urine, feces) of infected bats; or other clinical signs and symptoms (37). 3) indirectly by contamination of open wounds with bat fly Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2321 RESEARCH excreta. Several studies have reported detecting Bartonella References 1. Calisher CH, Childs JE, Field HE, Holmes KV, Schountz T. Bats: DNA in bat feces (12,46,47), and Dietrich et al. (47) detected important reservoir hosts of emerging viruses. Clin Microbiol Rev. Bartonella DNA in bat saliva and urine, providing support for 2006;19:531–45. routes 1 and 2 above, although no attempts have been made 2. Mickleburgh S, Waylen K, Racey P. Bats as bushmeat: a global to culture viable bacteria from these fluids. However, viable review. Oryx. 2009;43:217–34. S0030605308000938 Bartonella bacteria have been cultured from experimentally 3. Osinubi MO, Recuenco S, Kuzmin I, Haberling DL, Blau DM, infected ectoparasites, including fleas and bedbugs ( 48,49), Davis LB, et al. Knowledge, attitudes and practices among although such studies have yet to be performed for bat flies. populations exposed to bats in southern Nigeria. Revista de Nevertheless, evidence is accumulating that Bartonella could Educação Continuada em Medicina Veterinária e Zootecnia do CRMV-SP. 2012;10, n. 2/3. spread from infected mammalian hosts through multiple 4. Kwiecinski GG, Griffiths T A. Rousettus egyptiacus. Mammalian routes. Therefore, it may not be necessary for humans to in- Species. 1999;611:1–9. teract directly with live bats to be exposed to bat-associated 5. Kuzmin IV, Niezgoda M, Franka R, Agwanda B, Markotter W, Bartonella. Persons might be at risk when interacting with Beagley JC, et al. Lagos bat virus in Kenya. J Clin Microbiol. 2008;46:1451–61. bat carcasses, guano, or other contaminated products. Of 6. Towner JS, Amman BR, Sealy TK, Carroll SA, Comer JA, note, we do not provide definitive evidence of the route of Kemp A, et al. Isolation of genetically diverse Marburg viruses exposure for any of the 8 seropositive participants. Indeed, 3 from Egyptian fruit bats. PLoS Pathog. 2009;5:e1000536. of these participants reported no interactions at all with bats. 7. Nakamura S, Settai S, Hayashidani H, Urabe T, Namai S, Future studies should continue to evaluate the relative Une Y. Outbreak of yersiniosis in Egyptian rousette bats correlations of exposure routes, the pathobiology of bat- (Rousettus aegyptiacus) caused by Yersinia pseudotuberculosis borne B. rousetti in humans, and vector competency of bat serotype 4b. J Comp Pathol. 2013;148:410–3. flies for transmitting Bartonella. Results should provide 10.1016/j.jcpa.2012.07.007 8. Amman BR, Jones ME, Sealy TK, Uebelhoer LS, Schuh AJ, guidance to communities for mitigating the risks to humans Bird BH, et al. Oral shedding of Marburg virus in experimentally interacting with animals and their arthropod vectors. infected Egyptian fruit bats (Rousettus aegyptiacus). J Wildl Dis. 2015;51:113–24. Idanre Bat Festival Investigation Team members: Ivan V. Kuz min, 9. Chomel BB, Boulouis HJ, Breitschwerdt EB, Kasten RW, Dianna Blau, James Ellison, Lauren Greenberg, Marissa Person, Vayssier-Taussat M, Birtles RJ, et al. Ecological fitness and strategies of adaptation of Bartonella species to their hosts Ryan Wallace, Panayampalli S. Satheshkumar, Abimbola and vectors. Vet Res. 2009;40:29. Aman-Oloniyo, Elizabeth B. Adedire, Mariat O. Soleye, vetres/2009011 Gloria C. Okara, Sebastian Yennan, Mohammed Abdurrahman, 10. Chomel BB, Kasten RW, Floyd-Hawkins K, Chi B, Yamamoto K, Munir A. Sani, Solomon. W Audu, Maruf Lawal, and Roberts-Wilson J, et al. Experimental transmission of Bartonella henselae by the cat flea. J Clin Microbiol. 1996;34:1952–6. Philip P. Mshelbwala. 11. Lin EY, Tsigrelis C, Baddour LM, Lepidi H, Rolain JM, Patel R, et al. Candidatus Bartonella mayotimonensis and endocarditis. Emerg Infect Dis. 2010;16:500–3. Acknowledgments eid1603.081673 We thank J.D. Kirby, Mary Reynolds, Todd Smith, the Vice 12. Veikkolainen V, Vesterinen EJ, Lilley TM, Pulliainen AT. Chancellor and Management of Ahmadu Bello University, the Bats as reservoir hosts of human bacterial pathogen, Bartonella mayotimonensis. Emerg Infect Dis. 2014;20:960–7. Federal Ministry of Health (Abuja, Nigeria), the Owa of Idanre Oba Fredrick Adegunle Aroloye IV, and the chiefs of the Idanre 13. Lilley TM, Wilson CA, Bernard RF, Willcox EV, Vesterinen EJ, community, Ondo State, Nigeria, for their helpful comments and Webber QM, et al. Molecular detection of Candidatus Bartonella assistance with logistics. mayotimonensis in North American bats. Vector Borne Zoonotic Dis. 2017;17:243–6. This study was supported by the Biosecurity Engagement 14. Urushadze L, Bai Y, Osikowicz L, McKee C, Sidamonidze K, Program of the US Department of State, Bureau of International Putkaradze D, et al. Prevalence, diversity, and host associations of Bartonella strains in bats from Georgia (Caucasus). PLoS Negl Security and Nonproliferation, and the Office of Cooperative Trop Dis. 2017;11:e0005428. Threat Reduction’s Global Threat Reduction Programs; One journal.pntd.0005428 Health funding; and the Global Disease Detection Program of 15. Norman AF, Regnery R, Jameson P, Greene C, Krause DC. the Center for Global Health at CDC. Differentiation of Bartonella-like isolates at the species level by PCR-restriction fragment length polymorphism in the citrate About the Author synthase gene. J Clin Microbiol. 1995;33:1797–803. 16. Bai Y, Hayman DTS, McKee CD, Kosoy MY. Classification of Dr. Bai is a microbiologist in the Division of Vector-Borne Bartonella strains associated with straw-colored fruit bats Infectious Diseases, National Center for Emerging and Zoonotic (Eidolon helvum) across Africa using a multi-locus sequence typing platform. PLoS Negl Trop Dis. 2015;9:e0003478. Infectious Diseases, CDC, Fort Collins. Her research interests 10.1371/journal.pntd.0003478 include microbiology, epidemiology, and ecology of zoonotic 17. Theodor O. The Nycteribiidae of the Ethiopian region and infectious diseases. Madagascar. Parasitology. 1957;47:457–543. 2322 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Human Exposure to Novel Bartonella Species 18. Quetglas J, Balvín O, Lučan RK, Benda P. First records of the bat 35. Neuvonen MM, Tamarit D, Näslund K, Liebig J, Feldhaar H, bug Cacodmus vicinus (Heteroptera: Cimicidae) from Europe and Moran NA, et al. The genome of Rhizobiales bacteria in further data on its distribution. Vespertilio. 2012;16:243–8. predatory ants reveals urease gene functions but no genes for 19. Szalanski AL, Austin JW, Scheffrahn RH, Messenger MT. nitrogen fixation. Sci Rep. 2016;6:39197. Molecular diagnostics of the Formosan subterranean termite srep39197 (Isoptera: Rhinotermitidae). Fla Entomol. 2004;87:145–51. 36. Segers FH, Kešnerová L, Kosoy M, Engel P. Genomic changes[0145: associated with the evolutionary transition of an insect gut MDOTFS]2.0.CO;2 symbiont into a blood-borne pathogen. ISME J. 2017;11:1232–44. 20. Bai Y, Gilbert A, Fox K, Osikowicz L, Kosoy M. Bartonella rochalimae and B. vinsonii subsp. berkhoffii in wild carnivores 37. Kosoy M, Morway C, Sheff KW, Bai Y, Colborn J, Chalcraft L, from Colorado, USA. J Wildl Dis. 2016;52:844–9. et al. Bartonella tamiae sp. nov., a newly recognized pathogen 10.7589/2016-01-015 isolated from three human patients from Thailand. J Clin 21. Diniz PP, Maggi RG, Schwartz DS, Cadenas MB, Bradley JM, Microbiol. 2008;46:772–5. Hegarty B, et al. Canine bartonellosis: serological and molecular 38. Baneth G, Kordick DL, Hegarty BC, Breitschwerdt EB. prevalence in Brazil and evidence of co-infection with Comparative seroreactivity to Bartonella henselae and Bartonella Bartonella henselae and Bartonella vinsonii subsp. berkhoffii. Vet quintana among cats from Israel and North Carolina. Vet Microbiol. Res. 2007;38:697–710. 1996;50:95–103. 22. Kelly PJ, Rooney JJ, Marston EL, Jones DC, Regnery RL. 39. Vermeulen MJ, Verbakel H, Notermans DW, Reimerink JH, Bartonella henselae isolated from cats in Zimbabwe. Lancet. Peeters MF. Evaluation of sensitivity, specificity and cross- 1998;351:1706. reactivity in Bartonella henselae serology. J Med Microbiol. 23. Boutellis A, Veracx A, Angelakis E, Diatta G, Mediannikov O, 2010;59:743–5. Trape JF, et al. Bartonella quintana in head lice from Sénégal. 40. Iralu J, Bai Y, Crook L, Tempest B, Simpson G, Mckenzie T, Vector Borne Zoonotic Dis. 2012;12:564–7. et al. Rodent-associated Bartonella febrile illness, southwestern 10.1089/vbz.2011.0845 United States. Emerg Infect Dis. 2006;12:1081–6. 24. Kamani J, Morick D, Mumcuoglu KY, Harrus S. Prevalence 10.3201/eid1207.040397 and diversity of Bartonella species in commensal rodents and 41. Laudisoit A, Iverson J, Neerinckx S, Shako JC, Nsabimana JM, ectoparasites from Nigeria, West Africa. PLoS Negl Trop Dis. Kersh G, et al. Human seroreactivity against Bartonella species 2013;7:e2246. in the Democratic Republic of Congo. Asian Pac J Trop Med. 25. Kosoy M, Bai Y, Lynch T, Kuzmin IV, Niezgoda M, Franka R, 2011;4:320–2. et al. Bartonella spp. in bats, Kenya. Emerg Infect Dis. 42. Myint KS, Gibbons RV, Iverson J, Shrestha SK, Pavlin JA, 2010;16:1875–81. Mongkolsirichaikul D, et al. Serological response to Bartonella 26. La Scola B, Raoult D. Serological cross-reactions between species in febrile patients from Nepal. Trans R Soc Trop Med Hyg. Bartonella quintana, Bartonella henselae, and Coxiella burnetii. 2011;105:740–2. J Clin Microbiol. 1996;34:2270–4. 43. Mannerings AO, Osikowicz LM, Restif O, Nyarko E, Suu-Ire R, 27. Bai Y, Kosoy M, Recuenco S, Alvarez D, Moran D, Turmelle A, Cunningham AA, et al. Exposure to bat-associated Bartonella et al. Bartonella spp. in bats, Guatemala. Emerg Infect Dis. spp. among humans and other animals, Ghana. Emerg Infect Dis. 2011;17:1269–72. 2016;22:922–4. 28. Bai Y, Recuenco S, Gilbert AT, Osikowicz LM, Gómez J, 44. Billeter SA, Hayman DT, Peel AJ, Baker K, Wood JL, Rupprecht C, et al. Prevalence and diversity of Bartonella Cunningham A, et al. Bartonella species in bat flies (Diptera: spp. in bats in Peru. Am J Trop Med Hyg. 2012;87:518–23. Nycteribiidae) from western Africa. Parasitology. 2012;139:324–9. 29. Kamani J, Baneth G, Mitchell M, Mumcuoglu KY, Gutiérrez R, 45. Mosbacher M, Elliott SP, Shehab Z, Pinnas JL, Klotz JH, Klotz SA. Harrus S. Bartonella species in bats (Chiroptera) and bat flies Cat scratch disease and arthropod vectors: more to it than a scratch? (Nycteribiidae) from Nigeria, West Africa. Vector Borne Zoonotic J Am Board Fam Med. 2010;23:685–6. Dis. 2014;14:625–32. jabfm.2010.05.100025 30. Brook CE, Bai Y, Dobson AP, Osikowicz LM, Ranaivoson HC, 46. Banskar S, Bhute SS, Suryavanshi MV, Punekar S, Shouche YS. Zhu Q, et al. Bartonella spp. in fruit bats and blood-feeding Microbiome analysis reveals the abundance of bacterial pathogens ectoparasites in Madagascar. PLoS Negl Trop Dis. in Rousettus leschenaultii guano. Sci Rep. 2016;6:36948. 2015;9:e0003532. 31. Nartey NAN. Common parasites of fruit-eating bats in southern 47. Dietrich M, Kearney T, Seamark ECJ, Markotter W. The excreted Ghana. Accra (Ghana): University of Ghana; 2015. p. 1–144. microbiota of bats: evidence of niche specialisation based on 32. Charles AN. Haemoparasites and ectoparasites of fruit bat species multiple body habitats. FEMS Microbiol Lett. 2017;364:fnw284. in Amurum Forest Reserve and their effects on host physiologic 48. Kernif T, Leulmi H, Socolovschi C, Berenger J-M, Lepidi H, and morphometric parameters. Jos (Nigeria): University of Jos; Bitam I, et al. Acquisition and excretion of Bartonella quintana by 2015. p. 1–54. the cat flea, Ctenocephalides felis felis. Mol Ecol. 2014;23: 33. Leulmi H, Aouadi A, Bitam I, Bessas A, Benakhla A, Raoult D, 1204–12. et al. Detection of Bartonella tamiae, Coxiella burnetii and 49. Leulmi H, Bitam I, Berenger J-M, Lepidi H, Rolain J-M, rickettsiae in arthropods and tissues from wild and domestic Almeras L, et al. Competence of Cimex lectularius bed bugs for animals in northeastern Algeria. Parasit Vectors. 2016;9:27. the transmission of Bartonella quintana, the agent of trench fever. PLoS Negl Trop Dis. 2015;9:e0003789. 34. Kabeya H, Colborn JM, Bai Y, Lerdthusnee K, Richardson JH, journal.pntd.0003789 Maruyama S, et al. Detection of Bartonella tamiae DNA in ectoparasites from rodents in Thailand and their sequence Address for correspondence: Ying Bai, Centers for Disease Control similarity with bacterial cultures from Thai patients. Vector and Prevention, 3150 Rampart Rd, Fort Collins, CO 80521, USA; Borne Zoonotic Dis. 2010;10:429–34. email: vbz.2009.0124 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2323 HISTORICAL REVIEW Emergent Sand Fly–Borne Phleboviruses in the Balkan Region Nazli Ayhan, Remi N. Charrel Sand fly–borne phleboviruses are associated with febrile is included in the Sandfly fever Naples virus species. Like- diseases and nervous system infections in the Mediter- wise, Sicilian virus was renamed sandfly fever Sicilian virus ranean basin. Sandfly fever was first reported in the Bal - (SFSV), which is still a tentative species. SFSV and SFNV kan Peninsula at the end of the 19th century. Since then, are both responsible for sandfly fever, a self-limiting but accumulating data show that the Balkan Peninsula, as a incapacitating febrile illness. Toscana virus (TOSV), dis- transboundary region between Asia and Europe, plays a covered in 1971, was incriminated as causing central and major role in the emergence of vectorborne diseases in peripheral nervous system infections in 1983 (8,9). TOSV Europe. To provide an inclusive approach, we collected can cause aseptic meningitis and meningoencephalitis (9– published data on phleboviruses in the Balkan countries 12), as well as a number of other manifestations affecting and used them to evaluate the impact of these pathogens the central and peripheral nervous system. These viruses are from virologic, epidemiologic, and public health perspec- transmitted via bites of Phlebotomus spp. sand flies. tives. Recent findings show a high diversity of phlebo - viruses belonging to 3 species or serocomplexes circu - Data concerning the geographic distribution of SFSV, lating heavily in the Balkans. Focusing on undisputable SFNV, and TOSV have drastically increased during the human pathogens, we found direct and indirect labora- past 2 decades, resulting in a more accurate cartography of tory documentation for Toscana virus, Sandfly fever Sicil - their presence in the Mediterranean basin, the Middle East, ian virus, and Adria virus. These data demonstrate that and central Asia (12–16). the Balkans are a hotspot for phleboviruses transmitted The Balkan Peninsula is a principal region for sand- by sand flies. fly fever. It is located in southeastern Europe, and consists of Slovenia, Croatia, Bosnia-Herzegovina, the Republic of hleboviruses (genus Phlebovirus, family Phenuiviri- Macedonia, Albania, Bulgaria, Greece, Montenegro, Ro- Pdae, order Bunyavirales) are 80–120 nm in length mania, Serbia, and Kosovo. The Balkan region is composed and display helical symmetry. Their genome consists of 3 of 3 very different natural entities: the Adriatic littoral in segmented negative-sense single-stranded RNA: the large the southwest, the Pannonian plain in the northeast, and a segment encodes the viral RNA polymerase (RdRp), the broad expanse of mountainous regions in between. The first medium segment encodes envelope glycoproteins (Gn and record of sandfly fever originated in Bosnia-Herzegovina Gc), and the small segment encodes nucleocapsid protein at the end of the 19th century (online Technical Appen- (N) and nonstructural protein (NS) (1,2). The segmented dix Table, nature of the genome allows recombination and reassort- 1626-Techapp1.pdf). During World War I and World War ment to occur with the potential to generate new viruses II, sandfly fever affected great numbers of soldiers in the with distinct ancestors (3,4). Segment reassortment in Bu- region (17,18) (online Technical Appendix Table). In ad- nyavirales has been reported with increasing frequency, dition to historical data, recent reports show the activity especially in the genus Orthobunyavirus (5). Specifically, of several novel viruses with severe human infections. We reassortant viruses have been described in both Candiru reviewed all the published data for sand fly–borne phlebo- and Rift Valley fever species (5,6). viruses in the Balkan Peninsula to provide a comprehensive Two sand fly–borne phleboviruses in the Old World view of the current situation and of the public health effect were historically associated with cases of sandfly fever: on humans and vertebrate animals in the region. Sicilian virus and Naples virus (7). Later, Naples virus was renamed sandfly fever Naples virus (SFNV), which Methods We searched global web-based resources (PubMed [www. Author affiliations: Unité des Virus Emergents, Marseille, France], Google Scholar [https://schol - (N. Ayhan, R.N. Charrel); University of Florida, Gainesville,], and Web of Science [https://isiknowl - Florida, USA (R.N. Charrel)]) to collect all the sand fly–borne phlebovirus data from the Balkan region. In addition, we investigated DOI: 2324 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Sand Fly–Borne Phleboviruses in the Balkan Region libraries and other national resources to identify books During World War II, sandfly fever affected great and conference reports that are not accessible on web- numbers of foreign soldiers in all Mediterranean region based resources. We used the keywords “sand fly,” and Balkan countries during the summer seasons, when “Phlebovirus,” “Bunyaviridae,” “Phenuiviridae,” “sand sand fly activity peaks ( 17). The disease was called phle- fly fever,” “pappataci fever,” “three-day fever,” “sand - botomus fever, pappataci fever, or three-day fever. In fly fever,” “Toscana virus,” “Sicilian virus,” “Naples 1937, a massive outbreak occurred in Athens, Greece virus,” “SFSV,” and “SFNV” matched with “Balkan,” (online Technical Appendix Table). After World War II, “Balkan Peninsula,” “Yugoslavia,” “Slovenia,” “Croa- sandfly fever epidemics were recorded in Belgrade, Ser - tia,” “Bosnia-Herzegovina,” “Macedonia,” “Republic of bia, affecting thousands of persons and then expanding Macedonia,” “Former Yugoslav Republic of Macedonia,” into other regions of the Balkans (27) (online Technical “FYROM,” “RoM,” “Albania,” “Bulgaria,” “Greece,” Appendix Table). Although these articles could be the “Montenegro,” “Romania,” “Moldova,” “Serbia,” and first