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

Learn More →

Perfluorooctane Sulfonate Concentrations in Amniotic Fluid, Biomarkers of Fetal Leydig Cell Function, and Cryptorchidism and Hypospadias in Danish Boys (1980–1996)

Perfluorooctane Sulfonate Concentrations in Amniotic Fluid, Biomarkers of Fetal Leydig Cell... A Section 508–conformant HTML version of this article Research Children’s Health is available at http://dx.doi.org/10.1289/ehp.1409288. Perfluorooctane Sulfonate Concentrations in Amniotic Fluid, Biomarkers of Fetal Leydig Cell Function, and Cryptorchidism and Hypospadias in Danish Boys (1980–1996) 1,2 3 4 5 5 Gunnar Toft, Bo A.G. Jönsson, Jens Peter Bonde, Bent Nørgaard-Pedersen, David M. Hougaard, 5 3 6 6 7,8 Arieh Cohen, Christian H. Lindh, Richard Ivell, Ravinder Anand-Ivell, and Morten S. Lindhard 1 2 Department of Occupational Medicine, and Department of Clinical Epidemiology, Aarhus University Hospital, Aarhus, Denmark; 3 4 Division of Occupational and Environmental Medicine, Lund University, Lund, Sweden; Department of Occupational and Environmental Medicine, Copenhagen University Hospital, Bispebjerg, Copenhagen, Denmark; Danish Center for Neonatal Screening, Department of Clinical Biochemistry and Immunology, Statens Serum Institute, Copenhagen, Denmark; School of Biosciences, University of 7 8 Nottingham, Nottingham, United Kingdom; Department of Pediatrics, Regional Hospital of Randers, Randers, Denmark; Perinatal Epidemiology Research Unit, Department of Pediatrics, Aarhus University Hospital, Skejby, Denmark although these results were not corroborated Background : Exposure to perfluorooctane sulfonate (PFOS) may potentially disturb fetal Leydig in a recent study (Joensen et al. 2013), and cell hormone production and male genital development. other aspects of semen quality seem to be o Bsevitcej : We aimed to study the associations between levels of amniotic fluid PFOS, fetal steroid unaffected (Joensen et al. 2009, 2013; Toft hormone, and insulin-like factor 3 (INSL3) and the prevalence of cryptorchidism and hypospadias. et al. 2012). Apart from an inverse association between PFOS exposure and testosterone Methods : Using the Danish National Patient Registry, we selected 270 cryptorchidism cases, 75 hypospadias cases, and 300 controls with stored maternal amniotic fluid samples available in a level in one study (Joensen et al. 2013), Danish pregnancy-screening biobank (1980–1996). We used mass spectrometry to measure PFOS no significant associations between PFOS in amniotic fluid from 645 persons and steroid hormones in samples from 545 persons. INSL3 exposure and reproductive hormones in adult was measured by immunoassay from 475 persons. Associations between PFOS concentration in men have been observed (Joensen et al. 2009; amniotic fluid, hormone levels, and genital malformations were assessed by confounder-adjusted Raymer et al. 2012; Specht et al. 2012). linear and logistic regression. There have been fewer studies on the r esults : The highest tertile of PFOS exposure (> 1.4 ng/mL) in amniotic fluid was associated with effects of PFOS exposure on fetal steroidogen - a 40% (95% CI: –69, –11%) lower INSL3 level and an 18% (95% CI: 7, 29%) higher testosterone esis, and later reproductive function. A study level compared with the lowest tertile (< 0.8 ng/mL). Amniotic fluid PFOS concentration was not on rat fetal Leydig cell function indicated that associated with cryptorchidism or hypospadias. 20 mg/kg/day exposure of pregnant rats from c onclusions : Environmental PFOS exposure was associated with steroid hormone and INSL3 gestational days 11 to 19 was associated with concentrations in amniotic fluid, but was not associated with cryptorchidism or hypospadias in reduced testosterone production, reduced fetal our study population. Additional studies are needed to determine whether associations with fetal Leydig cell number, and decreased expres- hormone levels may have long-term implications for reproductive health. sion of steroidogenic enzymes (Zhao et al. c itation : Toft G, Jönsson BA, Bonde JP, Nørgaard-Pedersen B, Hougaard DM, Cohen A, 2014). Also, zebrafish embryos showed altered Lindh CH, Ivell R, Anand-Ivell R, Lindhard MS. 2016. Perfluorooctane sulfonate concentra- expression of stereoidogenic enzymes after tions in amniotic fluid, biomarkers of fetal Leydig cell function, and cryptorchidism and hypo- exposure to up to 500 μg/L PFOS from 4 spadias in Danish boys (1980–1996). Environ Health Perspect 124:151–156; http://dx.doi. to 120 hr past fertilization (Du et al. 2013). org/10.1289/ehp.1409288 Vested et al. (2013) studied whether prenatal exposure to PFOS (measured in maternal Introduction highest concentrations have been observed serum) was associated with adult male repro- Perfluorooctane sulfonate (PFOS) has among workers at facilities producing ductive function among 169 Danish men until recently been widely used in a variety PFOS, with mean PFOS concentrations of born in 1988–1989. Although prenatal PFOS of applications, especially in surface 1,000–2,000 ng/mL, whereas general popula- coatings used to make products water- and tions on average had concentrations of about Address correspondence to G. Toft, Department of oil- resistant. PFOS is highly biopersistent 35 ng/mL PFOS around the peak exposure Clinical Epidemiology, Aarhus University Hospital, Oluf Palmes Allé 43-45, 8200 Aarhus N, Denmark. with a half-life in humans of about 5 years period (Lau et al. 2004). Apart from the Telephone: 45 871 68202. E-mail: gunnar.toft@ (Olsen et al. 2007). PFOS was added to present study population (Jensen et al. 2012), clin.au.dk Annex B of the Stockholm Convention on PFOS has to our knowledge been measured We are grateful to Å. Amilon and A. Kristensen for Persistent Organochlorine Pollutants in in amniotic fluid only in one recent American skillful technical assistance. 2009, and the production and use of PFOS study of 28 women (Stein et al. 2012). at Th This study was initiated by generous grants from the have been regulated in Europe since 2008 study indicated a PFOS concentration in Danish Environmental Protection Agency, the Danish Ministry of Interior and Health, Research Centre for (European Commission 2006). In Norway, amniotic fluid about 20-fold less than in Environmental Health’s Fund, the Swedish Council PFOS levels rose through the mid-1990s, maternal serum. for Working Life and Social Research, Skåne County and have declined since 2000 (Haug et al. Experimental studies of the effects of Council’s Research and Development Foundation, 2009). However, model-based estimates PFOS on adult male reproductive function and the Medical Faculty at Lund University, Sweden. of future environmental PFOS exposures have reported reduced testosterone produc- The sponsors had no part in study design, data col- suggest a slower decay in temperate regions tion, altered gonadotropin secretion, and lection, analysis, or preparation of the manuscript, and they are not responsible for the scientific content (Armitage et al. 2009). PFOS has been reduced epididymal sperm in PFOS-treated and the conclusions expressed. detected in human populations from all rats and mice (López-Doval et al. 2014; The authors declare they have no actual or potential over the world, but considerable variation in Wan et al. 2011). Studies of human adults competing financial interests. exposure has been observed between popula- have shown associations between PFOS Received: 2 October 2014; Accepted: 1 June 2015; tions (Lau et al. 2004). Most previous studies exposure and the proportion of normal sperm Advance Publication: 5 June 2015; Final Publication: have measured PFOS in serum, and the cells (Joensen et al. 2009; Toft et al. 2012) 1 January 2016. | | Environmental Health Perspectives • volume 124 number 1 January 2016 151 Toft et al. exposure was not associated with reproduc- pregnancy-screening registry was recorded by detail by Jensen et al. (2012). Coefficients of tive hormones or semen quality, prenatal the personal identification number (unique to variation were 11% for PFOS and 9% for exposure to perfluorooctanoate (PFOA) was each Danish citizen) of the pregnant woman. cotinine; the limit of detection (LOD) was associated with reduced semen quality and We used these unique identifiers to obtain 0.20 ng/mL for PFOS and cotinine, deter- higher levels of luteinizing hormone (LH) obstetric data on the pregnancies from the mined as the concentrations corresponding and follicle- stimulating hormone (FSH) in Danish Medical Birth Registry (Knudsen to three times the standard deviation of the young adult men. These results suggest and Olsen 1998), including gestational age responses in chemical blanks. We analyzed that the prenatal period may be sensitive to at birth, singleton or multiple birth, maternal the samples using a liquid chromatograph environmental exposures to perfluoroalkyl parity, and birth weight and Apgar score of (LC; model UFLCXR, Shimadzu Corp). The substances (PFAS). the infant. In addition, we used the unique LC was connected to a hybrid triple quadru- Evidence of effects on fetal Leydig cell identifiers in the Danish Civil Registration pole linear ion trap tandem mass spectrom- hormone production has led others to System to identify male offspring from the eter (LC/MS/MS) equipped with a turbo ion hypothesize that PFOS exposure might pregnancies (Pedersen et al. 2006). spray source (QTRAP 5500; AB Sciex). affect androgen- and insulin-like factor 3 The Danish Regional Ethics Committee, We assayed the steroid hormones (INSL3)–dependent testicular descent (Bay the Danish National Board of Health, and testosterone, androstenedione, proges - et al. 2011) and scrotal fusion (Kalfa et al. the Danish Data Protection Agency approved terone, 17-OH-progesterone, and cortisol 2009). Only one previous study has evalu- the study. The use of the biobank for research in amniotic fluid at the Danish Center for ated the risk of cryptorchidism in relation to purposes has been approved, and additional Neonatal Screening, Department of Clinical PFAS exposure; this study did not observe informed consent from the study subjects for Biochemistry and Immunology, the State associations between PFAS level in cord this specific project was neither recommended Serum Institute, Copenhagen, Denmark. The blood and cryptorchidism (Vesterholm et al. nor required. online extraction LC-MS system consisted 2014). However, due to a limited study size Case–control definitions and ascertain- of an Aria TLX2 system (Thermo Scientific) of 59 cryptorchidism cases and 108 matched ment. Controls were randomly selected from with two Agilent 1100 binary pumps and controls, the power to show an association the 25,105 amniotic fluid samples belonging two Agilent 1200 quaternary pumps (Agilent) was limited. To our knowledge, no previous to live-born male offspring pregnancies in the connected to a Thermo TSQ Ultra triple quad - studies have evaluated the potential associa- screening database with complete obstetric rupole mass spectrometer equipped with an tion between in utero exposure to PFOS and data in the Danish Medical Birth Registry. APCI ion source. Extraction was performed fetal hormone level or hypospadias. The number of controls (n = 412) was chosen using a Cyclone P 0.5 × 50 mm Turboflow The aim of the present study was to to roughly equal the largest case group column (Thermo Scientific), and analytical determine whether in utero PFOS levels are consisting of 404 boys with cryptorchidism. separation was achieved using Kinetix 2.6 μ associated with altered Leydig cell function, Cryptorchidism cases had both a diagnosis 2.1 × 50 mm C18 columns (Phenomenex). as indicated by steroid hormone and INSL3 of undescended testis according to the All