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

Learn More →

Assessment of a six gene panel for the molecular detection of circulating tumor cells in the blood of female cancer patients

Assessment of a six gene panel for the molecular detection of circulating tumor cells in the... Background: The presence of circulating tumor cells (CTC) in the peripheral blood of cancer patients has been described for various solid tumors and their clinical relevance has been shown. CTC detection based on the analysis of epithelial antigens might be hampered by the genetic heterogeneity of the primary tumor and loss of epithelial antigens. Therefore, we aimed to identify new gene markers for the PCR-based detection of CTC in female cancer patients. Methods: Gene expression of 38 cancer cell lines (breast, ovarian, cervical and endometrial) and of 10 peripheral blood mononuclear cell (PBMC) samples from healthy female donors was measured using microarray technology (Applied Biosystems). Differentially expressed genes were identified using the maxT test and the 50% one-sided trimmed maxT-test. Confirmatory RT-qPCR was performed for 380 gene targets using the AB TaqMan® Low Density Arrays. Then, 93 gene targets were analyzed using the same RT-qPCR platform in tumor tissues of 126 patients with primary breast, ovarian or endometrial cancer. Finally, blood samples from 26 healthy women and from 125 patients (primary breast, ovarian, cervical, or endometrial cancer, and advanced breast cancer) were analyzed following OncoQuick enrichment and RNA pre-amplification. Likewise, hMAM and EpCAM gene expression was analyzed in the blood of breast and ovarian cancer patients. For each gene, a cut-off threshold value was set at three standard deviations from the mean expression level of the healthy controls to identify potential markers for CTC detection. Results: Six genes were over-expressed in blood samples from 81% of patients with advanced and 29% of patients with primary breast cancer. EpCAM gene expression was detected in 19% and 5% of patients, respectively, whereas hMAM gene expression was observed in the advanced group (39%) only. Multimarker analysis using the new six gene panel positively identified 44% of the cervical, 64% of the endometrial and 19% of the ovarian cancer patients. Conclusions: The panel of six genes was found superior to EpCAM and hMAM for the detection of circulating tumor cells in the blood of breast cancer, and they may serve as potential markers for CTC derived from endometrial, cervical, and ovarian cancers. Background cancer related deaths annually [1]. Although several Worldwide, more than two million women are diag- improvements have been made in early diagnosis during nosed with breast, cervical, endometrial or ovarian the past few decades, many patients still die of visceral cancer each year. These cancers contribute to 45% of metastasis, which is the main cause for tumor-related total female malignancies and approximately 880000 death. In these patients, the hematogenous spread of malignant cells remains undetected at the time of initial therapy. Since T. R. Ashworth first reported circulating * Correspondence: eva.obermayr@meduniwien.ac.at tumor cells (CTC) in the blood of cancer patients in Department of Obstetrics and Gynecology, Comprehensive Cancer Center, 1869 [2], the presence of CTC has been described for Medical University of Vienna, Vienna, Austria Full list of author information is available at the end of the article © 2010 Obermayr et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Obermayr et al. BMC Cancer 2010, 10:666 Page 2 of 12 http://www.biomedcentral.com/1471-2407/10/666 several solid tumors, such as colorectal, lung, kidney, cells [17]. EpCAM (epithelial cell adhesion molecule) is squamous oesophageal, liver, prostate and pancreatic not a perfect marker for CTC detection due to the high cancer [3]. Among cancers specific to women, the variation in its gene expression between tumor subtypes majority of CTC based research has been performed in and its illegitimate transcription from leukocytes [18],. breast cancer patients (reviewed in [3-6]), whereas few Likewise, the analysis of hMAM (human mammaglobin data exist for CTC in ovarian [7,8], cervical [9], and A), the most widely studied marker after CK19 (cytoker- endometrial cancer [10,11] patients. Recent studies have atin 19) in breast cancer patients, gene expression iden- demonstrated the prognostic role of CTC [12-14]; and tifies patients with nearly 100% specificity at the same the presence of tumor cells in the peripheral blood was sensitivity as CK19 (1 tumor cell in 10 peripheral blood considered to be established as an additional staging mononuclear cells) [19,20]. Nevertheless, mammaglobin parameter [15]. Hence, many efforts have been made to A gene expression is highly variable in female cancers develop reliable procedures for the sensitive and specific and is detected in the blood of approximately 10 to 30% detection of CTC, either at the protein level (antibody- of breast cancer patients [21]. Hence, there is a high based cell staining) or at the mRNA level (reverse tran- scientific and clinical need for the identification of new scription PCR). While the first approach is the gold markers for the detection of CTC. standard technique for the detection of tumor cells in In this study, we aimed to identify new gene markers the bone marrow of breast cancer patients, the latter for the RT-qPCR based detection of CTC in the blood of has been proven to be more sensitive and amenable to female cancer patients following a step-down strategy high-throughput analysis [6]. Nevertheless, the detection utilizing a whole genome analysis with oligonucleotide of CTC is often hampered by the heterogeneity of the microarrays (Applied Biosystems) and TaqMan® Low primary tumor and by the loss of epithelial antigens as Density Array (TLDA) based RT-qPCR using microflui- occurs during epithelial to mesenchymal transition [3]. dic technology. Based on the results of these experiments, It has been shown that normal-like breast cancer cells a panel of six candidate gene markers was selected for characterized by aggressive behaviour and worse treat- future routine diagnosis of circulating tumor cells. ment options are not recognized by the CellSearch cir- culating tumor cell test (Veridex LLC, San Diego, CA), Methods which uses EpCAM for cell isolation [16]. This test is Experimental plan the only diagnostic test that is currently approved by A step down strategy as depicted in Figure 1 was fol- the US Food and Drug Administration for the auto- lowed to identify gene markers for the detection of mated detection and enumeration of circulating tumor CTC. In the first step, microarray analysis of tumor cell Figure 1 Graphical scheme of the experimental plan. Following a step down strategy, six genes from initially 27.686 genes were identified as new gene markers for the RT-qPCR based detection of circulating tumor cells (CTC). In microarray analysis, we compared expression profiles of PBMC isolated by Ficoll gradient centrifugation from healthy individuals and various established cancer cell lines. In microarray validation, we compared expression profiles of PBMC isolated by Oncoquick from healthy individuals and cell lines. cDNA was amplified according to a published protocol [25]. For the experimental analysis of patients samples, we used Oncoquick only. cDNA was amplified using the TargetAmp™1-Round aRNA Amplification Kit. Obermayr et al. BMC Cancer 2010, 10:666 Page 3 of 12 http://www.biomedcentral.com/1471-2407/10/666 lines and peripheral blood mononuclear cells (PBMC) cancer patients) taken before the initial treatment (exci- from healthy female donors was performed. Second, the sion of the primary tumor or administration of systemic expression levels of a subset of all differentially neoadjuvant chemotherapy). Additionally, we analyzed expressed genes and of further known or supposed CTC one blood sample from 31 patients with recurrent breast markers were verified with RT-qPCR using the AB Taq- cancer and distant metastasis. Man® Low Density Array (TLDA) platform. In the third PB taken from 58 healthy female volunteers at the step, genes with absent or very low expression levels in MUW Department of Blood Group Serology and Trans- healthy PBMC were selected for the analysis of blood fusion Medicine, the MUW Department of Obstetrics samples taken from patients before initial surgery of the and Gynecology and ViennaLab Diagnostics GmbH primary tumor using the same RT-qPCR platform. As (Vienna, A) served as negative controls. All PB samples the number of circulating tumor cells was suspected to were collected in EDTA tubes and processed within be low, a RNA pre-amplification step was performed. 2 hours after venipuncture. The patient characteristics Again, a healthy control group was analysed. The aim of are given in Table 1. the third step was to identify new gene markers for the In the same time period, fresh frozen tissue samples RT-qPCR based detection of CTC. from patients with breast, ovarian, endometrial or cervi- cal carcinoma were kindly provided by the MUW Ethical considerations Department of Gynecopathology, Clinical Institute for The study was approved by the Ethics Committee from Pathology. Additional ovarian cancer tissues were col- the Medical University of Vienna, Austria (reference lected by the Department of Obstetrics and Gynecology numbers 366/2003 and 260/2003) and by the Institu- at the Charité-Universitätsmedizin Berlin (TOC = tional Review Board of the Charité Hospital. All periph- Tumorbank Ovarian cancer) (D). All tissue samples were eral blood and tumor tissue samples were collected with stored in liquid nitrogen prior to homogenization. The the patients’ written consent. study inclusion criteria were the same as for blood sam- ples; furthermore, recurrent patients and tissue samples Cell culture taken after neoadjuvant chemotherapy were excluded. Overall, 10 breast cancer cell lines (MCF-7, T-47 D, From a total of about 340 collected tumor tissues 50, 51 MDA-MB-231, Hs 578T, MDA-MB-435 S, MDA-MB- and 25 patients with primary breast, ovarian or endome- 453, BT-474, SK-BR-3, ZR-75-1, BT-549), 10 ovarian trial cancer, respectively were enrolled in the study. cancer cell lines (A2780, Caov-3, ES-2, NIHOVCAR-3, The patient characteristics are summarized in Addi- SK-OV-3, TOV-21G, TOV-112 D, OV-90, OV-MZ-01a, tional file 1. OV-MZ-6), 9 cervical cancer cell lines (HeLa, SW756, GH354, Ca Ski, C-4 I, C-33 A, HT-3, ME-180, SiHa), Cell spiking and 9 endometrial cancer cell lines (KLE, RL95-2, AN3 For sensitivity assays, a defined number of T-47 D CA, HEC-1-B, Ishikawa, Colo 684, HEC-50-B, EN, EJ) breast cancer cells were added to each 15 ml pre-cooled were cultivated according to the recommended proto- PB sample obtained from a healthy female donor and cols and harvested on at least three consecutive days. If provided by the Austrian National Red Cross Society. commercially available, the cell lines were purchased An unspiked blood sample from the same donor served from the American Type Culture Collection (ATCC, as a negative control. Each blood sample was spiked in http://www.atcc.org) or from the European Collection of duplicate. Samples were enriched by OncoQuick (Grei- Cell Cultures (ECACC, http://www.ecacc.org.uk). ner Bio-One, Frickenhausen, D) per the manufacturer’s instructions, resuspended in RLT-buffer (Qiagen RNA Peripheral blood and tumor tissues Isolation Kit), and the corresponding lysates pooled to From 2001 to 2006 peripheral blood (PB) samples were compensate for varying recovery rates of the enrichment collected from 567 patients with malign gynecological procedure. 