record of sandfly fever based on clinical and epide - “Kosovo” for the research. After gathering all the data, we miologic grounds, there is no scientific evidence to dem - discarded the irrelevant publications. We put the collected onstrate whether the disease described in the articles was data in order based on country, year, and the phlebovirus sandfly fever caused by phlebovirus. species complex. The seminal seroprevalence study using a neutral- We obtained all the accessible virus sequences from ization assay, by Tesh et al. in 1976, showed that SFNV Balkan countries from GenBank. We aligned 589 nt partial and SFSV had circulated and were likely to continue to nucleoprotein sequences and analyzed them with MEGA infect human populations in the tested regions (online software version 6 ( We Technical Appendix Table). Also in 1976, Gligić et al. constructed a phylogenetic tree using the neighbor-joining isolated a strain of SFNV (Yug Bogdanovac virus strain method (Figure 1) and tested the robustness of each node Yu 8/76) from P. perfiliewi sand flies in the Dobrič re - by 1,000 bootstrap replicates. gion of Serbia. Other strains of SFNV and SFSV were also isolated in Serbia from P. pappatasi sand flies, but Results no accessible sequence data are available (24). At the We collected 51 published articles: 2 articles from Alba- time of isolation, SFNV and SFSV identification was nia, 7 from Bosnia-Herzegovina, 11 from Croatia, 17 from done using mouse hyperimmune ascitis fluid for neutral - Greece, 5 from Kosovo, 1 from Republic of Macedonia, ization assays and acetone sucrose antigens for comple- and 7 from Serbia (online Technical Appendix Table). One ment fixation tests. SFNV strain YU-8-76 is available in reference from Bulgaria was not available (online Techni- the Yale University catalog, now stored at the University cal Appendix Table). We found no published data from of Texas Medical Branch at Galveston. Partial sequence Montenegro and Romania. Most of the references included of this strain has been determined and confirmed the data concerning seroprevalence studies conducted in hu- strain as belonging to the Sandfly fever Naples virus spe- mans or animals (online Technical Appendix Table). Sev- cies (28,29). eral articles reported results about either virus characteriza- In 1985, Corfou virus, closely related to but distinct tion or case reports/outbreak investigations (19–25) (online from SFSV, was isolated from P. neglectus sand flies col- Technical Appendix Table). lected in the island of Corfou, Greece (online Technical Appendix Table). Corfou and SFSV can be distinguished Historical Data on Phleboviruses in the Balkans only by neutralization assays, unlike other serologic assays Alois Pick made a clinical description of sandfly fever in (ELISA, hemagglutination inhibition [HI], indirect immu- Bosnia-Herzegovina military barracks from foreign sol- nofluorescence [IIF], complement fixation [CF]). Several diers at the end of the 19th century (Online Technical Ap- studies have confirmed the presence of antibodies against pendix Table ). Pick, an Austro-Hungarian military doctor both SFNV and SFSV in several areas of the Balkans (20) working in Trebinje (Herzegovina), characterized the syn- (online Technical Appendix Table). drome observed in cases of sandfly fever (26). Sandfly fe- ver was observed both in local populations and in visitors, Toscana Virus in the Balkan Region specifically foreign soldiers. In 1904, Taussig noticed the In 1993, a German traveler was infected with TOSV af- presence of “pappataci” sand flies in army barracks in Her- ter visiting Athens; diagnosis was established from im- zegovina and conducted a large clinical and epidemiologic munofluorescence serology results, and it is therefore study in the region (27). The “endemic disease” emerged classified as a probable case rather than a laboratory- only in places where sand flies were present. Subsequently, confirmed case (online Technical Appendix Table). Re - the causative agent was discovered as a filterable agent (vi- cent serologic studies have provided evidence of TOSV rus) that used pappataci sand flies as a vector ( 18). presence in Bosnia-Herzegovina, Kosovo, Croatia, and Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2325 HISTORICAL REVIEW Greece (online Technical Appendix Table). Several hu- line Technical Appendix Table). Subsequently, TOSV man cases documented serologically as TOSV infections lineage C was detected from a patient in Greece (on- have been reported in Greece (23) (online Technical Ap- line Technical Appendix Table); unfortunately, the virus pendix Table). was not isolated in both cases and therefore only par- In Croatia, TOSV RNA was detected in the cere- tial sequence data are available. Later, sequences ob- brospinal fluid of a patient infected with meningitis; tained from P. neglectus sand flies confirmed the pres- sequence analysis showed that he was infected with a ence of lineage C TOSV but also showed that lineage strain belonging to a genetic lineage that had not been B TOSV was present and that both genetic types previously recognized (subsequently named lineage C), were sympatric in Croatia (Figure 2; online Technical which was clearly distinct from lineages A and B (on- Appendix Table). Figure 1. Phylogenetic relationships between sand fly–borne phleboviruses in the Old World based on 589 nt partial nucleoprotein sequence. Phylogenetic tree was constructed using the neighbor-joining method with MEGA software version 6 ( Black circles indicate viruses identified in the Balkan region. Boxes (top to bottom) include all viruses belonging to the Sandfly fever Naples virus species, viruses belonging to the Salehabad phlebovirus species, and viruses belonging to the Sandfly fever Sicilian and Corfou tentative virus species. The robustness of each node was tested by 1,000 bootstrap replicates. GenBank accession numbers are provided. Scale bar indicates nucleotide substitutions per site. 2326 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Sand Fly–Borne Phleboviruses in the Balkan Region Table. Characteristics of sand fly–borne phleboviruses in the Balkan region Probable Human/animal Virus Taxonomy Source Virus isolation Distribution vector species infections Sand fly fever Sandfly fever Naples Field-collected Yes, sand fly pools Serbia P. perfiliewi Yes Naples virus* virus species sand flies Corfou virus Sandfly fever Sicilian Field-collected Yes, sand fly pools Greece P. neglectus Probable virus species† sand flies Adria virus Salahabad virus Field-collected No, partial L Albania, Greece Phlebotomus Yes species sand flies, segment sequence spp. patient blood available Balkan virus Sandfly fever Naples Field-collected No, partial L and S Albania, Bosnia- P. neglectus Unknown virus species sand flies segment sequences Herzegovina, available Croatia Bregalaka virus Salahabad virus Field-collected Yes, sandfly pools Republic of P. perfiliewi Unknown species sand flies Macedonia Zaba virus Salahabad virus Field-collected Yes, sandfly pools Croatia P. neglectus Unknown species sand flies Toscana virus Sandfly fever Naples Field collected No, partial L and S Croatia, Greece P. neglectus Yes virus species sand-flies, CSF segment sequences available *Yug Bogdanovac virus strain Yu 4/76. †Tentative species. New Phleboviruses Identified from Partial Human and Animal Exposure to Phleboviruses Genomic Sequences SFSV and SFNV are both responsible for a febrile illness A novel phlebovirus, Adria virus, was detected in 2 pools that is self-limited but incapacitating, with signs that are of sand flies collected in Albania in 2005 (online Techni - commonly observed in arboviral diseases, such as fever, cal Appendix Table). Adria virus is most closely related to headache, malaise, photophobia, myalgia, and retroorbital Arbia virus, which was isolated in Italy (8); both belong to pain. From a clinical perspective, it is impossible to distin- the Salehabad phlebovirus species. Adria virus RNA was guish SFNV from SFSV infections, and also to discrimi- detected in the blood of a 2.5-year-old patient with febrile nate between SFNV/SFSV and other arboviral infections. seizures in Greece (online Technical Appendix Table). As mentioned previously, historic records were based on This evidence showed that a virus within the Salehabad clinical and epidemiologic evidence, but virological docu- phlebovirus species could be associated with human dis- mentation was lacking for studies before the 1950s (Figure ease (Figure 2). 2; online Technical Appendix Table). Balkan virus (BALKV) was detected from 2 pools Although SFSV and SFNV infections are clinically of P. neglectus sand flies in Albania in 2014, 1 pool from indistinguishable from each other, they are caused by ge- Bosnia-Herzegovina in 2014–15, and 4 pools from Croa- netically and antigenically different viruses. Infection with tia in 2015 (online Technical Appendix Table). Sequence SFNV does not induce cross-protection against SFSV and data analysis showed that BALKV belongs to the Sand- vice versa (33). As mentioned previously, neutralization fly fever Naples virus species, where it clusters with sub- test is the only technique that permits undisputable identi- group I together with Tehran, Zerdali, Fermo, and SFNV fication at the specific and intraspecific levels. Other tech - YU 8–76 viruses respectively discovered in Iran, Turkey, niques, such as ELISA, CF, HI, and IFA, which are prone Italy, and Serbia (30–32) (Figure 2; online Technical Ap- to cross-reactions, cannot achieve unambiguous identifica- pendix Table). tion at the intraspecific or at the interspecific level. Seroprevalence studies conducted in the Balkans from New Phleboviruses Identified from Complete 1976 onward have described antibodies in human popula- Genomic Sequences tions confirming exposure to several phleboviruses trans - Bregalaka virus (BREV) was isolated in P. perfiliewi mitted by sand flies. Complement-fixation tests showed sand flies from the Republic of Macedonia in 2015. Se - antibodies against SFNV in Bosnia-Herzegovina (online quence analysis demonstrated that BREV is most closely Technical Appendix Table), and HI tests showed anti- related to Adana virus, which was isolated in Turkey bodies against SFNV and SFSV in the islands of Croatia from field-collected sand flies in 2012. In Croatia, Zaba (Table; online Technical Appendix Table). In Greece, virus (ZABAV) was isolated from P. neglectus sand neutralizing antibodies against SFNV and SFSV were de- flies. ZABAV is most closely related to Adria virus and scribed; 36% of persons >30 of age showed positive results Salehabad virus. Both BREV and ZABAV belong to the for SFSV, and 13% showed positive results for SFNV. Salehabad phlebovirus species (online Technical Ap- Persons <30 years of age had much lower rates, suggest- pendix Table). ing that the antimalarial campaign had drastically reduced Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2327 HISTORICAL REVIEW the sand fly population and therefore the exposure to vi - In Kosovo, 9.6% of the 104 human serum samples tested ruses transmitted by sand flies (online Technical Appendix were positive for neutralizing antibodies against SFSV and Table). Presence of neutralizing antibodies against SFSV 27.9% of the serum samples were positive for neutralizing showed wide circulation (71.9%) in mainland and island antibodies against SFNV (online Technical Appendix Ta- regions of Greece in dogs used as sentinel animals; in the ble). With the same technique, 58.5% of cattle and 22.2% same study, TOSV and Arbia virus neutralizing antibod- of sheep were positive (online Technical Appendix Table). ies were also found at lower rates: 4.4% for TOSV and CF antibodies were found for SFNV in 19.4% of human se- 2.6% for Arbia virus (online Technical Appendix Table). rum samples in Serbia (online Technical Appendix Table). Figure 2. Current distribution of sand fly–borne phleboviruses in the Balkan region. A) Toscana virus, B) sandfly fever Naples virus, C) sandfly fever Sicilian virus, and D) Salehabad virus. Pictograms refer to virus isolation or sequence data demonstrating the presence of the virus in that area. Human/animal seroprevalence refers to studies reporting the presence of specific antibodies against the virus mentioned in the panel. 2328 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Sand Fly–Borne Phleboviruses in the Balkan Region TOSV was discovered in 1971, but it was identified as a Italy after the discovery of TOSV neurotropism. human pathogen 12 years later, which prevented early in- Corfou virus was isolated from P. major sand flies in clusion in the seroprevalence studies; thus, almost no data the eponymous Greek island (online Technical Appendix exist for TOSV before the 1990s. Table). Although Corfou/SFSV circulation was assessed Recent data have confirmed the circulation of TOSV by seroprevalence studies, Corfou virus remains the only and associated human cases in Kosovo, Greece, and Croatia. SFSV-like virus isolated in the Balkans. In Croatia, 2 risk factors were associated with TOSV posi- Adria virus was the first member of the Salehabad virus tive serology: living on an island and age (online Technical complex to be associated with human disease. Because of this Appendix Table). Possible presence of TOSV was assessed finding, and in light of the newly discovered viruses within in Bosnia-Herzegovina through immune-line assays (online this species (BREV and ZABAV), future actions should be Technical Appendix Table). TOSV neutralizing antibodies directed at implementing direct and indirect diagnosis of Sale- were detected in cats and dogs in Greece and in cattle and habad phlebovirus species in clinical microbiology laborato- sheep in Kosovo (Figure 2; online Technical Appendix Table). ries to better understand their potential public health impact. The fall of communism, the breakup of the former Yu- Discussion goslavia, and the following civil war and other climatic and Sand fly–borne diseases are widespread in the Balkan region environmental changes resulted in an increase of zoonotic because of the favorable climate and socioeconomic condi- infections that emerged or reemerged in the Balkans (28). tions in that area. After the first record of sandfly fever in As previously suggested for North Africa and in Turkey, it Bosnia-Herzegovina at the end of the 19th century (online is time to organize systematic testing of patients with CNS Technical Appendix Table), several outbreaks occurred in infections or unexplained febrile illness for such viruses in the whole Balkan region. Epidemics of sandfly fever and clinical microbiology laboratories in hospitals (35–37). leishmaniasis prompted faunistic and ecologic investigations In summary, when historical and recent data are com- of sand flies from 1947 through the 1970s ( 33). The number piled, it appears that the Balkan region is a hotspot for vi- of studies on sand fly fauna has decreased since that time ruses transmitted by sand flies, including those that cause because of the decline in recorded sandfly fever cases. For diseases in humans. The variety of different viruses is some Balkan countries, almost nothing is known about sand higher than in other regions that were investigated, and cer- fly distribution; when data are available, they are too old to tain areas display sympatric circulation of several viruses. reflect the current situation accurately. The collapse of the Circulation of these viruses must be assessed by studies former Yugoslavia and subsequent armed conflicts have also conducted in human populations and vertebrates, and diag- contributed to the lack of sustained studies on sand fly–borne nosis of human infections caused by sand fly–borne viruses pathogens in this region. However, recent data show that the must now be implemented using molecular and serologic Balkan region is still a major hotspot for arboviral diseases. tools in clinical microbiology laboratories. Most virus studies are based on serosurveillance. This work was supported in part by the European Virus Archive The seminal neutralization-based seroprevalence study, Goes Global (EVAg) project, which has received funding from performed by Tesh et al. in the 1970s, identified antibod- the European Union’s Horizon 2020 research and innovation ies against SFNV and SFSV in human populations from program under grant agreement no. 653316. Croatia, Greece, and Kosovo (online Technical Appendix Table). Successive studies confirmed the presence of anti- bodies against phleboviruses in most parts of the Balkans, About the Authors and recent serologic studies show the circulation of TOSV Dr. Ayhan is a postdoctoral fellow at the Institute of Research in Bosnia-Herzegovina, Kosovo, Croatia, and Greece (on- for Development in Marseille, France. Her primary research line Technical Appendix Table). interest is the study of phleboviruses transmitted by sand flies in SFNV Yu 8/76 was the first phlebovirus isolated in the the Old World. Balkans (online Technical Appendix Table). It was isolated from P. perfiliewi sand flies; before this finding, other vi- Dr. Charrel is a professor of virology at Aix Marseille ruses had been isolated from P. papatasi sand flies, which University, Marseille, France. His primary research interest were believed to be the unique vector competent for SFNV is in the field of arboviruses, with specific interest in viruses and SFSV (17). The recent discovery of BALKV should transmitted by sand flies, and, more generally, study of stimulate studies to address possible human pathogenicity mechanisms of viral emergence. (online Technical Appendix Table). Recent evidence for the presence of at least 2 different lineages of TOSV calls References for studies to measure its involvement in summer menin- 1. Elliott RM. Molecular biology of the Bunyaviridae. J Gen Virol. gitis and other neurologic infections, as were performed in 1990;71:501–22. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2329 HISTORICAL REVIEW 2. Adams MJ, Lefkowitz EJ, King AMQ, Harrach B, Harrison RL, 19. Gligić A, Mišcević Z, Tesh RB, et al. First isolations of Naples Knowles NJ, et al. Changes to taxonomy and the International sandfly fever virus in Yugoslavia. Acta Biol Jug Mikrobiol. Code of Virus Classification and Nomenclature ratified by the 1982;19:167–75. International Committee on Taxonomy of Viruses (2017). Arch Virol. 20. Borcić B, Punda V. Sandfly fever epidemiology in Croatia. Acta 2017;162:2505–38. Med Iugosl. 