calibrators, controls, internal standards, concentrations in amniotic fluid, and to International Classification of Diseases, 8th and and micro-titer plates were purchased from evaluate whether these potential effects on 10th Revisions [ICD-8: 75210, 75211, 75219; PerkinElmer via their CHS steroid profiling kit hormone levels were associated with genital ICD-10: Q53, Q531(A), Q532(A), Q539] for mass spectrometry (PerkinElmer). Formic malformation in offspring. and a corrective surgical procedure according acid was purchased from Merck. Ammonium to the Surgery and Treatment Classification acetate and zinc sulphate heptahydrate Methods of the Danish National Board of Health were purchased from Sigma-Aldrich. Water Study population and amniotic fluid samples. (STC: 55600, 55640) or the Nordic was purified using a water purification unit The study population has been described Classic fi ation of Surgical Procedures (NCSP: from Millipore. We used a volume of 50 μL previously in detail (Jensen et al. 2012). KKFH00, KKFH01, KKFH10) recorded amniotic fluid for the assay. The LODs and Briefly, we used amniotic fluid samples from in the Danish National Patient Registry the intra- and interassay coefficients of varia - a Danish biobank maintained at the State (DNPR). Boys with a registry entry of tion were as follows: testosterone, 0.1 nmol/L, Serum Institute in Copenhagen (http://www. inguinal hernia repair (STC: 40620, 40640; 9 and 10%, respectively; androstenedione, ssi.dk). The biobank holds samples from a NCSP: KJAB00-KJAB97) were excluded 0.3 nmol/L, 8 and 9%, respectively; proges- pregnancy-screening registry, including from this case group to avoid iatrogenic terone, 0.4 nmol/L, 10 and 11%, respectively; information on amniotic fluid samples from cryptorchidism secondary to hernia repair. 17-OH-progesterone, 0.4 nmol/L, 9 and 10%, 63,882 pregnancies covering the period We included all boys that fulfilled these respectively; and cortisol, 4.4 nmol/L, 10 and 1980–1996. After restriction to live-born criteria (404 of 25,105; 1.61%) to maximize 12%, respectively. singleton boys with complete obstetric data, the cryptorchidism case group and overall INSL3 in amniotic fluid was measured 25,105 pregnancies were eligible (Jensen study size. For the hypospadias case group we using a semicompetitive time-resolved fluo- et al. 2012). The amniotic fluid samples included all 109 of the 25,105 boys (0.43%) rometric immunoassay (TRFIA) slightly were centrifuged before routine diagnostic with a diagnosis of hypospadias in the DNPR modified from that described in detail by analyses, and the supernatants were kept (ICD-8: 75220, 75221, 75222, 75228, Anand-Ivell et al. (2006). The laboratory work frozen at –20°C until the present analyses 75229; ICD-10: Q540, Q541, Q542, Q548, was performed at Leibniz Institute for Farm were carried out. Indications for amniocen- Q549). All boys were followed for the afore- Animal Biology, Dummerstorf, Germany. tesis included age ≥ 35 years and/or increased mentioned diagnoses and surgery entries in The modifications involved the replacement risk of severe malformations or Down the DNPR from birth until November 2008. of the original rat antiserum by a new poly- syndrome based on results from maternal Measurement of chemical compounds clonal antiserum (no. #RIA5) raised in rabbits serum analyses. We calculated the gestational and hormones. During all chemical analyses, (IMVS Antibody Services) against the same week of amniocentesis based on the date of laboratory technicians were blinded to chemically synthesized human INSL3 as amniocentesis, the date of birth, and the esti- case–control status and to levels of analytes previously, at a final dilution of 1:10,000. As mated gestational age at birth as described measured by others. We assayed PFOS and tracer, we used the same Europium-labeled by Jensen et al. (2012). Each sample in the cotinine in amniotic fluid as described in human INSL3 as described (Anand-Ivell | | 152 volume 124 number 1 January 2016 • Environmental Health Perspectives PFOS, Leydig cells, and genital malformations et al. 2006). Accordingly, also the secondary multiple linear or logistic regression analyses evaluate whether year of amniocentesis influ - goat-anti-rabbit antibody (Rockland for continuous and categorical outcomes, enced these associations, we made supplemen- Immunochemicals Inc.) used to coat the plates respectively. We divided PFOS exposure into tary stratified analysis by year of amniocentesis substituted for the original anti-rat antibody. tertiles to quantify the difference between the (1980–1986, 1987–1992, and 1992–1996). All other conditions were similar. Standard highest and lowest third of the population We also restricted the analyses to boys curves for measurement of amniotic fluid were and to evaluate whether any marked devia- with no other congenital malformations to constructed using serial dilutions of human tions from a linear association was evident. exclude cases with syndromes and chromo- INSL3 in EDTA–phosphate-buffered saline Differences in median hormone level somal abnormalities. (Anand-Ivell et al. 2008). A volume of 100 μL between the PFOS tertiles were evaluated PFOS and all hormone data were trans- amniotic fluid was used for the assay, and by the nonparametric Kruskal–Wallis test. formed by the natural logarithm (ln) to the limit of detection for the modified assay Then adjusted linear regression analyses were improve normality of their distribution of was 0.01 ng/mL, with inter- and intraplate performed evaluating tertile differences. We residuals in the regression analyses. coefficients of variation of < 8% and < 1%, additionally performed regression analyses Results respectively, across the range. There was no on a continuous PFOS variable to test for a cross-reactivity detectable across the physi- linear trend. Of the original 925 included subjects in the ological range with the structurally related Data are presented as difference from the cohort, we could not locate samples in the peptides, insulin, IGF-1 (insulin-like growth lowest tertile and β [95% confidence interval biobank for 60 subjects, and 220 had insuf- factor-1), and relaxin. There was also no (CI)] for linear regression models both unad- ficient volume for PFOS measurements. cross-reactivity detectable with rat INSL3, justed and adjusted. We performed inter- Thus, the study population consists of 645 and only 10% cross-reactivity with bovine action analyses including a ln-PFOS × case/ mother–child pairs with information on INSL3 at the highest values. Spiking experi- control group term in the regression models fetal PFOS exposure and case status of the ments into normal male and female human to evaluate whether combined analyses of children, including 270 cases of cryptorchi- sera from previous studies (Anand-Ivell et al. cases of cryptorchidism, hypospadias, and dism, 75 cases of hypospadias, and a random 2006, 2013) indicated 122.9 + 1.5% and control groups were statistically justifiable. control group of 300. The mean and standard 96.5 + 2.7% recovery, respectively. INSL3 In supplementary analyses, we additionally deviation of maternal characteristics including measurements of human serum following evaluated the association between PFOS and age and gestational age of amniocentesis and five successive freeze–thaw cycles showed no hormones separately for controls, cryptorchi- distribution of smoking prevalence did not significant change (data not shown). dism cases, and hypospadias cases to evaluate differ markedly between the case and control Statistical analysis. We imputed values whether the association varied across groups. groups (Table 1). As expected, the cryptor- below the LOD of our chemical assays with a We a priori decided to adjust all hormone chidism and hypospadias cases weighed less random value between the LOD and LOD/2 analyses for gestational age of amniocentesis and were on average born slightly earlier as a simplified method of maintaining vari - (continuous, weeks), maternal age (years), and than the control group. Also, the distribu- ability in values below LOD. For the INSL3 smoking (using amniotic fluid specific levels tion of calendar year of amniocentesis differed assay, 58 of 543 (10.7%) samples were below as a biomarker for nonsmoker: < 25 ng/mL between case groups (Table 1). the LOD (0.01 ng/mL). For the PFOS assay, cotinine; passive smoker: 25–85 ng/mL; and Amniotic fluid steroid hormone levels were 10 of 645 (1.6%) samples were below the smoker: > 85 ng/mL) (Jauniaux et al. 1999). measured on 545 pregnant women with suffi - LOD (0.2 ng/mL). For the steroid assay Analysis of combined associations between cient sample volume (84% of cryptorchidism (n = 574), we used the instrument readout PFOS and hormone levels across case and cases, 100% of hypospadias cases, and 81% for values below the LOD; two (0.3%) testos- control groups were additionally adjusted for of control samples). Testosterone level was terone values < 0.1 nmol/L, no androstene- case–control status. The logistic regression positively associated with PFOS exposure with dione values < 0.3 nmol/L, no progesterone analyses of cryptorchidism and hypospadias 18% higher testosterone (95% CI: 7, 29%) values < 0.4 nmol/L, no 17-OH-progesterone was adjusted for gestational age of amnio- in the highest PFOS tertile compared with values < 0.4 nmol/L, and six (1.0%) cortisol centesis (continuous, weeks), year of amnio- the lowest in the overall population (all three values < 4.4 nmol/L. centesis (three groups), maternal age (years), groups combined). Also, the linear trend test INSL3 levels are highly dependent on gestational age (weeks), birth weight (grams), showed a significant positive association in both gestational age at amniocentesis with a peak and smoking (cotinine groups). To further the unadjusted and adjusted analysis (Table 2). around week 15 (Anand-Ivell et al. 2008). We calculated multiple of the median Table 1. Characteristics of the included pregnancies and boys by case–control status among Danish (MoM) values for INSL3 by gestational age pregnant women with amniocentesis (1980–1996). at amniocentesis to reduce this dependence Control Cryptorchidism Hypospadias (Knight and Palomaki 2003). We used the Characteristics (n = 300) (n = 270) (n = 75) random sample control group to estimate the Maternal age at birth [years (mean ± SD)] 32.6 ± 5.3 32.8 ± 5.3 31.3 ± 5.8 median level for each week, and all INSL3 Gestational week of amniocentesis (mean ± SD) 15.7 ± 1.3 15.9 ± 1.8 15.6 ± 1.6 values (cases and controls) were then divided Gestational age at birth [weeks (mean ± SD)] 39.6 ± 1.5 39.2 ± 2.2 38.7 ± 2.8 Birth weight [g (mean ± SD)] 3,535 ± 561 3,370 ± 682 3,216 ± 827 by the median of the corresponding week to Year of amniocentesis (%) calculate the MoM value. Weeks 11–13 were 1980–1984 34 40 17 pooled because of limited number of observa- 1985–1990 34 32 37 tions, and the median value from week 21 1991–1996 33 29 45 (controls) was applied to week 22 (cases) Smoking (%) because there were no controls in week 22. Nonsmoker: cotinine < 25 ng/mL 61 60 58 All statistical analyses were presented for both Passive smoker: cotinine 25–85 ng/mL 6 4 7 the raw and the MoM INSL3 measures. Smoker: cotinine ≥ 85 ng/mL 33 36 35 Congentital malformations (%) 6 14 20 Associations between PFOS exposure and the included outcomes were evaluated by Malformations other than cryptorchidism and hypospadias. | | Environmental Health Perspectives • volume 124 number 1 January 2016 153 Toft et al. Also, androstendione, progesterone, and 1992–1996) also did not indicate any possibility that prenatal PFOS exposure may 17-OH-progesterone, and cortisol, but not association between PFOS exposure and increase fetal steroid hormone production. DHEAS (dehydroepiandrosterone sulfate), odds for cryptorchidism or hypospadias The inverse association between prenatal were positively associated with PFOS exposure (data not shown). Furthermore, analyses concentrations of PFOS and INSL3 may be (Table 2). INSL3 was measured in amniotic restricted to boys with no other congenital consistent with a direct effect on fetal Leydig fluid samples from 475 pregnancies (74% malformations (numbers given in Table 1) cells, because, based on the current knowl- of cryptorchidism cases, 84% of hypospa- produced essentially unchanged results (data edge, INSL3 is produced only by male fetal dias cases, and 71% of controls). INSL3 not shown). Leydig cells, whereas steroid hormones are was negatively associated with PFOS, with a also produced by the fetal adrenal gland Discussion 40% lower INSL3 concentration (95% CI: (Anand-Ivell and Ivell 2014). Fetal urine is –69, –11%) estimated for the highest tertile PFOS concentrations measured in amniotic believed to be the primary source of steroid compared with the lowest based on the overall fluid were associated with higher steroid hormones in amniotic fluid from the second adjusted model, and approximately the same hormone levels and lower INSL3 in the trimester onward (Schindler 1982). The magnitude when evaluated as INSL3 MoM. combined study population but were not asso- source or sources of steroids in the amniotic The inverse associations were confirmed by ciated with cryptorchidism or hypospadias fluid are unknown, but could include the linear trend analyses (Table 2). (270 and 75 cases, respectively, compared umbilical cord or placenta (Schindler 1982). The association between PFOS exposure, with 300 controls). Progesterone and estradiol are produced testosterone, and INSL3 was similar in the Our results are consistent with those of in considerable amounts in the placenta, case and control groups as indicated by Vesterholm et al. (2014), who reported no and may be partly transferred to the fetus p-values of product interaction terms in the association of cord blood concentrations of (Pasqualini 2005). A recent in vitro study regression models (Table 2) and estimates PFOS or other PFAS with cryptochidism in using human placental cells indicated that (Table 3). However, in the stratified analysis, 215 Danish and Finnish boys. placental aromatase activity may be altered fewer statistical significant associations were To our knowledge, the present study is the after exposure to PFOS (Gorrochategui et al. observed, probably because of a lower number first human study evaluating potential asso- 2014). Thus, higher testosterone levels in of cases (Table 3). ciation between fetal exposure to PFOS and amniotic fluid in our study population might PFOS concentrations in amniotic fluid biomarkers of human fetal steroid and INSL3 be attributable to inhibited aromatase activity were not associated with cryptorchidism levels. A study using the human adrenocor- in the placenta. However, whereas andros- or hypospadias based on adjusted logistic tical carcinoma (H295R) in vitro cell assay tendione and DHEAS concentrations were regression models (Table 4). Supplementary reported increased testosterone, progesterone, weakly correlated between second-trimester analysis stratifying these associations by and estradiol secretion after PFOS exposure amniotic fluid and maternal serum samples sampling year (1980–1986, 1987–1992, (Kraugerud et al. 2011), supporting the from mothers expecting male offspring, Table 2. PFOS exposure (ng/mL) and markers of Leydig cell function in amniotic fluid among Danish pregnant women with amniocentesis (1980–1996). a a Hormone and PFOS exposure n Median hormone level % difference (95% CI) Hormone and PFOS exposure n Median hormone level % difference (95% CI) Testosterone (nmol/L) 545 17-OH-Progesterone (nmol/L) 545 < 0.8 ng/mL PFOS 0.73 Reference < 0.8 ng/mL PFOS 4.94 Reference 0.8–1.4 ng/mL PFOS 0.81 9 (–2, 20) 0.8–1.4 ng/mL PFOS 5.22 7 (–1, 13) > 1.4 ng/mL PFOS 0.91 18 (7, 29) > 1.4 ng/mL PFOS 6.18 18 (11, 26) p-Value 0.002 p-Value < 0.001 c c Ln-PFOS (crude) 0.14 (0.08, 0.22) Ln-PFOS (crude) 0.17 (0.12, 0.22) c c Ln-PFOS (adjusted) 0.16 (0.09, 0.23) Ln-PFOS (adjusted) 0.15 (0.11, 0.20) d d p-Value for interaction 0.68 p-Value for interaction 0.92 DHEAS (nmol/L) 545 Cortisol (nmol/L) 545 < 0.8 ng/mL PFOS 17.0 Reference < 0.8 ng/mL PFOS 15.9 Reference 0.8–1.4 ng/mL PFOS 17.4 5 (–10, 20) 0.8–1.4 ng/mL PFOS 17.0 9 (–0, 17) > 1.4 ng/mL PFOS 17.4 2 (–14, 17) > 1.4 ng/mL PFOS 22.4 28 (19, 37) b b p-Value 0.93 p-Value < 0.001 c c Ln-PFOS (crude) 0.06 (–0.03, 0.16) Ln-PFOS (crude) 0.26 (0.20, 0.32) c c Ln-PFOS (adjusted) 0.07 (–0.03, 0.16) Ln-PFOS (adjusted) 0.19 (0.13, 0.25) d d p-Value for interaction 0.03 p-Value for interaction 0.96 Androstendione (nmol/L) 545 INSL3 (ng/mL) 475 < 0.8 ng/mL PFOS 2.62 Reference < 0.8 ng/mL PFOS 0.14 Reference 0.8–1.4 ng/mL PFOS 2.72 8 (0, 17) 0.8–1.4 ng/mL PFOS 0.12 –21 (–48, 7) > 1.4 ng/mL PFOS 3.07 17 (8, 25) > 1.4 ng/mL PFOS 0.09 –40 (–69,–11) b b p-Value 0.001 p-Value < 0.001 c c Ln-PFOS (crude) 0.15 (0.10, 0.20) Ln-PFOS (crude) –0.35 (–0.54, –0.16) c c Ln-PFOS (adjusted) 0.15 (0.10, 0.21) Ln-PFOS (adjusted) –0.21 (–0.40, –0.02) d d p-Value for interaction 0.22 p-Value for interaction 0.98 Progesterone (nmol/L) 545 INSL3 (MoM) 475 < 0.8 ng/mL PFOS 163.7 Reference < 0.8 ng/mL PFOS 1.45 Reference 0.8–1.4 ng/mL PFOS 170.8 11 (0, 23) 0.8–1.4 ng/mL PFOS 1.13 –21(–49, 8) > 1.4 ng/mL PFOS 186.4 22 (11, 34) > 1.4 ng/mL PFOS 0.85 –39 (–68,–10) b b p-Value 0.001 p-Value 0.07 c c Ln-PFOS (crude) 0.21 (0.14, 0.28) Ln-PFOS (crude) –0.17 (–0.36, 0.01) Ln-PFOS (adjusted) 0.21 (0.14, 0.29) Ln-PFOS (adjusted) –0.20 (–0.38, –0.01) p-Value for interaction 0.07 p-Value for interaction 0.82 a b c Adjusted for gestational age of amniocentesis, maternal age, smoking (cotinine groups), and case or control status unless otherwise indicated. Kruskal–Wallis rank test. Estimates represent the percent difference in hormone concentrations with a 1% increase in PFOS. Interaction terms (lnPFOS × case–control status) in adjusted linear regression models. Multiple of the median (MoM) values are calculated as specified in the methods. | | 154 volume 124 number 1 January 2016 • Environmental Health Perspectives PFOS, Leydig cells, and genital malformations testosterone concentrations were not corre- as a proxy for fetal exposure (Vested et al. socio-occupational class (Hougaard et al. lated, suggesting limited maternal to fetal 2013; Vesterholm et al. 2014). Although the 2014). However, our register-based ascertain- transfer of testosterone relative to other steroid presence in cord blood indicates that PFOS ment of cryptorchidism and hypospadias did hormones (van de Beek et al. 2004). is transferred across the placenta to the fetus not include milder forms of these outcomes In utero exposure of rats to di(n-butyl) (Vesterholm et al. 2014), little is known that may spontaneously disappear during phthalate, a compound with antiandrogenic about actual exposure level of the fetus, and the first years of life, so we cannot exclude properties, caused reduced INSL3 expression we propose that measurement in amniotic that PFOS exposure may be related to milder in fetal Leydig cells and an increase in cryptor- fluid is the closest we can get to actual forms of genital malformations. chidism, and INSL3 has been suggested as an measurement of fetal exposure at the relevant PFOS and hormone concentrations were endogenous marker of endocrine disruption time window of exposure (Jensen et al. 2012). measured concurrently in amniotic fluid (Anand-Ivell and Ivell 2014). A recent study The strengths of the present study include samples, so we cannot confirm the temporal from our group using the same study popula- the use of a large biobank of amniotic fluid relation between PFOS exposure and tion as in the present study indicated that di(2- samples linked with medical birth registries the outcomes. ethylhexyl) phthalate (DEHP) metabolites but to produce a nested case–control sample The analysis of associations between PFOS not diisononyl phthalate (DiNP) metabolites of cryptorchidism and hypospadias cases and hormone levels in the combined case and were also related to decreased INSL3 level selected purely based on clinical diagnoses control groups may have produced biased (Jensen et al. 2015). However, regarding that are unlikely to be related to the exposure. estimates of associations in the source popula- PFOS exposure, a recent study indicated that Although it might be speculated that socio- tion, if associations in the case groups differed fetal rat Leydig cell INSL3 production was occupational class could be related to PFOS from the population as a whole. Our strati- unaltered after exposure to 20 mg/kg/day exposure and detection rate of cryptorchi- fied and interaction analyses did not indicate PFOS from gestational day 11 to 19 (Zhao dism and hypospadias, we have previously major differences among the case and control et al. 2014). There may be species differences shown that among Danish boys the time to groups, though sample sizes were small and in effects of PFOS on INSL3 production, detection of cryptorchidism is unrelated to estimates were imprecise, particularly for the partly because of the much shorter half-life of PFOS in rats than in humans (Lau et al. Table 3. PFOS exposure (ng/mL) and markers of Leydig cell function in amniotic fluid among Danish 2004). In our study population, prenatal pregnant women with amniocentesis (1980–1996) stratified by case–control group. PFOS exposure was associated with lower Controls Cryptorchism Hypospadias INSL3 in amniotic fluid, but was not asso- % difference % difference % difference ciated with cryptorchidism. Whether lower a a a Hormone and PFOS exposure n (95% CI) n (95% CI) n (95% CI) INSL3 levels can be associated with long-term Testosterone (nmol/L) 242 228 75 effects on male reproductive health is not < 0.8 ng/mL PFOS Reference Reference Reference known. However, it is of interest to note that 0.8–1.4 ng/mL PFOS 10 (–5, 24) 14 (–4, 32) –12 (–47, 24) a recent follow-up study found lower sperm > 1.4 ng/mL PFOS 19 (4, 33) 20 (3, 38) –6 (–47, 36) Ln-PFOS (crude) 0.12 (0.03, 0.21) 0.18 (0.08, 0.28) 0.13 (–0.11, 0.37) counts and increased LH and FSH levels Ln-PFOS (adjusted) 0.14 (0.04, 0.23) 0.20 (0.09, 0.31) 0.00 (–0.28, 0.28) in 19- to 21-year-old males at the highest INSL3 213 199 63 tertile of PFOA exposure in utero compared < 0.8 ng/mL PFOS Reference Reference Reference with the lowest (Vested et al. 2013). In that 0.8–1.4 ng/mL PFOS –28 (–72, 16) –5 (–47, 38) –49 (–118, 20) study, PFOA and PFOS was estimated from > 1.4 ng/mL PFOS –37 (–81, 8) –38 (–82, 6) –21 (–110, 69) maternal pregnancy serum samples. Although Ln-PFOS (crude) –0.26 (–0.54, 0.02) –0.39 (–0.68, –0.11) –0.62 (–1.23, 0.00) no association with PFOS was found in the Ln-PFOS (adjusted) –0.22 (–0.51, 0.06) –0.20 (–0.48, 0.07) 0.02 (–0.60, 0.64) Vested et al. (2013) study, PFOS and PFOA INSL3 MoM 213 199 63 < 0.8 ng/mL PFOS Reference Reference Reference were strongly correlated in maternal serum 0.8–1.4 ng/mL PFOS –30 (–73,12) 0 (–44, 45) –60 (–135, 15) samples (r = 0.73). We did not measure > 1.4 ng/mL PFOS –43 (–86, –0) –30 (–75, 15) –20 (–117, 77) PFOA in the present study; thus it cannot be Ln-PFOS (crude) –0.24 (–0.51, 0.03) –0.09 (–0.37, 0.18) –0.28 (–0.87, 0.30) excluded that the associations with INSL3 Ln-PFOS (adjusted) –0.27 (0.54, 0.00) –0.12 (–0.40, 0.17) –0.01 (–0.68, 0.65) may be related to an effect of PFOA rather Adjusted for gestational age of amniocentesis, maternal age, smoking (cotinine groups), and case or control status than PFOS on INSL3. Future studies should unless otherwise indicated. Estimates represent the percent difference in hormone concentrations with a 1% increase evaluate whether low fetal INSL3 level is asso- in PFOS. Multiple of the median (MoM) values are calculated as specified in the methods. ciated with long-term consequences for male reproductive health. Table 4. Odds ratios (95% CI) for cryptorchidism or hypospadias in relation to amniotic fluid PFOS concentration among Danish pregnant women with amniocentesis (1980–1996). We measured PFOS, steroid hormones, and INSL3 in amniotic fluid for several n n Unadjusted Adjusted reasons. Most amniotic fluid samples were Malformation control cases OR (95% CI) OR (95% CI) taken in gestational weeks 15 to 17, close to Cryptorchidism the time window of male genital development 1st tertile 102 86 1.0 (reference) 1.0 (reference) 2nd tertile PFOS 101 91 1.11 (0.74, 1.67) 1.08 (0.71, 1.63) in gestational weeks 8 to 15 (Scott et al. 2009). 