1/6 of the extracted total RNA (Qiagen RNA diseases at the Department of Obstetrics and Gynecol- Isolation Kit) was pre-amplified in triplicate reactions ogy and at the Department of Medicine I, Division of employing the TargetAmp™1-Round aRNA Amplifica- Oncology (all located at the MUW Medical University tion Kit (Epicentre, Madison WI, USA) according to of Vienna, A). Patients with tumors of low malignant manufacturer instructions. The pre-amplified RNA was potential (i.e. borderline tumor of the ovaries), with con- converted into cDNA with M-MLV Reverse Transcrip- comitant or previous malignant tumors other than from tase, RNase H Minus (Promega, Madison WI, USA) and the breast, the ovaries or the uterus, transplanted random hexamers as primers. To assess the sensitivity patients, and pregnant patients were excluded. Finally, of the TLDA platform to detect circulating tumor cells, we included one blood sample from 94 patients (21 RT-qPCR was performed using the TLDA format 96a as breast, 23 ovarian, and each 25 cervical and endometrial described below. Obermayr et al. BMC Cancer 2010, 10:666 Page 4 of 12 http://www.biomedcentral.com/1471-2407/10/666 Table 1 Base line characteristics of patients included into Table 1 Base line characteristics of patients included into the RT-qPCR analysis of peripheral blood the RT-qPCR analysis of peripheral blood (Continued) Venipuncture median 53 AB range 37-78 Total number of patients 94 31 Histology Breast cancer Serous 72.7% Number 21 31 Mucinous 12.1% Age (yrs) Others/unknown 15.1% median 54 50 FIGO Stage range 35-78 25-75 I 10.0% Histology II 10.0% IDC 100.0% 64.5% III 65.0% ILC 0 12.9% IV 15.0% Others/unknown 0 22.6% Peripheral blood was taken from 94 patients before initial treatment of the primary tumor (A). From 31 recurrent breast cancer patients (B) the blood was TNM Stage * taken during disease progression (* TNM stage refers to the primary tumor). I 38.1% 3.2% II 33.3% 48.4% Sample processing III 23.8% 19.4% For the microarray-based gene expression studies, IV 4.8% 3.2% PBMC were isolated from 50 ml healthy female blood Unknown 0 22.6% by a density gradient using Ficoll-Paque™Plus (GE Endometrial cancer Healthcare Bio-Sciences AB, Uppsala, S) per the Number 25 0 standard procedure. For gene expression analysis with RT- qPCR, which required an enhanced depletion of leuko- Age (yrs) cytes than is provided by Ficoll, mononuclear cells from median 64 15-25 ml PB taken from healthy females and patients were range 30-83 enriched using OncoQuick® tubes (Greiner Bio-One, Histology Frickenhausen, D) according to the manufacturer’s Endometrioid 100.0% instructions. FIGO Stage 100 mg of fresh frozen tumor tissue was pulverized I 60.0% for 2 min at 2000 rpm using a microdismembrator (B. II 8.0% Braun Biotech., Melsungen, D) and further homogenized III 28.0% in lysis solution by intense vortexing. IV 4.0% RNA extraction Cervical Cancer Total RNA was extracted with two commercially avail- Number 25 0 able kits depending on the amount of cells in the start- Age (yrs) ing material: First, the Total RNA Isolation Mini Kit median 48 (Agilent Technologies, Waldbronn, D) was used for range 29-78 RNA extraction from cultivated tumor cells, from Histology homogenized tumor tissue and from PBMC enriched Non-keratinizing 48.0% by Ficoll-Paque™Plus density gradient centrifugation. Keratinizing 40.0% Total RNA samples were spectrophotometrically quan- Others/unknown 12.0% tified and examined for residual genomic DNA by PCR employing primers which span exon 9 of the breast FIGO Stage cancer 2, early onset gene BRCA2 (sense primer: 5’- I 8.0% ATA ACT GAA ATC ACC AAA AGT G-3’;antisense II 44.0% primer: 5’-CTG TAG TTC AAC TAA ACA GAG G- III 28.0% 3’). Residual genomic DNA was digested by DNase I. IV 20.0% Finally, quality assessment of the cell line- and PBMC- Ovarian cancer RNA was performed with the RNA 6000 Nano LabChip Number 23 0 Kit run on the 2100 Bioanalyzer (Agilent Technologies, Age (yrs) Waldbronn, D) and of RNA samples isolated from Obermayr et al. BMC Cancer 2010, 10:666 Page 5 of 12 http://www.biomedcentral.com/1471-2407/10/666 tumor tissues with denaturing agarose gel electrophor- identify genes, which are over-expressed in only a sub- esis. The total RNAs extracted from at least three conse- group of the tumor cell lines. cutive cell line harvests were combined to compensate Finally, 356 genes with a mean difference between for differences in expression that may result from vary- tumor cell lines compared to the healthy control sam- ing culture conditions. Each of the RNA pools and the ples of greater than 10 were selected for confirmatory RNA samples extracted from healthy PBMC were preci- gene expression profiling by RT-qPCR using the AB pitated to reach a minimal final concentration of 1.5 μg/ TaqMan® Low Density Array (TLDA) platform. Addi- μl. Second, the RNeasy Micro Kit (Qiagen, Hilden, D) tionally, the selected 356 genes were supplemented with was used for RNA extraction from cells enriched by 15 known or supposed markers for CTC detection. Oncoquick® gradient. Because we expected low RNA yields, we restrained from losing further material by Verification of microarray results with RT-qPCR assessing the RNA quality or quantity in these samples. The expression levels of the 356 genes selected from the microarray analyses and of the 15 known or supposed Expression profiling using Human Genome Microarrays CTC markers were verified with RT-qPCR in a subset A total of 48 Human Genome Survey Microarrays Hs.v1 of each five breast, ovarian and endometrial cancer cell (Applied Biosystems, Foster City CA, USA) were per- lines and in blood samples from 19 healthy females. RT- formed to compare the gene expression of 38 tumor cell qPCR was performed on the AB 7900HT Fast Real-time lines and 10 healthy control samples at GeneSys Labora- PCR System per manufacturer instructions using the tories GmbH (Muenster, D) under standard conditions TLDA format 384 for the analysis of 380 gene targets in using kits, reagents and the chemiluminescent microar- single reactions and of one mandatory endogenous con- ray analyzer 1700 from AB. In brief, 20 μg of total RNA trol gene (glyceraldehyde-3-phosphate dehydrogenase was used to prepare digoxigenin-labeled cDNA, which [GAPDH]) in a quadruplicate reaction. The 380 gene developed a chemiluminescent signal after hybridizing targets consisted of the 3 additional TaqMan® Endogen- to the 60-mer oligonucleotide probes spotted onto the ous Controls (beta-2-microglobulin [B2M], TAT-box microarray platform. Primary analysis and quality con- binding protein [TBP], and phosphoglyceratekinase 1 trol were performed using the AB Navigator Software [PGK]) and 377 TaqMan® Gene Expression Assays spe- Version 1.0.0.3. After background correction, data were cific for the 15 known or supposed CTC marker and normalized using the ABI 1700 Chemiluminescent Ana- specific for the previously selected differentially lyzer first by feature, then by spatial effects in the slide. expressed genes according to a mapping of microarray Finally a global normalization per slide was performed. probe IDs to assay IDs provided by AB. The RNA AB provides the normalized data in the column assay extracted from tumor cell lines was converted into normalized signal (ANS). The log base 2 ANS was con- cDNA with M-MLV Reverse Transcriptase, RNase H sidered for further analysis. Microarray expression mea- Minus (Promega, Madison WI, USA) and random hex- surements with a flag of greater than 5000 indicating a amers as primers. Blood mononuclear cells were low quality spot were filtered out. These measurements enriched with Oncoquick density gradient centrifuga- generally correlate with spots that have a signal to noise tion. Then, the extracted RNA was amplified following a ratio smaller than or equal to 3. Since we are interested modified version of a protocol published by Klein et al. in genes that are not expressed in healthy controls, only [25]. In short, the RNA was first converted into cDNA those gene probes with an average ANS in the control with M-MLV Reverse Transcriptase, RNase H Minus samples that was smaller than 1.5 were subjected to sta- (Promega, Madison WI, USA) and random primers con- tistical analysis. We performed two statistical tests in taining a 5’-oligo-dC flanking region (5’-[CCC] TGC parallel to identify differentially expressed genes. For the AGG N -3’; VBC Genomics, Vienna, A). After generat- maxT test from the multtest Bioconductor package [22], ing a 3’-oligo-dG flanking region, the flanked cDNA was genes were considered differentially expressed if they primed with CP2 (5’-TCA GAA TTC ATG [CCC] -3’; contained a corrected p-value ≤ 0.05 [23] Additionally VBC Genomics) and amplified with Super Taq (HT Bio- we used a 50% one-sided trimmed maxT-test [24] with technology Ltd., Cambridge, GB). The TLDA were a familywise error rate of 0.05 and 1000 permutations. loaded with the sample-specific PCR mix containing the This test resembles the ordinary maxT test but replaces template cDNA as recommended by the manufacturer the t-statistic with a trimmed t-statistic in both the ori- (2 ng per well). Raw data were analyzed with the AB ginal and the permuted data. For each gene of the origi- 7900 Sequence Detection Software version 2.2.2 using nal and each permuted data set, the trimmed t-statistic automatic baseline correction and a manual cycle is computed from only those data values, which are threshold (C ) setting. Resulting C data was exported t t greater than the group medians. In contrast to the for further analysis. To downsize the number of poten- maxT test, the 50% one-sided trimmed maxT-test can tial candidate genes from initially more than 27.000 Obermayr et al. BMC Cancer 2010, 10:666 Page 6 of 12 http://www.biomedcentral.com/1471-2407/10/666 genes to about 100 genes, all genes with expression average expression of gene X to the average expression levels beyond the RT-qPCR detection limit (i.e. C 50) in of the endogenous control gene GAPDH. If only one the healthy control samples were excluded. The remain- healthy control sample revealed detectable gene expres- ing genes were sorted in descending order by their aver- sion, the one dC was taken as cut-off threshold value. tX age C value obtained from the 15 tumor cell lines. The A tumor patient was considered positive for the molecu- first 93 genes were selected for RT-qPCR analysis of lar analysis of gene × if dC was below the defined tX blood and tissue samples taken from tumor patients threshold value T . using the TLDA 96a format. Additionally, three genes Additionally, hMAM-and EpCAM-specific RT-qPCR (B2M, GAPDH and PGK) were selected as internal was performed for the same set of breast and ovarian reference genes. cancer blood samples and for healthy female control samples after cell enrichment and RNA pre-amplifica- Gene expression analysis of tumor tissue samples tion as described above using individual AB TaqMan® The expression of the previously selected 93 genes was Pre-Developed Assay Reagents (Hs00267190_m1 and measured in tumor tissue samples from patients with Hs00158980_m1). primary breast (N = 50), ovarian (N = 51) and endome- trial cancer (N = 25) with RT-qPCR using the TLDA Results 96a format to verify their use as candidate markers for RNA quality assessment the detection of CTC in the blood of cancer patients. Prior to microarray hybridization and RT-qPCR analysis, RNA was converted into cDNA by Omniscript Reverse the RNA extracted from the tumor cell lines and the Transcriptase (Quiagen, Hilden, D) using an oligo- healthy PBMC was checked for quality with the RNA dT-flanked primer. Loading the microfluidic cards, 6000 Nano LabChip Kit run on the Agilent 2100 Bioa- RT-qPCR amplification, and raw data analysis were per- nalyzer. As a result, 85% of the RNA samples were of formed as described in the last preceding section. All very good RNA quality (RIN≥8), 60% of which were samples were analyzed in duplicates. considered to have an excellent quality (RIN≥9). Gene expression analysis of patients’ blood samples Differentially expressed genes in tumor cell lines The expression of the same 93 genes was evaluated in compared to healthy PBMC blood samples from healthy female volunteers (N = 26) We compared the gene expression profile of 38 estab- and in peripheral blood samples from patients with lished cancer cell lines to those of PBMC taken from 10 breast (N = 52), ovarian (N = 23), cervical and endome- healthy donors to identify genes that were not expressed trial cancer (25 patients each), using the TLDA 96a RT- or expressed at very low level in the peripheral blood of qPCR platform. After cell enrichment with OncoQuick healthy females but appeared very highly expressed in density gradient centrifugation 1/6 of the total RNA was the cancer cell lines. From the 18151 (54.8%) genes with amplified employing the TargetAmp™1-Round aRNA an average ANS < 1.5 in the healthy control samples Amplification Kit (Epicentre, Madison WI, USA) per maxT-test identified 457, 534, 526, and 503 genes differ- manufacturer instructions. The amplified RNA was con- entially expressed for the breast, cervical, endometrial, verted into cDNA with M-MLV Reverse Transcriptase, and ovarian cancer cell lines, respectively. These genes RNase H Minus (Promega, Madison WI, USA) and ran- comprised 54, 81, 63, and 60 genes with cancer-type dom hexamers as primers. Loading of the microfluidic specific expression for the respective cancer cell lines. cards, RT-qPCR amplification, and raw data analysis Additionally, the 50% one-sided trimmed maxT-test were performed as described in the microarray verifica- identified further 25, 27, 20 and 29 genes, which were tion section. All samples were analyzed in duplicate. differentially expressed in the breast, cervical, endome- The mean of the resulting duplicate C values was used trial and ovarian cancer cell lines compared to the as a quantitative value. If only one of the duplicates was healthy controls. positive (i.e. C < 50), the positive C value was taken. Finally, 356 differentially expressed genes were chosen t t Low-level expression of many genes in the peripheral for confirmatory gene expression profiling with RT- blood of the healthy control group decreased the overall qPCR using the TLDA 384 format (microarray data are specificity of the assay and required the introduction of provided in Additional file 2). This consisted of 337 a cut-off threshold value to separate the cancer patient genes identified by the maxT-test, 19 by the 50% one- group from the healthy control group: sidedtrimmedmaxT-test only, and the 4 genes: As proposed by Mikhitarian et al. [26], a threshold EFEMP1, EPS8L1, CRYZL1 and PCDHG represented value T for each gene X was set to three standard with more than one TaqMan® Assay. Additionally we deviations from the mean dC value in the control decided to analyze nine markers of well-known tumor tX group. dC values were