1987;41:89–97. 3. Steinhauer DA, Holland JJ. Rapid evolution of RNA viruses. 21. Hertig M, Sabin AB. Sandfly fever . In: Coates JB, editor. Annu Rev Microbiol. 1987;41:409–31. Preventive medicine in World War II, vol. VII. Communicable annurev.mi.41.100187.002205 diseases. Washington: US Government Printing Office; 1964. 4. Briese T, Calisher CH, Higgs S. Viruses of the family p. 109–74. Bunyaviridae: are all available isolates reassortants? Virology. 22. Papadopoulos O. Arbovirus problems in Greece. In: 2013;446:207–16. Vesenjak-Hirjan J. editor Arboviruses in the Mediterranean 5. Palacios G, Tesh R, Travassos da Rosa A, Savji N, Sze W, Jain K, countries. 6th FEMS symposium. Stuttgart: Gustav Fisher et al. Characterization of the Candiru antigenic complex Verlag; 1980. (Bunyaviridae: Phlebovirus), a highly diverse and reassorting 23. Papa A, Kontana A, Tsergouli K. Phlebovirus infections in group of viruses affecting humans in tropical America. J Virol. Greece. J Med Virol. 2015;87:1072–6. 2011;85:3811–20. jmv.24163 6. Freire CC, Iamarino A, Soumaré POL, Faye O, Sall AA, 24. Guelmino DJ, Jevtic M. An epidemiological and hematological Zanotto PM. Reassortment and distinct evolutionary dynamics of study of sandfly fever in Serbia. Acta Trop. 1955;12:179–82. Rift Valley fever virus genomic segments. Sci Rep. 2015;5:11353. 25. Drenski P, Drenski K. Contribution to the study of genus Phebotomus (Dipt.) and the three-day fever in Bulgaria. Proceed 7. Sabin AB. Experimental studies on Phlebotomus (pappataci, Bulg Entomol Soc. 1928;4:31–56. sandfly) fever during World War II. Arch Gesamte Virusforsch. 26. Pick A. On the pathology and therapy of a peculiar endemic 1860 1951;4:367–410. disease form [in German]. Wien Med Wochenschr. 1886;33:1141–5. 8. Verani P, Ciufolini MG, Caciolli S, Renzi A, Nicoletti L, 27. Taussig S. Die Hundskrankheit, endemischer Magenkatarrh in der Sabatinelli G, et al. Ecology of viruses isolated from sand flies Herzegowina. Wien Klin Wochenschr. 1905;50:164. in Italy and characterized of a new Phlebovirus (Arbia virus). 28. Hukić M, Numanović F, Sis̆ irak M, Moro A, Dervović E, Am J Trop Med Hyg. 1988;38:433–9. Jakovec S, et al. Surveillance of wildlife zoonotic diseases in the ajtmh.1988.38.433 Balkans region. Med Glas (Zenica). 2010;7:96–105. 9. Ehrnst A, Peters CJ, Niklasson B, Svedmyr A, Holmgren B. 29. Liu DY, Tesh RB, Travassos Da Rosa AP, Peters CJ, Yang Z, Neurovirulent Toscana virus (a sandfly fever virus) in Guzman H, et al. Phylogenetic relationships among members Swedish man after visit to Portugal. Lancet. 1985;325:1212–3. of the genus Phlebovirus (Bunyaviridae) based on partial M segment sequence analyses. J Gen Virol. 2003;84:465–73. 10. Dionisio D, Esperti F, Vivarelli A, Valassina M. Epidemiological, clinical and laboratory aspects of sandfly fever. Curr Opin Infect 30. Karabatos N. International catalogue of arboviruses including Dis. 2003;16:383–8. certain other viruses of vertebrates, third ed. San Antonio (TX): 00001432-200310000-00003 American Society of Tropical Medicine and Hygiene; 1985. 11. Charrel RN, Gallian P, Navarro-Mari JM, Nicoletti L, Papa A, Sán- 31. Alkan C, Erisoz Kasap O, Alten B, de Lamballerie X, Charrel RN. chez-Seco MP, et al. Emergence of Toscana virus in Europe. Emerg In- Sandfly-borne phlebovirus isolations from Turkey: New insight fect Dis. 2005;11:1657–63. into the sandfly fever Sicilian and sandfly fever Naples species. 12. Depaquit J, Grandadam M, Fouque F, Andry PE, Peyrefitte C. PLoS Negl Trop Dis. 2016;10:e0004519. Arthropod-borne viruses transmitted by Phlebotomine sandflies in journal.pntd.0004519 Europe: a review. Euro Surveill. 2010;15:19507. 32. Remoli ME, Fortuna C, Marchi A, Bucci P, Argentini C, 13. Gaidamovich SY, Khutoretskaya NV, Asyamov YV, et al. Sandfly Bongiorno G, et al. Viral isolates of a novel putative phlebovirus in fever in central Asia and Afghanistan. In: Calisher CH, editor. the Marche region of Italy. Am J Trop Med Hyg. 2014;90:760–3. Hemorrhagic fever with renal syndrome, tick- and mosquito-borne viruses. Basel (Switzerland): Springer; 1991. p. 287–93. 33. Sabin AB. Recent advances in our knowledge of dengue and 14. Al-Hazmi M, Ayoola EA, Abdurahman M, Banzal S, Ashraf J, sandfly fever. Am J Trop Med Hyg. 1955;4:198–207. El-Bushra A, et al. Epidemic Rift Valley fever in Saudi Arabia: a clinical study of severe illness in humans. Clin Infect Dis. 34. Simić Č, Živković V . Wildlife sandflies in Yugoslavia. I. The 2003;36:245–52. phlebotomids of Macedonia, southern Serbia and the Kosovo and 15. Çarhan A, Uyar Y, Özkaya E, Ertek M, Dobler G, Dilcher M, et al. Metohia region [in French]. Glas Srpske akademije nauka CXCIV, Characterization of a sandfly fever Sicilian virus isolated during Odeljenje medicinskih nauka. 1949;1:151–81. a sandfly fever epidemic in Turkey. J Clin Virol. 2010;48:264–9. 35. Alkan C, Allal-Ikhlef AB, Alwassouf S, Baklouti A, Piorkowski G, de Lamballerie X, et al. Virus isolation, genetic characterization 16. Ayhan N, Baklouti A, Prudhomme J, Walder G, Amaro F, and seroprevalence of Toscana virus in Algeria. Clin Microbiol Alten B, et al. Practical guidelines for studies on sandfly-borne - Infect. 2015;21:1040.e1–9. phleboviruses: part I: important points to consider ante field 2015.07.012 work. Vector Borne Zoonotic Dis. 2017;17:73–80. 36. Charrel R. The tree that hides the forest: not only West Nile virus, but also Toscana virus and Usutu virus. Vet Ital. 2017; 53:367–8. 17. Alkan C, Bichaud L, de Lamballerie X, Alten B, Gould EA, 37. Ergunay K, Ayhan N, Charrel RN. Novel and emergent sandfly- Charrel RN. Sandfly-borne phleboviruses of Eurasia and Africa: borne phleboviruses in Asia Minor: a systematic review. Rev Med epidemiology, genetic diversity, geographic range, control Virol. 2017;27:e1898. measures. Antiviral Res. 2013;100:54–74. Address for correspondence: Nazli Ayhan, Unite des Virus Emergents, 10.1016/j.antiviral.2013.07.005 18. Doerr R, Franz K, Taussing S. Das Papatatsi Fieber. Leipzig-Wien: School of Medicine, 27 blvd Jean Moulin, Marseille 13005, France; Franz Deuticke. 1909. email: 2330 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 DISPATCHES Isolation of Burkholderia pseudomallei from a Pet Green Iguana, Belgium Tom Hellebuyck, Pierre Wattiau, Filip Boyen, of the intrinsic resistance of the bacterium to many anti- Ilse Moeremans, Nancy H. Roosens, microbial drugs, combined with the inability to provide Kevin Vanneste, An Garmyn, Veronique Saey, appropriate medical care in disease-endemic developing Frank Pasmans, Freddy Haesebrouck countries, overall case-fatality rates might exceed 70% (1,4,9). B. pseudomallei is rarely reported in animals other We isolated Burkholderia pseudomallei, the causative than cattle, goats, and swine (10). agent of melioidosis, from liver granulomas of a pet green In reptiles, isolation of B. pseudomallei has been anec- iguana (Iguana iguana) in Belgium. This case highlights dotally documented in crocodiles (10), and clinical infec- a risk for imported green iguanas acting as a reservoir tions have been reported in 2 pet green iguanas in Cali- for introduction of this high-threat, zoonotic pathogen into fornia, USA (11) and a pet green iguana in Prague (Czech nonendemic regions. Republic) (4). Dermal abscesses were observed in the igua- na from the Czech Republic and in 1 of the iguanas from he highly pathogenic, gram-negative bacterium Burk- California, and hepatic masses were observed in the second Tholderia pseudomallei is the causative agent of melioi- iguana from California. dosis, which is endemic to countries in Southeast Asia and in Because of the variable clinical manifestations of northern Australia and an emerging infectious disease in sev- melioidosis and limited value of conventional bacterial eral tropical developing countries (1,2). Human cases in Eu- methods for identification of B. pseudomallei, diagnosis of rope are limited to patients who traveled to disease-endemic melioidosis can be challenging (2,11). If one considers the regions. In Belgium, the last case was documented in 2013 highly pathogenic and zoonotic nature of B. pseudomallei, in a 44-year-old man who had traveled to Madagascar (3). use of appropriate molecular detection methods is crucial B. pseudomallei is classified as a tier 1 overlap select agent to warrant correct identification and discrimination of B. by the US Federal Select Agent Program (4). Postexposure pseudomallei from other Burkholderia species (11). Next- prophylaxis and postexposure monitoring should be planned generation sequencing might be a valuable supplement to for persons who have had high-risk exposures, such as cer- current identification and diagnostic methods. tain laboratory procedures with the organism that were not conducted under Biosafety Level 3 conditions (5). The Study Infected humans and importing of infected animals can A 5-year-old female green iguana (Iguana iguana) showed introduce melioidosis into nonendemic areas (2,6). Howev- acute onset of lethargy, anorexia, and general weakness. er, importing of infected animals has not yet been associat- The iguana had been purchased 4.5 years earlier by private ed with epizootic transmission. Infection commonly occurs owners from a pet shop in the Netherlands that imported through cutaneous inoculation, ingestion of contaminated the iguana from a captive breeding operation in Central soil or water, or inhalation of aerosolized bacteria (7,8). Al- America. Serum biochemical and hematologic tests showed though the incubation period in humans is typically 1–21 hyperuricemia, hyperphosphatemia, an increased level of days, clinical disease might develop years after infection aspartate aminotransferase, hyperglobulinemia, nonregen- (9). The incubation period in naturally infected animals is erative anemia, and severe leukocytosis in comparison not known (10). with physiologic reference ranges for these conditions (12). Acute melioidosis results predominantly in pneumo- Ultrasonography showed hepatomegaly, multiple hyper- nia and septicemia. Chronic infection is associated with echoic hepatic masses, and severely enlarged, hyperechoic abscesses of the liver, lungs, spleen, and skin (9). Because kidneys. A presumptive diagnosis of hepatitis, kidney fail- ure, and septicemia was made. Because the iguana did not Author affiliations: Ghent University, Merelbeke, Belgium respond to supportive treatment and its general condition (T. Hellebuyck, F. Boyen, I. Moeremans, A. Garmyn, V. Saey, continued to deteriorate, the owners agreed to euthanize the F. Pasmans, F. Haesebrouck); Veterinary and Agrochemical iguana on the fifth day after signs began. Research Centre, Brussels, Belgium (P. Wattiau); Sciensano, During necropsy, pronounced renomegaly and hepato- Ixelles, Belgium (N.H. Roosens, K. Vanneste) megaly, as well as granulomatous hepatitis was observed. DOI: Histologic evaluation of hematoxylin and eosin–stained Emerging Infectious Diseases • •Vol. 24, No. 12, December 2018 2331 DISPATCHES sections of liver showed coalescent granulomas throughout with other imported animals (e.g., in pet shops or in captive the liver parenchyma, with central necrosis surrounded by ac- collections in nonendemic countries) cannot be excluded. tivated macrophages and giant cells (Figure). Gram, periodic After we identified the isolate as B. pseudomallei, we acid–Schiff, and acid-fast staining did not show intralesional collected 100 mL of water from a plastic water bath and a gram-positive or acid-fast bacteria or fungi. Microbiological swab specimen from the emptied water reservoir from the examination of liver tissue yielded a pure and abundant cul- housing of the iguana and tested these samples by using ture of small colonies of gram-negative rods that could not be real-time PCR. All samples showed negative results. identified by using standard biochemical identification tests. The zoonotic potential of the bacterium was dis- Sequencing of the 16S rRNA gene of the isolate cussed with the owners and personnel who came into con- showed 100% identity with B. pseudomallei (1,382 bp). tact with the iguana or samples that were collected from Sequence data were deposited in GenBank under acces- the lizard. Although the owners did not have a clinical sion no. MF523223. After culturing of the isolate under history that could indicate B. pseudomallei infection, they Biosafety Level 3 conditions, we performed multilocus were advised to consult with their physicians about poten- sequence typing–derived PCR and real-time PCR as de- tial exposure. Although the owners refused postexposure scribed by Wattiau et al. (13), which confirmed identifica- prophylaxis and postexposure monitoring, postexposure tion of the isolate as B. pseudomallei. prophylaxis was given to a selected number of staff mem - We used next-generation sequencing to obtain the ge- bers as recommended by Lipsitz et al. (5) because of po- nome sequence of the isolate, which was deposited in the tential exposure to the pathogen. Short Read Archive under accession no. SRR6056996. The genome sequence was used to infer its multilocus se- Conclusions quence type (ST) as ST518. Isolate information was depos- Previously reported B. pseudomallei–infected iguanas (4,11) ited accordingly in the B. pseudomallei database (https:// and the 1 reported in this study were presumably imported under ID 5121 (14). Of 5,600 deposited from disease-endemic regions, highlighting the potential isolates that we accessed on June 18, 2018, ST518 matched role of this species as a reservoir of B. pseudomallei. On the only with B. pseudomallei isolates (ID 3330) obtained from basis of the relatedness of ST518 with an isolate obtained a human who was infected in Costa Rica in 2009 (ID 1928) from a human in Costa Rica (14) and with 2 isolates from (14) and with 2 isolates obtained from abscesses in pet igua- iguanas that presumably became infected in Central America nas in 2007 (ID 5008) and 2012 (ID 3330) (11). We found (11), whose genomes were later confirmed to be closely re - that ID 3330 differed from the isolates from Central Amer - lated (15), our findings support the hypothesis that the iguana ica by only 3 single-nucleotide polymorphisms, which we report became infected in the captive breeding facility in strongly suggested that the iguanas were infected in Central Central America before importation to Europe. America before transport abroad (15). However, whether If one considers the long incubation period observed captive iguanas might acquire infections through contact for green iguanas, these reptiles might shed the bacterium unnoticed for years before nonspecific chronic disease develops. Transmission from infected captive iguanas to humans might occur through contact with stool, infected tissue, or biting and scratching lesions. Indirect infection might originate from wound infection through contaminat- ed soil or water or inhalation of aerosolized bacteria. In conclusion, traders, (para)veterinarians, and labora- tory staff members who handle green iguanas or samples obtained from this species are susceptible to infection. In addition, physicians who are consulted by pet green iguana owners should be aware that these animals and their envi- ronment could potentially harbor B. pseudomallei. About the Author Dr. Hellebuyck is head of the clinic at the Division of Poultry, Exotic Animals, Wildlife, and Experimental Animals, Figure. Coalescent granuloma in liver parenchyma of a pet green Department of Pathology, Bacteriology and Avian Diseases, iguana (Iguana iguana) infected with Burkholderia pseudomallei, Faculty of Veterinary Medicine, Ghent University, Merelbeke, Belgium. Hematoxylin and eosin stain shows central necrosis Belgium. His research interests are infectious and noninfectious surrounded by activated macrophages and giant cells. Scale bar indicates 200 µm. diseases in reptiles, amphibians, and birds. 2332 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 B. pseudomallei from a Pet Green Iguana References 1. Aldhous P. Tropical medicine: melioidosis? Never heard of it.... EID Podcast: Nature. 2005;434:692–3. 2. Limmathurotsakul D, Golding N, Dance DA, Messina JP, Pigott DM, Moyes CL, et al. Predicted global distribution Visions of Matchstick of Burkholderia pseudomallei and burden of melioidosis. Nat Microbiol. 2016; 1:15008. Men and Icons of nmicrobiol.2015.8 3. ProMED-mail. Melioidosis—Belgium ex Madagascar. 2013 [cited 2018 Aug 1]., archive no. Industrialization 20130503.1687746 4. Elschner MC, Hnizdo J, Stamm I, El-Adawy H, Mertens K, Byron Breedlove, managing editor of Melzer F. Isolation of the highly pathogenic and zoonotic agent Burkholderia pseudomallei from a pet green iguana in Prague, the journal, discusses and reads his Czech Republic. BMC Vet Res. 2014;10:283–7. 10.1186/s12917-014-0283-7 November 2017 cover art essay. 5. Lipsitz R, Garges S, Aurigemma R, Baccam P, Blaney DD, This cover (Going to Work, 1943) is Cheng AC, et al. Workshop on treatment of and postexposure prophylaxis for Burkholderia pseudomallei and B. mallei infection, by English artist Laurence Stephen 2010. Emerg Infect Dis. 2012;18:e2. eid1812.120638 Lowry (1887–1976) who died of 6. Galimand M, Dodin A. Pseudomonas pseudomallei distribution pneumonia in 1976. in France and worldwide melioidosis [in French]. Bulletin de la Societé Vétérinaire Pratique de France. 1982;66:651–7. 7. Cheng AC, Jacups SP, Gal D, Mayo M, Currie BJ. Extreme weather events and environmental contamination are associated with case-clusters of melioidosis in the Northern Territory of Australia. Int J Epidemiol. 2006;35:323–9. dyi271 8. Limmathurotsakul D, Kanoksil M, Wuthiekanun V, Kitphati R, deStavola B, Day NP, et al. Activities of daily living associated with acquisition of melioidosis in northeast Thailand: a matched case–control study. PLoS Negl Trop Dis. 2013;7:e2072. 9. Limmathurotsakul D, Peacock SJ. Melioidosis: a clinical overview. Br Med Bull. 2011;99:125–39. bmb/ldr007 10. Sprague LD, Neubauer H. Melioidosis in animals: a review on epizootiology, diagnosis and clinical presentation. J Vet Med B Infect Dis Vet Public Health. 2004;51:305–20. 10.1111/j.1439-0450.2004.00797.x 11. Zehnder AM, Hawkins MG, Koski MA, Lifland B, Byrne BA, Swanson AA, et al. Burkholderia pseudomallei isolates in 2 pet iguanas, California, USA. Emerg Infect Dis. 2014;20:304–6. 12. Divers SJ, Redmayne G, Aves EK. Haematological and biochemical values of 10 green iguanas (Iguana iguana). Vet Rec. 1996;138:203–5. 