3rd tertile PFOS 97 93 1.09 (0.73, 1.63) 1.01 (0.66, 1.53) Contaminant concentrations in amniotic fluid Ln-PFOS 300 270 1.05 (0.81, 1.35) 0.99 (0.75, 1.30) during the first half of pregnancy, when the Hypospadias amniotic u fl id is composed largely of exudates 1st tertile 102 27 1.0 (reference) 1.0 (reference) from fetal blood and fluids, are consid- 2nd tertile PFOS 101 26 1.01 (0.55, 1.81) 0.97 (0.51, 1.87) ered a suitable proxy measure of intrafetal 3rd tertile PFOS 97 22 0.82 (0.44, 1.54) 0.69 (0.35, 1.38) contaminant levels (Beall et al. 2007). Ln-PFOS 300 75 0.97 (0.65, 1.44) 0.87 (0.57, 1.34) Other studies on fetal PFOS exposures a Adjusted for gestational age of amniocentesis, year of amniocentesis (groups), maternal age, gestational age, birth have used maternal serum or cord blood weight, and smoking (cotinine groups). | | Environmental Health Perspectives • volume 124 number 1 January 2016 155 Toft et al. hypospadias group. However, estimates based MacLeod M, Cousins IT. 2009. Modeling the global affect steroidogenesis and viability in the human fate and transport of perfluorooctane sulfonate adrenocortical carcinoma (H295R) in  vitro cell only on the controls, who should be represen- (PFOS) and precursor compounds in relation to assay. Toxicol Lett 205:62–68. tative of the overall population, were similar to temporal trends in wildlife exposure. Environ Sci Lau C, Butenhoff JL, Rogers JM. 2004. The develop- the estimates based on the combined case and Technol 43:9274–9280. mental toxicity of perfluoroalkyl acids and their control groups (Table 3). Bay K, Main KM, Toppari J, Skakkebæk NE. 2011. derivatives. Toxicol Appl Pharmacol 198:231–241. One may speculate that evaporation Testicular descent: INSL3, testosterone, genes López-Doval S, Salgado R, Pereiro N, Moyano  R, from stored samples or differential decay may and the intrauterine milieu. Nat Rev Urol 8:187–196. Lafuente A. 2014. Perfluorooctane sulfonate Beall MH, van den Wijngaard JP, van Gemert MJ, effects on the reproductive axis in adult male rats. explain some of the associations between Ross MG. 2007. Regulation of amniotic fluid Environ Res 134:158–168. hormone level and PFOS in amniotic fluid volume. Placenta 28:824–832. Olsen GW, Burris JM, Ehresman DJ, Froehlich JW, samples. However, PFOS, testosterone, and Du G, Hu J, Huang H, Qin Y, Han X, Wu D, et al. 2013. Seacat  AM, Butenhoff JL, et  al. 2007. Half-life INSL3 levels were only weakly correlated with Perfluorooctane sulfonate (PFOS) affects hormone of serum elimination of perfluorooctanesulfonate, year of amniocentesis (r-values of 0.07, –0.06, receptor activity, steroidogenesis, and expression perfluorohexanesulfonate, and perfluorooctanoate in and –0.05, respectively) and the estimated of endocrine-related genes in  vitro and in  vivo. retired fluorochemical production workers. Environ Environ Toxicol Chem 32:353–360. Health Perspect 115:1298–1305; doi:10.1289/ehp.10009. volume left in the biobank (r-values of 0.06, European Commission. 2006. Directive 2006/122/EC of Pasqualini JR. 2005. Enzymes involved in the forma- –0.06, and –0.02, respectively), suggesting the European Parliament and of the Council of 12 tion and transformation of steroid hormones in that any influence of storage on the observed amending for the 30th Time Council Directive 76/769/ the fetal and placental compartments. J Steroid results would have been minor. EEC on the Approximation of the Laws, Regulations Biochem Mol Biol 97:401–415. The concentration of PFOS in amniotic and Administrative Provisions of the Member States Pedersen CB, Gøtzsche H, Møller JO, Mortensen PB. fluid samples in the present sample was relating to Restrictions on the Marketing and Use of 2006. The Danish Civil Registration System. A cohort Certain Dangerous Substances and Preparations of eight million persons. Dan Med Bull 53:441–449. somewhat higher than a previous U.S. study of (Perfluorooctane Sulfonates). December 2006. Raymer JH, Michael LC, Studabaker WS, Olsen GW, pregnant women from New York, 2005–2008, Official Journal of the European Union 49:32–34. Sloan  CS, Wilcosky T, et  al. 2012. Concentrations where a median of 0.4 ng/mL PFOS was Gorrochategui E, Pérez-Albaladejo E, Casas J, of perfluorooctane sulfonate (PFOS) and perfluoro- measured in amniotic fluid (Stein et al. 2012) Lacorte S, Porte C. 2014. Perfluorinated chemicals: octanoate (PFOA) and their associations with human compared with 1.1 ng/mL PFOS in the present differential toxicity, inhibition of aromatase activity semen quality measurements. Reprod Toxicol study. This discrepancy probably stems from and alteration of cellular lipids in human placental 33:419–427. cells. Toxicol Appl Pharmacol 277:124–130. Schindler AE. 1982. Hormones in Human Amniotic Fluid. the phaseout of PFOS in the early 2000s, after Haug LS, Thomsen C, Becher G. 2009. Time trends and Berlin, Germany:Springer-Verlag. our sample collection (1980–1996). Because the influence of age and gender on serum concen- Scott HM, Mason JI, Sharpe RM. 2009. Steroidogenesis of the higher concentration, we were therefore trations of perfluorinated compounds in archived in the fetal testis and its susceptibility to disruption also able to detect PFOS in 98% of our samples human samples. Environ Sci Technol 43:2131–2136. by exogenous compounds. Endocr Rev 30:883–925. compared with 32% in the recent U.S. study Hougaard KS, Larsen AD, Hannerz H, Andersen  AM, Specht IO, Hougaard KS, Spanò M, Bizzaro  D, (Stein et al. 2012) despite a similar limit of Jørgensen  KT, Toft GV, et  al. 2014. Socio- Manicardi  GC, Lindh CH, et  al. 2012. Sperm DNA occupational class, region of birth and maternal integrity in relation to exposure to environmental detection and methodology used. age: influence on time to detection of cryptorchidism perfluoroalkyl substances—a study of spouses of (undescended testes): a Danish nationwide register pregnant women in three geographical regions. Conclusions study. BMC Urol 14:23; doi:10.1186/1471-2490-14-23. Reprod Toxicol 33:577–583. Associations of PFOS with INSL3 and steroid Jauniaux E, Gulbis B, Acharya G, Thiry P, Rodeck C. Stein CR, Wolff MS, Calafat AM, Kato K, Engel SM. hormone concentrations in amnionic fluid 1999. Maternal tobacco exposure and cotinine 2012. Comparison of polyfluoroalkyl compound suggest that prenatal PFOS exposure may have levels in fetal fluids in the first half of pregnancy. concentrations in maternal serum and amniotic Obstet Gynecol 93:25–29. fluid: a pilot study. Reprod Toxicol 34:312–316. affected fetal Leydig cell function in our study Jensen MS, Anand-Ivell R, Nørgaard-Pedersen B, Toft G, Jönsson BA, Lindh CH, Giwercman A, Spano M, population. However, prenatal PFOS concen- Jönsson  BAG, Bonde JP, Hougaard DM, et  al. Heederik D, et al. 2012. Exposure to perfluorinated trations were not associated with cryptorchi- 2015. Amniotic fluid phthalate levels and male fetal compounds and human semen quality in Arctic and dism or hypospadias. Additional studies are gonad function. Epidemiology 26:91–99. European populations. Hum Reprod 27:2532–2540. needed to confirm associations between PFOS Jensen MS, Nørgaard-Pedersen  B, Toft  G, van de Beek C, Thijssen JH, Cohen-Kettenis PT, van and fetal INSL3 and steroid hormone levels, Hougaard  DM, Bonde JP, Cohen A, et  al. 2012. Goozen  SH, Buitelaar JK. 2004. Relationships Phthalates and perfluoro octanesulfonic acid in between sex hormones assessed in amniotic fluid, evaluate potential mechanisms, and determine human amniotic fluid: temporal trends and timing and maternal and umbilical cord serum: what is the whether the estimated differences in fetal of amniocentesis in pregnancy. Environ Health best source of information to investigate the effects hormone levels are associated with long-term Perspect 120:897–903; doi:10.1289/ehp.1104522. of fetal hormonal exposure? Horm Behav 46:663–669. consequences for male reproductive health. Joensen UN, Bossi R, Leffers H, Jensen AA, Vested A, Ramlau-Hansen CH, Olsen SF, Bonde  JP, Skakkebæk NE, Jørgensen N. 2009. Do perfluoroalkyl Kristensen  SL, Halldorsson TI, et  al. 2013. RefeRences compounds impair human semen quality? Environ Associations of in utero exposure to perfluorinated Health Perspect 117:923–927; doi:10.1289/ehp.0800517. alkyl acids with human semen quality and repro- Joensen UN, Veyrand B, Antignac JP, Blomberg ductive hormones in adult men. Environ Health Anand-Ivell R, Ivell R. 2014. Insulin-like factor 3 as a Jensen  M, Petersen JH, Marchand P, et  al. 2013. Perspect 121:453–458; doi:10.1289/ehp.1205118. monitor of endocrine disruption. Reproduction PFOS (perfluorooctanesulfonate) in serum is nega- Vesterholm JD, Christensen J, Virtanen HE, 147:R87–R95. tively associated with testosterone levels, but not Skakkebæk  NE, Main KM, Toppari  J, et  al. 2014. Anand-Ivell R, Ivell R, Driscoll D, Manson J. 2008. with semen quality, in healthy men. Hum Reprod No association between exposure to perfluori- Insulin-like factor 3 levels in amniotic fluid of human 28:599–608. nated compounds and congenital cryptorchidism: male fetuses. Hum Reprod 23:1180–1186. Kalfa N, Philibert P, Sultan C. 2009. Is hypospadias a a nested case–control study among 215 boys from Anand-Ivell R, Tremellen K, Dai Y, Heng K, Yoshida M, genetic, endocrine or environmental disease, or Denmark and Finland. Reproduction 147:411–417. Knight PG, et al. 2013. Circulating Insulin-like factor still an unexplained malformation? Int J Androl Wan HT, Zhao YG, Wong MH, Lee KF, Yeung WS, 3 (INSL3) in healthy and infertile women. Hum 32:187–197. Giesy JP, et al. 2011. Testicular signaling is the poten- Reprod 28:3093–3102. Knight GJ, Palomaki GE. 2003. Epidemiologic monitoring tial target of perfluorooctanesulfonate-mediated Anand-Ivell R, Wohlgemuth J, Haren MT, Hope PJ, of prenatal screening for neural tube defects and subfertility in male mice. Biol Reprod 84:1016–1023. Hatzinikolas  G, Wittert G, et  al. 2006. Peripheral Down syndrome. Clin Lab Med 23:531–551. Zhao B, Li L, Liu J, Li H, Zhang C, Han P, et  al. 2014. INSL3 concentrations decline with age in a Knudsen LB, Olsen J. 1998. The Danish Medical Birth Exposure to perfluorooctane sulfonate in  utero large population of Australian men. Int J Androl Registry. Dan Med Bull 45:320–323. reduces testosterone production in rat fetal Leydig 29:618–626. Kraugerud M, Zimmer KE, Ropstad E, Verhaegen S. cells. PLoS One 9:e78888; doi:10.1371/journal. Armitage JM, Schenker U, Scheringer M, Martin JW, 2011. Perfluorinated compounds differentially pone.0078888. | | 156 volume 124 number 1 January 2016 • Environmental Health Perspectives http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Environmental Health Perspectives Pubmed Central

Perfluorooctane Sulfonate Concentrations in Amniotic Fluid, Biomarkers of Fetal Leydig Cell Function, and Cryptorchidism and Hypospadias in Danish Boys (1980–1996)

Loading next page...