calculated by normalizing the specificity (ERBB2, ESR1, PGR, PLAT, SCGB2A1, tX Obermayr et al. BMC Cancer 2010, 10:666 Page 7 of 12 http://www.biomedcentral.com/1471-2407/10/666 SCGB2A2, SERPINE1, SERPINE2 and TFF1)andsix Although background expression of EMP2, PPIC, candidate markers for CTC detection that were pre- DKFZp762E1312,and SLC6A8 was detected in the viously identified by our research group (COL3A1, GHR, unspiked blood, increasing expression levels of the CALB1, LPHN1, FN1 and EDNRA) [27]. respective genes were observed when tumor cells had been added to the blood, with a detection limit of 3 Verification of microarray results with RT-qPCR (EMP2, PPIC) and 26 tumor cells per ml of blood 146 genes of the TLDA 384 gene set were identified as (DKFZp762E1312, SLC6A8). Furthermore, the spiking potential markers for the detection of CTC in the blood experiments revealed that RT-qPCR might be less sensi- of cancer patients with expression levels below the detec- tive using the TLDA platform than using conventional tion limit of RT-qPCR (i.e. C 50) in the healthy control PCR tubes, because linear amplification patterns distin- group. The genes were sorted in descending order by guishing each 10-fold dilution were only observed with their average C value obtained from the 15 tumor cell C values smaller than 35 (data not shown). t t lines, and the first 93 genes were selected for further gene expression analysis of patients’ samples using the Gene expression in tumor tissues TLDA 96a format (see Additional file 3). None of the 15 The gene expression of the previously selected 93 genes known or supposed markers for CTC detection was con- was confirmed in tumor samples from patients with pri- sidered for further investigations either due to detectable mary breast, ovarian and endometrial cancer. We expression levels (ERBB2, ESR1, SERPINE1, SERPINE2 observed that the house-keeping gene expression levels and FN1) in healthy controls or due to inadequate gene were lower in ovarian cancer tissues than in tumor tis- expression in the tumor cell lines. sues of breast and endometrial cancer patients (GAPDH 24.2 ± 2.6, 22.2 ± 1.2, 22.7 ± 1.4 (SD) C ; B2 M 22.1 ± Sensitivity 3.4, 18.1 ± 1.5, 17.7 ± 1.9 (SD) C ; PGK 25.5 ± 2.7, 23.5 To assess the applicability of the TLDA platform for the ± 1.1, 22.4 ± 3.0 (SD) C in the respective tumor RT-qPCR based detection of circulating tumor cells, the patients). Two of the 93 genes were found to be tumor- expression levels of the specified 93 genes were mea- site specific: PLEKHC1 (pleckstrin homology domain sured in healthy female blood samples spiked with T-47 containing, family C [with FERM domain] member 1) D breast cancer cells. CCNE2 and MAL2 transcripts and SGCB (sarcoglycan beta) transcripts were detected were detected in blood samples spiked with at least 26 only in ovarian cancer patients (see Additional file 4), and 3 tumor cells per ml blood, respectively (Figure 2), although they were also detected in cancer cell lines of but they were not detected in the unspiked blood. breast and endometrial origin either. Interestingly, Figure 2 Sensitivity of RT-qPCR using TLDA platform. Expression levels of 93 candidate genes were analyzed using cDNA generated from total RNA isolated from peripheral blood samples from a healthy female donor and the same blood spiked with 4, 40 and 400 T47-D tumor cells after cell enrichment. RNA was pre-amplified using the TargetAmp™1-Round aRNA Amplification Kit. Average C values obtained from RT- qPCR amplification of CCNE2, DKFZp762E1312, EMP2, MAL2, PPIC, and SLC6A8 transcripts using the TLDA format are shown. MAL2 and CCNE2 gene expression was below the detection limit of RT-qPCR in the unspiked blood. The detection sensitivity of the respective marker gene was estimated to be 40 and 400 tumor cells per 15 ml whole blood. Obermayr et al. BMC Cancer 2010, 10:666 Page 8 of 12 http://www.biomedcentral.com/1471-2407/10/666 expression of the selected 93 genes was detected in ovarian cancer groups, the percentage of positive more ovarian cancer patients than in breast and endo- patients was found to be 44%, 64% and 19%, respectively metrial cancer patients (median percentage of positive (see Table 2 and Figure 3). patients in the respective tumor groups was 78.4%, Additionally, hMAM-specific RT-qPCR performed for 64.0% and 32.0%). the same set of breast and ovarian cancer blood samples confirmed the tissue specific expression of mammaglo- Gene markers for CTC detection bin A. Transcripts were only detected in recurrent The expression of the previously selected 93 genes was breast cancer patients with an incidence of 38.7%, but evaluated in blood samples from cancer patients, to iden- neither in primary breast cancer patients, ovarian cancer tify the most promising markers for CTC detection. At patients, nor in the healthy controls. Likewise, EpCAM primary diagnosis, each 17 (68.0%) cervical and endome- gene over-expression was detected in the blood of trial cancer, 6 (26.1%) ovarian cancer and 8 (38.1%) neither ovarian cancer patients nor healthy females. In breast cancer patients over-expressed at least one out of the blood of breast cancer patients, we found EpCAM the 93 potential candidate genes at levels above the over-expression in 5.0% of the patients at primary diag- defined threshold. At the time-point of disease recur- nosis and in 19.4% of the patients with clinical evidence rence, 27 (87.1%) breast cancer patients were positive for of disease recurrence (see Table 2). at least one gene. Of the 93 candidate genes, 40 were able to identify patients using the defined respective thresh- Discussion olds. 33 of these genes were capable to identify patients Using a stepwise approach combining genome-wide with primary breast cancer, and this number was reduced expression profiling and TaqMan® based RT-qPCR we to 15 for patients with advanced disease stage. 14 of these identified six genes (CCNE2, DKFZp762E1312, EMP2, genes could identify patients with cervical and endome- MAL2, PPIC,and SLC6A8) as potential markers for the trial cancer and four of the 40 genes identified ovarian detection of circulating tumor cells in the peripheral cancer patients. The remaining 55 genes did not provide blood of patients with breast cancer and gynecological any value due to similar expression levels in both the malignancies. Although each of these genes is implicated healthy control and cancer patient groups. in cancer, they have not previously been specified for The purpose of this study was to identify a panel of the detection of circulating tumor cells in cancer genes for future multi-marker RT-qPCR based analysis patients. to increase the sensitivity to detect circulating tumor Initial screening of candidate gene markers for CTC cells. For this purpose, we selected those genes, which detection was performed using a microarray-based gene were over-expressed in more than 10% of the patients expression analysis of human cancer cell lines and with recurrent breast cancer, since circulating tumor mononuclear blood cells obtained from healthy females. cells are more likely in advanced disease. According to After verification of the microarray results, a set of 93 this criterion, six genes (CCNE2, DKFZp762E1312, gene markers was selected for the RT-qPCR analysis of EMP2, MAL2, PPIC and SLC6A8)werechosenfor a blood samples from healthy females and from patients RT-qPCR marker panel. Using this panel 81% of the with breast, ovarian, endometrial, and cervical cancer. breast cancer patients with recurrence and 29% of the Due to background gene expression in the healthy breast cancer patients at initial diagnosis were positive blood samples, a rigorous cut-off threshold value was for at least one gene. In the cervical, endometrial and introduced to separate the patients from the healthy Table 2 Marker gene expression in peripheral blood Positive blood samples (%) Patients Panel CCNE2 MAL2 EMP2 SLC6A8 DKFZ PPIC hMAM EpCAM rec. BC (N = 31) 80.6 32.3 19.4 32.3 45.2 25.8 19.4 38.7 19.4 BC (N = 21) 28.6 23.8 0 4.8 0 4.8 0 0 5.0 OC (N = 23) 19.0 13.0 4.3 0 0 0 0 0 0 EC (N = 25) 64.0 36.0 20.0 12.0 12.0 8.0 8.0 0 0 CC (N = 25) 44.0 40.0 4.0 4.0 4.0 4.0 0 0 0 Healthy (N = 26) 0 0 0 0 0 0 0 0 0 The percentage of patients with RT-qPCR positive blood samples is shown. RT-qPCR positivity was defined as gene expression beyond the cut-off threshold, which was set for each gene marker at three standard deviations from the mean expression in healthy control blood samples. Positivity in percentage shown for the “panel” (CCNE2, DKFZp762E1312, EMP2, MAL2, PPIC and SLC6A8) is defined as positivity for at least one of the markers. (BC: breast cancer, rec. BC: recurrent breast cancer, OC: ovarian cancer, EC: endometrial cancer, CC: cervical cancer, ND: not done) Obermayr et al. BMC Cancer 2010, 10:666 Page 9 of 12 http://www.biomedcentral.com/1471-2407/10/666 Figure 3 RT-qPCR analysis of marker gene expression in peripheral blood. Gene expression was analyzed in blood samples taken from patients (triangles) with recurrent breast cancer (A), and in blood samples taken at first diagnosis from breast (B), endometrial (C), cervical (D) and ovarian (E) cancer patients. Blood from healthy females (circles) served as a control group. Mononuclear cells were enriched with the Oncoquick density gradient. RT-qPCR was performed following a RNA pre-amplification step. Average C values obtained from duplicates were normalized to GAPDH gene expression. Cut-off threshold values calculated from the mean average normalized gene expression in healthy female blood as indicated by horizontal lines for the respective gene markers (DKFZp762E1312 1.39, SLC6A8 2.92, PPIC 3.61, EMP2 6.84, MAL2 14.61, CCNE2 16.83). controls. We assumed that the over-expression of at their blood samples were chosen to identify new gene least one gene marker in relation to the defined thresh- markers for CTC detection. A panel of six genes: old value indicated the presence of circulating tumor CCNE2, DKFZp762E1312, EMP2, MAL2, PPIC,and cells. As patients with recurrent breast cancer are most SLC6A8 that were over-expressed in the blood of 81% likely to harbor circulating tumor cells in their blood, of patients with recurrent breast cancer was then chosen Obermayr et al. BMC Cancer 2010, 10:666 Page 10 of 12 http://www.biomedcentral.com/1471-2407/10/666 as gene markers for the molecular detection of CTC. In blood of patients with breast cancer or gynecologic contrast, at initial diagnosis using the six gene panel malignancies is useful for the detection of circulating only 29% of the breast cancer patients were RT-qPCR tumor cells, alone or combined with other markers positive. In addition, the new gene panel identified such as hMAM or EpCAM. Interestingly, the patients with other female cancers (i.e. cervical, endome- DKFZp762E1312, EMP2, PPIC,and SLC6A8 transcripts, trial and ovarian cancer). but not CCNE2 and MAL2 transcripts were detected in In tumor cell spiking experiments the sensitivity of the the blood of healthy females. Therefore, we suppose that applied RT-qPCR was estimated to be 3 to 26 tumor cells the detection of CCNE2 and MAL2 transcripts in the per ml whole blood; similar sensitivities are reported for blood of cancer patients is indicative for CTC presence RT-qPCR- and immuno-mediated detection (reviewed by (which had not been verified by immunocytochemistry). Gervasoni et al. [28]). However, we found out that Taq- However, the observed increase of CCNE2 mRNA levels Man® Low Density Arrays are typically not the method of in the diseased group compared to the healthy control choice for the detection of rare template molecules. group, which are reported to be undetectable in normal In the present study, all blood samples were taken quiescent cells arrested in G [30], is in conflict with the before removal of the tumor masses. To estimate supposed non-proliferative nature of circulating tumor whether the six gene panel is useful to detect minimal cells [31]. Interestingly, both CCNE2 and MAL2 are residual disease, further experiments should include located on chromosome 8q, a region which is frequently blood samples from cancer patients taken after the exci- increased in copy number in breast [32] and other can- sion of the primary tumor. Although we have already cer types [32,33], and one of the most important target analysed several blood samples taken from breast cancer genes affected by gains and amplifications of 8q is the patients with no evidence of disease six months after MYC oncogene. completion of their adjuvant chemotherapy, the follow- The frequency of hMAM gene expression in the blood up time is yet too short to make any conclusions con- of breast cancer patients is in line with the frequencies cerning the patient outcome. reported by Roncella et al. [20]. 