13. Wattiau P, Van Hessche M, Neubauer H, Zachariah R, Wernery U, Imberechts H. Identification of Burkholderia pseudomallei and related bacteria by multiple-locus sequence typing-derived PCR and real-time PCR. J Clin Microbiol. 2007;45:1045–8. 14. Aanensen DM, Spratt BG. The multilocus sequence typing network: Nucleic Acids Res. 2005;33:W728-33. Visit our website to listen: 15. Gee JE, Gulvik CA, Elrod MG, Batra D, Rowe LA, Sheth M, et al. Phylogeography of Burkholderia pseudomallei isolates, Western Hemisphere. Emerg Infect Dis. 2017;23:1133–8. podcasts/player.asp?f=8647173 Address for correspondence: Tom Hellebuyck, Department of Pathology, Bacteriology and Avian Diseases, Faculty of Veterinary Medicine, Ghent University, Salisburylaan 133, B-9820 Merelbeke, Belgium; email: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2333 DISPATCHES Neglected Hosts of Small Ruminant Morbillivirus calculate whether PPRV RNA loads in secretions and ex- Claudia Schulz, Christine Fast, Kore Schlottau, cretions (oral, nasal, and fecal swab samples) collected Bernd Hoffmann, Martin Beer over time from pigs, wild boar, goats, and sheep differed Eradication of small ruminant morbillivirus (PPRV) is target- significantly and to determine correlations between the ed for 2030. PPRV lineage IV is found in much of Asia and results of virus isolation and PCR assays by using swab Africa. We used PPRV lineage IV strain Kurdistan/2011 in samples and purified leukocytes as sample materials (on- transmission trials to investigate the role of pigs, wild boar, line Technical Appendix, and small ruminants as PPRV reservoirs. Suids were a pos- article/24/12/18-0507- Techapp1.pdf). sible source of infection. As expected, goats showed the typical moderate to severe clinical signs (trials 1 and 3) reported previously este des petits ruminants is one of the most serious (7–9). Clinical signs in PPRV-infected sheep (trial 4) were P(economically and clinically) transboundary animal generally mild to moderate, as previously described (3,8). diseases (1–3). Of 4 lineages, small ruminant morbilli- Contact controls showed similar clinical signs. One PPRV- virus (previously called peste des petits ruminants virus; infected sheep showed severe clinical signs similar to those PPRV) lineage IV (PPRV-LIV) has spread the most widely of the goats. Surprisingly, all PPRV infected pigs and wild in the past decade, particularly in Asia, and increasingly boar (trials 1–3) showed various mild to moderate clinical dominates the PPRV lineages in Africa (2,4). Morbidity signs, including fever and reduced general condition (all and mortality rates for goats are high, up to 100%; how- suids), diarrhea (pig 1, boar 1–4), and ocular (pigs 1–3) and ever, sheep can be subclinically infected and play a ma- nasal (boar 4) discharge typical for PPRV infection (Fig- jor role in the silent spread of PPRV over large distances ure 1; Figure 2, panels A–C; online Technical Appendix). and across borders (3). The role of other wild and domestic PPRV-induced immunosuppression may predispose af- Artiodactyls in the epidemiology of PPRV is unknown or fected animals to secondary infections (3,9) as reflected by insufficiently understood (3). Pigs are considered dead-end distinct severe leukocytopenia in pigs and goats a few days hosts for PPRV (5). In an experimental infection study, pigs after inoculation. Different expressions of clinical signs af- infected with PPRV lineage II (LII) strains did not trans- ter PPRV infection might have been caused by concurrent mit PPRV to goats, but they can transmit the closely re- infections with other pathogens or differences in individual lated Rinderpest morbillivirus to cattle (6). To determine resistance to PPRV infection (9). In the 4 wild boar, for the pathogenesis of PPRV-LIV infection in pigs and wild example, Balantidium coli, detected by histopathologic ex- boar and the capability of these suids to transmit PPRV in amination (data not shown), might have been an additional comparison with that of goats and sheep, we conducted 4 factor causing the diarrhea (10). Nevertheless, similar to independent transmission trials during 2015–2016. The ex- the lack of clinical signs reported for pigs infected with a perimental protocol was reviewed by a state ethics com- PPRV-LII strain (6), the 2 pigs in trial 3 showed only mild mission and approved by the State Office for Agriculture, clinical signs. Food Safety and Fisheries of Mecklenburg-Vorpommern, Contact transmission of PPRV from intranasally in- Rostock, Germany (LALLF M-V/TSD/7221.3-1-018/14). fected pigs to 1 contact goat and 1 pig was noted (trial 1). This pig was refractory to intranasal infection but was The Study infected by contact at a similar time as one of the goats. In 4 trials (trials 1–4; Table), we intranasally inoculated Furthermore, PPRV was transmitted from intranasally suids with a recent PPRV-LIV strain (Kurdistan/2011 infected goats to contact pigs (trial 3) (Table). Hence, in [7,8]). Contact control animals were added 2 days later. contrast to the findings of Nawathe and Taylor (6), who We recorded clinical signs and temperature regularly and reported contact transmission of a PPRV-LII strain from collected samples to evaluate the suitability of different experimentally infected goats to contact pigs but not vice virologic, serologic, and pathological methods for detect- versa, our transmission trials demonstrated that a complete ing PPRV infection. We conducted statistical analyses to interspecies transmission cycle of a PPRV-LIV strain be- tween goats and pigs, and possibly also intraspecies trans- Author affiliation: Friedrich-Loeffler-Institut, Greifswald–Insel mission between pigs, can be maintained. The virulence of Riems, Germany Current affiliation: University of Veterinary Medicine Hannover, DOI: Hannover, Germany. 2334 Emerging Infectious Diseases • •Vol. 24, No. 12, December 2018 Neglected Hosts of Small Ruminant Morbillivirus Table. Design and outcomes of PPRV transmission trials, Germany* Outcomes No. Seroconversion, Excretion of Excretion of Contact transmission (no. Trial inoculated No. contact total no. by PPRV RNA, total infectious PPRV, contact-infected/total no. no.† Trial animals controls species no. by species total no. by species in contact) 1 P-GP 3P‡ 2G, 1P‡ 3P,‡ 2G 3P,‡ 2G 1P, 2G Yes (1/2G;§ 1/1P‡) 2 W-GP 4W 2G, 2P 4W 4W 2W No (0/2G; 0/2P) 3 G-P 2G 2P 2G, 2P 2G, 2P 2G Yes (2/2P) 4¶ S-S 5S 5S 5S 5S 5S No (0/5S) *P, pig; PPRV, small ruminant morbillivirus (formerly called peste des petits ruminants virus); W, wild boar; G, goat; GP, goats and pigs; S, sheep. †For trials 1–3, animals were experimentally infected by intranasal inoculation with PPRV strain Kurdistan/2011 for independent transmission trials conducted in the containment facility of the Friedrich-Loeffler-Institut, Isle of Riems, Germany. Contact control animals were added 2 d after experimental infection. In 2 of the trials, PPRV transmission was documented from pigs to 1 goat and 1 pig (trial 1) and from goats to 2 pigs (trial 3). Infectious PPRV excretion was detected in >1 animal of each species, and PPRV RNA and seroconversion were detected in all experimentally infected or contact-infected animals (further details in online Technical Appendix Figure 1, For trial 4, a 1-to-1 (pairwise) study design was chosen to estimate the reproductive ratio. The results of the sheep trial are presented in this study to enable comprehensive comparison with the PPRV pathogenesis in suids. ‡One of 3 pigs was probably not infected by experimental intranasal PPRV inoculation but by contact infection. §One contact goat was infected by pigs; however the source of infection (goat or pig) cannot be determined for the second contact goat. ¶In each of 5 stables, 2 sheep were kept together: 1 experimentally infected sheep and 1 contact control sheep. the PPRV lineage or strain is possibly a factor influencing (Figure 1). Statistically significantly higher PPRV RNA the susceptibility to PPRV infection and the probability of loads over time were found in PPRV-infected goats than PPRV transmission (9,11). in suids and sheep. Peak viral loads in goat samples were From 2 of 4 wild boar (trial 2), PPRV was isolated up to 1 log step (PCR) and 2.5 log steps (virus isolation) 7 6.0 from a few fecal swab samples but was not transmitted to higher (9.3 × 10 copies/mL; 10 TCID [50% tissue cul- the contact goats or pigs. Unexpectedly, none of the in- ture infective dose]/mL) than in pig and wild boar samples 7 3.5 tranasally infected sheep transmitted PPRV to any of the (1.5 × 10 copies/mL; 10 TCID /mL). Of note, peak contact sheep. The considerable differences in transmis - viral loads in sheep (10 TCID /mL) were only slightly sion efficiency between goats and the other Artiodactyls higher than those in pigs and wild boar, which may explain can be explained by higher PPRV loads excreted by goats why none of the sheep transmitted PPRV to the contact Figure 1. Progression of virologic, serologic, and clinical parameters analyzed in pigs (A), wild boar (B), goats (C), and sheep (D) in Germany after experimental infection with PPRV lineage IV strain Kurdistan/2011. Results are shown for reverse transcription quantitative PCR (solid black lines), endpoint dilution assay (dashed black lines), competitive ELISA (dark gray lines), and clinical score sheets (light gray lines). A detailed description of the infection experiment is provided in the online Technical Appendix ( article/24/12/18-0507-Techapp1. pdf). Abs, antibodies; CS, clinical signs; pi, postinfection; PPRV, small ruminant morbillivirus (formerly called peste des petits ruminants virus). Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2335 DISPATCHES Figure 2. Clinical signs observed in wild boar and pigs and small ruminant morbillivirus (formerly called peste des petits ruminants virus; PPRV) antigen detection in a pig tonsil in experimental study of PPRV transmission, Germany. A) Purulent nasal discharge in wild boar 4 at 8 days after infection; B) diarrhea in wild boar 4 at 7 days after infection; C) swollen eyelids in pig 3 at 10 days after infection; D) PPRV antigen (red) in the tonsil of pig 1 at 30 days after infection (≈22 days after contact infection of pig 1), by immunohistochemical staining with monocloncal mouse anti-PPRV; scale bar indicates 50 µm. Clinical signs in the 3 pigs in trial 1 included a transient rise in body temperature, ruffling bristles, diarrhea, reduced activity and food intake/ slight emaciation, swelling of the eyelids, mild to severe conjunctivitis, and mucous to purulent ocular discharge in the first days after infection. In the 4 wild boar in trial 2, clinical signs included transiently increased body temperature, diarrhea (including fresh blood), reduced general condition, inappetence, and mucopurulent nasal discharge. A detailed description of the infection experiments is provided in the online Technical Appendix ( control sheep. The higher viral loads in goats could also might have been associated with PPRV-infected immune explain the ≈4 days earlier contact infection of the contact cells, as has been reported for ferrets infected with closely pigs in trial 3 than the contact goat and pig in trial 1. Be- related canine morbillivirus (14). Immunohistochemistry sides a higher innate susceptibility to PPRV infection ob- demonstrated that PPRV antigens in other tissues were served for goats compared with sheep and suids, the infec- often associated with immune cells. For PPRV diagno- tive dose may play a role in the efficiency of transmission sis in the examined species, tissue of the lymphoreticular and infection dynamics of PPRV in suids as previously system, in particular tonsils (Figure 2, panel D), head and reported for goats (9) and camelids (12). lung–associated lymph nodes, mesenteric lymph nodes, We detected seroconversion in all PPRV-infected and small intestinal Peyer’s patches, were found most animals by using competitive ELISA and neutralization suitable for postmortem diagnosis with PCR and immu- tests. Neutralizing antibody titers were moderate to high nohistochemistry. PCR was the most sensitive virologic in suids and goats (2.16–2.96 log ND [virus neutral- method independent from the sample material, and com- 10 50 ization in 50% of replicates]) and slightly lower in sheep petitive ELISA proved reliable for serologic PPRV diag- (1.76–2.56 log ND ). After seroconversion, no PPRV nosis (online Technical Appendix). 10 50 could be isolated from swab and purified leukocyte sam - ples, but PPRV RNA was detected in swab samples for at Conclusions least 3–4 weeks after infection in all species, with indi- Transmission trials with a virulent PPRV-LIV strain re- vidual differences (Figure 1; online Technical Appendix). vealed that suids are an unexpected possible source for Correlation analyses revealed a poor to excellent correla- PPRV infection. Therefore, domestic pigs and wild boar tion of PCR and virus isolation results before seroconver- should be considered as possible PPRV reservoir hosts. sion, depending on the animal species. Possible reasons This finding is especially relevant to stringent control pro- for (transient) PPRV RNA persistence are infection of grams. The epidemiologic role of suids in the spread of neurons followed by transsynaptic spread (13). Indeed, PPRV, as maintenance or spillover hosts (15), should be PPRV RNA was detected in single or multiple brain sam- further investigated in field and experimental studies using ples of 2 sheep, 4 goats, and in the choroid plexus of 1 pig, different PPRV lineages and strains at different environ- 1 wild boar, and 3 goats. PPRV RNA in the choroid plexus mental and experimental conditions. 2336 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Neglected Hosts of Small Ruminant Morbillivirus Animal_Health_in_the_World/docs/pdf/Disease_cards/PESTE_ Acknowledgments DES_PETITS_RUMINANTS.pdf We thank Holger Freese, Christian Korthase, Isabell Metz, 6. Nawathe DR, Taylor WP. Experimental infection of domestic pigs Karin Pinger, Anja Schulz, and Susanne Zahnow for their with the virus of peste des petits ruminants. Trop Anim Health excellent technical assistance; the staff of the experimental Prod. 1979;11:120–2. 7. Hoffmann B, W iesner H, Maltzan J, Mustefa R, Eschbaumer M, animal facilities for great assistance and dedicated animal care; Arif FA, et al. Fatalities in wild goats in Kurdistan associated with Sandra Blome, Bernd Haas, and Jana Pietschmann for scientific peste des petits ruminants virus. Transbound Emerg Dis. 2012; and organizational advice; Reiner Ulrich and Jan Schinköthe 59:173–6. for excellent support with gross pathology; and Jörn Gethmann, 8. Wernike K, Eschbaumer M, Breithaupt A, Maltzan J, Wiesner H, Beer M, et al. Experimental infection of sheep and goats with a Mario Ziller, Carola Sauter-Louis, and Hans-Hermann Thulke recent isolate of peste des petits ruminants virus from Kurdistan. for biostatistical advice. Vet Microbiol. 2014;172:140–5. j.vetmic.2014.05.010 We also thank the European Commission Animal Health, 9. Pope RA, Parida S, Bailey D, Brownlie J, Barrett T, the Welfare European Research Area Network Fund, and the Banyard AC. Early events following experimental infection German Federal Office for Agriculture and Food (grant no. with peste-des-petits ruminants virus suggest immune cell 2813ERA074) for providing support for epidemiologic research targeting. PLoS One. 2013;8:e55830. journal.pone.0055830 under the Improved Understanding of the Epidemiology of 10. Deplazes P, Eckert J, Zahner H. Lehrbuch der Parasitologie für die Peste-des-Petits Ruminants project. Tiermedizin. 3rd ed. Stuttgart (Germany): EnkeVerlag; 2012. 11. Couacy-Hymann E, Bodjo C, Danho T, Libeau G, Diallo A. About the Author Evaluation of the virulence of some strains of peste-des-petits- ruminants virus (PPRV) in experimentally infected West African Dr. Schulz is a postdoctoral researcher at the University of dwarf goats. Vet J. 2007;173:178–83. Veterinary Medicine Hannover, Germany. Her primary research j.tvjl.2005.08.020 interests are the pathogenesis and epidemiology of emerging and 12. Wernery U. Peste des petits ruminants (PPR) in camelids with own vectorborne diseases in wild and domestic animals. investigations. J Camel Pract Res. 2011;18:219–23. 13. Lin WH, Kouyos RD, Adams RJ, Grenfell BT, Griffin DE. Prolonged persistence of measles virus RNA is characteristic of primary infection dynamics. Proc Natl Acad Sci U S A. 2012;109: References 14989–94. 1. Banyard AC, Parida S, Batten C, Oura C, Kwiatek O, Libeau G. 14. Ludlow M, Rennick LJ, Nambulli S, de Swart RL, Duprex WP. Global distribution of peste des petits ruminants virus and prospects Using the ferret model to study morbillivirus entry, spread, for improved diagnosis and control. J Gen Virol. 2010;91:2885–97. transmission and cross-species infection. Curr Opin Virol. 2014;4:15–23. 2. Munir M. Role of wild small ruminants in the epidemiology of 15. Palmer MV, Thacker TC, Waters WR, Gortazar C, Corner LA. peste des petits ruminants. Transbound Emerg Dis. 2014;61:411–24. Mycobacterium bovis: a model pathogen at the interface of 3. European Food Safety Authority. Scientific opinion on peste des livestock, wildlife, and humans. Vet Med Int. 2012;2012. petits ruminants: EFSA Panel on Animal Health and Welfare (AHAW). EFSA J. 2015;13:3985. 4. Baron MD, Parida S, Oura CA. Peste des petits ruminants: a suitable candidate for eradication? Vet Rec. 2011;169:16–21. Address for correspondence: Claudia Schulz, University of Veterinary Medicine Hannover, Bünteweg 17, 30539 Hannover, Germany; email: 5. World Organisation for Animal Health. Peste des petits ruminants [cited 2018 Oct 5]. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2337 DISPATCHES Vaccinia Virus among Domestic Dogs and Wild Coatis, Brazil, 2013–2015 Galileu Barbosa Costa, Lara Ribeiro de Almeida, To determine the potential role of domestic and wild Aline Gabriele Ribeiro Cerqueira, animals as a source of VACV infection for humans, we Wander Ulisses Mesquita, Jaqueline Silva investigated VACV circulation among domestic dogs and de Oliveira, Júlia Bahia Miranda, Ana Teresa wild coatis, animals that live at the intersection of urban Saraiva-Silva, Jônatas Santos Abrahão, and wild environments in Brazil. The capture of wild ani- Betânia Paiva Drumond, Erna Geessien Kroon, mals was authorized by the Brazilian Institute of Environ- Pedro Lúcio Lithg Pereira, Danielle Ferreira de ment and Renewable Natural Resources, and the study was Magalhães Soares, Giliane de Souza Trindade approved by the Ethics Committee in Animal Experimenta- tion of Universidade Federal de Minas Gerais. To determine their potential role as a source of human infec- tion, we tested domestic dogs (urban) and wild coatis (wild) The Study in Brazil for vaccinia virus. Our findings of positive neutraliz - We analyzed serum and anal swab samples collected dur- ing antibodies and quantitative PCR results for 35/184 dogs ing 2013–2015 from 184 domestic dogs and 90 wild coatis and 13/90 coatis highlight a potential public health risk. in the city of Belo Horizonte (19°55′15′′S, 43°56′16′′W) in the state of Minas Gerais, Brazil (Figure 1). Swab samples ince smallpox was declared eradicated in 1980, after a of lesions, if present, were also collected. To determine the Smassive effort led by the World Health Organization, presence of neutralizing antibodies in serum, we used an other orthopoxviruses have gained notoriety as zoonotic orthopoxvirus plaque reduction neutralization test as previ- agents worldwide (1). Over the past 17 years in Brazil, ously described (12). Serum titers were defined as the high- many zoonotic outbreaks of vaccinia virus (VACV) infec- est dilutions that inhibited >70% of virus plaques compared tion have been recorded throughout the country, becoming with negative controls. a burden for the dairy industry and public health (2). The To detect VACV DNA from serum and anal swab most affected hosts during outbreaks are dairy cattle and samples, we performed real-time PCR targeting the C11R humans (2). Recent studies assessing the role of wildlife or A56R gene (12). We directly sequenced A56R fragments in the maintenance cycle of VACV in nature have corrobo- in both orientations and in triplicate by using the ABI3130 rated previous findings that rodents and marsupials serve platform (Applied Biosystems, Waltham, MA, USA). Se- as links between natural and anthropic environments (2–4). quences were aligned with other reference sequences from Indeed, the increased frequency of reported VACV de- GenBank by using MEGA 7.0 (http://www.megasoftware. tection in several species of mammals points toward new net). Statistical analyses were conducted by using Epi Info insights into the circulation and maintenance of VACV in software version (; wild (forest) and rural (farm) environments (2,5–8). Stud- χ and Fisher exact tests were applied with significance ies conducted in Latin America suggest that wildlife, espe- set at 5%. We also calculated relative odds ratios (ORs) cially small and medium-sized mammals, plays a role in and 95% CIs. virus transmission and maintenance of orthopoxviruses in We detected orthopoxvirus neutralizing antibodies in nature (9). Furthermore, some studies have shown the pres- 35 dogs (prevalence rate 19.0%, 95% CI 14.0%–25.5%; ti- ence of VACV in urban environments, emphasizing the ters 100–400 neutralizing units/mL) and in 13 coatis (prev- risks for humans (especially those not vaccinated against alence rate 14.4%, 95% CI 8.5%–23.3%; titers 100–800 smallpox) (10,11). neutralizing units/mL) (Table 1). Univariate analyses indi- cated significant associations between presence of neutral - Author affiliations: Universidade Federal de Minas Gerais, Belo izing antibodies and the following: male dogs (OR 2.6; p Horizonte, Brazil (G.B. Costa, L.R. de Almeida, A.G.R. Cerqueira, = 0.02), dogs 6–10 years of age (OR 5.2; p = 0.04), coatis J.S. de Oliveira, J.B. Miranda, A.T. Saraiva-Silva, J.S. Abrahão, captured in 2013 (OR 11.2; p = 0.002), juvenile coatis (<1 B.P. Drumond, E.G. Kroon, P.L.L. Pereira, D.F. de Magalhães y of age) (OR 35; p = 0.001), and adult coatis (>2 y of age) Soares, G. de Souza Trindade); Universidade Federal de Ouro (OR 5.1; p = 0.04). Preto, Ouro Preto, Brazil (W.U. Mesquita) Samples from all seropositive animals were submitted DOI: for quantitative PCR (qPCR) to detect VACV DNA (Table 2338 Emerging Infectious Diseases • •Vol. 24, No. 12, December 2018 Vaccinia Virus among Dogs and Coatis, Brazil Figure 1. Area of study of vaccinia virus among domestic dogs and wild coatis, Brazil, 2013–2015. A) Countries in South America where vaccinia virus has been detected in recent years. B) Belo Horizonte (red locator), located in Minas Gerais state, Brazil. C) Regions of Belo Horizonte; green indicates area in wild environment where coatis were captured. D) Google Earth map from 2017 of studied area, showing details of the wild and urban environments. Green dots indicate where coatis were captured; blue dots indicate where dogs were sampled. Source: 2). Overall, serum samples from 7 dogs and 6 coatis were human cowpox virus infections have mostly occurred in positive for the C11R gene; of these, anal swab samples urban areas of Europe. Cowpox virus is transmitted to hu- were positive for this gene for 3 dogs and 4 coatis. Samples mans mainly by domestic cats that are in contact with ro- from the C11R-positive animals were submitted for an- dents, the natural cowpox virus reservoirs (13). However, other qPCR targeting the A56R gene. Serum samples were some authors have hypothesized that domestic dogs could positive for the A56R gene for 5 dogs and 4 coatis; of these, be implicated in the transmission cycle of VACV, acting as anal swab samples were positive for A56R for 1 dog and 3 a link between the natural reservoirs and humans in urban coatis. No lesion swab samples were positive by qPCR for environments (14,15). Indeed, our molecular findings sup- both C11R and A56R genes. port exposure and possible VACV infection of these ani- Alignment of the amplified A56R fragments showed mals, thereby indicating that they are a potential source of high similarity to the homologous gene of VACV isolates VACV exposure for humans in urban areas. from Brazil (online Technical Appendix Figure, https:// The seroprevalence of orthopoxvirus neutralizing an- tibodies in dogs in Brazil has been described. Peres et al. pdf). Furthermore, 5 sequenced samples (from 4 dogs and found that, along with other farm animals, 22.8% of 114 1 coatis) showed an 18-nt signature deletion, which is dogs tested were seropositive for orthopoxviruses (7); this present in sequences of mouse nonvirulent VACV strains seroprevalence differs from that observed in our study, from Brazil (group 1 VACV). This deletion was not which was 3.8% lower. In addition, most animals tested by present in samples from 4 animals (1 dog and 3 coatis), Peres et al. were from rural areas where no bovine vaccinia grouping with mouse virulent VACV strains from Brazil outbreaks had been officially reported, and 96% of farm- (group 2 VACV). ers declared that their domestic animals have contact with wild animals (9). These findings indicate that dogs could Conclusions be exposed to VACV through contact with wild animals, We assessed VACV exposure of 2 interacting species of corroborating our hypothesis. animal: domestic dogs from an urban area and coatis from We also detected orthopoxvirus neutralizing antibod- a bordering wild area. In contrast to VACV infections, ies in wild coatis, which is consistent with results of a Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2339 DISPATCHES Table 1. Associations between neutralizing antibodies against Orthopoxvirus and demographic characteristics of domestic dogs and wild coatis, Belo Horizonte, Brazil, 2013–2015* Variable No. (%)† No. (%) positive‡ No. (%) negative‡ p value Odds ratio (95% CI) Domestic dogs Year of sampling 2014 123 (66.8) 24 (19.5) 99 (80.5) 2015 61 (33.2) 11 (18.0) 50 (82.0) 1.00 Sex F 85 (46.4) 23 (27.1) 62 (72.9) Reference 0.02 2.6 (1.2–5.6) M 96 (52.5) 12 (12.5) 84 (87.5) Age, y <1 24 (13.1) 7 (29.2) 17 (70.8) Reference 2–5 82 (44.8) 16 (19.5) 66 (80.5) 0.4 6–10 41 (22.4) 3 (7.3) 38 (92.7) 0.04 5.2 (1.2–22.6) >10 18 (9.8) 4 (22.2) 14 (87.8) 0.9 Size Small 75 (41.0) 13 (18.8) 56 (81.2) Reference Medium 69 (37.7) 11 (20.0) 44 (80.0) 1.00 Large 30 (16.4) 4 (14.3) 24 (85.7) 0.8 Confinement status Always inside home 41 (22.4) 8 (19.5) 33 (80.5) Reference Always in backyard 115 (62.8) 18 (15.6) 97 (84.3) 0.7 Home and backyard 25 (13.7) 9 (36.0) 16 (64.0) 0.2 Outdoors access† Yes 83 (45.3) 19 (22.9) 64 (77.1) 0.3 No 98 (53.6) 16 (16.3) 82 (83.7) Reference Access to MMP Yes 18 (9.8) 6 (33.3) 12 (66.7) 0.2 No 101 (55.2) 18 (17.8) 83 (82.2) Reference Wild coatis Year of capture 2013 57 (52.8) 12 (21.0) 34 (59.6) 0.002 14.8 (1.8–119.8) 2014 51 (47.2) 1 (1.9) 42 (82.3) Reference Sex F 64 (59.3) 10 (15.6) 44 (68.7) Reference M 44 (40.7) 3 (6.8) 32 (72.7) 0.3 Age group Juvenile, <1 y 44 (40.7) 1 (2.3) 35 (79.5) Reference 0.01 0.05 (0.006–0.5) Subadult, 1–2 y 18 (16.7) 5 (27.8) 10 (55.6) Adult, >2 y 46 (42.6) 7 (15.2) 31 (67.4) 0.04 5.1 (1.2–22.6) *Totals may not add up to 100% because of missing information. Boldface indicates significance; odds ratios are shown only for significant results. MMP, Mangabeiras Municipal Park. †Includes access beyond backyard. ‡By plaque reduction neutralization test. previous study that described the seroprevalence of or- for domestic dogs and humans and serve as a link between thopoxviruses in procyonids from Mexico (9). Our detec- wild and urban environments. However, future studies to tion of VACV DNA in anal swab samples from coatis determine if viable virus is shed are needed to confirm indicate that these animals could act as a source of virus this possibility. Table 2. Diagnostic results for 7 domestic dogs and 6 wild coatis with neutralizing antibodies for vaccinia virus, Belo Horizonte, Brazil, 2031–2015* qPCR C11R qPCR A56R Animal PRNT titer (NU/mL) Serum sample Anal swab sample Serum sample Anal swab sample Strain Dog 2 1:40 (100) + – + – Group 1 Dog 58 1:80 (200) + + + – Group 1 Dog 41 1:40 (100) + – – – Dog 77 1:80 (200) + + + – Group 1 Dog 86 1:40 (100) + – – – Dog 121 1:160 (400) + – + – Group 1 Dog 128 1:160 (400) + + + + Group 2 Coati 5 1:40 (100) + + + + Group 2 Coatis 17 1:40 (100) + – – – Coatis 27 1:80 (200) + – – – Coatis 39 1:160 (400) + + + + Group 2 Coatis 48 1:40 (100) + + + – Group 1 Coatis 50 1:80 (200) + + + + Group 2 *NU, neutralizing units; PRNT70, 70% plaque reduction neutralization test; qPCR, quantitative PCR. 2340 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Vaccinia Virus among Dogs and Coatis, Brazil Figure 2. A hypothetical model developed to visualize the role of domestic animals and wildlife in the natural cycle of vaccinia virus (VACV). The model illustrates the dynamics of VACV circulation in urban and wild areas of Brazil. In urban areas, wild coatis could promote the transmission of VACV between domestic animals or humans because they are in direct contact with domestic dogs and circulate among urban residences. Domestic dogs could also promote the transmission of VACV to humans because of direct contact or possibly indirect contact thought contaminated feces. In the wild environment, coatis can interact with other mammals such as wild rodents, which are believed to be VACV reservoirs, and acquire the infection (this potential interaction is still under investigation). Financial support was provided by the Conselho Nacional To impart information about the role of domestic ani- de Desenvolvimento Científico e Tecnológico, Coordenação mals and wildlife in the natural cycle of VACV, we de- de Aperfeiçoamento de Pessoal de Nível Superior, veloped a hypothetical model based on previous studies Fundação de Amparo à Pesquisa do Estado de Minas (2,3,5), which could illustrate the dynamics of VACV cir- Gerais, and Pró-Reitoria de Pesquisa/ Universidade Federal culation in urban areas (Figure 2). Because coatis can cir- de Minas Gerais. J.S.A., E.G.K., and G.S.T. are researchers culate in wild environments and surrounding urban areas, from Conselho Nacional de Desenvolvimento Científico they could act as a bridge promoting the transmission of e Tecnológico. VACV between wild animals (mainly rodents) and dogs or humans. Domestic dogs could transmit VACV directly About the Author to humans through close contact or indirectly thought con- Mr. Costa is a PhD candidate in microbiology at the taminated feces (Figure 2). These data raise questions about Laboratório de Vírus, Microbiology Department, Instituto de VACV circulation in Brazil and open discussions about the Ciências Biológicas, Universidade Federal de Minas Gerais. role of dogs and coatis in the VACV epidemiologic cycle. His research is focused on the diagnosis, epidemiology, control, and prevention of infection with orthopoxviruses and Acknowledgments other emerging viruses. We thank the Grupo de Epidemiologia e Conservação de Animais Silvestres and Residência Multiprofissional em Saúde, from Escola de Veterinária, Universidade Federal de Minas References Gerais. We are thankful to the Gerência de Controle de Zoonoses 1. Shchelkunov SN. An increasing danger of zoonotic team and Regional Centro Sul da Prefeitura Municipal de Belo orthopoxvirus infections. PLoS Pathog. 2013;9:e1003756. Horizonte, members from Projeto Quatis, do Parque das 2. Oliveira JS, Figueiredo PO, Costa GB, Assis FL, Drumond BP, Mangabeiras, and all the staff and partners for their support da Fonseca FG, et al. Vaccinia virus natural infections in Brazil: during sample collection. We also thank colleagues from the the good, the bad, and the ugly. Viruses. 2017;9:E340. Laboratório de Vírus for their excellent technical support. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2341 DISPATCHES 3. Miranda JB, Borges IA, Campos SPS, Vieira FN, de Ázara TMF, Marques FA, et al. Serologic and molecular evidence of vaccinia virus circulation among small mammals from different biomes, Brazil. Emerg Infect Dis. 2017;23:931–8. Spotlight: eid2306.161643 4. Peres MG, Bacchiega TS, Appolinário CM, Vicente AF, Mioni MSR, Ribeiro BLD, et al. Vaccinia virus in feces and urine HIV-AIDS of wild rodents from São Paulo state, Brazil. Viruses. 2018;10:E51. 5. Abrahão JS, Guedes MIM, Trindade GS, Fonseca FG, Campos RK, Mota BF, et al. One more piece in the VACV ecological puzzle: could peridomestic rodents be the link between wildlife and bovine vaccinia outbreaks in Brazil? PLoS One. 2009;4:e7428. 6. Abrahão JS, Silva-Fernandes AT, Lima LS, Campos RK, Guedes MI, Cota MM, et al. Vaccinia virus infection in monkeys, Brazilian Amazon. Emerg Infect Dis. 2010;16:976–9. 7. Peres MG, Bacchiega TS, Appolinário CM, Vicente AF, Allendorf SD, Antunes JM, et al. Serological study of vaccinia virus reservoirs in areas with and without official reports of outbreaks in cattle and humans in São Paulo, Brazil. Arch Virol. 2013;158:2433–41. s00705-013-1740-5 8. Peres MG, Barros CB, Appolinário CM, Antunes JM, Mioni MS, Bacchiega TS, et al. Dogs and opossums positive for vaccinia virus during outbreak affecting cattle and humans, São Paulo state, Brazil. Emerg Infect Dis. 2016;22:271–3. 10.3201/eid2202.140747 9. Gallardo-Romero NF, Aréchiga-Ceballos N, Emerson GL, Martínez-Martínez FO, Doty JB, Nakazawa YJ, et al. Endemic orthopoxvirus circulating in procyonids in Mexico. J Wildl Dis. 2016;52:609–15. 2015-10-291 10. Dutra LA, de Freitas Almeida GM, Oliveira GP, Abrahão JS, Kroon EG, Trindade GS. Molecular evidence of orthopoxvirus World AIDS Day is held on DNA in capybara (Hydrochoerus hydrochaeris) stool samples. Arch Virol. 2017;162:439–48. December 1 of each year s00705-016-3121-3 11. Costa GB, Miranda JB, Almeida GG, Silva de Oliveira J, and is an opportunity for Pinheiro MS, Gonçalves SA, et al. Detection of vaccinia virus in urban domestic cats, Brazil. Emerg Infect Dis. 2017;23:360–2. people worldwide to unite in 12. Geessien Kroon E, Santos Abrahão J, de Souza Trindade GS, the fight against HIV, show Pereira Oliveira G, Moreira Franco-Luiz AP, Barbosa Costa G, et al. Natural vaccinia virus infection: diagnosis, isolation, and their support for people living characterization. Curr Protoc Microbiol. 2016;42:14A.5. 1–14A.5.43. with HIV, and commemorate 13. Essbauer S, Pfeffer M, Meyer H. Zoonotic poxviruses. Vet Microbiol. 2010;140:229–36. people who have died. World j.vetmic.2009.08.026 14. Smith KC, Bennett M, Garrett DC. Skin lesions caused by AIDS Day was the first ever orthopoxvirus infection in a dog. J Small Anim Pract. 1999; 40:495–7. tb03003.x global health day, held for 15. von Bomhard W, Mauldin EA, Breuer W, Pfleghaar S, Nitsche A. Localized cowpox infection in a 5-month-old the first time in 1988. Rottweiler. Vet Dermatol. 2011;22:111–4. 