 
/lp/pubmed-central/perfluorooctane-sulfonate-concentrations-in-amniotic-fluid-biomarkers-ZVF2tVXAYA

References (42)

Publisher
Pubmed Central
ISSN
0091-6765
eISSN
1552-9924
DOI
10.1289/ehp.1409288
Publisher site
See Article on Publisher Site

Abstract

A Section 508–conformant HTML version of this article Research Children’s Health is available at http://dx.doi.org/10.1289/ehp.1409288. Perfluorooctane Sulfonate Concentrations in Amniotic Fluid, Biomarkers of Fetal Leydig Cell Function, and Cryptorchidism and Hypospadias in Danish Boys (1980–1996) 1,2 3 4 5 5 Gunnar Toft, Bo A.G. Jönsson, Jens Peter Bonde, Bent Nørgaard-Pedersen, David M. Hougaard, 5 3 6 6 7,8 Arieh Cohen, Christian H. Lindh, Richard Ivell, Ravinder Anand-Ivell, and Morten S. Lindhard 1 2 Department of Occupational Medicine, and Department of Clinical Epidemiology, Aarhus University Hospital, Aarhus, Denmark; 3 4 Division of Occupational and Environmental Medicine, Lund University, Lund, Sweden; Department of Occupational and Environmental Medicine, Copenhagen University Hospital, Bispebjerg, Copenhagen, Denmark; Danish Center for Neonatal Screening, Department of Clinical Biochemistry and Immunology, Statens Serum Institute, Copenhagen, Denmark; School of Biosciences, University of 7 8 Nottingham, Nottingham, United Kingdom; Department of Pediatrics, Regional Hospital of Randers, Randers, Denmark; Perinatal Epidemiology Research Unit, Department of Pediatrics, Aarhus University Hospital, Skejby, Denmark although these results were not corroborated Background : Exposure to perfluorooctane sulfonate (PFOS) may potentially disturb fetal Leydig in a recent study (Joensen et al. 2013), and cell hormone production and male genital development. other aspects of semen quality seem to be o Bsevitcej : We aimed to study the associations between levels of amniotic fluid PFOS, fetal steroid unaffected (Joensen et al. 2009, 2013; Toft hormone, and insulin-like factor 3 (INSL3) and the prevalence of cryptorchidism and hypospadias. et al. 2012). Apart from an inverse association between PFOS exposure and testosterone Methods : Using the Danish National Patient Registry, we selected 270 cryptorchidism cases, 75 hypospadias cases, and 300 controls with stored maternal amniotic fluid samples available in a level in one study (Joensen et al. 2013), Danish pregnancy-screening biobank (1980–1996). We used mass spectrometry to measure PFOS no significant associations between PFOS in amniotic fluid from 645 persons and steroid hormones in samples from 545 persons. INSL3 exposure and reproductive hormones in adult was measured by immunoassay from 475 persons. Associations between PFOS concentration in men have been observed (Joensen et al. 2009; amniotic fluid, hormone levels, and genital malformations were assessed by confounder-adjusted Raymer et al. 2012; Specht et al. 2012). linear and logistic regression. There have been fewer studies on the r esults : The highest tertile of PFOS exposure (> 1.4 ng/mL) in amniotic fluid was associated with effects of PFOS exposure on fetal steroidogen - a 40% (95% CI: –69, –11%) lower INSL3 level and an 18% (95% CI: 7, 29%) higher testosterone esis, and later reproductive function. A study level compared with the lowest tertile (< 0.8 ng/mL). Amniotic fluid PFOS concentration was not on rat fetal Leydig cell function indicated that associated with cryptorchidism or hypospadias. 20 mg/kg/day exposure of pregnant rats from c onclusions : Environmental PFOS exposure was associated with steroid hormone and INSL3 gestational days 11 to 19 was associated with concentrations in amniotic fluid, but was not associated with cryptorchidism or hypospadias in reduced testosterone production, reduced fetal our study population. Additional studies are needed to determine whether associations with fetal Leydig cell number, and decreased expres- hormone levels may have long-term implications for reproductive health. sion of steroidogenic enzymes (Zhao et al. c itation : Toft G, Jönsson BA, Bonde JP, Nørgaard-Pedersen B, Hougaard DM, Cohen A, 2014). Also, zebrafish embryos showed altered Lindh CH, Ivell R, Anand-Ivell R, Lindhard MS. 2016. Perfluorooctane sulfonate concentra- expression of stereoidogenic enzymes after tions in amniotic fluid, biomarkers of fetal Leydig cell function, and cryptorchidism and hypo- exposure to up to 500 μg/L PFOS from 4 spadias in Danish boys (1980–1996). Environ Health Perspect 124:151–156; http://dx.doi. to 120 hr past fertilization (Du et al. 2013). org/10.1289/ehp.1409288 Vested et al. (2013) studied whether prenatal exposure to PFOS (measured in maternal Introduction highest concentrations have been observed serum) was associated with adult male repro- Perfluorooctane sulfonate (PFOS) has among workers at facilities producing ductive function among 169 Danish men until recently been widely used in a variety PFOS, with mean PFOS concentrations of born in 1988–1989. Although prenatal PFOS of applications, especially in surface 1,000–2,000 ng/mL, whereas general popula- coatings used to make products water- and tions on average had concentrations of about Address correspondence to G. Toft, Department of oil- resistant. PFOS is highly biopersistent 35 ng/mL PFOS around the peak exposure Clinical Epidemiology, Aarhus University Hospital, Oluf Palmes Allé 43-45, 8200 Aarhus N, Denmark. with a half-life in humans of about 5 years period (Lau et al. 2004). Apart from the Telephone: 45 871 68202. E-mail: gunnar.toft@ (Olsen et al. 2007). PFOS was added to present study population (Jensen et al. 2012), clin.au.dk Annex B of the Stockholm Convention on PFOS has to our knowledge been measured We are grateful to Å. Amilon and A. Kristensen for Persistent Organochlorine Pollutants in in amniotic fluid only in one recent American skillful technical assistance. 2009, and the production and use of PFOS study of 28 women (Stein et al. 2012). at Th This study was initiated by generous grants from the have been regulated in Europe since 2008 study indicated a PFOS concentration in Danish Environmental Protection Agency, the Danish Ministry of Interior and Health, Research Centre for (European Commission 2006). In Norway, amniotic fluid about 20-fold less than in Environmental Health’s Fund, the Swedish Council PFOS levels rose through the mid-1990s, maternal serum. for Working Life and Social Research, Skåne County and have declined since 2000 (Haug et al. Experimental studies of the effects of Council’s Research and Development Foundation, 2009). However, model-based estimates PFOS on adult male reproductive function and the Medical Faculty at Lund University, Sweden. of future environmental PFOS exposures have reported reduced testosterone produc- The sponsors had no part in study design, data col- suggest a slower decay in temperate regions tion, altered gonadotropin secretion, and lection, analysis, or preparation of the manuscript, and they are not responsible for the scientific content (Armitage et al. 2009). PFOS has been reduced epididymal sperm in PFOS-treated and the conclusions expressed. detected in human populations from all rats and mice (López-Doval et al. 2014; The authors declare they have no actual or potential over the world, but considerable variation in Wan et al. 2011). Studies of human adults competing financial interests. exposure has been observed between popula- have shown associations between PFOS Received: 2 October 2014; Accepted: 1 June 2015; tions (Lau et al. 2004). Most previous studies exposure and the proportion of normal sperm Advance Publication: 5 June 2015; Final Publication: have measured PFOS in serum, and the cells (Joensen et al. 2009; Toft et al. 2012) 1 January 2016. | | Environmental Health Perspectives • volume 124 number 1 January 2016 151 Toft et al. exposure was not associated with reproduc- pregnancy-screening registry was recorded by detail by Jensen et al. (2012). Coefficients of tive hormones or semen quality, prenatal the personal identification number (unique to variation were 11% for PFOS and 9% for exposure to perfluorooctanoate (PFOA) was each Danish citizen) of the pregnant woman. cotinine; the limit of detection (LOD) was associated with reduced semen quality and We used these unique identifiers to obtain 0.20 ng/mL for PFOS and cotinine, deter- higher levels of luteinizing hormone (LH) obstetric data on the pregnancies from the mined as the concentrations corresponding and follicle- stimulating hormone (FSH) in Danish Medical Birth Registry (Knudsen to three times the standard deviation of the young adult men. These results suggest and Olsen 1998), including gestational age responses in chemical blanks. We analyzed that the prenatal period may be sensitive to at birth, singleton or multiple birth, maternal the samples using a liquid chromatograph environmental exposures to perfluoroalkyl parity, and birth weight and Apgar score of (LC; model UFLCXR, Shimadzu Corp). The substances (PFAS). the infant. In addition, we used the unique LC was connected to a hybrid triple quadru- Evidence of effects on fetal Leydig cell identifiers in the Danish Civil Registration pole linear ion trap tandem mass spectrom- hormone production has led others to System to identify male offspring from the eter (LC/MS/MS) equipped with a turbo ion hypothesize that PFOS exposure might pregnancies (Pedersen et al. 2006). spray source (QTRAP 5500; AB Sciex). affect androgen- and insulin-like factor 3 The Danish Regional Ethics Committee, We assayed the steroid hormones (INSL3)–dependent testicular descent (Bay the Danish National Board of Health, and testosterone, androstenedione, proges - et al. 2011) and scrotal fusion (Kalfa et al. the Danish Data Protection Agency approved terone, 17-OH-progesterone, and cortisol 2009). Only one previous study has evalu- the study. The use of the biobank for research in amniotic fluid at the Danish Center for ated the risk of cryptorchidism in relation to purposes has been approved, and additional Neonatal Screening, Department of Clinical PFAS exposure; this study did not observe informed consent from the study subjects for Biochemistry and Immunology, the State associations between PFAS level in cord this specific project was neither recommended Serum Institute, Copenhagen, Denmark. The blood and cryptorchidism (Vesterholm et al. nor required. online extraction LC-MS system consisted 2014). However, due to a limited study size Case–control definitions and ascertain- of an Aria TLX2 system (Thermo Scientific) of 59 cryptorchidism cases and 108 matched ment. Controls were randomly selected from with two Agilent 1100 binary pumps and controls, the power to show an association the 25,105 amniotic fluid samples belonging two Agilent 1200 quaternary pumps (Agilent) was limited. To our knowledge, no previous to live-born male offspring pregnancies in the connected to a Thermo TSQ Ultra triple quad - studies have evaluated the potential associa- screening database with complete obstetric rupole mass spectrometer equipped with an tion between in utero exposure to PFOS and data in the Danish Medical Birth Registry. APCI ion source. Extraction was performed fetal hormone level or hypospadias. The number of controls (n = 412) was chosen using a Cyclone P 0.5 × 50 mm Turboflow The aim of the present study was to to roughly equal the largest case group column (Thermo Scientific), and analytical determine whether in utero PFOS levels are consisting of 404 boys with cryptorchidism. separation was achieved using Kinetix 2.6 μ associated with altered Leydig cell function, Cryptorchidism cases had both a diagnosis 2.1 × 50 mm C18 columns (Phenomenex). as indicated by steroid hormone and INSL3 of undescended testis according to the All calibrators, controls, internal