10 of the 12 hMAM There are further limitations that need to be acknowl- positive blood samples (83%) were also positive when edged and addressed regarding the experimental design analyzed using the six gene panel, and 52% of the recur- of the present study. First, when we evaluated various rent breast cancer blood samples were solely identified approaches for the enrichment of circulating tumor cells by CCNE2, DKFZp762E1312, EMP2, MAL2, PPIC,or in in the course of the project, we found out that Onco- SLC6A8. Similarly, the six gene panel identified all of quick may insufficiently recover spiked tumor cells, in the EpCAM positive blood samples. particular when only a few tumor cells were added to the blood (i.e. ≤ 20 tumor cells per 15 ml blood) [29]. Conclusions For this reason, false-negative RT-qPCR results are likely In this study, we identified new gene markers for the to occur for cancer patients with low CTC counts. Sec- assessment of circulating tumor cells. We have shown ond, the density of the tumor cells depends on their dif- that the RT-qPCR-based multi-marker analysis using ferentiation state. Therefore, undifferentiated tumor the six genes: CCNE2, DKFZp762E1312, EMP2, MAL2, cells having a higher density might pass through the PPIC,or SLC6A8 more than doubled the number of Oncoquick density gradient. Finally, we cannot exclude positive patients with recurrent breast cancer compared false-positive cases due to non-malignant epithelial cells, to the analysis of hMAM or EpCAM gene expression which may contaminate the blood samples during veni- alone. Therefore, we suggest that the significantly higher puncture and which express the targeted transcripts. expression of these six genes in the peripheral blood Nevertheless, we decided in favour of the Oncoquick indicates the presence of circulating tumor cells. This density gradient, because it dramatically reduced back- multi marker analysis may provide a tool for clinical ground gene expression of the selected targets in healthy monitoring and treatment control of breast cancer and PBMC samples. To enhance the sensitivity and specifi- of gynecological malignancies. Eventually it may also be city of the approach, future experiments should primar- useful for the early detection. ily aim at improving the recovery rate of the tumour cell enrichment. Further evaluation of the six CTC mar- Additional material kers should be done without RNA pre-amplification and using the conventional PCR tube format instead of Taq- Additional file 1: Base line characteristics of patients included into the RT-qPCR analysis of tumor tissue. Man® Low Density Array format. Additional file 2: Microarray data of 356 differentially expressed Despite these limitations, we suggest that the RT- genes. qPCR based analysis of CCNE2, DKFZp762E1312, EMP2, MAL2, PPIC, and SLC6A8 gene expression in the Obermayr et al. BMC Cancer 2010, 10:666 Page 11 of 12 http://www.biomedcentral.com/1471-2407/10/666 5. Lacroix M: Significance, detection and markers of disseminated breast Additional file 3: Gene identifiers of the TLDA 96a platform.93 cancer cells. Endocr Relat Cancer 2006, 13(4):1033-1067. genes were selected as CTC candidate genes for the RT-qPCR analysis of 6. Zieglschmid V, Hollmann C, Bocher O: Detection of disseminated tumor blood and tumor tissue samples from cancer patients. Additionally, three cells in peripheral blood. Crit Rev Clin Lab Sci 2005, 42(2):155-196. house-keeping genes (B2M, GAPDH, and PGK1) were chosen as an 7. Marth C, Kisic J, Kaern J, Trope C, Fodstad O: Circulating tumor cells in the internal reference. peripheral blood and bone marrow of patients with ovarian carcinoma Additional file 4: Gene expression in tumor tissues. The percentage do not predict prognosis. Cancer 2002, 94(3):707-712. of breast, endometrial and ovarian cancer patients with gene expression 8. Wimberger P, Heubner M, Otterbach F, Fehm T, Kimmig R, Kasimir-Bauer S: detected by RT-qPCR is shown for each of the 93 candidate genes and Influence of platinum-based chemotherapy on disseminated tumor cells for the three internal reference genes (B2M, GAPDH, and PGK1). in blood and bone marrow of patients with ovarian cancer. Gynecol Oncol 2007, 107(2):331-338. 9. Diddle AW, Sholes DM, Hollingsworth J, Kinlaw S: Cervical carcinoma; cancer cells in the circulating blood. Am J Obstet Gynecol 1959, 78:582-585. Acknowledgements 10. Yabushita H, Shimazu M, Yamada H, Sawaguchi K, Noguchi M, Nakanishi M, This work was supported by the GEN-AU project “Cancer Transcriptomics” Kawai M: Occult lymph node metastases detected by cytokeratin and “Bioinformatics Integration Network II” (BIN II) of the Austrian Federal immunohistochemistry predict recurrence in node-negative endometrial Ministry of Science and Research. Keiichi Isaka (Department of Obstetrics and cancer. Gynecol Oncol 2001, 80(2):139-144. Gynecology at the Tokyo Medical University, J) kindly provided the tumor 11. Ji XQ, Sato H, Tanaka H, Konishi Y, Fujimoto T, Takahashi O, Tanaka T: Real- cell lines EN and EJ. Volker Möbus (Department of Obstetrics and time quantitative RT-PCR detection of disseminated endometrial tumor Gynecology, University of Ulm, D) gave the cell lines OV-MZ-01a and OV-MZ- cells in peripheral blood and lymph nodes using the LightCycler System. 6 by, and Hiroyuki Kuramoto (Department of Clinical Cytology Graduate Gynecol Oncol 2006, 100(2):355-360. School of Medical Sciences, School of Medicine, Kitasato University, 12. Cristofanilli M, Budd GT, Ellis MJ, Stopeck A, Matera J, Miller MC, Reuben JM, Sagamihara, Kanagawa, J) gave HEC-50-B. We thank Nicola Tidow (GeneSys Doyle GV, Allard WJ, Terstappen LW, et al: Circulating tumor cells, disease Laboratories GmbH, Muenster, D) for performing the microarray experiments. progression, and survival in metastatic breast cancer. N Engl J Med 2004, We are particularly grateful to Gerhard G. Thallinger (Institute for Genomics 351(8):781-791. and Bioinformatics, Graz University of Technology, Graz, A) for assisting 13. Cristofanilli M, Hayes DF, Budd GT, Ellis MJ, Stopeck A, Reuben JM, statistical analysis of the microarray data and to Ingrid Schiebel (Department Doyle GV, Matera J, Allard WJ, Miller MC, et al: Circulating tumor cells: a of Obstetrics and Gynecology, Medical University of Vienna, A) for cultivating novel prognostic factor for newly diagnosed metastatic breast cancer. J tumor cells. Clin Oncol 2005, 23(7):1420-1430. 14. Pachmann K, Camara O, Kavallaris A, Schneider U, Schunemann S, Author details 1 Hoffken K: Quantification of the response of circulating epithelial cells to Department of Obstetrics and Gynecology, Comprehensive Cancer Center, 2 neodadjuvant treatment for breast cancer: a new tool for therapy Medical University of Vienna, Vienna, Austria. Institute for Genomics and 3 monitoring. Breast Cancer Res 2005, 7(6):R975-979. Bioinformatics, Graz University of Technology, Graz, Austria. Department of 15. Cristofanilli M, Broglio KR, Guarneri V, Jackson S, Fritsche HA, Islam R, Medicine I, Comprehensive Cancer Center, Medical University of Vienna, 4 Dawood S, Reuben JM, Kau SW, Lara JM, et al: Circulating tumor cells in Vienna, Austria. Department of Blood Group Serology and Transfusion 5 metastatic breast cancer: biologic staging beyond tumor burden. Clin Medicine, Medical University of Vienna, Vienna, Austria. Department of Breast Cancer 2007, 7(6):471-479. Gynecology, European Competence Center for Ovarian Cancer, Charité - 6 16. Sieuwerts AM, Kraan J, Bolt J, van der Spoel P, Elstrodt F, Schutte M, University Medicine of Berlin, Berlin, Germany. Clinical Institute of Pathology, Martens JW, Gratama JW, Sleijfer S, Foekens JA: Anti-epithelial cell Comprehensive Cancer Center, Medical University of Vienna, Vienna, Austria. 7 adhesion molecule antibodies and the detection of circulating normal- Section of Clinical Biometrics, Center for Medical Statistics, Informatics and like breast tumor cells. J Natl Cancer Inst 2009, 101(1):61-66. Intelligent Systems, Comprehensive Cancer Center, Medical University of 8 17. Medical devices; immunology and microbiology devices; classification of Vienna, Vienna, Austria. Ludwig Boltzmann Gesellschaft - Cluster the immunomagnetic circulating cancer cell selection and enumeration Translational Oncology, A-1090 Vienna, Austria. system. Final rule. Fed Regist 2004, 69(91):26036-26038. 18. Zhong XY, Kaul S, Eichler A, Bastert G: Evaluating GA733-2 mRNA as a Authors’ contributions marker for the detection of micrometastatic breast cancer in peripheral EO performed and supervised sample processing, carried out the RT-qPCR blood and bone marrow. Arch Gynecol Obstet 1999, 263(1-2):2-6. analysis and data evaluation, and drafted the manuscript. FSC and GH 19. Stathopoulou A, Mavroudis D, Perraki M, Apostolaki S, Vlachonikolis I, performed statistical analysis of microarray data. CFS, MKT, AR, MK, MBF, RH Lianidou E, Georgoulias V: Molecular detection of cancer cells in the and JH coordinated the collection of patients’ blood and tissue samples. DT peripheral blood of patients with breast cancer: comparison of CK-19, and RZ designed the study and contributed to data interpretation. RZ CEA and maspin as detection markers. Anticancer Res 2003, served as mentor for the entire project. All authors read and approved the 23(2C):1883-1890. final manuscript. 20. Stathopoulou A, Angelopoulou K, Perraki M, Georgoulias V, Malamos N, Lianidou E: Quantitative RT-PCR luminometric hybridization assay with Competing interests an RNA internal standard for cytokeratin-19 mRNA in peripheral blood ZR, having ZR, DT and EO as inventors, filed a patent application based of patients with breast cancer. Clin Biochem 2001, 34(8):651-659. upon this manuscript. 21. Roncella S, Ferro P, Bacigalupo B, Pronzato P, Tognoni A, Falco E, Gianquinto D, Ansaldo V, Dessanti P, Fais F, et al: Human mammaglobin Received: 22 February 2010 Accepted: 3 December 2010 mRNA is a reliable molecular marker for detecting occult breast cancer Published: 3 December 2010 cells in peripheral blood. J Exp Clin Cancer Res 2005, 24(2):265-271. 22. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, Ellis B, References Gautier L, Ge Y, Gentry J, et al: Bioconductor: open software development 1. GLOBOCAN 2002: Cancer Incidence, Mortality and Prevalence for computational biology and bioinformatics. Genome Biol 2004, 5(10): Worldwide. [http://www-dep.iarc.fr]. R80. 2. Ashworth T: A case of cancer in which cells similar to those in the 23. Ge YDS, Speed TP: Resampling-based multiple testing for microarray data tumours were seen in the blood after death. Aus Med J 1869, 14:146-149. analysis. Test 2003, 12:1-44. 3. Paterlini-Brechot P, Benali NL: Circulating tumor cells (CTC) detection: 24. Gleiss A, Sanchez-Cabo F, Perco P, Tong D, Heinze G: Adaptive trimmed t- clinical impact and future directions. Cancer Lett 2007, 253(2):180-204. statistics for identifying predominantly high expression in a microarray 4. Smerage JB, Hayes DF: The measurement and therapeutic implications of experiment. Stat Med . circulating tumour cells in breast cancer. Br J Cancer 2006, 94(1):8-12. Obermayr et al. BMC Cancer 2010, 10:666 Page 12 of 12 http://www.biomedcentral.com/1471-2407/10/666 25. Klein CA, Seidl S, Petat-Dutter K, Offner S, Geigl JB, Schmidt-Kittler O, Wendler N, Passlick B, Huber RM, Schlimok G, et al: Combined transcriptome and genome analysis of single micrometastatic cells. Nat Biotechnol 2002, 20(4):387-392. 26. Mikhitarian K, Martin RH, Ruppel MB, Gillanders WE, Hoda R, Schutte del H, Callahan K, Mitas M, Cole DJ: Detection of mammaglobin mRNA in peripheral blood is associated with high grade breast cancer: interim results of a prospective cohort study. BMC Cancer 2008, 8:55. 27. Zeillinger R, Fabjani G: Method and kit for diagnosis of a cancerous disease, method for determining the reaction of a patient to the treatment for a cancerous disease, medicament for the prophylaxis or treatment of a cancerous disease. European Patent No. 1786929 2006. 28. Gervasoni A, Monasterio Munoz RM, Wengler GS, Rizzi A, Zaniboni A, Parolini O: Molecular signature detection of circulating tumor cells using a panel of selected genes. Cancer Lett 2008, 263(2):267-279. 29. Konigsberg R, Gneist M, Jahn-Kuch D, Pfeiler G, Hager G, Hudec M, Dittrich C, Zeillinger R: Circulating tumor cells in metastatic colorectal cancer: efficacy and feasibility of different enrichment methods. Cancer Lett 293(1):117-123. 30. Lauper N, Beck AR, Cariou S, Richman L, Hofmann K, Reith W, Slingerland JM, Amati B: Cyclin E2: a novel CDK2 partner in the late G1 and S phases of the mammalian cell cycle. Oncogene 1998, 17(20):2637-2643. 31. Muller V, Stahmann N, Riethdorf S, Rau T, Zabel T, Goetz A, Janicke F, Pantel K: Circulating tumor cells in breast cancer: correlation to bone marrow micrometastases, heterogeneous response to systemic therapy and low proliferative activity. Clin Cancer Res 2005, 11(10):3678-3685. 32. Forozan F, Karhu R, Kononen J, Kallioniemi A, Kallioniemi OP: Genome screening by comparative genomic hybridization. Trends Genet 1997, 13(10):405-409. 33. Knuutila S, Bjorkqvist AM, Autio K, Tarkkanen M, Wolf M, Monni O, Szymanska J, Larramendy ML, Tapper J, Pere H, et al: DNA copy number amplifications in human neoplasms: review of comparative genomic hybridization studies. Am J Pathol 1998, 152(5):1107-1123. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2407/10/666/prepub doi:10.1186/1471-2407-10-666 Cite this article as: Obermayr et al.: Assessment of a six gene panel for the molecular detection of circulating tumor cells in the blood of female cancer patients. BMC Cancer 2010 10:666. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BMC Cancer Springer Journals

Assessment of a six gene panel for the molecular detection of circulating tumor cells in the blood of female cancer patients

Loading next page...