10.1111/j.1365-3164.2010.00923.x Address for correspondence: Galileu Barbosa Costa or Giliane de Souza Trinidade, Universidade Federal de Minas Gerais, Microbiologia, Avenida Presidente Antônio Carlos 6627, Campus Pampulha, Belo ® Horizonte, Minas Gerais, Brazil 31270-90; email: or 2342 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Highly Pathogenic Avian Influenza A(H5N6) in Domestic Cats, South Korea 1 1 KyungHyun Lee, Eun-Kyoung Lee, enlarged 2-fold. The pancreas showed spotty hemorrhage HyunKyoung Lee, Gyeong-Beom Heo, Yu-Na Lee, and white pinpoint foci. Ji-Youl Jung, You-chan Bae, ByungJae So, Histopathologic examination revealed severe lesions Youn-Jeong Lee, Eun-Jin Choi in brain, lungs, and liver in the examined cats. We observed necrosis and loss of neurons, lymphocytic perivascular In December 2016, highly pathogenic avian influenza cuffing, and gliosis (Figure 1, panel C) in the cerebellum (HPAI) infection with systemic pathologic lesions was found and cerebrum, and especially severe necrosis in the hip- in cats in South Korea. Genetic analyses indicated that the pocampus. The lungs showed marked congestion, edema, feline isolates were similar to HPAI H5N6 viruses isolated hemorrhage, and severe interstitial pneumonia (Figure 1, in chicken farms nearby. This finding highlights the need for panel D), and thrombus in the alveolar capillaries. The liver monitoring of domestic mammals during HPAI outbreaks. showed severe necrotic foci and hepatitis. We observed in- fluenza viral antigen in neurons (Figure 1, panel E), glial ighly pathogenic avian influenza (HPAI) H5N6 has cells, and alveolar macrophages (Figure 1, panel F). Table Hspread across Asia, Europe, and Africa. Since a novel 1 describes histopathologic lesions and Table 2 immuno- influenza A(H5N6) virus emerged in China in late 2013 (1), histochemical reactivity. H5N6 viruses have been subsequently reported in South- We recognized H5N6 virus infection in a domestic east Asia. In China, HPAI A(H5N6) virus caused the earli- male cat (cat 1) and juvenile outdoor cats (cats 2 and 3). est reported human infection in 2014 and became one of We observed necrotic lesions and influenza viral antigens the dominant subtypes in poultry farms and live poultry in multiple visceral organs, suggesting that the virus caused markets (2). These viruses caused a potential threat to other systemic infection. It seems likely that the neurotropism of mammals, including pigs and cats (3,4). We report H5N6 H5N6 virus was a key factor contributing to the sudden virus infection in cats during 2016–17 HPAI outbreaks in death of these cats. The results of this study are consistent domestic poultry in South Korea (5). with those of other studies of HPAI pathogenicity in ex- perimentally infected dogs (6,7). The Study The histopathologic findings and the localization of The 2016–17 winter season saw epidemics of HPAI H5N6 virus antigen to the lungs and liver, but not to the A(H5N6) in domestic poultry and wild birds in South brain, in cats have been reported (8). In this case, we ob- Korea (5). At the end of December 2016, three carcasses served meningoencephalitis. Moreover, the 3 cats showed of cats were submitted from areas near H5N6 virus–in- neurologic symptoms such as salivation and convulsion, fected chicken farms in Pocheon. The cats had shown which may be related to necrosis and loss of neurons. The sudden clinical signs of salivation, lethargy, convulsion, severity of the lesions was consistent with the number of and bloody discharge around the mouth and jaws and cells that reacted with influenza viral antigen. A few studies died within 4 days after illness onset despite antimicrobial reported that H9 and H10 influenza viruses were nephro- drug treatment. After necropsy, we processed representa- tropic in chickens with low pathogenicity (9,10) and that tive tissues for histopathology and immunohistochemis- HPAI H5 virus causes acute renal lesions in mammals and try (online Technical Appendix, primates, including humans (11). The results of our study EID/article/24/12/18-0290-Techapp1.pdf). The necropsy suggest that the HPAI H5N6 virus affects cats differently findings included bloody nasal discharge (Figure 1, panel than do other HPAI viruses; therefore, further studies are A), severe pulmonary congestion and edema, and white- needed to experimentally infect cats with other HPAI H5 colored foci in the liver (Figure 1, panel B). The lungs were subtypes, including the isolate from this study, for com- red and yellow in color and incompletely collapsed and had plete clarification. accumulated a small amount of frothy fluid. The spleen was Previous studies have shown that avian viruses pref- erentially recognize α-2,3 linkage (SA α 2,3Gal) and bind Author affiliation: Animal and Plant Quarantine Agency, Gimcheon, to type II alveolar cells, which are abundant in the lower South Korea These authors contributed equally to this article. DOI: Emerging Infectious Diseases • •Vol. 24, No. 12, December 2018 2343 DISPATCHES Figure 1. Gross, microscopic, and immunohistochemical (IHC) findings in 3 cats with highly pathogenic avian influenza A(H5N6) virus infection, South Korea. A) Bloody nasal discharge. B) Severe congestion and edema in the lungs and white-colored foci in the liver. C) Gliosis in the brain. Hematoxylin and eosin stain; original magnification ×100. D) Interstitial pneumonia with degenerated pneumocytes. Hematoxylin and eosin; original magnification ×40. E) Influenza viral antigens in neurons. IHC testing; original magnification ×100. F) Influenza viral antigens in alveolar macrophages. IHC testing; original magnification ×100. respiratory tract of mammals (12,13). These findings are RNA samples extracted from organs of the cats were consistent with our observations of severe pneumonia with positive for influenza H5 and N6 subtypes by reverse lung edema in the infected cats. transcription PCR. We selected 2 HPAI H5N6 viruses: Table 1. Pathologic lesions in various tissues of 3 cats diagnosed with highly pathogenic avian influenza, South Korea Tissue Lesions Cat 1 Cat 2 Cat 3 Brain Neuronal necrosis Severe Severe Mild Meningoencephalitis Moderate Moderate Minimal Gliosis Moderate Moderate Minimal Trachea Lymphocytic tracheitis Minimal Minimal Minimal Lung Interstitial pneumonia Severe Severe Severe Pneumocytic necrosis Severe Moderate Severe Heart Myocytic necrosis Minimal Minimal Minimal Lymphocytic myocarditis Minimal Minimal Minimal Spleen Lymphocytic necrosis Minimal Minimal Minimal Lymphocytic depletion Mild Minimal Minimal Pancreas Acinar epithelial necrosis Minimal None None Intestine Enterocytic necrosis Minimal None None Enteritis Minimal None None Liver Hepatic necrosis Severe Severe Severe Kidney Tubular necrosis None None None Nephritis None None None 2344 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Influenza A(H5N6) in Cats, South Korea Table 2. Reactivity to influenza viral nucleoprotein in various tissues of 3 cats with highly pathogenic avian influenza, South Korea Reactivity Tissue Cells Cat 1 Cat 2 Cat 3 Brain Neurons Numerous Numerous Moderate Glial cells Moderate Numerous Minimal Ependymal cells Numerous Numerous Minimal Trachea Epithelial cells Minimal Minimal Minimal Lung Macrophages Numerous Numerous Numerous Vascular endothelial cells Numerous Numerous Numerous Heart Myocytes Minimal Minimal Minimal Epicardial cells Minimal Minimal Minimal Spleen Ellipsoid capillary endothelium Minimal Minimal Minimal Macrophages and necrotic debris Moderate Minimal Minimal Pancreas Acinar epithelium Minimal None None Intestine Crypt epithelium Minimal None None Mesenteric ganglial neurons Minimal None None Liver Kupffer cells and necrotic debris Numerous Numerous Numerous Kidney Tubule epithelium Minimal Minimal Unknown Glomeruli Minimal None None Figure 2. Maximum-likelihood phylogenetic tree of the hemagglutinin (A) and neuraminidase (B) gene segments for highly pathogenic avian influenza A(H5N6) viruses from cats, South Korea, and comparison viruses. Black circles indicate isolates from cats and triangles indicate chicken isolates from this study. Virus sequences from the GISAID EpiFlu database ( and GenBank were used for each phylogenetic comparison. The genetic subclades are annotated to the right of the tree. The genetic clusters major, minor, and G1.1.9, were designated according to the criteria of Bi et al. (2). The number at each branch indicates a bootstrap value. Scale bars indicate nucleotide substitutions per site. Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2345 DISPATCHES A/feline/Korea/H646-1/2016(H5N6) from the domestic Acknowledgments male cat and A/feline/Korea/H646-2/2016(H5N6) from We thank the Animal and Plant Quarantine Agency (APQA); 1 juvenile outdoor cat. We performed virus isolation, se- Ministry of Agriculture, Food and Rural Affairs (MAFRA); quencing, and phylogenetic analysis as described (online and regional office for animal disease control for their efforts to Technical Appendix). Phylogenetic analyses showed control HPAI. that the H5 genes of the cat isolates belonged to clade This study was supported by a grant from APQA, MAFRA, Re- and were very closely related to H5N6 viruses public of Korea (project code nos. N-1543069-2015-99-01 and detected in poultry in areas within a radius of 1 km of the B-1543418-2018-19-01). cats’ location (Figure 2, panel A). In a previous study, clade H5N6 viruses from the 2016–17 epi- demic in South Korea were divided into 5 distinct geno- About the Author groups (C-1 to C-5) (5). The feline isolates showed high Dr. KyungHyun Lee is a veterinary researcher at the Animal and similarity with H5N6 viruses in genogroup C-4, which Plant Quarantine Agency, South Korea. Her research interests was detected in domestic poultry nearby during 2016–17 include pathology and emerging infectious diseases in animals. HPAI A(H5N6) outbreaks (Figure 2; online Technical Dr. Eun-Kyoung Lee is a veterinary researcher at the Animal Appendix Figure). and Plant Quarantine Agency. Her research interests include the Epidemiologic studies show that the cats might be in- molecular epidemiology and surveillance of avian influenza. fected by feeding on or by contact with infected wild birds, although the virus was not isolated from wild birds around References this area. The affected domestic cat lived in a house near 1. Qi X, Cui L, Yu H, Ge Y, Tang F. Whole-genome sequence of a a small stream where migratory birds were observed and reassortant H5N6 avian influenza virus isolated from a live poultry market in China, 2013. Genome Announc. 2014;2:e00706-14. a wide main road. Across the main road, H5N6 virus–af- fected chicken farms were located within 1 km. In previous 2. Bi Y, Chen Q, Wang Q, Chen J, Jin T, Wong G, et al. Genesis, reports, cats and tigers were naturally infected by feeding evolution and prevalence of H5N6 avian influenza viruses in China. on infected bird carcasses (8,14). In China, H5N6 virus in- Cell Host Microbe. 2016;20:810–21. 10.1016/j.chom.2016.10.022 fection in cats has been reported in regions such as Suchuan 3. Yu Z, Gao X, Wang T, Li Y, Li Y, Xu Y, et al. Fatal H5N6 avian and Guangdong Provinces (3). influenza virus infection in a domestic cat and wild birds in China. We compared each gene of the feline and chicken Sci Rep. 2015;5:10704. H5N6 isolates (online Technical Appendix Table). The 4. Li X, Fu Y, Yang J, Guo J, He J, Guo J, et al. Genetic and biological characterization of two novel reassortant H5N6 swine hemagglutinin (HA) genes of the viruses contained mul- influenza viruses in mice and chickens. Infect Genet Evol. tiple basic amino acid residues at the HA cleavage site 2015;36:462–6. (PLRERRRKR). The amino acid residues on the recep- 5. Lee EK, Song BM, Lee YN, Heo GB, Bae YC, Joh SJ, et al. tor binding sites of the HA gene of H5N6 viruses were Multiple novel H5N6 highly pathogenic avian influenza viruses, South Korea, 2016. Infect Genet Evol. 2017;51:21–3. Q226 and G228 (H3 numbering), indicating an avian- like (α2,3-SA) receptor-binding preference. T160A mu - 6. Son K, Kim YK, Oem JK, Jheong WH, Sleeman JM, Jeong J. tation in the HA gene suggested a possible increased Experimental infection of highly pathogenic avian influenza viral affinity for human-like (α2,6-SA) receptor binding, viruses, clade H5N6 and H5N8, in mandarin ducks from South Korea. Transbound Emerg Dis. 2017;65:899–903. shown in feline isolates. The neuraminidase genes of fe- 7. Lyoo KS, Na W, Phan LV, Yoon SW, Yeom M, Song D, et al. line isolates also had 11 aa deletions at positions 59–69, Experimental infection of clade 1.1.2 (H5N1), clade which were often observed in avian influenza virus lin - (H5N1) and clade (H5N6) highly pathogenic avian eages adapted to poultry and may increase the virulence influenza viruses in dogs. Transbound Emerg Dis. 2017;64:1669– 75. to mammals (2). We did not observe amino acid substi- 8. Songserm T, Amonsin A, Jam-on R, Sae-Heng N, Meemak N, tution at position E627K of the polybasic 2 gene in the Pariyothorn N, et al. Avian influenza H5N1 in naturally infected feline isolates. domestic cat. Emerg Infect Dis. 2006;12:681–3. 10.3201/eid1204.051396 9. Swayne DE, Alexander DJ. Confirmation of nephrotropism and Conclusions nephropathogenicity of three low-pathogenic chicken-origin Our results demonstrate that cats can be directly infected influenza viruses for chickens. Avian Pathol. 1994;23:345–52. by HPAI H5N6 virus. Cats are companion animals and may act as a vector for influenza transmission to hu - 10. Bonfante F, Fusaro A, Zanardello C, Patrono LV, De Nardi R, Maniero S, et al. Lethal nephrotropism of an H10N1 avian mans. Despite the low probability of H5N6 virus infec- influenza virus stands out as an atypical pathotype. Vet Microbiol. tion in companion animals, avian influenza surveillance 2014;173:189–200. will be needed for mammals, including cats, during 11. Cardona CJ, Xing Z, Sandrock CE, Davis CE. Avian H5N6 outbreaks. influenza in birds and mammals. Comp Immunol Microbiol 2346 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Influenza A(H5N6) in Cats, South Korea Infect Dis. 2009;32:255–73. 14. Keawcharoen J, Oraveerakul K, Kuiken T, Fouchier RA, j.cimid.2008.01.001 Amonsin A, Payungporn S, et al. Avian influenza H5N1 in 12. van Riel D, Munster VJ, de Wit E, Rimmelzwaan GF, Fouchier RA, tigers and leopards. Emerg Infect Dis. 2004;10:2189–91. Osterhaus AD, et al. H5N1 virus attachment to lower respiratory tract. Science. 2006;312:399. science.1125548 Address for correspondence: Youn-Jeong Lee and Eun-Jin Choi, Animal 13. Shinya K, Ebina M, Yamada S, Ono M, Kasai N, Kawaoka Y. and Plant Quarantine Agency, hyuksin8ro 177 Gimcheon, Gimcheon Influenza virus receptors in the human airway. Nature. 39660, South Korea; email:; 2006;440:435–6. September 2017 Zoonoses • Bioinformatic Analyses of Whole-Genome Sequence Data in • Group A Rotavirus Associated with Encephalitis in Red Fox a Public Health Laboratory • Protective Eect of V ff al -PrP against Bovine Spongiform • Convergence of Humans, Bats, Trees, and Culture in Nipah Encephalopathy but not Variant Creutzfeldt-Jakob Disease Virus Transmission, Bangladesh • Imported Infections with Mansonella perstans Nematodes, • Processes Underlying Rabies Virus Incursions across Italy US–Canada Border as Revealed by Whole-Genome • Genetic Diversity of Highly Pathogenic Avian Influenza Phylogeography A(H5N8/H5N5) Viruses in Italy, 2016–17 • Real-Time Whole-Genome Sequencing for Surveillance of • Microcephaly Caused by Lymphocytic Choriomeningitis Virus Listeria monocytogenes, France • Influenza A(H3N2) Virus in Swine at Agricultural Fairs and • Role of Food Insecurity in Outbreak of Anthrax Infections Transmission to Humans, Michigan and Ohio, USA, 2016 among Humans and Hippopotamuses Living in a Game • Use of Blood Donor Screening to Monitor Prevalence of HIV Reserve Area, Rural Zambia and Hepatitis B and C Viruses, South Africa • Molecular Antimicr obial Resistance Surveillance for • Emergence of Plasmid-Mediated Fosfomycin-Resistance Neisseria gonorrhoeae, Northern Territory, Australia Genes among Escherichia coli Isolates, France • Estimated Annual Numbers of Foodborne Pathogen– • Determination of the Ferret Enteric Coronavirus Genome in Associated Illnesses, Hospitalizations, and Deaths, France, Laboratory Ferrets 2008–2013 • Myocarditis Caused by Human Parechovirus in Adult • Epidemiology of Salmonella enterica Serotype Dublin Infections among Humans, United States,1968–2013 • Cost of Nosocomial Outbreak Caused by NDM-1–Containing Klebsiella pneumoniae in the Netherlands, 2015–2016 • Prevalence of Yersinia enterocolitica Bioserotype 3/O:3 among Children with Diarrhea, China, 2010–2015 • Evaluation of 5 Commercially Available Zika Virus Immunoassays • Risk for Low Pathogenicity Avian Influenza Virus on Poultry Farms, the Netherlands, 2007–2013 • Epidemiology of Neisseria gonorrhoeae Gyrase A Genotype, Los Angeles, California, USA • Paerns of Human Plague in Ug tt anda, 2008–2016 • Conveyance Contact Investigation for Imported Middle East • Serologic Evidence for Influenza C and D Virus among Respiratory Syndrome Cases, United States, May 2014 Ruminants and Camelids, Africa, 1991–2015 • Possible Role of Fish as Transport Hosts for Dracunculus • Norovirus in Bottled Water Associated with Gastroenteritis spp. Larvae Outbreak, Spain, 2016 To revisit the September 2017 issue, go to: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2347 DISPATCHES Candidatus Cryptoplasma Associated with Green Lizards and Ixodes ricinus Ticks, Slovakia, 2004–2011 Božena Kočíková, Igor Majláth, The Study Bronislava Víchová, Lenka Maliničová, We conducted this study in the Slovak Karst National Park Peter Pristaš, Vincent A. Connors, in Slovakia (48°36′N, 20°52′E) during 2004–2011. We Viktória Majláthová carried out lizard capture and sample collections with offi- cial permits (6103/2007-2.1 and 5498/2011-2.2) issued by During 2004–2011, we collected green lizards and Ixodes the Ministry of Environment of the Slovak Republic. We ricinus ticks in Slovak Karst National Park in Slovakia; captured 103 green lizards (Lacerta viridis) and collected 90% (36/40) of lizards and 37% of ticks removed from blood from 40 (30 males and 10 females). We collected lizards were infected with family Anaplasmataceae bac- 235 I. ricinus ticks (118 larvae and 117 nymphs) from 63 teria. Only Candidatus Cryptoplasma sp. REP (reptile) green lizards and 271 questing I. ricinus ticks (132 nymphs, was identified in these samples. Green lizards transmit 76 males, and 63 females) from the same area and immedi- this bacterium. ately stored them in 70% ethanol. We isolated DNA from lizard blood using a DNeasy he family Anaplasmataceae (Rickettsiales; Alphapro- Blood & Tissue Kit (QIAGEN, Hilden Germany) and iso- Tteobacteria) comprises bacteria that are able to invade lated DNA from ticks by alkaline hydrolysis. We performed and infect their vertebrate host’s blood cells, bone mar- PCR amplification in 25-μL (total) reaction mixtures us- row–derived phagocytic cells, and endothelial cells; these ing the MasterTaq DNA Polymerase Kit (Eppendorf AG, bacteria can also infect cells of insects, helminths, and Hamburg, Germany). We amplified sequences using the arthropod reproductive tissues (1–3). Tickborne family primer combinations EHR747 plus EHR521 or fD1 plus members include bacteria of Anaplasma, Ehrlichia, Can- rP2 (9), which spanned almost the entire 16S rRNA se- didatus Neoehrlichia sp., and Candidatus Cryptoplasma quence (Table 1). We examined the ≈250-bp gene fragment californiense (4). of 16S rRNA by single-strand conformation polymorphism Although reptiles play a role as hosts for ixodid and (SSCP) analysis to determine Anaplasmataceae species argasid ticks, their role in maintaining tickborne Anaplas- type (10). We performed SSCP analysis following the pro- mataceae bacteria in the environment has not been de- tocol of Derdakova et al. (11). We ran positive control sam- scribed. Nieto et al. (5) suggested that lizards and snakes ples A. phagocytophilum, A. ovis, Wolbachia sp., and Can- in the far western part of the United States could become didatus N. mikurensis with each reaction. We purified the exposed to Anaplasma phagocytophilum when fed on by PCR products obtained using the GenElute PCR Clean-Up infected ticks. Moreover, Rejmanek et al. detected 2 highly Kit (Sigma-Aldrich, Buchs, Switzerland) and sequenced dissimilar strains of A. phagocytophilum in the same lizard both strands. We edited variants obtained in this study species (6). In Europe, an undescribed Anaplasma sp. was (1,410 bp) using MEGA 4.0.2 ( detected in Ixodes ricinus ticks feeding on sand lizards and and checked by eye. We made comparisons to sequences sand lizard blood samples (7,8). In our study, we sought in GenBank with BLASTn 2.2.26 (https://pods.iplantcol- to confirm these previous findings by determining whether For family Anaplasmataceae bacteria were present in lizards phylogenetic analysis of our variant (GenBank accession and their feeding ticks in Slovakia. no. MG924904), we aligned 17 related sequences obtained from the GenBank database and constructed a phylogenetic tree using the Bayesian inference method (12). Author affiliations: Slovak Academy of Sciences, Košice, Slovakia We examined all blood and ticks collected for the (B. Kočíková, B. Víchová, P. Pristaš, V. Majláthová); Pavol Jozef presence of family Anaplasmataceae bacteria. Anaplas- Šafárik University in Košice, Košice (I. Majláth, L. Maliničová, mataceae family members were present in 36 lizards (28 P. Pristaš, V. Majláthová); University of South Carolina Upstate, males and 8 females). Of the ticks removed from lizards, Spartanburg, South Carolina, USA (V.A. Connors) 87 (37%) were infected, and of questing ticks, 18 (6.6%) DOI: were infected (Table 2). 