standards, concentrations in amniotic fluid, and to International Classification of Diseases, 8th and and micro-titer plates were purchased from evaluate whether these potential effects on 10th Revisions [ICD-8: 75210, 75211, 75219; PerkinElmer via their CHS steroid profiling kit hormone levels were associated with genital ICD-10: Q53, Q531(A), Q532(A), Q539] for mass spectrometry (PerkinElmer). Formic malformation in offspring. and a corrective surgical procedure according acid was purchased from Merck. Ammonium to the Surgery and Treatment Classification acetate and zinc sulphate heptahydrate Methods of the Danish National Board of Health were purchased from Sigma-Aldrich. Water Study population and amniotic fluid samples. (STC: 55600, 55640) or the Nordic was purified using a water purification unit The study population has been described Classic fi ation of Surgical Procedures (NCSP: from Millipore. We used a volume of 50 μL previously in detail (Jensen et al. 2012). KKFH00, KKFH01, KKFH10) recorded amniotic fluid for the assay. The LODs and Briefly, we used amniotic fluid samples from in the Danish National Patient Registry the intra- and interassay coefficients of varia - a Danish biobank maintained at the State (DNPR). Boys with a registry entry of tion were as follows: testosterone, 0.1 nmol/L, Serum Institute in Copenhagen (http://www. inguinal hernia repair (STC: 40620, 40640; 9 and 10%, respectively; androstenedione, ssi.dk). The biobank holds samples from a NCSP: KJAB00-KJAB97) were excluded 0.3 nmol/L, 8 and 9%, respectively; proges- pregnancy-screening registry, including from this case group to avoid iatrogenic terone, 0.4 nmol/L, 10 and 11%, respectively; information on amniotic fluid samples from cryptorchidism secondary to hernia repair. 17-OH-progesterone, 0.4 nmol/L, 9 and 10%, 63,882 pregnancies covering the period We included all boys that fulfilled these respectively; and cortisol, 4.4 nmol/L, 10 and 1980–1996. After restriction to live-born criteria (404 of 25,105; 1.61%) to maximize 12%, respectively. singleton boys with complete obstetric data, the cryptorchidism case group and overall INSL3 in amniotic fluid was measured 25,105 pregnancies were eligible (Jensen study size. For the hypospadias case group we using a semicompetitive time-resolved fluo- et al. 2012). The amniotic fluid samples included all 109 of the 25,105 boys (0.43%) rometric immunoassay (TRFIA) slightly were centrifuged before routine diagnostic with a diagnosis of hypospadias in the DNPR modified from that described in detail by analyses, and the supernatants were kept (ICD-8: 75220, 75221, 75222, 75228, Anand-Ivell et al. (2006). The laboratory work frozen at –20°C until the present analyses 75229; ICD-10: Q540, Q541, Q542, Q548, was performed at Leibniz Institute for Farm were carried out. Indications for amniocen- Q549). All boys were followed for the afore- Animal Biology, Dummerstorf, Germany. tesis included age ≥ 35 years and/or increased mentioned diagnoses and surgery entries in The modifications involved the replacement risk of severe malformations or Down the DNPR from birth until November 2008. of the original rat antiserum by a new poly- syndrome based on results from maternal Measurement of chemical compounds clonal antiserum (no. #RIA5) raised in rabbits serum analyses. We calculated the gestational and hormones. During all chemical analyses, (IMVS Antibody Services) against the same week of amniocentesis based on the date of laboratory technicians were blinded to chemically synthesized human INSL3 as amniocentesis, the date of birth, and the esti- case–control status and to levels of analytes previously, at a final dilution of 1:10,000. As mated gestational age at birth as described measured by others. We assayed PFOS and tracer, we used the same Europium-labeled by Jensen et al. (2012). Each sample in the cotinine in amniotic fluid as described in human INSL3 as described (Anand-Ivell | | 152 volume 124 number 1 January 2016 • Environmental Health Perspectives PFOS, Leydig cells, and genital malformations et al. 2006). Accordingly, also the secondary multiple linear or logistic regression analyses evaluate whether year of amniocentesis influ - goat-anti-rabbit antibody (Rockland for continuous and categorical outcomes, enced these associations, we made supplemen- Immunochemicals Inc.) used to coat the plates respectively. We divided PFOS exposure into tary stratified analysis by year of amniocentesis substituted for the original anti-rat antibody. tertiles to quantify the difference between the (1980–1986, 1987–1992, and 1992–1996). All other conditions were similar. Standard highest and lowest third of the population We also restricted the analyses to boys curves for measurement of amniotic fluid were and to evaluate whether any marked devia- with no other congenital malformations to constructed using serial dilutions of human tions from a linear association was evident. exclude cases with syndromes and chromo- INSL3 in EDTA–phosphate-buffered saline Differences in median hormone level somal abnormalities. (Anand-Ivell et al. 2008). A volume of 100 μL between the PFOS tertiles were evaluated PFOS and all hormone data were trans- amniotic fluid was used for the assay, and by the nonparametric Kruskal–Wallis test. formed by the natural logarithm (ln) to the limit of detection for the modified assay Then adjusted linear regression analyses were improve normality of their distribution of was 0.01 ng/mL, with inter- and intraplate performed evaluating tertile differences. We residuals in the regression analyses. coefficients of variation of < 8% and < 1%, additionally performed regression analyses Results respectively, across the range. There was no on a continuous PFOS variable to test for a cross-reactivity detectable across the physi- linear trend. Of the original 925 included subjects in the ological range with the structurally related Data are presented as difference from the cohort, we could not locate samples in the peptides, insulin, IGF-1 (insulin-like growth lowest tertile and β [95% confidence interval biobank for 60 subjects, and 220 had insuf- factor-1), and relaxin. There was also no (CI)] for linear regression models both unad- ficient volume for PFOS measurements. cross-reactivity detectable with rat INSL3, justed and adjusted. We performed inter- Thus, the study population consists of 645 and only 10% cross-reactivity with bovine action analyses including a ln-PFOS × case/ mother–child pairs with information on INSL3 at the highest values. Spiking experi- control group term in the regression models fetal PFOS exposure and case status of the ments into normal male and female human to evaluate whether combined analyses of children, including 270 cases of cryptorchi- sera from previous studies (Anand-Ivell et al. cases of cryptorchidism, hypospadias, and dism, 75 cases of hypospadias, and a random 2006, 2013) indicated 122.9 + 1.5% and control groups were statistically justifiable. control group of 300. The mean and standard 96.5 + 2.7% recovery, respectively. INSL3 In supplementary analyses, we additionally deviation of maternal characteristics including measurements of human serum following evaluated the association between PFOS and age and gestational age of amniocentesis and five successive freeze–thaw cycles showed no hormones separately for controls, cryptorchi- distribution of smoking prevalence did not significant change (data not shown). dism cases, and hypospadias cases to evaluate differ markedly between the case and control Statistical analysis. We imputed values whether the association varied across groups. groups (Table 1). As expected, the cryptor- below the LOD of our chemical assays with a We a priori decided to adjust all hormone chidism and hypospadias cases weighed less random value between the LOD and LOD/2 analyses for gestational age of amniocentesis and were on average born slightly earlier as a simplified method of maintaining vari - (continuous, weeks), maternal age (years), and than the control group. Also, the distribu- ability in values below LOD. For the INSL3 smoking (using amniotic fluid specific levels tion of calendar year of amniocentesis differed assay, 58 of 543 (10.7%) samples were below as a biomarker for nonsmoker: < 25 ng/mL between case groups (Table 1). the LOD (0.01 ng/mL). For the PFOS assay, cotinine; passive smoker: 25–85 ng/mL; and Amniotic fluid steroid hormone levels were 10 of 645 (1.6%) samples were below the smoker: > 85 ng/mL) (Jauniaux et al. 1999). measured on 545 pregnant women with suffi - LOD (0.2 ng/mL). For the steroid assay Analysis of combined associations between cient sample volume (84% of cryptorchidism (n = 574), we used the instrument readout PFOS and hormone levels across case and cases, 100% of hypospadias cases, and 81% for values below the LOD; two (0.3%) testos- control groups were additionally adjusted for of control samples). Testosterone level was terone values < 0.1 nmol/L, no androstene- case–control status. The logistic regression positively associated with PFOS exposure with dione values < 0.3 nmol/L, no progesterone analyses of cryptorchidism and hypospadias 18% higher testosterone (95% CI: 7, 29%) values < 0.4 nmol/L, no 17-OH-progesterone was adjusted for gestational age of amnio- in the highest PFOS tertile compared with values < 0.4 nmol/L, and six (1.0%) cortisol centesis (continuous, weeks), year of amnio- the lowest in the overall population (all three values < 4.4 nmol/L. centesis (three groups), maternal age (years), groups combined). Also, the linear trend test INSL3 levels are highly dependent on gestational age (weeks), birth weight (grams), showed a significant positive association in both gestational age at amniocentesis with a peak and smoking (cotinine groups). To further the unadjusted and adjusted analysis (Table 2). around week 15 (Anand-Ivell et al. 2008). We calculated multiple of the median Table 1. Characteristics of the included pregnancies and boys by case–control status among Danish (MoM) values for INSL3 by gestational age pregnant women with amniocentesis (1980–1996). at amniocentesis to reduce this dependence Control Cryptorchidism Hypospadias (Knight and Palomaki 2003). We used the Characteristics (n = 300) (n = 270) (n = 75) random sample control group to estimate the Maternal age at birth [years (mean ± SD)] 32.6 ± 5.3 32.8 ± 5.3 31.3 ± 5.8 median level for each week, and all INSL3 Gestational week of amniocentesis (mean ± SD) 15.7 ± 1.3 15.9 ± 1.8 15.6 ± 1.6 values (cases and controls) were then divided Gestational age at birth [weeks (mean ± SD)] 39.6 ± 1.5 39.2 ± 2.2 38.7 ± 2.8 Birth weight [g (mean ± SD)] 3,535 ± 561 3,370 ± 682 3,216 ± 827 by the median of the corresponding week to Year of amniocentesis (%) calculate the MoM value. Weeks 11–13 were 1980–1984 34 40 17 pooled because of limited number of observa- 1985–1990 34 32 37 tions, and the median value from week 21 1991–1996 33 29 45 (controls) was applied to week 22 (cases) Smoking (%) because there were no controls in week 22. Nonsmoker: cotinine < 25 ng/mL 61 60 58 All statistical analyses were presented for both Passive smoker: cotinine 25–85 ng/mL 6 4 7 the raw and the MoM INSL3 measures. Smoker: cotinine ≥ 85 ng/mL 33 36 35 Congentital malformations (%) 6 14 20 Associations between PFOS exposure and the included outcomes were evaluated by Malformations other than cryptorchidism and hypospadias. | | Environmental Health Perspectives • volume 124 number 1 January 2016 153 Toft et al. Also, androstendione, progesterone, and 1992–1996) also did not indicate any possibility that prenatal PFOS exposure may 17-OH-progesterone, and cortisol, but not association between PFOS exposure