 
/lp/springer-journals/assessment-of-a-six-gene-panel-for-the-molecular-detection-of-e7rJGmDbtE

References (34)

Publisher
Springer Journals
Copyright
Copyright © 2010 by Obermayr et al; licensee BioMed Central Ltd.
Subject
Biomedicine; Cancer Research; Oncology; Surgical Oncology; Health Promotion and Disease Prevention; Biomedicine general; Medicine/Public Health, general
ISSN
1471-2407
eISSN
1471-2407
DOI
10.1186/1471-2407-10-666
pmid
21129172
Publisher site
See Article on Publisher Site

Abstract

Background: The presence of circulating tumor cells (CTC) in the peripheral blood of cancer patients has been described for various solid tumors and their clinical relevance has been shown. CTC detection based on the analysis of epithelial antigens might be hampered by the genetic heterogeneity of the primary tumor and loss of epithelial antigens. Therefore, we aimed to identify new gene markers for the PCR-based detection of CTC in female cancer patients. Methods: Gene expression of 38 cancer cell lines (breast, ovarian, cervical and endometrial) and of 10 peripheral blood mononuclear cell (PBMC) samples from healthy female donors was measured using microarray technology (Applied Biosystems). Differentially expressed genes were identified using the maxT test and the 50% one-sided trimmed maxT-test. Confirmatory RT-qPCR was performed for 380 gene targets using the AB TaqMan® Low Density Arrays. Then, 93 gene targets were analyzed using the same RT-qPCR platform in tumor tissues of 126 patients with primary breast, ovarian or endometrial cancer. Finally, blood samples from 26 healthy women and from 125 patients (primary breast, ovarian, cervical, or endometrial cancer, and advanced breast cancer) were analyzed following OncoQuick enrichment and RNA pre-amplification. Likewise, hMAM and EpCAM gene expression was analyzed in the blood of breast and ovarian cancer patients. For each gene, a cut-off threshold value was set at three standard deviations from the mean expression level of the healthy controls to identify potential markers for CTC detection. Results: Six genes were over-expressed in blood samples from 81% of patients with advanced and 29% of patients with primary breast cancer. EpCAM gene expression was detected in 19% and 5% of patients, respectively, whereas hMAM gene expression was observed in the advanced group (39%) only. Multimarker analysis using the new six gene panel positively identified 44% of the cervical, 64% of the endometrial and 19% of the ovarian cancer patients. Conclusions: The panel of six genes was found superior to EpCAM and hMAM for the detection of circulating tumor cells in the blood of breast cancer, and they may serve as potential markers for CTC derived from endometrial, cervical, and ovarian cancers. Background cancer related deaths annually [1]. Although several Worldwide, more than two million women are diag- improvements have been made in early diagnosis during nosed with breast, cervical, endometrial or ovarian the past few decades, many patients still die of visceral cancer each year. These cancers contribute to 45% of metastasis, which is the main cause for tumor-related total female malignancies and approximately 880000 death. In these patients, the hematogenous spread of malignant cells remains undetected at the time of initial therapy. Since T. R. Ashworth first reported circulating * Correspondence: eva.obermayr@meduniwien.ac.at tumor cells (CTC) in the blood of cancer patients in Department of Obstetrics and Gynecology, Comprehensive Cancer Center, 1869 [2], the presence of CTC has been described for Medical University of Vienna, Vienna, Austria Full list of author information is available at the end of the article © 2010 Obermayr et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Obermayr et al. BMC Cancer 2010, 10:666 Page 2 of 12 http://www.biomedcentral.com/1471-2407/10/666 several solid tumors, such as colorectal, lung, kidney, cells [17]. EpCAM (epithelial cell adhesion molecule) is squamous oesophageal, liver, prostate and pancreatic not a perfect marker for CTC detection due to the high cancer [3]. Among cancers specific to women, the variation in its gene expression between tumor subtypes majority of CTC based research has been performed in and its illegitimate transcription from leukocytes [18],. breast cancer patients (reviewed in [3-6]), whereas few Likewise, the analysis of hMAM (human mammaglobin data exist for CTC in ovarian [7,8], cervical [9], and A), the most widely studied marker after CK19 (cytoker- endometrial cancer [10,11] patients. Recent studies have atin 19) in breast cancer patients, gene expression iden- demonstrated the prognostic role of CTC [12-14]; and tifies patients with nearly 100% specificity at the same the presence of tumor cells in the peripheral blood was sensitivity as CK19 (1 tumor cell in 10 peripheral blood considered to be established as an additional staging mononuclear cells) [19,20]. Nevertheless, mammaglobin parameter [15]. Hence, many efforts have been made to A gene expression is highly variable in female cancers develop reliable procedures for the sensitive and specific and is detected in the blood of approximately 10 to 30% detection of CTC, either at the protein level (antibody- of breast cancer patients [21]. Hence, there is a high based cell staining) or at the mRNA level (reverse tran- scientific and clinical need for the identification of new scription PCR). While the first approach is the gold markers for the detection of CTC. standard technique for the detection of tumor cells in In this study, we aimed to identify new gene markers the bone marrow of breast cancer patients, the latter for the RT-qPCR based detection of CTC in the blood of has been proven to be more sensitive and amenable to female cancer patients following a step-down strategy high-throughput analysis [6]. Nevertheless, the detection utilizing a whole genome analysis with oligonucleotide of CTC is often hampered by the heterogeneity of the microarrays (Applied Biosystems) and TaqMan® Low primary tumor and by the loss of epithelial antigens as Density Array (TLDA) based RT-qPCR using microflui- occurs during epithelial to mesenchymal transition [3]. dic technology. Based on the results of these experiments, It has been shown that normal-like breast cancer cells a panel of six candidate gene markers was selected for characterized by aggressive behaviour and worse treat- future routine diagnosis of circulating tumor cells. ment options are not recognized by the CellSearch cir- culating tumor cell test (Veridex LLC, San Diego, CA), Methods which uses EpCAM for cell isolation [16]. This test is Experimental plan the only diagnostic test that is currently approved by A step down strategy as depicted in Figure 1 was fol- the US Food and Drug Administration for the auto- lowed to identify gene markers for the detection of mated detection and enumeration of circulating tumor CTC. In the first step, microarray analysis of tumor cell Figure 1 Graphical scheme of the experimental plan. Following a step down strategy, six genes from initially 27.686 genes were identified as new gene markers for the RT-qPCR based detection of circulating tumor cells (CTC). In microarray analysis, we compared expression profiles of PBMC isolated by Ficoll gradient centrifugation from healthy individuals and various established cancer cell lines. In microarray validation, we compared expression profiles of PBMC isolated by Oncoquick from healthy individuals and cell lines. cDNA was amplified according to a published protocol [25]. For the experimental analysis of patients samples, we used Oncoquick only. cDNA was amplified using the TargetAmp™1-Round aRNA Amplification Kit. Obermayr et al. BMC Cancer 2010, 10:666 Page 3 of 12 http://www.biomedcentral.com/1471-2407/10/666 lines and peripheral blood mononuclear cells (PBMC) cancer patients) taken before the initial treatment (exci- from healthy female donors was performed. Second, the sion of the primary tumor or administration of systemic expression levels of a subset of all differentially neoadjuvant chemotherapy). Additionally, we analyzed expressed genes and of further known or supposed CTC one blood sample from 31 patients with recurrent breast markers were verified with RT-qPCR using the AB Taq- cancer and distant metastasis. Man® Low Density Array (TLDA) platform. In the third PB taken from 58 healthy female volunteers at the step, genes with absent or very low expression levels in MUW Department of Blood Group Serology and Trans- healthy PBMC were selected for the analysis of blood fusion Medicine, the MUW Department of Obstetrics samples taken from patients before initial surgery of the and Gynecology and ViennaLab Diagnostics GmbH primary tumor using the same RT-qPCR platform. As (Vienna, A) served as negative controls. All PB samples the number of circulating tumor cells was suspected to were collected in EDTA tubes and processed within be low, a RNA pre-amplification step was performed. 2 hours after venipuncture. The patient characteristics Again, a healthy control group was analysed. The aim of are given in Table 1. the third step was to identify new gene markers for the In the same time period, fresh frozen tissue samples RT-qPCR based detection of CTC. from patients with breast, ovarian, endometrial or cervi- cal carcinoma were kindly provided by the MUW Ethical considerations Department of Gynecopathology, Clinical Institute for The study was approved by the Ethics Committee from Pathology. Additional ovarian cancer tissues were col- the Medical University of Vienna, Austria (reference lected by the Department of Obstetrics and Gynecology numbers 366/2003 and 260/2003) and by the Institu- at the Charité-Universitätsmedizin Berlin (TOC = tional Review Board of the Charité Hospital. All periph- Tumorbank Ovarian cancer) (D). All tissue samples were eral blood and tumor tissue samples were collected with stored in liquid nitrogen prior to homogenization. The the patients’ written consent. study inclusion criteria were the same as for blood sam- ples; furthermore, recurrent patients and tissue samples Cell culture taken after neoadjuvant chemotherapy were excluded. Overall, 10 breast cancer cell lines (MCF-7, T-47 D, From a total of about 340 collected tumor tissues 50, 51 MDA-MB-231, Hs 578T, MDA-MB-435 S, MDA-MB- and 25 patients with primary breast, ovarian or endome- 453, BT-474, SK-BR-3, ZR-75-1, BT-549), 10 ovarian trial cancer, respectively were enrolled in the study. cancer cell lines (A2780, Caov-3, ES-2, NIHOVCAR-3, The patient characteristics are summarized in Addi- SK-OV-3, TOV-21G, TOV-112 D, OV-90, OV-MZ-01a, tional file 1. OV-MZ-6), 9 cervical cancer cell lines (HeLa, SW756, GH354, Ca Ski, C-4 I, C-33 A, HT-3, ME-180, SiHa), Cell spiking and 9 endometrial cancer cell lines (KLE, RL95-2, AN3 For sensitivity assays, a defined number of T-47 D CA, HEC-1-B, Ishikawa, Colo 684, HEC-50-B, EN, EJ) breast cancer cells were added to each 15 ml pre-cooled were cultivated according to the recommended proto- PB sample obtained from a healthy female donor and cols and harvested on at least three consecutive days. If provided by the Austrian National Red Cross Society. commercially available, the cell lines were purchased An unspiked blood sample from the same donor served from the American Type Culture Collection (ATCC, as a negative control. Each blood sample was spiked in http://www.atcc.org) or from the European Collection of duplicate. Samples were enriched by OncoQuick (Grei- Cell Cultures (ECACC, http://www.ecacc.org.uk). ner Bio-One, Frickenhausen, D) per the manufacturer’s instructions, resuspended in RLT-buffer (Qiagen RNA Peripheral blood and tumor tissues Isolation Kit), and the corresponding lysates pooled to From 2001 to 2006 peripheral blood (PB) samples were compensate for varying recovery rates of the enrichment collected from 567 patients with malign gynecological procedure. 