2348 Emerging Infectious Diseases • •Vol. 24, No. 12, December 2018 Candidatus Cryptoplasma and Lizards and Ticks Table 1. Primers used to amplify 16S rRNA gene of Candidatus Cryptoplasma sp. found in green lizards and Ixodes ricinus ticks, Slovakia, 2004–2011 Primer Length of amplified Organism name Sequences, 5  3 fragment, bp Reference Family Anaplasmataceae EHR747 GCACTCATCGTTTACAGCGTG 247 (10) EHR521 TGTAGGCGGTTCGGTAAGTTAAAG Most eubacteria fD1 CCGAATTCGTCGACAACAGAGTTTGATCCTGGCTCAG 1,500 (9) rP2 CCCGGGATCCAAGCTTACGGCTACCTTGTTACGACTT Denatured and electrophoresed PCR products from interaction between reptiles and Anaplasmataceae fam- samples demonstrated several SSCP profiles, of which 1 ily members has only been investigated in a few studies. was clearly distinguishable from the profiles of the Ana- Our findings expand knowledge on this research topic. plasmataceae species used as controls (Figure 1). We de- Only limited information about the reptile–Anaplasma tected this unique profile in all lizard blood samples, all relationship exists. Ekner et al. suggested that sand liz- ticks feeding on lizards, and some questing ticks. We se- ards could potentially serve as a reservoir host for spe- quenced representatives of this unidentified SSCP profile cies of the Anaplasmataceae family when she discovered (≈247 bp; GenBank accession nos. KY031322–3) and com- that ticks collected from these lizards in Poland were in- pared them with DNA fragments in the GenBank database. fected with Anaplasma-like pathogens (8). Although A. The closest related (99% identity) 16S rRNA sequences phagocytophilum might be transmitted by reptiles to a were from uncultured Anaplasma sp. isolates from quest- limited extent (5), the Anaplasma-like species detected ing I. ricinus ticks from Morocco (GenBank accession no. in reptiles could also be a novel species, as suggested by AY672415), Tunisia (GenBank accession no. AY672420), Rejmanek et al. (6). and France (GenBank accession no. GU734325). Sequenc- Despite the fact that lizards are exposed to a number ing of a longer (1,410-bp) fragment of the 16S rRNA gene of family Anaplasmataceae bacteria through infected ticks, revealed 99.1% similarity with the Candidatus C. californi- our findings suggest that, except for Canditatus Crypto- ense isolate from I. pacificus ticks in California (Figure 2). plasma sp. REP, green lizards do not acquire infections The 16S rRNA sequence obtained in this study was found with these species. In short, we detected Canditatus Cryp- to share a maximum of 94% identity with A. phagocytophi- toplasma sp. REP in 90% of examined lizards, 37% of ticks lum Norway variant 2 (GenBank accession no. CP015376). feeding on lizards, and 6.6% of questing ticks in localities The phylogenetic tree we constructed using 16S rRNA with lizards. gene sequences showed that the reptile-associated Candi- On the basis of our results, we cautiously speculate datus Cryptoplasma sp. REP (reptile) clustered in a sepa- that Canditatus Cryptoplasma sp. REP is selected for and rate branch with Candidatus C. californiense, indicating other genospecies selected against in ticks feeding on liz- the isolate represents a lineage distinct from other known ards. The Canditatus Cryptoplasma sp. REP variant had a Anaplasmataceae species (e.g., A. phagocytophilum, A. high homology (100%) with a sequence obtained from an marginale, A. platys, Ehrlichia muris, E. chaffeensis, and Apodemus agrarius mouse from Slovakia (13), which in- E. ewingii). dicates that rodents or other mammals might also become infected with this bacterium and contribute (to a lesser ex- Conclusions tent) to the circulation of these bacteria in nature. The role of ectotherm animals, especially lizards, in the In conclusion, we found a yet to be named species of maintenance of vectorborne pathogens is not clear. The Canditatus Cryptoplasma sp. (Canditatus Cryptoplasma sp. Table 2. Prevalence of family Anaplasmataceae bacteria in Ixodes ricinus ticks collected from green lizards and surrounding vegetation, Slovakia, 2004–2011 No. (%) positive ticks No. ticks No. (%) positive Candidatus Anaplasma Wolbachia Candidatus Tick source, type examined ticks Cryptoplasma phagocytophilum pipientis Neoehrlichia mikurensis Lizards Larvae 118 43 (36.4) 43 (100) – – – Nymphs 117 44 (37.6) 44 (100) – – – Total 235 87 (37) 87 (100) – – – Vegetation Nymphs 132 8 (6.1) 2 3 3 0 Males 76 4 (5.3) 1 2 0 1 Females 63 6 (9.5) 3 2 1 0 Adults 139 10 (7.2) 4 4 1 1 Total 271 18 (6.6) 6 7 4 1 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2349 DISPATCHES REP) in questing I. ricinus ticks, I. ricinus ticks collected from and feeding on green lizards, and the blood of green lizards in Slovakia. These results indicate that green lizards serve as an intermediate host for this bacterium and that liz- ards can influence the enzootic maintenance and circulation of bacteria in the environment. However, other hosts besides reptiles could be involved in the Canditatus Cryptoplasma sp. REP lifecycle as well, though probably to a lesser extent. This study was conducted within the framework of the project Environmental Protection against Parasitozoonoses under the Influence of Global Climate and Social Changes (project no. 26220220116), which is supported by the Research and Development Operating Program funded by the European Fund for Regional Development (0.4). Additional support was provid- ed by the Scientific Grant Agency of the Ministry of Education of the Slovak Republic, Slovak Academy of Sciences VEGA 2/0113/18 (to I.M.), and EurNegVec COST Action TD1303. We thank the Fulbright Foundation (USA, to V.A.C.) for support. About the Author Figure 1. Single-strand conformation polymorphism profile Dr. Kočíková is a postdoctoral researcher at the Department of of Anaplasmataceae isolate from reptiles, Slovakia, 2004– 2011. The 247-bp 16S rRNA PCR fragments from the isolate Vector-Borne Diseases, Institute of Parasitology, Slovak from reptiles and known Anaplasmataceae species were Academy of Sciences, Košice, Slovakia. Her research interests denatured and electrophoresed. Lane 1, 100-bp ladder include ecology and epidemiology of ticks and tickborne marker; lane 2, Candidatus Neoehrlichia mikurensis; lane pathogens, the role of reptiles in transmission of vectorborne 3, Anaplasma phagocytophilum; lane 4, isolate Candidatus microorganisms, and molecular diversity of emerging parasitic Cryptoplasma sp. REP (reptile) obtained in this study; lane 5, A. ovis; and lane 6, Wolbachia . and infectious disease agents in light of global changes. Figure 2. Phylogenetic relatedness of Candidatus Cryptoplasma sp. REP (reptile; bold), Slovakia, 2004–2011, to other Anaplasmataceae sp. family members. We constructed the tree using 16S rRNA sequences and the Bayesian inference method. The Rickettsia parkeri sequence was used as an outgroup. Scale bar indicates nucleotide substitutions per site. 2350 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Candidatus Cryptoplasma and Lizards and Ticks References 1. Dumler JS. Anaplasma and Ehrlichia infection. Ann N Y Acad Sci. 2005;1063:361–73. EID SPOTLIGHT TOPIC annals.1355.069 2. Rikihisa Y. Mechanisms to create a safe haven by members of the family Anaplasmataceae. Ann N Y Acad Sci. 2003;990:548–55. TICKS tb07425.x 3. Rar V, Golovljova I. Anaplasma, Ehrlichia, and “Candidatus Neoehrlichia” bacteria: pathogenicity, biodiversity, and molecular genetic characteristics, a review. Infect Genet Ticks transmit a variety of different Evol. 2011;11:1842–61. j.meegid.2011.09.019 pathogens, including bacteria, protozoa, 4. Eshoo MW, Carolan HE, Massire C, Chou DM, Crowder CD, Rounds MA, et al. Survey of Ixodes pacificus ticks in and viruses, which can produce serious California reveals a diversity of microorganisms and a and even fatal disease in humans and novel and widespread Anaplasmataceae species. PLoS One. 2015;10:e0135828. animals. Tens of thousands of cases of journal.pone.0135828 tickborne disease are reported each year, 5. Nieto NC, Foley JE, Bettaso J, Lane RS. Reptile infection with Anaplasma phagocytophilum, the causative agent of including Lyme disease. granulocytic anaplasmosis. J Parasitol. 2009;95:1165–70. 6. Rejmanek D, Bradburd G, Foley J. Molecular characterization Lyme disease is the most well-known reveals distinct genospecies of Anaplasma phagocytophilum tickborne disease. However, other tick - from diverse North American hosts. J Med Microbiol. 2012;61:204–12. borne illnesses such as Rocky Mountain 7. Tijsse-Klasen E, Fonville M, Reimerink JHJ, Spitzen-van der Sluijs A, Sprong H. Role of sand lizards in the ecology of spotted fever, tularemia, babesiosis, and Lyme and other tick-borne diseases in the Netherlands. ehrlichiosis also contribute to severe Parasit Vectors. 2010;3:42. 1756-3305-3-42 morbidity and more mortality each year. 8. Ekner A, Dudek K, Sajkowska Z, Majláthová V, Majláth I, Tryjanowski P. Anaplasmataceae and Borrelia burgdorferi sensu lato in the sand lizard Lacerta agilis and co-infection of these bacteria in hosted Ixodes ricinus ticks. Parasit Vectors. 2011;4:182. 9. Weisburg WG, Barns SM, Pelletier DA, Lane DJ. 16S ribosomal DNA amplification for phylogenetic study. J Bacteriol. 1991;173:697–703. jb.173.2.697-703.1991 10. Pancholi P, Kolbert CP, Mitchell PD, Reed KD Jr, Dumler JS, Bakken JS, et al. Ixodes dammini as a potential vector of human granulocytic ehrlichiosis. J Infect Dis. 1995;172:1007–12. 11. Derdáková M, Beati L, Pet’ko B, Stanko M, Fish D. Genetic variability within Borrelia burgdorferi sensu lato genospecies established by PCR-single-strand conformation polymorphism analysis of the rrfA-rrlB intergenic spacer in Ixodes ricinus ticks from the Czech Republic. Appl Environ Microbiol. 2003;69:509–16. 516.2003 12. Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. MEGA6: Molecular Evolutionary Genetics Analysis version 6.0. Mol Biol Evol. 2013;30:2725–9. molbev/mst197 13. Štefančíková A, Derdáková M, Lenčáková D, Ivanová R, Stanko M, Čisláková L, et al. Serological and molecular page/tick-spotlight detection of Borrelia burgdorferi sensu lato and Anaplasmataceae in rodents. Folia Microbiol (Praha). 2008;53:493–9. Address for correspondence: Viktória Majláthová, Institute of Parasitology, Slovak Academy of Sciences, Hlinkova 3, 040 01 Košice, Slovakia; email: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2351 DISPATCHES Excess Mortality and Causes Associated with Chikungunya, Puerto Rico, 2014–2015 André Ricardo Ribas Freitas, Maria Rita The Study Donalisio, Pedro María Alarcón-Elbal We studied excess mortality associated with chikungunya in Puerto Rico by comparing monthly expected deaths and During 2014–2015, a total of 31 deaths were associated actual deaths during the epidemic (5–7). We calculated ex- with the first chikungunya epidemic in Puerto Rico. We ana - pected deaths by the average age-adjusted mortality rate lyzed excess mortality from various causes for the same for each month for 2010 to 2013 and projected them to the months during the previous 4 years and detected 1,310 estimated population for 2014 and 2015 (6). We considered deaths possibly attributable to chikungunya. Our findings the difference between observed and expected deaths for raise important questions about increased mortality rates the months in which observed deaths exceeded the upper associated with chikungunya. limit of the 99% CI (6) as excess deaths associated with the chikungunya epidemic (Figure 1). n December 2013, the first locally acquired chikungunya We estimated the mortality rate expected for the sec- Ivirus infections in the Americas were reported in Saint ond half of 2014, the worst period of the epidemic, and Martin. Since that report, the virus has spread to 45 coun- compared it with the observed mortality rate. This second tries and territories in North, Central, and South America, group of calculations was based on the all-cause age-spe- causing ≈2.4 million suspected and confirmed cases and cific mortality rate for the 10 leading causes of death in 440 deaths through December 2016 (1). Puerto Rico and main causes of deaths associated with se- Chikungunya has emerged worldwide since 2004. vere chikungunya in Réunion Island (2). Several gaps in knowledge exist about the disease and its We obtained data on estimated population from the consequences. Until recently chikungunya was considered US Census Bureau (8) and on deaths from the Centers a nonlethal disease, but severe forms and deaths have been for Disease Control and Prevention’s National Center for described since an epidemic on Réunion Island during Health Statistics (9). In mid-September 2014, Puerto Rico 2005–2006 (2). issued an official administrative order stating that only In Puerto Rico, the chikungunya epidemic began in persons hospitalized with suspected chikungunya should May 2014 as the first occurrence of the virus in the coun - be reported to health authorities in Puerto Rico (3), Thus, try (3). Official surveillance reported 28,327 suspected we monitored the occurrence of chikungunya, dengue, and chikungunya cases, of which 4,339 were laboratory- other viral diseases in Puerto Rico using secondary data confirmed; 31 persons died (0.9 deaths/100,000 popula - from published studies evaluating the etiology of acute fe- tion). The chikungunya mortality rate was significantly brile illness of any patient, such as sentinel surveillance at lower than that observed in epidemics on other islands, St. Luke’s Episcopal Hospital in the cities of Ponce and such as Réunion Island (25.9 deaths/100,000 population Guayama (10,11). in 2006), Martinique (20.5/100,000 population in 2014), We determined an excess of 1,310 deaths concurrent and Guadaloupe (14.4/100,000 population in 2014) (1,4). with the peak of the chikungunya epidemic of 2014 in These differences could be a consequence of the diffi - Puerto Rico. We found no substantial occurrence of den- culty of recognizing the etiology of severe clinical forms gue, influenza, or leptospirosis during the chikungunya epi- and deaths. demic, according to data from St. Luke’s Episcopal Hospi- tal (Figure 1) and official Puerto Rico surveillance (10,12), Author affiliations: Secretaria de Saúde de Campinas, Campinas, reinforcing the possibility that chikungunya might be etio- Brazil (A.R.R. Freitas); Faculdade de Medicina São Leopoldo logically associated with the deaths. Although the most af- Mandic, Instituto de Pesquisas São Leopoldo Mandic, Campinas fected age group was persons >75 years of age, a statisti- (A.R.R. Freitas); Universidade Estadual de Campinas, Campinas cally significant proportion of deaths occurred in persons (M.R. Donalisio); Universidad Iberoamericana, Instituto de 24–55 years of age, suggesting chikungunya-associated Medicina Tropical & Salud Global, Santo Domingo, Dominican deaths are not exclusive to elderly persons (Figure 2). Republic (P.M. Alarcón-Elbal) The estimated mortality rate was 30.1 deaths/100,000 DOI: population, compatible with the excess deaths observed in 2352 Emerging Infectious Diseases • •Vol. 24, No. 12, December 2018 Excess Mortality and Chikungunya, Puerto Rico Figure 1. Expected and observed monthly deaths in Puerto Rico, 2014–2015. A) Excess number of monthly deaths and number of chikungunya, dengue, and influenza A and B cases diagnosed among patients with acute febrile illness in St. Luke’s Episcopal Hospital, Puerto Rico, January 2014–December 2015. Data for acute febrile illness were discontinued in April 2015. B) Number of monthly deaths, observed deaths, and expected deaths and upper limit of 99% CI, Puerto Rico, January 2014– December 2015. Réunion Island in 2006 (33.8/100,000 population) using conditions participate only as additional risk factors for death. similar methods (5) and slightly below the estimated rate of The most frequent clinically worsening conditions described 47.9 deaths/100,000 population in Pernambuco, the most in laboratory-confirmed chikungunya in other settings were affected state of Brazil, during the chikungunya epidemic heart failure, multiple organ failure, acute hepatitis, encephali- of 2016 (7). We estimated excess deaths as 42 times greater tis, epidermolysis bullosa, myocarditis, respiratory failure, and than the 31 deaths identified by the official surveillance de- renal failure (2). Many of these causes of death on Réunion veloped during the chikungunya epidemic in Puerto Rico Island (3) also increased over previous years in Puerto Rico (3). In a similar study conducted in northeastern Brazil, we during the epidemic (Table 2). In this study, volume deple- found 7,231 excess deaths in the region during the chikun- tion, emphysema, arrhythmias, asthma, diabetes mellitus and gunya epidemic of 2016, when the official surveillance sys- chronic ischemic heart disease led to a higher mortality rate tem confirmed only 120 deaths, 60 times lower ( 7). during the epidemic than during previous years; all of these We detected a 1-month lag between the peak of chi- kungunya identified by St. Luke’s Episcopal Hospital and the peak of excess deaths. Other lags also were observed in similar studies in Ahmedabad, India; Mauritius; and re- cently in Brazil (6,7); these lags can be explained by pro- longed hospitalization before death (2). Deaths classified as “all other forms of chronic isch - emic heart disease” (International Classification of Dis - eases, 10th