and increase fetal steroid hormone production. DHEAS (dehydroepiandrosterone sulfate), odds for cryptorchidism or hypospadias The inverse association between prenatal were positively associated with PFOS exposure (data not shown). Furthermore, analyses concentrations of PFOS and INSL3 may be (Table 2). INSL3 was measured in amniotic restricted to boys with no other congenital consistent with a direct effect on fetal Leydig fluid samples from 475 pregnancies (74% malformations (numbers given in Table 1) cells, because, based on the current knowl- of cryptorchidism cases, 84% of hypospa- produced essentially unchanged results (data edge, INSL3 is produced only by male fetal dias cases, and 71% of controls). INSL3 not shown). Leydig cells, whereas steroid hormones are was negatively associated with PFOS, with a also produced by the fetal adrenal gland Discussion 40% lower INSL3 concentration (95% CI: (Anand-Ivell and Ivell 2014). Fetal urine is –69, –11%) estimated for the highest tertile PFOS concentrations measured in amniotic believed to be the primary source of steroid compared with the lowest based on the overall fluid were associated with higher steroid hormones in amniotic fluid from the second adjusted model, and approximately the same hormone levels and lower INSL3 in the trimester onward (Schindler 1982). The magnitude when evaluated as INSL3 MoM. combined study population but were not asso- source or sources of steroids in the amniotic The inverse associations were confirmed by ciated with cryptorchidism or hypospadias fluid are unknown, but could include the linear trend analyses (Table 2). (270 and 75 cases, respectively, compared umbilical cord or placenta (Schindler 1982). The association between PFOS exposure, with 300 controls). Progesterone and estradiol are produced testosterone, and INSL3 was similar in the Our results are consistent with those of in considerable amounts in the placenta, case and control groups as indicated by Vesterholm et al. (2014), who reported no and may be partly transferred to the fetus p-values of product interaction terms in the association of cord blood concentrations of (Pasqualini 2005). A recent in vitro study regression models (Table 2) and estimates PFOS or other PFAS with cryptochidism in using human placental cells indicated that (Table 3). However, in the stratified analysis, 215 Danish and Finnish boys. placental aromatase activity may be altered fewer statistical significant associations were To our knowledge, the present study is the after exposure to PFOS (Gorrochategui et al. observed, probably because of a lower number first human study evaluating potential asso- 2014). Thus, higher testosterone levels in of cases (Table 3). ciation between fetal exposure to PFOS and amniotic fluid in our study population might PFOS concentrations in amniotic fluid biomarkers of human fetal steroid and INSL3 be attributable to inhibited aromatase activity were not associated with cryptorchidism levels. A study using the human adrenocor- in the placenta. However, whereas andros- or hypospadias based on adjusted logistic tical carcinoma (H295R) in vitro cell assay tendione and DHEAS concentrations were regression models (Table 4). Supplementary reported increased testosterone, progesterone, weakly correlated between second-trimester analysis stratifying these associations by and estradiol secretion after PFOS exposure amniotic fluid and maternal serum samples sampling year (1980–1986, 1987–1992, (Kraugerud et al. 2011), supporting the from mothers expecting male offspring, Table 2. PFOS exposure (ng/mL) and markers of Leydig cell function in amniotic fluid among Danish pregnant women with amniocentesis (1980–1996). a a Hormone and PFOS exposure n Median hormone level % difference (95% CI) Hormone and PFOS exposure n Median hormone level % difference (95% CI) Testosterone (nmol/L) 545 17-OH-Progesterone (nmol/L) 545 < 0.8 ng/mL PFOS 0.73 Reference < 0.8 ng/mL PFOS 4.94 Reference 0.8–1.4 ng/mL PFOS 0.81 9 (–2, 20) 0.8–1.4 ng/mL PFOS 5.22 7 (–1, 13) > 1.4 ng/mL PFOS 0.91 18 (7, 29) > 1.4 ng/mL PFOS 6.18 18 (11, 26) p-Value 0.002 p-Value < 0.001 c c Ln-PFOS (crude) 0.14 (0.08, 0.22) Ln-PFOS (crude) 0.17 (0.12, 0.22) c c Ln-PFOS (adjusted) 0.16 (0.09, 0.23) Ln-PFOS (adjusted) 0.15 (0.11, 0.20) d d p-Value for interaction 0.68 p-Value for interaction 0.92 DHEAS (nmol/L) 545 Cortisol (nmol/L) 545 < 0.8 ng/mL PFOS 17.0 Reference < 0.8 ng/mL PFOS 15.9 Reference 0.8–1.4 ng/mL PFOS 17.4 5 (–10, 20) 0.8–1.4 ng/mL PFOS 17.0 9 (–0, 17) > 1.4 ng/mL PFOS 17.4 2 (–14, 17) > 1.4 ng/mL PFOS 22.4 28 (19, 37) b b p-Value 0.93 p-Value < 0.001 c c Ln-PFOS (crude) 0.06 (–0.03, 0.16) Ln-PFOS (crude) 0.26 (0.20, 0.32) c c Ln-PFOS (adjusted) 0.07 (–0.03, 0.16) Ln-PFOS (adjusted) 0.19 (0.13, 0.25) d d p-Value for interaction 0.03 p-Value for interaction 0.96 Androstendione (nmol/L) 545 INSL3 (ng/mL) 475 < 0.8 ng/mL PFOS 2.62 Reference < 0.8 ng/mL PFOS 0.14 Reference 0.8–1.4 ng/mL PFOS 2.72 8 (0, 17) 0.8–1.4 ng/mL PFOS 0.12 –21 (–48, 7) > 1.4 ng/mL PFOS 3.07 17 (8, 25) > 1.4 ng/mL PFOS 0.09 –40 (–69,–11) b b p-Value 0.001 p-Value < 0.001 c c Ln-PFOS (crude) 0.15 (0.10, 0.20) Ln-PFOS (crude) –0.35 (–0.54, –0.16) c c Ln-PFOS (adjusted) 0.15 (0.10, 0.21) Ln-PFOS (adjusted) –0.21 (–0.40, –0.02) d d p-Value for interaction 0.22 p-Value for interaction 0.98 Progesterone (nmol/L) 545 INSL3 (MoM) 475 < 0.8 ng/mL PFOS 163.7 Reference < 0.8 ng/mL PFOS 1.45 Reference 0.8–1.4 ng/mL PFOS 170.8 11 (0, 23) 0.8–1.4 ng/mL PFOS 1.13 –21(–49, 8) > 1.4 ng/mL PFOS 186.4 22 (11, 34) > 1.4 ng/mL PFOS 0.85 –39 (–68,–10) b b p-Value 0.001 p-Value 0.07 c c Ln-PFOS (crude) 0.21 (0.14, 0.28) Ln-PFOS (crude) –0.17 (–0.36, 0.01) Ln-PFOS (adjusted) 0.21 (0.14, 0.29) Ln-PFOS (adjusted) –0.20 (–0.38, –0.01) p-Value for interaction 0.07 p-Value for interaction 0.82 a b c Adjusted for gestational age of amniocentesis, maternal age, smoking (cotinine groups), and case or control status unless otherwise indicated. Kruskal–Wallis rank test. Estimates represent the percent difference in hormone concentrations with a 1% increase in PFOS. Interaction terms (lnPFOS × case–control status) in adjusted linear regression models. Multiple of the median (MoM) values are calculated as specified in the methods. | | 154 volume 124 number 1 January 2016 • Environmental Health Perspectives PFOS, Leydig cells, and genital malformations testosterone concentrations were not corre- as a proxy for fetal exposure (Vested et al. socio-occupational class (Hougaard et al. lated, suggesting limited maternal to fetal 2013; Vesterholm et al. 2014). Although the 2014). However, our register-based ascertain- transfer of testosterone relative to other steroid presence in cord blood indicates that PFOS ment of cryptorchidism and hypospadias did hormones (van de Beek et al. 2004). is transferred across the placenta to the fetus not include milder forms of these outcomes In utero exposure of rats to di(n-butyl) (Vesterholm et al. 2014), little is known that may spontaneously disappear during phthalate, a compound with antiandrogenic about actual exposure level of the fetus, and the first years of life, so we cannot exclude properties, caused reduced INSL3 expression we propose that measurement in amniotic that PFOS exposure may be related to milder in fetal Leydig cells and an increase in cryptor- fluid is the closest we can get to actual forms of genital malformations. chidism, and INSL3 has been suggested as an measurement of fetal exposure at the relevant PFOS and hormone concentrations were endogenous marker of endocrine disruption time window of exposure (Jensen et al. 2012). measured concurrently in amniotic fluid (Anand-Ivell and Ivell 2014). A recent study The strengths of the present study include samples, so we cannot confirm the temporal from our group using the same study popula- the use of a large biobank of amniotic fluid relation between PFOS exposure and tion as in the present study indicated that di(2- samples linked with medical birth registries the outcomes. ethylhexyl) phthalate (DEHP) metabolites but to produce a nested case–control sample The analysis of associations between PFOS not diisononyl phthalate (DiNP) metabolites of cryptorchidism and hypospadias cases and hormone levels in the combined case and were also related to decreased INSL3 level selected purely based on clinical diagnoses control groups may have produced biased (Jensen et al. 2015). However, regarding that are unlikely to be related to the exposure. estimates of associations in the source popula- PFOS exposure, a recent study indicated that Although it might be speculated that socio- tion, if associations in the case groups differed fetal rat Leydig cell INSL3 production was occupational class could be related to PFOS from the population as a whole. Our strati- unaltered after exposure to 20 mg/kg/day exposure and detection rate of cryptorchi- fied and interaction analyses did not indicate PFOS from gestational day 11 to 19 (Zhao dism and hypospadias, we have previously major differences among the case and control et al. 2014). There may be species differences shown that among Danish boys the time to groups, though sample sizes were small and in effects of PFOS on INSL3 production, detection of cryptorchidism is unrelated to estimates were imprecise, particularly for the partly because of the much shorter half-life of PFOS in rats than in humans (Lau et al. Table 3. PFOS exposure (ng/mL) and markers of Leydig cell function in amniotic fluid among Danish 2004). In our study population, prenatal pregnant women with amniocentesis (1980–1996) stratified by case–control group. PFOS exposure was associated with lower Controls Cryptorchism Hypospadias INSL3 in amniotic fluid, but was not asso- % difference % difference % difference ciated with cryptorchidism. Whether lower a a a Hormone and PFOS exposure n (95% CI) n (95% CI) n (95% CI) INSL3 levels can be associated with long-term Testosterone (nmol/L) 242 228 75 effects on male reproductive health is not < 0.8 ng/mL PFOS Reference Reference Reference known. However, it is of interest to note that 0.8–1.4 ng/mL PFOS 10 (–5, 24) 14 (–4, 32) –12 (–47, 24) a recent follow-up study found lower sperm > 1.4 ng/mL PFOS 19 (4, 33) 20 (3, 38) –6 (–47, 36) Ln-PFOS (crude) 0.12 (0.03, 0.21) 0.18 (0.08, 0.28) 0.13 (–0.11, 0.37) counts and increased LH and FSH levels Ln-PFOS (adjusted) 0.14 (0.04, 0.23) 0.20 (0.09, 0.31) 0.00 (–0.28, 0.28) in 19- to 21-year-old males at the highest INSL3 213 199 63 tertile of PFOA exposure in utero compared < 0.8 ng/mL PFOS Reference Reference Reference with the lowest (Vested et al. 2013). In that 0.8–1.4 ng/mL PFOS –28 (–72, 16) –5 (–47, 38) –49 (–118, 20) study, PFOA and PFOS was estimated from > 1.4 ng/mL PFOS –37 (–81, 8) –38 (–82, 6) –21 (–110, 69) maternal pregnancy serum samples. Although Ln-PFOS (crude) –0.26 (–0.54, 0.02) –0.39 (–0.68, –0.11) –0.62 (–1.23, 0.00) no association with PFOS was found in the Ln-PFOS (adjusted) –0.22 (–0.51, 0.06) –0.20 (–0.48, 0.07) 0.02 (–0.60, 0.64) Vested et al. (2013) study, PFOS and PFOA INSL3 MoM 213 199 63 < 0.8 ng/mL PFOS Reference Reference Reference were strongly correlated in maternal serum 0.8–1.4 ng/mL PFOS –30 (–73,12) 0 (–44, 45) –60 (–135, 15) samples (r = 0.73). We did not measure > 1.4 ng/mL PFOS –43 (–86, –0) –30 (–75, 15) –20 (–117, 77) PFOA in the present study; thus it cannot be Ln-PFOS (crude) –0.24 (–0.51, 0.03) –0.09 (–0.37, 0.18) –0.28 (–0.87, 0.30) excluded that the associations with INSL3 Ln-PFOS (adjusted) –0.27 (0.54, 0.00) –0.12 (–0.40, 0.17) –0.01 (–0.68, 0.65) may be related to an effect of PFOA rather Adjusted for gestational age of amniocentesis, maternal age, smoking (cotinine groups), and case or control status than PFOS on INSL3. Future studies should unless otherwise indicated. Estimates represent the percent difference in hormone concentrations with a 1% increase evaluate whether low fetal INSL3 level is asso- in PFOS. Multiple of the median (MoM) values are calculated as specified in the methods. ciated with long-term consequences for male reproductive health. Table 4. Odds ratios (95% CI) for cryptorchidism or hypospadias in relation to amniotic fluid PFOS concentration among Danish pregnant women with amniocentesis (1980–1996). We measured PFOS, steroid hormones, and INSL3 in amniotic fluid for several n n Unadjusted Adjusted reasons. Most amniotic fluid samples were Malformation control cases OR (95% CI) OR (95% CI) taken in gestational weeks 15 to 17, close to Cryptorchidism the time window of male genital development 1st tertile 102 86 1.0 (reference) 1.0 (reference) 2nd tertile PFOS 101 91 1.11 (0.74, 1.67) 1.08 (0.71, 1.63) in gestational weeks 8 to 15 (Scott et al. 2009). 