1/6 of the extracted total RNA (Qiagen RNA diseases at the Department of Obstetrics and Gynecol- Isolation Kit) was pre-amplified in triplicate reactions ogy and at the Department of Medicine I, Division of employing the TargetAmp™1-Round aRNA Amplifica- Oncology (all located at the MUW Medical University tion Kit (Epicentre, Madison WI, USA) according to of Vienna, A). Patients with tumors of low malignant manufacturer instructions. The pre-amplified RNA was potential (i.e. borderline tumor of the ovaries), with con- converted into cDNA with M-MLV Reverse Transcrip- comitant or previous malignant tumors other than from tase, RNase H Minus (Promega, Madison WI, USA) and the breast, the ovaries or the uterus, transplanted random hexamers as primers. To assess the sensitivity patients, and pregnant patients were excluded. Finally, of the TLDA platform to detect circulating tumor cells, we included one blood sample from 94 patients (21 RT-qPCR was performed using the TLDA format 96a as breast, 23 ovarian, and each 25 cervical and endometrial described below. Obermayr et al. BMC Cancer 2010, 10:666 Page 4 of 12 http://www.biomedcentral.com/1471-2407/10/666 Table 1 Base line characteristics of patients included into Table 1 Base line characteristics of patients included into the RT-qPCR analysis of peripheral blood the RT-qPCR analysis of peripheral blood (Continued) Venipuncture median 53 AB range 37-78 Total number of patients 94 31 Histology Breast cancer Serous 72.7% Number 21 31 Mucinous 12.1% Age (yrs) Others/unknown 15.1% median 54 50 FIGO Stage range 35-78 25-75 I 10.0% Histology II 10.0% IDC 100.0% 64.5% III 65.0% ILC 0 12.9% IV 15.0% Others/unknown 0 22.6% Peripheral blood was taken from 94 patients before initial treatment of the primary tumor (A). From 31 recurrent breast cancer patients (B) the blood was TNM Stage * taken during disease progression (* TNM stage refers to the primary tumor). I 38.1% 3.2% II 33.3% 48.4% Sample processing III 23.8% 19.4% For the microarray-based gene expression studies, IV 4.8% 3.2% PBMC were isolated from 50 ml healthy female blood Unknown 0 22.6% by a density gradient using Ficoll-Paque™Plus (GE Endometrial cancer Healthcare Bio-Sciences AB, Uppsala, S) per the Number 25 0 standard procedure. For gene expression analysis with RT- qPCR, which required an enhanced depletion of leuko- Age (yrs) cytes than is provided by Ficoll, mononuclear cells from median 64 15-25 ml PB taken from healthy females and patients were range 30-83 enriched using OncoQuick® tubes (Greiner Bio-One, Histology Frickenhausen, D) according to the manufacturer’s Endometrioid 100.0% instructions. FIGO Stage 100 mg of fresh frozen tumor tissue was pulverized I 60.0% for 2 min at 2000 rpm using a microdismembrator (B. II 8.0% Braun Biotech., Melsungen, D) and further homogenized III 28.0% in lysis solution by intense vortexing. IV 4.0% RNA extraction Cervical Cancer Total RNA was extracted with two commercially avail- Number 25 0 able kits depending on the amount of cells in the start- Age (yrs) ing material: First, the Total RNA Isolation Mini Kit median 48 (Agilent Technologies, Waldbronn, D) was used for range 29-78 RNA extraction from cultivated tumor cells, from Histology homogenized tumor tissue and from PBMC enriched Non-keratinizing 48.0% by Ficoll-Paque™Plus density gradient centrifugation. Keratinizing 40.0% Total RNA samples were spectrophotometrically quan- Others/unknown 12.0% tified and examined for residual genomic DNA by PCR employing primers which span exon 9 of the breast FIGO Stage cancer 2, early onset gene BRCA2 (sense primer: 5’- I 8.0% ATA ACT GAA ATC ACC AAA AGT G-3’;antisense II 44.0% primer: 5’-CTG TAG TTC AAC TAA ACA GAG G- III 28.0% 3’). Residual genomic DNA was digested by DNase I. IV 20.0% Finally, quality assessment of the cell line- and PBMC- Ovarian cancer RNA was performed with the RNA 6000 Nano LabChip Number 23 0 Kit run on the 2100 Bioanalyzer (Agilent Technologies, Age (yrs) Waldbronn, D) and of RNA samples isolated from Obermayr et al. BMC Cancer 2010, 10:666 Page 5 of 12 http://www.biomedcentral.com/1471-2407/10/666 tumor tissues with denaturing agarose gel electrophor- identify genes, which are over-expressed in only a sub- esis. The total RNAs extracted from at least three conse- group of the tumor cell lines. cutive cell line harvests were combined to compensate Finally, 356 genes with a mean difference between for differences in expression that may result from vary- tumor cell lines compared to the healthy control sam- ing culture conditions. Each of the RNA pools and the ples of greater than 10 were selected for confirmatory RNA samples extracted from healthy PBMC were preci- gene expression profiling by RT-qPCR using the AB pitated to reach a minimal final concentration of 1.5 μg/ TaqMan® Low Density Array (TLDA) platform. Addi- μl. Second, the RNeasy Micro Kit (Qiagen, Hilden, D) tionally, the selected 356 genes were supplemented with was used for RNA extraction from cells enriched by 15 known or supposed markers for CTC detection. Oncoquick® gradient. Because we expected low RNA yields, we restrained from losing further material by Verification of microarray results with RT-qPCR assessing the RNA quality or quantity in these samples. The expression levels of the 356 genes selected from the microarray analyses and of the 15 known or supposed Expression profiling using Human Genome Microarrays CTC markers were verified with RT-qPCR in a subset A total of 48 Human Genome Survey Microarrays Hs.v1 of each five breast, ovarian and endometrial cancer cell (Applied Biosystems, Foster City CA, USA) were per- lines and in blood samples from 19 healthy females. RT- formed to compare the gene expression of 38 tumor cell qPCR was performed on the AB 7900HT Fast Real-time lines and 10 healthy control samples at GeneSys Labora- PCR System per manufacturer instructions using the tories GmbH (Muenster, D) under standard conditions TLDA format 384 for the analysis of 380 gene targets in using kits, reagents and the chemiluminescent microar- single reactions and of one mandatory endogenous con- ray analyzer 1700 from AB. In brief, 20 μg of total RNA trol gene (glyceraldehyde-3-phosphate dehydrogenase was used to prepare digoxigenin-labeled cDNA, which [GAPDH]) in a quadruplicate reaction. The 380 gene developed a chemiluminescent signal after hybridizing targets consisted of the 3 additional TaqMan® Endogen- to the 60-mer oligonucleotide probes spotted onto the ous Controls (beta-2-microglobulin [B2M], TAT-box microarray platform. Primary analysis and quality con- binding protein [TBP], and phosphoglyceratekinase 1 trol were performed using the AB Navigator Software [PGK]) and 377 TaqMan® Gene Expression Assays spe- Version 1.0.0.3. After background correction, data were cific for the 15 known or supposed CTC marker and normalized using the ABI 1700 Chemiluminescent Ana- specific for the previously selected differentially lyzer first by feature, then by spatial effects in the slide. expressed genes according to a mapping of microarray Finally a global normalization per slide was performed. probe IDs to assay IDs provided by AB. The RNA AB provides the normalized data in the column assay extracted from tumor cell lines was converted into normalized signal (ANS). The log base 2 ANS was con- cDNA with M-MLV Reverse Transcriptase, RNase H sidered for further analysis. Microarray expression mea- Minus (Promega, Madison WI, USA) and random hex- surements with a flag of greater than 5000 indicating a amers as primers. Blood mononuclear cells were low quality spot were filtered out. These measurements enriched with Oncoquick density gradient centrifuga- generally correlate with spots that have a signal to noise tion. Then, the extracted RNA was amplified following a ratio smaller than or equal to 3. Since we are interested modified version of a protocol published by Klein et al. in genes that are not expressed in healthy controls, only [25]. In short, the RNA was first converted into cDNA those gene probes with an average ANS in the control with M-MLV Reverse Transcriptase, RNase H Minus samples that was smaller than 1.5 were subjected to sta- (Promega, Madison WI, USA) and random primers con- tistical analysis. We performed two statistical tests in taining a 5’-oligo-dC flanking region (5’-[CCC] TGC parallel to identify differentially expressed genes. For the AGG N -3’; VBC Genomics, Vienna, A). After generat- maxT test from the multtest Bioconductor package [22], ing a 3’-oligo-dG flanking region, the flanked cDNA was genes were considered differentially expressed if they primed with CP2 (5’-TCA GAA TTC ATG [CCC] -3’; contained a corrected p-value ≤ 0.05 [23] Additionally VBC Genomics) and amplified with Super Taq (HT Bio- we used a 50% one-sided trimmed maxT-test [24] with technology Ltd., Cambridge, GB). The TLDA were a familywise error rate of 0.05 and 1000 permutations. loaded with the sample-specific PCR mix containing the This test resembles the ordinary maxT test but replaces template cDNA as recommended by the manufacturer the t-statistic with a trimmed t-statistic in both the ori- (2 ng per well). Raw data were analyzed with the AB ginal and the permuted data. For each gene of the origi- 7900 Sequence Detection Software version 2.2.2 using nal and each permuted data set, the trimmed t-statistic automatic baseline correction and a manual cycle is computed from only those data values, which are threshold (C ) setting. Resulting C data was exported t t greater than the group medians. In contrast to the for further analysis. To downsize the number of poten- maxT test, the 50% one-sided trimmed maxT-test can tial candidate genes from initially more than 27.000 Obermayr et al. BMC Cancer 2010, 10:666 Page 6 of 12 http://www.biomedcentral.com/1471-2407/10/666 genes to about 100 genes, all genes with expression average expression of gene X to the average expression levels beyond the RT-qPCR detection limit (i.e. C 50) in of the endogenous control gene GAPDH. If only one the healthy control samples were excluded. The remain- healthy control sample revealed detectable gene expres- ing genes were sorted in descending order by their aver- sion, the one dC was taken as cut-off threshold value. tX age C value obtained from the 15 tumor cell lines. The A tumor patient was considered positive for the molecu- first 93 genes were selected for RT-qPCR analysis of lar analysis of gene × if dC was below the defined tX blood and tissue samples taken from tumor patients threshold value T . using the TLDA 96a format. Additionally, three genes Additionally, hMAM-and EpCAM-specific RT-qPCR (B2M, GAPDH and PGK) were selected as internal was performed for the same set of breast and ovarian reference genes. cancer blood samples and for healthy female control samples after cell enrichment and RNA pre-amplifica- Gene expression analysis of tumor tissue samples tion as described above using individual AB TaqMan® The expression of the previously selected 93 genes was Pre-Developed Assay Reagents (Hs00267190_m1 and measured in tumor tissue samples from patients with Hs00158980_m1). primary breast (N = 50), ovarian (N = 51) and endome- trial cancer (N = 25) with RT-qPCR using the TLDA Results 96a format to verify their use as candidate markers for RNA quality assessment the detection of CTC in the blood of cancer patients. Prior to microarray hybridization and RT-qPCR analysis, RNA was converted into cDNA by Omniscript Reverse the RNA extracted from the tumor cell lines and the Transcriptase (Quiagen, Hilden, D) using an oligo- healthy PBMC was checked for quality with the RNA dT-flanked primer. Loading the microfluidic cards, 6000 Nano LabChip Kit run on the Agilent 2100 Bioa- RT-qPCR amplification, and raw data analysis were per- nalyzer. As a result, 85% of the RNA samples were of formed as described in the last preceding section. All very good RNA quality (RIN≥8), 60% of which were samples were analyzed in duplicates. considered to have an excellent quality (RIN≥9). Gene expression analysis of patients’ blood samples Differentially expressed genes in tumor cell lines The expression of the same 93 genes was evaluated in compared to healthy PBMC blood samples from healthy female volunteers (N = 26) We compared the gene expression profile of 38 estab- and in peripheral blood samples from patients with lished cancer cell lines to those of PBMC taken from 10 breast (N = 52), ovarian (N = 23), cervical and endome- healthy donors to identify genes that were not expressed trial cancer (25 patients each), using