Revision, codes I20, I25.1–I25.9) and diabetes mellitus (E10–E14) increased significantly, but deaths as - sociated with other important causes of death, such as acute myocardial infarction (I21, I22) and neoplasms (C00–D49), did not (Table 1). These findings suggest that chikungu - Figure 2. Excess deaths and difference between observed and nya contributes substantially to severity through its own expected deaths during chikungunya epidemic, Puerto Rico, July– December 2014. Mortality rate is deaths per 100,000 population. pathophysiologic mechanisms and that some preexisting Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2353 DISPATCHES Table 1. Expected and observed deaths during a chikungunya epidemic, Puerto Rico, July–December 2014 Observed Expected Relative Cause of death (code)* deaths deaths Difference† risk 95% CI 99% CI p value Acute myocardial infarction (I21–I22) 735 783 0.938 0.866–1.017 0.844–1.043 0.06 48 All other forms of chronic ischemic heart 963 787 176 1.224 1.139–1.315 1.114–1.346 <0.01 disease (I20, I25.1–I25.9) Alzheimer's disease (G30) 1,050 1,020 30 1.030 0.962–1.102 0.942–1.126 0.20 Diabetes mellitus (E10–E14) 1,761 1,623 138 1.085 1.030–1.144 1.012–1.163 <0.01 Hypertensive heart disease (I11) 366 376 10 0.974 0.869–1.092 0.838–1.132 0.33 Other chronic obstructive pulmonary 382 410 28 0.932 0.833–1.041 0.805–1.078 0.11 disease (J44) Pneumonia due to other or unspecified 392 396 0.990 0.886–1.105 0.856–1.145 0.43 4 organisms (J16, J18) Renal failure (N17–N19) 500 479 21 1.045 0.947–1.153 0.918–1.189 0.19 Septicemia (A40–A41) 456 418 38 1.091 0.984–1.210 0.952–1.250 0.05 Neoplasms (C00–D49) 2,849 2,849 0 1.000 0.960–1.042 0.947–1.055 0.50 *Causes of death were grouped according to the List of 358 Selected Causes of Death of the National Center for Health Statistics ( Codes are from the International Classification of Diseases, 10th Revision. †Observed deaths (no. cases)/expected deaths (no. cases). diseases were found as cause of death or preexisting condition deaths is a tool that should be included in the assessment of in patients with severe chikungunya (2,13). chikungunya epidemics. The results of our study are subject to several limita- Conclusions tions. We conducted an ecologic analysis, which does not We found substantial excess mortality in Puerto Rico dur- enable establishment of causality, and the investigation was ing the 2014 chikungunya epidemic, which should no lon- based on secondary data, which may result in some inaccu- ger be treated as a nonlethal disease. In addition to elderly racies. Therefore, some of the excess deaths we calculated persons, excess deaths occurred in other age groups. The might have resulted from other diseases, particularly vec- main causes of death in patients with laboratory-confirmed torborne diseases, which have seasonal patterns of occur- chikungunya in hospital-based studies were similar to those rence similar to those of chikungunya. However, we tried in our study (2,13). to reduce these limitations using several sources of informa- Our study reinforces the hypothesis of the association tion and accounting for other diseases that could interfere in of chikungunya with severe manifestations and deaths. the deaths in the region, including other viral diseases. The Chikungunya-related death is critical to defining public method used in this study is already well known and widely health priorities, such as investment in research, vaccine used to estimate deaths and hospitalizations associated with development, and vector control. The evaluation of excess respiratory viruses and extreme weather phenomena (14,15). Table 2. Number of deaths from specific causes observed during chikungunya epidemic period, Puerto Rico, July–December 2010– 2015, selected according to clinical presentation of severe chikungunya deaths in Réunion Island, 2005–2006* Difference between Year 2014 and mean of Change, Cause of death (code)† 2010 2011 2012 2013 2014 2015 2010–2013 % Volume depletion, disorders of fluid, electrolyte and 117 117 134 122 170 93 48 39 acid-base balance (E86–E87) Emphysema (J43) 82 66 76 51 104 70 35 51 Conduction disorders and cardiac dysrhythmias (I44– 155 146 140 154 182 142 33 22 I49) Asthma (J45–J46) 33 45 13 24 53 37 24 84 Pneumonia (J12–J18) 419 401 339 364 402 283 21 6 Other and unspecified diseases of skin and sub- 85 98 81 95 106 90 16 18 cutaneous tissue (L10–L98) Other chronic liver disease and cirrhosis (K73–K74) 61 73 62 58 77 77 14 21 Alcoholic liver disease (K70) 45 37 49 50 58 51 13 28 Other and unspecified heart failure (I50.1–I50.9) 63 60 57 55 71 56 12 21 Diseases of pericardium and acute myocarditis (I30– 3 2 2 4 6 3 3 118 I31, I40) Meningitis and encephalitis (A83–A84, A85.2, G00, 5 3 1 5 5 4 2 43 G03) Congestive heart failure (I50.0) 229 199 207 222 204 162 10 5 *See reference (2). †Except for pneumonia, the causes of death were grouped according to the List of 358 Selected Causes of Death of the National Center for Health Statistics ( Codes are from the International Classification of Diseases, 10th Revision. 2354 Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 Excess Mortality and Chikungunya, Puerto Rico Island, France, 2005-2006. Am J Trop Med Hyg. 2007;77:727–31. Although limited, our results raise important ques- tions about the occurrence of severe disease and increased 5. Josseran L, Paquet C, Zehgnoun A, Caillere N, Le Tertre A, mortality rates associated with chikungunya. Therefore, Solet J-L, et al. Chikungunya disease outbreak, Reunion Island. fundamental research is needed about chikungunya patho- Emerg Infect Dis. 2006;12:1994–5. eid1212.060710 physiologic mechanisms involving severe forms, exacer- 6. Mavalankar D, Shastri P, Bandyopadhyay T, Parmar J, Ramani KV. bation of preexisting conditions, and deaths. In addition to Increased mortality rate associated with chikungunya epidemic, clinical studies, systematic diagnostic research of recent Ahmedabad, India. Emerg Infect Dis. 2008;14:412–5. infection, including chikungunya, in all severe hospital- 7. Freitas ARR, Cavalcanti L, Von Zuben AP, Donalisio MR. Excess ized patients during outbreaks could answer some impor- mortality related to chikungunya epidemics in the context of tant questions. co-circulation of other arboviruses in Brazil. PLoS Curr. 2017;9:pii: ecurrents.outbreaks.14608e586cd321d8d5088652d7a0d884. 8. US Census Bureau. Puerto Rico Commonwealth population totals Acknowledgments tables: 2010–2016 [cited 2017 Apr 8]. We thank Luana Hughes Freitas, Nicole Montenegro de data/tables/2016/demo/popest/total-puerto-rico.html Medeiros, Bruno Coelho, and Marcela Montenegro for their 9. Centers for Disease Control and Prevention, National Center for support in revising the manuscript. Health Statistics. Vital statistics online. Mortality multiple cause files [cited 2017 Apr 8]. vitalstatsonline.htm#Mortality_Multiple 10. Tomashek KM, Lorenzi OD, Andújar-Pérez DA, About the Author Torres-Velásquez BC, Hunsperger EA, Munoz-Jordan JL, et al. Dr. Freitas is a medical epidemiologist who works in the Clinical and epidemiologic characteristics of dengue and other epidemiologic surveillance of Campinas, Campinas, São Paulo, etiologic agents among patients with acute febrile illness, Puerto Brazil, and is a professor of epidemiology, biostatistics, and Rico, 2012-2015. PLoS Negl Trop Dis. 2017;11:e0005859. scientific methodology at Faculdade de Medicina São Leopoldo 11. Rohatgi A. WebPlotDigitizer [cited 2017 Apr 8]. Mandic de Campinas. His primary research interests include studies of mortality and communicable diseases, with emphasis 12. Departamento de Salud de Puerto Rico. Informe semanal de on arboviruses and respiratory viruses. bigilancia del dengue—semana 52/2014 [cited 2017 Apr 8]. Pages/Dengue.aspx References 13. Crosby L, Perreau C, Madeux B, Cossic J, Armand C, Herrmann-Storke C, et al. Severe manifestations of chikungunya 1. Pan American Health Organization, World Health Organization. virus in critically ill patients during the 2013-2014 Caribbean Chikungunya: communicable diseases and health analysis. Data, maps and statistics [cited 2016 Dec 13]. outbreak. Int J Infect Dis. 2016;48:78–80. 10.1016/j.ijid.2016.05.010 hq/index.php?option=com_topics&view=readall&cid=5927&Itemi d=40931&lang=en 14. Freitas ARR, Donalisio MR. Excess of mortality in adults and elderly and circulation of subtypes of influenza virus in southern 2. Economopoulou A, Dominguez M, Helynck B, Sissoko D, Brazil. Front Immunol. 2018;8:1903. Wichmann O, Quenel P, et al. Atypical chikungunya virus infections: clinical manifestations, mortality and risk factors fimmu.2017.01903 15. Hoshiko S, English P, Smith D, Trent R. A simple method for for severe disease during the 2005–2006 outbreak on Réunion. Epidemiol Infect. 2009;137:534–41. estimating excess mortality due to heat waves, as applied to the 2006 California heat wave. Int J Public Health. 2010;55:133–7. S0950268808001167 3. Sharp TM, Ryff KR, Alvarado L, Shieh W-J, Zaki SR, Margolis HS, et al. Surveillance for chikungunya and dengue during the first year of chikungunya virus circulation in Puerto Address for correspondence: André Ricardo Ribas Freitas, Faculdade Rico. J Infect Dis. 2016;214(suppl 5):S475–81. São Leopoldo Mandic, Instituto de Pesquisas São Leopoldo Mandic, área 10.1093/infdis/jiw245 de Epidemiologia, R. Dr. José Rocha Junqueira, 13, Swift, Campinas, SP 4. Renault P, Solet JL, Sissoko D, Balleydier E, Larrieu S, Filleul L, et al. A major epidemic of chikungunya virus infection on Reunion São Paulo 13045-755, Brazil; email: Emerging Infectious Diseases • • Vol. 24, No. 12, December 2018 2355 DISPATCHES Borrelia miyamotoi Infections in Small Mammals, California, USA 1 1 Daniel J. Salkeld, Nathan C. Nieto, the basis of local B. miyamotoi prevalence of 3.6%–10.7% in Denise L. Bonilla, Melissa H. Yoshimizu, I. pacificus nymphs (5 ). Trapping occurred in June to coincide Kerry A. Padgett with the peak abundance of nymphal black-legged ticks and so perhaps increase the chance that B. miyamotoi would be Surveillance to investigate the wildlife–vector transmission circulating in animal populations (11). Animals were captured cycle of the human pathogen Borrelia miyamotoi in Cali- using Sherman live traps baited with peanut butter and oats. fornia, USA, revealed infections in dusky-footed woodrats, We anesthetized captured animals with isoflurane, brush mice, and California mice. Phylogenetic analyses identified them by morphology using taxonomic guides, suggest a single, well-supported clade of B. miyamotoi is and examined them for ticks. We obtained blood and ear- circulating in California. tissue biopsy samples from each individual and tested both sample types for Borrelia spp. because different Borrelia orrelia miyamotoi is a spirochete that causes a relaps- species may vary in tissue tropism (9). Bing febrile illness and is transmitted by hard Ixodes We extracted DNA from all samples (i.e., whole blood, species ticks (1,2). B. miyamotoi is prevalent in western ear punch biopsies, and ticks) using DNeasy Blood and Tis- black-legged tick (I. pacificus) populations in California, sue Kits (QIAGEN, Valencia, CA, USA) and assayed for USA (3,4); in some locations, B. miyamotoi prevalence in the presence of Borrelia using quantitative PCR (qPCR) ticks is comparable with or higher than the prevalence of (9), which is able to detect as few as 10 spirochetes. We se- the Lyme disease agent, B. burgdorferi sensu stricto (3–5). quenced all qPCR-positive samples using a primer set tar- There is mounting evidence that human infections occur in geting the intergenic spacer rrs-rrlA locus, which allowed northern California (6). for differentiation of Borrelia genospecies (12). Alignments Surveillance of B. miyamotoi in California has focused were made in ClustalX ( We com- on ticks, and little is known about infection in wildlife hosts. pared our sequences from I. pacificus ticks and wild-caught B. miyamotoi has been identified from spleen samples of birds rodents (GenBank accession nos. MH342008–31) to repre- and rodents in Europe (7), from blood and bladder samples of sentative GenBank sequences from isolates found in other rodents in Japan (8), and from white-footed mice (Peromyscus sites in California (accession nos. KT343321, KT343334, leucopus) in the eastern United States (9). In California, sur- KT343337, KU184505, KF957668), elsewhere in the United veillance in Alameda County (east of the San Francisco Bay) States (accession nos. HQ658901, HQ658902, AY374140, observed B. miyamotoi in tick populations but failed to detect AY37139, AY374138, AY363706, GU993308, KY293400, the spirochete in mammals (10). We investigated B. miya- KY293399, KY293398, KY293397, KY293396, GQ856588, motoi infection status in small mammals at 3 California sites GU993309, GQ856589), Japan (accession nos. AY363703, where the bacterium is present in tick populations (3,5). AY363704), and Sweden (accession no. AY363705). We conducted phylogenetic reconstruction using MrBayes The Study ( and visualizations using Fig- We captured animals on 2 consecutive nights at each of 3 Tree (http:// sites in the San Francisco Bay area of California during June We captured a total of 117 small mammals from 5 spe- 2014: Windy Hill (37.37315°, –122.22466°) and Thorne- cies (Table 1). Our surveillance efforts demonstrate that B. wood (37.39086°, –122.25066°) Open Space Preserves miyamotoi infects woodrats (Neotoma fuscipes), brush mice (OSP) in San Mateo County, and Foothills Park (37.36243°, (Peromyscus boylii), and California mice (P. californicus) –122.17362°) in Santa Clara County. We chose these sites on (Table 2). At sites where B. miyamotoi was present in small Author affiliations: Colorado State University, Fort Collins, mammals, B. miyamotoi prevalence was 6%–33% in dif- Colorado, USA (D.J. Salkeld); Northern Arizona University, ferent host species (Table 2). These data reflect B. miyamo- Flagstaff, Arizona, USA (N.C. Nieto); California Department of toi prevalence in small-mammal hosts in other geographic Public Health, Richmond, California, USA (D.L. Bonilla, These authors contributed equally to this article. M.H. Yoshimizu, K.A. Padgett) Current affiliation: United States Department of Agriculture, DOI: Fort Collins, Colorado, USA. 2356 Emerging Infectious Diseases • •Vol. 24, No. 12, December 2018 Borrelia miyamotoi in Mammals, California, USA Table 1. Numbers of mammals captured and tested for Borrelia spp., by species and location, California, USA* Thornewood OSP Redwood Oak–madrone Species Foothills Park habitat woodland Windy Hill OSP Total Dusky-footed woodrat (Neotoma fuscipes) 4 0 1 1 6 Brush mouse (Peromyscus boylii) 27 9 17 18 71 California mouse (Peromyscus californicus) 6 3 9 6 24 Deer mouse (Peromyscus maniculatus) 6 4 2 1 13 Pinyon mouse (Peromyscus truei) 3 0 0 0 3 Total 46 16 29 26 117 *OSP, Open Space Preserves. regions: 10.7% of voles and mice (n = 65) in the Netherlands, Co-infections of B. burgdorferi and B. miyamotoi have where nymphal infection prevalence (NIP) of B. miyamotoi previously been reported from mice and ticks in the in I. ricinus ticks is 2.5% (84/3360) (7), and 6.5% of white- northeastern United States (9) and from ticks in Marin footed mice (P. leucopus) in the northeastern United States, County, California (4). where NIP in I. scapularis ticks is 5.5% (38/689) (9). We did Five mammals were infested with I. angustus ticks, of not observe B. miyamotoi in pinyon mice or deer mice, either which we observed all 3 life stages. In redwood habitat at because of small sample sizes or because these species are Thornewood OSP, 1 California mouse hosted 2 adult females not involved in B. miyamotoi transmission. In nearby Alam- and a nymph, 1 brush mouse was infested with 2 adult fe- eda County, B. miyamotoi was not observed in small mam- males and 2 nymphs, and 1 brush mouse carried 1 female mals (10); possible reasons are that the spirochete is rarer in and 1 nymphal tick. Of 2 brush mice captured at Windy Hill, this locality (NIP = 0.4% in Alameda study sites), that mam- 1 harbored 3 larvae and the other 5 larvae. None of the 17 I. mal capture periods were dispersed across multiple years and angustus ticks tested positive for Borrelia spp., nor did the not as coincident with nymphal tick activity, or that brush host animals from which the ticks were removed. We found mice and California mice were not captured at that location. no I. pacificus ticks infesting the small mammals. Other identified Borrelia species included B. bissettiae in Phylogenetic analyses suggest that B. miyamotoi in 1 pinyon mouse (P. truei) and 1 brush mouse, which mirrors California is a single strain, separate from B. miyamotoi earlier studies of B. bissettiae from farther north in California in the eastern United States and from strains circulating in (13,14). Woodrats, California voles (Microtus californicus), Asia and Europe (Figure), corroborating an earlier study deer mice (P. maniculatus), and black rats (Rattus rattus) have (15). Sequences from I. pacificus ticks previously col- also been observed infected with B. bissettiae (10,13). During lected in the San Francisco Bay area were identical to the previous tick sampling efforts at our study sites, we did not de - sequences obtained from rodent infections. tect B. bissettiae in questing western black-legged ticks (3,5). We did not find B. burgdorferi sensu stricto, although Conclusions we observed uncharacterized B. burgdorferi sensu lato The identification of B. miyamotoi in small mammals in infection in 1 California mouse. None of the animals cap- California mirrors research from other locations that have tured in the redwood habitat (Thornewood OSP) were in- documented the spirochete in small rodents. It is premature fected with Borrelia spp., although the sample size was to claim these infected species as B. miyamotoi reservoirs small at this location (Table 1). We found a co-infection (i.e., responsible for maintenance of the pathogen and act- of B. burgdorferi sl and B. miyamotoi in a brush mouse. ing as a source for of zoonotic transmission), in part because Table 2. Prevalence of Borrelia spp. in small mammal species in the San Francisco Bay area, California, USA Sample Prevalence at site, no. Prevalence across sites, Borrelia species Mammal species Site source tested/no. positive (%) no. tested/no. positive (%) B. miyamotoi Dusky-footed woodrat Foothills Blood 1/4 (25) 1/6 (17) Brush mouse Thornewood Ear 1/17 (6) 2/71 (3) woodland Brush mouse* Windy Hill Ear