3rd tertile PFOS 97 93 1.09 (0.73, 1.63) 1.01 (0.66, 1.53) Contaminant concentrations in amniotic fluid Ln-PFOS 300 270 1.05 (0.81, 1.35) 0.99 (0.75, 1.30) during the first half of pregnancy, when the Hypospadias amniotic u fl id is composed largely of exudates 1st tertile 102 27 1.0 (reference) 1.0 (reference) from fetal blood and fluids, are consid- 2nd tertile PFOS 101 26 1.01 (0.55, 1.81) 0.97 (0.51, 1.87) ered a suitable proxy measure of intrafetal 3rd tertile PFOS 97 22 0.82 (0.44, 1.54) 0.69 (0.35, 1.38) contaminant levels (Beall et al. 2007). Ln-PFOS 300 75 0.97 (0.65, 1.44) 0.87 (0.57, 1.34) Other studies on fetal PFOS exposures a Adjusted for gestational age of amniocentesis, year of amniocentesis (groups), maternal age, gestational age, birth have used maternal serum or cord blood weight, and smoking (cotinine groups). | | Environmental Health Perspectives • volume 124 number 1 January 2016 155 Toft et al. hypospadias group. However, estimates based MacLeod M, Cousins IT. 2009. Modeling the global affect steroidogenesis and viability in the human fate and transport of perfluorooctane sulfonate adrenocortical carcinoma (H295R) in  vitro cell only on the controls, who should be represen- (PFOS) and precursor compounds in relation to assay. Toxicol Lett 205:62–68. tative of the overall population, were similar to temporal trends in wildlife exposure. Environ Sci Lau C, Butenhoff JL, Rogers JM. 2004. The develop- the estimates based on the combined case and Technol 43:9274–9280. mental toxicity of perfluoroalkyl acids and their control groups (Table 3). Bay K, Main KM, Toppari J, Skakkebæk NE. 2011. derivatives. Toxicol Appl Pharmacol 198:231–241. One may speculate that evaporation Testicular descent: INSL3, testosterone, genes López-Doval S, Salgado R, Pereiro N, Moyano  R, from stored samples or differential decay may and the intrauterine milieu. Nat Rev Urol 8:187–196. Lafuente A. 2014. Perfluorooctane sulfonate Beall MH, van den Wijngaard JP, van Gemert MJ, effects on the reproductive axis in adult male rats. explain some of the associations between Ross MG. 2007. Regulation of amniotic fluid Environ Res 134:158–168. hormone level and PFOS in amniotic fluid volume. Placenta 28:824–832. Olsen GW, Burris JM, Ehresman DJ, Froehlich JW, samples. However, PFOS, testosterone, and Du G, Hu J, Huang H, Qin Y, Han X, Wu D, et al. 2013. Seacat  AM, Butenhoff JL, et  al. 2007. Half-life INSL3 levels were only weakly correlated with Perfluorooctane sulfonate (PFOS) affects hormone of serum elimination of perfluorooctanesulfonate, year of amniocentesis (r-values of 0.07, –0.06, receptor activity, steroidogenesis, and expression perfluorohexanesulfonate, and perfluorooctanoate in and –0.05, respectively) and the estimated of endocrine-related genes in  vitro and in  vivo. retired fluorochemical production workers. Environ Environ Toxicol Chem 32:353–360. Health Perspect 115:1298–1305; doi:10.1289/ehp.10009. volume left in the biobank (r-values of 0.06, European Commission. 2006. Directive 2006/122/EC of Pasqualini JR. 2005. Enzymes involved in the forma- –0.06, and –0.02, respectively), suggesting the European Parliament and of the Council of 12 tion and transformation of steroid hormones in that any influence of storage on the observed amending for the 30th Time Council Directive 76/769/ the fetal and placental compartments. J Steroid results would have been minor. EEC on the Approximation of the Laws, Regulations Biochem Mol Biol 97:401–415. The concentration of PFOS in amniotic and Administrative Provisions of the Member States Pedersen CB, Gøtzsche H, Møller JO, Mortensen PB. fluid samples in the present sample was relating to Restrictions on the Marketing and Use of 2006. The Danish Civil Registration System. A cohort Certain Dangerous Substances and Preparations of eight million persons. Dan Med Bull 53:441–449. somewhat higher than a previous U.S. study of (Perfluorooctane Sulfonates). December 2006. Raymer JH, Michael LC, Studabaker WS, Olsen GW, pregnant women from New York, 2005–2008, Official Journal of the European Union 49:32–34. Sloan  CS, Wilcosky T, et  al. 2012. Concentrations where a median of 0.4 ng/mL PFOS was Gorrochategui E, Pérez-Albaladejo E, Casas J, of perfluorooctane sulfonate (PFOS) and perfluoro- measured in amniotic fluid (Stein et al. 2012) Lacorte S, Porte C. 2014. Perfluorinated chemicals: octanoate (PFOA) and their associations with human compared with 1.1 ng/mL PFOS in the present differential toxicity, inhibition of aromatase activity semen quality measurements. Reprod Toxicol study. This discrepancy probably stems from and alteration of cellular lipids in human placental 33:419–427. cells. Toxicol Appl Pharmacol 277:124–130. Schindler AE. 1982. Hormones in Human Amniotic Fluid. the phaseout of PFOS in the early 2000s, after Haug LS, Thomsen C, Becher G. 2009. Time trends and Berlin, Germany:Springer-Verlag. our sample collection (1980–1996). Because the influence of age and gender on serum concen- Scott HM, Mason JI, Sharpe RM. 2009. Steroidogenesis of the higher concentration, we were therefore trations of perfluorinated compounds in archived in the fetal testis and its susceptibility to disruption also able to detect PFOS in 98% of our samples human samples. Environ Sci Technol 43:2131–2136. by exogenous compounds. Endocr Rev 30:883–925. compared with 32% in the recent U.S. study Hougaard KS, Larsen AD, Hannerz H, Andersen  AM, Specht IO, Hougaard KS, Spanò M, Bizzaro  D, (Stein et al. 2012) despite a similar limit of Jørgensen  KT, Toft GV, et  al. 2014. Socio- Manicardi  GC, Lindh CH, et  al. 2012. Sperm DNA occupational class, region of birth and maternal integrity in relation to exposure to environmental detection and methodology used. age: influence on time to detection of cryptorchidism perfluoroalkyl substances—a study of spouses of (undescended testes): a Danish nationwide register pregnant women in three geographical regions. Conclusions study. BMC Urol 14:23; doi:10.1186/1471-2490-14-23. Reprod Toxicol 33:577–583. Associations of PFOS with INSL3 and steroid Jauniaux E, Gulbis B, Acharya G, Thiry P, Rodeck C. Stein CR, Wolff MS, Calafat AM, Kato K, Engel SM. hormone concentrations in amnionic fluid 1999. Maternal tobacco exposure and cotinine 2012. Comparison of polyfluoroalkyl compound suggest that prenatal PFOS exposure may have levels in fetal fluids in the first half of pregnancy. concentrations in maternal serum and amniotic Obstet Gynecol 93:25–29. fluid: a pilot study. Reprod Toxicol 34:312–316. affected fetal Leydig cell function in our study Jensen MS, Anand-Ivell R, Nørgaard-Pedersen B, Toft G, Jönsson BA, Lindh CH, Giwercman A, Spano M, population. However, prenatal PFOS concen- Jönsson  BAG, Bonde JP, Hougaard DM, et  al. Heederik D, et al. 2012. Exposure to perfluorinated trations were not associated with cryptorchi- 2015. Amniotic fluid phthalate levels and male fetal compounds and human semen quality in Arctic and dism or hypospadias. Additional studies are gonad function. Epidemiology 26:91–99. European populations. Hum Reprod 27:2532–2540. needed to confirm associations between PFOS Jensen MS, Nørgaard-Pedersen  B, Toft  G, van de Beek C, Thijssen JH, Cohen-Kettenis PT, van and fetal INSL3 and steroid hormone levels, Hougaard  DM, Bonde JP, Cohen A, et  al. 2012. Goozen  SH, Buitelaar JK. 2004. Relationships Phthalates and perfluoro octanesulfonic acid in between sex hormones assessed in amniotic fluid, evaluate potential mechanisms, and determine human amniotic fluid: temporal trends and timing and maternal and umbilical cord serum: what is the whether the estimated differences in fetal of amniocentesis in pregnancy. Environ Health best source of information to investigate the effects hormone levels are associated with long-term Perspect 120:897–903; doi:10.1289/ehp.1104522. of fetal hormonal exposure? Horm Behav 46:663–669. consequences for male reproductive health. Joensen UN, Bossi R, Leffers H, Jensen AA, Vested A, Ramlau-Hansen CH, Olsen SF, Bonde  JP, Skakkebæk NE, Jørgensen N. 2009. Do perfluoroalkyl Kristensen  SL, Halldorsson TI, et  al. 2013. RefeRences compounds impair human semen quality? Environ Associations of in utero exposure to perfluorinated Health Perspect 117:923–927; doi:10.1289/ehp.0800517. alkyl acids with human semen quality and repro- Joensen UN, Veyrand B, Antignac JP, Blomberg ductive hormones in adult men. Environ Health Anand-Ivell R, Ivell R. 2014. Insulin-like factor 3 as a Jensen  M, Petersen JH, Marchand P, et  al. 2013. Perspect 121:453–458; doi:10.1289/ehp.1205118. monitor of endocrine disruption. Reproduction PFOS (perfluorooctanesulfonate) in serum is nega- Vesterholm JD, Christensen J, Virtanen HE, 147:R87–R95. tively associated with testosterone levels, but not Skakkebæk  NE, Main KM, Toppari  J, et  al. 2014. Anand-Ivell R, Ivell R, Driscoll D, Manson J. 2008. with semen quality, in healthy men. Hum Reprod No association between exposure to perfluori- Insulin-like factor 3 levels in amniotic fluid of human 28:599–608. nated compounds and congenital cryptorchidism: male fetuses. Hum Reprod 23:1180–1186. Kalfa N, Philibert P, Sultan C. 2009. Is hypospadias a a nested case–control study among 215 boys from Anand-Ivell R, Tremellen K, Dai Y, Heng K, Yoshida M, genetic, endocrine or environmental disease, or Denmark and Finland. Reproduction 147:411–417. Knight PG, et al. 2013. Circulating Insulin-like factor still an unexplained malformation? Int J Androl Wan HT, Zhao YG, Wong MH, Lee KF, Yeung WS, 3 (INSL3) in healthy and infertile women. Hum 32:187–197. Giesy JP, et al. 2011. Testicular signaling is the poten- Reprod 28:3093–3102. Knight GJ, Palomaki GE. 2003. Epidemiologic monitoring tial target of perfluorooctanesulfonate-mediated Anand-Ivell R, Wohlgemuth J, Haren MT, Hope PJ, of prenatal screening for neural tube defects and subfertility in male mice. Biol Reprod 84:1016–1023. Hatzinikolas  G, Wittert G, et  al. 2006. Peripheral Down syndrome. Clin Lab Med 23:531–551. Zhao B, Li L, Liu J, Li H, Zhang C, Han P, et  al. 2014. INSL3 concentrations decline with age in a Knudsen LB, Olsen J. 1998. The Danish Medical Birth Exposure to perfluorooctane sulfonate in  utero large population of Australian men. Int J Androl Registry. Dan Med Bull 45:320–323. reduces testosterone production in rat fetal Leydig 29:618–626. Kraugerud M, Zimmer KE, Ropstad E, Verhaegen S. cells. PLoS One 9:e78888; doi:10.1371/journal. Armitage JM, Schenker U, Scheringer M, Martin JW, 2011. Perfluorinated compounds differentially pone.0078888. | | 156 volume 124 number 1 January 2016 • Environmental Health Perspectives

Journal

Environmental Health PerspectivesPubmed Central

Published: Jun 5, 2015

There are no references for this article.