the TLDA 96a RT- or expressed at very low level in the peripheral blood of qPCR platform. After cell enrichment with OncoQuick healthy females but appeared very highly expressed in density gradient centrifugation 1/6 of the total RNA was the cancer cell lines. From the 18151 (54.8%) genes with amplified employing the TargetAmp™1-Round aRNA an average ANS < 1.5 in the healthy control samples Amplification Kit (Epicentre, Madison WI, USA) per maxT-test identified 457, 534, 526, and 503 genes differ- manufacturer instructions. The amplified RNA was con- entially expressed for the breast, cervical, endometrial, verted into cDNA with M-MLV Reverse Transcriptase, and ovarian cancer cell lines, respectively. These genes RNase H Minus (Promega, Madison WI, USA) and ran- comprised 54, 81, 63, and 60 genes with cancer-type dom hexamers as primers. Loading of the microfluidic specific expression for the respective cancer cell lines. cards, RT-qPCR amplification, and raw data analysis Additionally, the 50% one-sided trimmed maxT-test were performed as described in the microarray verifica- identified further 25, 27, 20 and 29 genes, which were tion section. All samples were analyzed in duplicate. differentially expressed in the breast, cervical, endome- The mean of the resulting duplicate C values was used trial and ovarian cancer cell lines compared to the as a quantitative value. If only one of the duplicates was healthy controls. positive (i.e. C < 50), the positive C value was taken. Finally, 356 differentially expressed genes were chosen t t Low-level expression of many genes in the peripheral for confirmatory gene expression profiling with RT- blood of the healthy control group decreased the overall qPCR using the TLDA 384 format (microarray data are specificity of the assay and required the introduction of provided in Additional file 2). This consisted of 337 a cut-off threshold value to separate the cancer patient genes identified by the maxT-test, 19 by the 50% one- group from the healthy control group: sidedtrimmedmaxT-test only, and the 4 genes: As proposed by Mikhitarian et al. [26], a threshold EFEMP1, EPS8L1, CRYZL1 and PCDHG represented value T for each gene X was set to three standard with more than one TaqMan® Assay. Additionally we deviations from the mean dC value in the control decided to analyze nine markers of well-known tumor tX group. dC values were calculated by normalizing the specificity (ERBB2, ESR1, PGR, PLAT, SCGB2A1, tX Obermayr et al. BMC Cancer 2010, 10:666 Page 7 of 12 http://www.biomedcentral.com/1471-2407/10/666 SCGB2A2, SERPINE1, SERPINE2 and TFF1)andsix Although background expression of EMP2, PPIC, candidate markers for CTC detection that were pre- DKFZp762E1312,and SLC6A8 was detected in the viously identified by our research group (COL3A1, GHR, unspiked blood, increasing expression levels of the CALB1, LPHN1, FN1 and EDNRA) [27]. respective genes were observed when tumor cells had been added to the blood, with a detection limit of 3 Verification of microarray results with RT-qPCR (EMP2, PPIC) and 26 tumor cells per ml of blood 146 genes of the TLDA 384 gene set were identified as (DKFZp762E1312, SLC6A8). Furthermore, the spiking potential markers for the detection of CTC in the blood experiments revealed that RT-qPCR might be less sensi- of cancer patients with expression levels below the detec- tive using the TLDA platform than using conventional tion limit of RT-qPCR (i.e. C 50) in the healthy control PCR tubes, because linear amplification patterns distin- group. The genes were sorted in descending order by guishing each 10-fold dilution were only observed with their average C value obtained from the 15 tumor cell C values smaller than 35 (data not shown). t t lines, and the first 93 genes were selected for further gene expression analysis of patients’ samples using the Gene expression in tumor tissues TLDA 96a format (see Additional file 3). None of the 15 The gene expression of the previously selected 93 genes known or supposed markers for CTC detection was con- was confirmed in tumor samples from patients with pri- sidered for further investigations either due to detectable mary breast, ovarian and endometrial cancer. We expression levels (ERBB2, ESR1, SERPINE1, SERPINE2 observed that the house-keeping gene expression levels and FN1) in healthy controls or due to inadequate gene were lower in ovarian cancer tissues than in tumor tis- expression in the tumor cell lines. sues of breast and endometrial cancer patients (GAPDH 24.2 ± 2.6, 22.2 ± 1.2, 22.7 ± 1.4 (SD) C ; B2 M 22.1 ± Sensitivity 3.4, 18.1 ± 1.5, 17.7 ± 1.9 (SD) C ; PGK 25.5 ± 2.7, 23.5 To assess the applicability of the TLDA platform for the ± 1.1, 22.4 ± 3.0 (SD) C in the respective tumor RT-qPCR based detection of circulating tumor cells, the patients). Two of the 93 genes were found to be tumor- expression levels of the specified 93 genes were mea- site specific: PLEKHC1 (pleckstrin homology domain sured in healthy female blood samples spiked with T-47 containing, family C [with FERM domain] member 1) D breast cancer cells. CCNE2 and MAL2 transcripts and SGCB (sarcoglycan beta) transcripts were detected were detected in blood samples spiked with at least 26 only in ovarian cancer patients (see Additional file 4), and 3 tumor cells per ml blood, respectively (Figure 2), although they were also detected in cancer cell lines of but they were not detected in the unspiked blood. breast and endometrial origin either. Interestingly, Figure 2 Sensitivity of RT-qPCR using TLDA platform. Expression levels of 93 candidate genes were analyzed using cDNA generated from total RNA isolated from peripheral blood samples from a healthy female donor and the same blood spiked with 4, 40 and 400 T47-D tumor cells after cell enrichment. RNA was pre-amplified using the TargetAmp™1-Round aRNA Amplification Kit. Average C values obtained from RT- qPCR amplification of CCNE2, DKFZp762E1312, EMP2, MAL2, PPIC, and SLC6A8 transcripts using the TLDA format are shown. MAL2 and CCNE2 gene expression was below the detection limit of RT-qPCR in the unspiked blood. The detection sensitivity of the respective marker gene was estimated to be 40 and 400 tumor cells per 15 ml whole blood. Obermayr et al. BMC Cancer 2010, 10:666 Page 8 of 12 http://www.biomedcentral.com/1471-2407/10/666 expression of the selected 93 genes was detected in ovarian cancer groups, the percentage of positive more ovarian cancer patients than in breast and endo- patients was found to be 44%, 64% and 19%, respectively metrial cancer patients (median percentage of positive (see Table 2 and Figure 3). patients in the respective tumor groups was 78.4%, Additionally, hMAM-specific RT-qPCR performed for 64.0% and 32.0%). the same set of breast and ovarian cancer blood samples confirmed the tissue specific expression of mammaglo- Gene markers for CTC detection bin A. Transcripts were only detected in recurrent The expression of the previously selected 93 genes was breast cancer patients with an incidence of 38.7%, but evaluated in blood samples from cancer patients, to iden- neither in primary breast cancer patients, ovarian cancer tify the most promising markers for CTC detection. At patients, nor in the healthy controls. Likewise, EpCAM primary diagnosis, each 17 (68.0%) cervical and endome- gene over-expression was detected in the blood of trial cancer, 6 (26.1%) ovarian cancer and 8 (38.1%) neither ovarian cancer patients nor healthy females. In breast cancer patients over-expressed at least one out of the blood of breast cancer patients, we found EpCAM the 93 potential candidate genes at levels above the over-expression in 5.0% of the patients at primary diag- defined threshold. At the time-point of disease recur- nosis and in 19.4% of the patients with clinical evidence rence, 27 (87.1%) breast cancer patients were positive for of disease recurrence (see Table 2). at least one gene. Of the 93 candidate genes, 40 were able to identify patients using the defined respective thresh- Discussion olds. 33 of these genes were capable to identify patients Using a stepwise approach combining genome-wide with primary breast cancer, and this number was reduced expression profiling and TaqMan® based RT-qPCR we to 15 for patients with advanced disease stage. 14 of these identified six genes (CCNE2, DKFZp762E1312, EMP2, genes could identify patients with cervical and endome- MAL2, PPIC,and SLC6A8) as potential markers for the trial cancer and four of the 40 genes identified ovarian detection of circulating tumor cells in the peripheral cancer patients. The remaining 55 genes did not provide blood of patients with breast cancer and gynecological any value due to similar expression levels in both the malignancies. Although each of these genes is implicated healthy control and cancer patient groups. in cancer, they have not previously been specified for The purpose of this study was to identify a panel of the detection of circulating tumor cells in cancer genes for future multi-marker RT-qPCR based analysis patients. to increase the sensitivity to detect circulating tumor Initial screening of candidate gene markers for CTC cells. For this purpose, we selected those genes, which detection was performed using a microarray-based gene were over-expressed in more than 10% of the patients expression analysis of human cancer cell lines and with recurrent breast cancer, since circulating tumor mononuclear blood cells obtained from healthy females. cells are more likely in advanced disease. According to After verification of the microarray results, a set of 93 this criterion, six genes (CCNE2, DKFZp762E1312, gene markers was selected for the RT-qPCR analysis of EMP2, MAL2, PPIC and SLC6A8)werechosenfor a blood samples from healthy females and from patients RT-qPCR marker panel. Using this panel 81% of the with breast, ovarian, endometrial, and cervical cancer. breast cancer patients with recurrence and 29% of the Due to background gene expression in the healthy breast cancer patients at initial diagnosis were positive blood samples, a rigorous cut-off threshold value was for at least one gene. In the cervical, endometrial and introduced to separate the patients from the healthy Table 2 Marker gene expression in peripheral blood Positive blood samples (%) Patients Panel CCNE2 MAL2 EMP2 SLC6A8 DKFZ PPIC hMAM EpCAM rec. BC (N = 31) 80.6 32.3 19.4 32.3 45.2 25.8 19.4 38.7 19.4 BC (N = 21) 28.6 23.8 0 4.8 0 4.8 0 0 5.0 OC (N = 23) 19.0 13.0 4.3 0 0 0 0 0 0 EC (N = 25) 64.0 36.0 20.0 12.0 12.0 8.0 8.0 0 0 CC (N = 25) 44.0 40.0 4.0 4.0 4.0 4.0 0 0 0 Healthy (N = 26) 0 0 0 0 0 0 0 0 0 The percentage of patients with RT-qPCR positive blood samples is shown. RT-qPCR positivity was defined as gene expression beyond the cut-off threshold, which was set for each gene marker at three standard deviations from the mean expression in healthy control blood samples. Positivity in percentage shown for the “panel” (CCNE2, DKFZp762E1312, EMP2, MAL2, PPIC and SLC6A8) is defined as positivity for at least one of the markers. (BC: breast cancer, rec. BC: recurrent breast cancer, OC: ovarian cancer, EC: endometrial cancer, CC: cervical cancer, ND: not done) Obermayr et al. BMC Cancer 2010, 10:666 Page 9 of 12 http://www.biomedcentral.com/1471-2407/10/666 Figure 3 RT-qPCR analysis of marker gene expression in peripheral blood. Gene expression was analyzed in blood samples taken from patients (triangles) with recurrent breast cancer (A), and in blood samples taken at first diagnosis from breast (B), endometrial (C), cervical (D) and ovarian (E) cancer patients. Blood from healthy females (circles) served as a control group. Mononuclear cells were enriched with the Oncoquick density gradient. RT-qPCR was performed following a RNA pre-amplification step. Average C values obtained from duplicates were normalized to GAPDH gene expression. Cut-off threshold values calculated from the mean average normalized gene expression in healthy female blood as indicated by horizontal lines for the respective gene markers (DKFZp762E1312 1.39, SLC6A8 2.92, PPIC 3.61, EMP2 6.84, MAL2 14.61, CCNE2 16.83). controls. We assumed that the over-expression of at their blood samples were chosen to identify new gene least one gene marker in relation to the defined thresh- markers for CTC detection. A panel of six genes: old value indicated the presence of circulating tumor CCNE2, DKFZp762E1312, EMP2, MAL2, PPIC,and cells. As patients with recurrent breast cancer are most SLC6A8 that were over-expressed in the blood of 81% likely to harbor circulating tumor cells in their blood, of patients with recurrent breast cancer was then chosen Obermayr et al. BMC Cancer 2010, 10:666 Page 10 of 12 http://www.biomedcentral.com/1471-2407/10/666 as gene markers for the molecular detection of CTC. In blood of patients with breast cancer or gynecologic contrast, at initial diagnosis using the six gene panel malignancies is useful for the detection of circulating only 29% of the breast cancer patients were RT-qPCR tumor cells, alone or combined with other markers positive. In addition, the new gene panel identified such as hMAM or EpCAM. Interestingly, the patients with other female cancers (i.e. cervical, endome- DKFZp762E1312, EMP2, PPIC,and SLC6A8 transcripts, trial and ovarian cancer). but not CCNE2 and MAL2 transcripts were detected in In tumor cell spiking experiments the sensitivity of the the blood of healthy females. Therefore, we suppose that applied RT-qPCR was estimated to be 3 to 26 tumor cells the detection of CCNE2 and MAL2 transcripts in the per ml whole blood; similar sensitivities are reported for blood of cancer patients is indicative for CTC presence RT-qPCR- and immuno-mediated detection (reviewed by (which had not been verified by immunocytochemistry). Gervasoni et al. [28]). However, we found out that Taq- However, the observed increase of CCNE2 mRNA levels Man® Low Density Arrays are typically not the method of in the diseased group compared to the healthy control choice for the detection of rare template molecules. group, which are reported to be undetectable in normal In the present study, all blood samples were taken quiescent cells arrested in G [30], is in conflict with the before removal of the tumor masses. To estimate supposed non-proliferative nature of circulating tumor whether the six gene panel is useful to detect minimal cells [31]. Interestingly, both CCNE2 and MAL2 are residual disease, further experiments should include located on chromosome 8q, a region which is frequently blood samples from cancer patients taken after the exci- increased in copy number in breast [32] and other can- sion of the primary tumor. Although we have already cer types [32,33], and one of the most important target analysed several blood samples taken from breast cancer genes affected by gains and amplifications of 8q is the patients with no evidence of disease six months after MYC oncogene. completion of their adjuvant chemotherapy, the follow- The frequency of hMAM gene expression in the blood up time is yet too short to make any conclusions con- of breast cancer patients is in line with the frequencies cerning the patient outcome. reported by Roncella et al. [20]. 10 of the 12 hMAM There are further limitations that need to be acknowl- positive blood samples (83%) were also positive when edged and addressed regarding the experimental design analyzed using the six gene panel, and 52% of the recur- of the present study. First, when we evaluated various rent breast cancer blood samples were solely identified approaches for the enrichment of circulating tumor cells by CCNE2, DKFZp762E1312, EMP2, MAL2, PPIC,or in in the course of the project, we found out that Onco- SLC6A8. Similarly, the six gene panel identified all of quick may insufficiently recover spiked tumor cells, in the EpCAM positive blood samples. particular when only a few tumor cells were added to the blood (i.e. ≤ 20 tumor cells per 15 ml blood) [29]. Conclusions For this reason, false-negative RT-qPCR results are likely In this study, we identified new gene markers for the to occur for cancer patients with low CTC counts. Sec- assessment of circulating tumor cells. We have shown ond, the density of the tumor cells depends on their dif- that the RT-qPCR-based multi-marker analysis using ferentiation state. Therefore, undifferentiated tumor the six genes: CCNE2, DKFZp762E1312, EMP2, MAL2, cells having a higher density might pass through the PPIC,or SLC6A8 more than doubled the number of Oncoquick density gradient. Finally, we cannot exclude positive patients with recurrent breast cancer compared false-positive cases due to non-malignant epithelial cells, to the analysis of hMAM or EpCAM gene expression which may contaminate the blood samples during veni- alone. Therefore, we suggest that the significantly higher puncture and which express the targeted transcripts. expression of these six genes in the peripheral blood Nevertheless, we decided in favour of the Oncoquick indicates the presence of circulating tumor cells. This density gradient, because it dramatically reduced back- multi marker analysis may provide a tool for clinical ground gene expression of the selected targets in healthy monitoring and treatment control of breast cancer and PBMC samples. To enhance the sensitivity and specifi- of gynecological malignancies. Eventually it may also be city of the approach, future experiments should primar- useful for the early detection. ily aim at improving the recovery rate of the tumour cell enrichment. Further evaluation of the six CTC mar- Additional material kers should be done without RNA pre-amplification and using the conventional PCR tube format instead of Taq- Additional file 1: Base line characteristics of patients included into the RT-qPCR analysis of tumor tissue. Man® Low Density Array format. Additional file 2: Microarray data of 356 differentially expressed Despite these limitations, we suggest that the RT- genes. qPCR based analysis of CCNE2, DKFZp762E1312, EMP2, MAL2, PPIC, and SLC6A8 gene expression in the Obermayr et al. BMC Cancer 2010, 10:666 Page 11 of 12 http://www.biomedcentral.com/1471-2407/10/666 5. Lacroix M: Significance, detection and markers of disseminated breast Additional file 3: Gene identifiers of the TLDA 96a platform.93 cancer cells. Endocr Relat Cancer 2006, 13(4):1033-1067. genes were selected as CTC candidate genes for the RT-qPCR analysis of 6. Zieglschmid V, Hollmann C, Bocher O: Detection of disseminated tumor blood and tumor tissue samples from cancer patients. Additionally, three cells in peripheral blood. Crit Rev Clin Lab Sci 2005, 42(2):155-196. house-keeping genes (B2M, GAPDH, and PGK1) were chosen as an 7. Marth C, Kisic J, Kaern J, Trope C, Fodstad O: Circulating tumor cells in the internal reference. peripheral blood and bone marrow of patients with ovarian carcinoma Additional file 4: Gene expression in tumor tissues. The percentage do not predict prognosis. Cancer 2002, 94(3):707-712. of breast, endometrial and ovarian cancer patients with gene expression 8. Wimberger P, Heubner M, Otterbach F, Fehm T, Kimmig R, Kasimir-Bauer S: detected by RT-qPCR is shown for each of the 93 candidate genes and Influence of platinum-based chemotherapy on disseminated tumor cells for the three internal reference genes (B2M, GAPDH, and PGK1). in blood and bone marrow of patients with ovarian cancer. Gynecol Oncol 2007, 107(2):331-338. 9. Diddle AW, Sholes DM, Hollingsworth J, Kinlaw S: Cervical carcinoma; cancer cells in the circulating blood. Am J Obstet Gynecol 1959, 78:582-585. Acknowledgements 10. Yabushita H, Shimazu M, Yamada H, Sawaguchi K, Noguchi M, Nakanishi M, This work was supported by the GEN-AU project “Cancer Transcriptomics” Kawai M: Occult lymph node metastases detected by cytokeratin and “Bioinformatics Integration Network II” (BIN II) of the Austrian Federal immunohistochemistry predict recurrence in node-negative endometrial Ministry of Science and Research. Keiichi Isaka (Department of Obstetrics and cancer. Gynecol Oncol 2001, 80(2):139-144. Gynecology at the Tokyo Medical University, J) kindly provided the tumor 11. Ji XQ, Sato H, Tanaka H, Konishi Y, Fujimoto T, Takahashi O, Tanaka T: Real- cell lines EN and EJ. Volker Möbus (Department of Obstetrics and time quantitative RT-PCR detection of disseminated endometrial tumor Gynecology, University of Ulm, D) gave the cell lines OV-MZ-01a and OV-MZ- cells in peripheral blood and lymph nodes using the LightCycler System. 6 by, and Hiroyuki Kuramoto (Department of Clinical Cytology Graduate Gynecol Oncol 2006, 100(2):355-360. School of Medical Sciences, School of Medicine, Kitasato University, 12. Cristofanilli M, Budd GT, Ellis MJ, Stopeck A, Matera J, Miller MC, Reuben JM, Sagamihara, Kanagawa, J) gave HEC-50-B. We thank Nicola Tidow (GeneSys Doyle GV, Allard WJ, Terstappen LW, et al: Circulating tumor cells, disease Laboratories GmbH, Muenster, D) for performing the microarray experiments. progression, and survival in metastatic breast cancer. N Engl J Med 2004, We are particularly grateful to Gerhard G. Thallinger (Institute for Genomics 351(8):781-791. and Bioinformatics, Graz University of Technology, Graz, A) for assisting 13. Cristofanilli M, Hayes DF, Budd GT, Ellis MJ, Stopeck A, Reuben JM, statistical analysis of the microarray data and to Ingrid Schiebel (Department Doyle GV, Matera J, Allard WJ, Miller MC, et al: Circulating tumor cells: a of Obstetrics and Gynecology, Medical University of Vienna, A) for cultivating novel prognostic factor for newly diagnosed metastatic breast cancer. J tumor cells. Clin Oncol 2005, 23(7):1420-1430. 14. Pachmann K, Camara O, Kavallaris A, Schneider U, Schunemann S, Author details 1 Hoffken K: Quantification of the response of circulating epithelial cells to Department of Obstetrics and Gynecology, Comprehensive Cancer Center, 2 neodadjuvant treatment for breast cancer: a new tool for therapy Medical University of Vienna, Vienna, Austria. Institute for Genomics and 3 monitoring. Breast Cancer Res 2005, 7(6):R975-979. Bioinformatics, Graz University of Technology, Graz, Austria. Department of 15. Cristofanilli M, Broglio KR, Guarneri V, Jackson S, Fritsche HA, Islam R, Medicine I, Comprehensive Cancer Center, Medical University of Vienna, 4 Dawood S, Reuben JM, Kau SW, Lara JM, et al: Circulating tumor cells in Vienna, Austria. Department of Blood Group Serology and Transfusion 5 metastatic breast cancer: biologic staging beyond tumor burden. Clin Medicine, Medical University of Vienna, Vienna, Austria. Department of Breast Cancer 2007, 7(6):471-479. Gynecology, European Competence Center for Ovarian Cancer, Charité - 6 16. Sieuwerts AM, Kraan J, Bolt J, van der Spoel P, Elstrodt F, Schutte M, University Medicine of Berlin, Berlin, Germany. Clinical Institute of Pathology, Martens JW, Gratama JW, Sleijfer S, Foekens JA: Anti-epithelial cell Comprehensive Cancer Center, Medical University of Vienna, Vienna, Austria. 7 adhesion molecule antibodies and the detection of circulating normal- Section of Clinical Biometrics, Center for Medical Statistics, Informatics and like breast tumor cells. J Natl Cancer Inst 2009, 101(1):61-66. Intelligent Systems, Comprehensive Cancer Center, Medical University of 8 17. Medical devices; immunology and microbiology devices; classification of Vienna, Vienna, Austria. Ludwig Boltzmann Gesellschaft - Cluster the immunomagnetic circulating cancer cell selection and enumeration Translational Oncology, A-1090 Vienna, Austria. system. Final rule. Fed Regist 2004, 69(91):26036-26038. 18. Zhong XY, Kaul S, Eichler A, Bastert G: Evaluating GA733-2 mRNA as a Authors’ contributions marker for the detection of micrometastatic breast cancer in peripheral EO performed and supervised sample processing, carried out the RT-qPCR blood and bone marrow. Arch Gynecol Obstet 1999, 263(1-2):2-6. analysis and data evaluation, and drafted the manuscript. FSC and GH 19. Stathopoulou A, Mavroudis D, Perraki M, Apostolaki S, Vlachonikolis I, performed statistical analysis of microarray data. CFS, MKT, AR, MK, MBF, RH Lianidou E, Georgoulias V: Molecular detection of cancer cells in the and JH coordinated the collection of patients’ blood and tissue samples. DT peripheral blood of patients with breast cancer: comparison of CK-19, and RZ designed the study and contributed to data interpretation. RZ CEA and maspin as detection markers. Anticancer Res 2003, served as mentor for the entire project. All authors read and approved the 23(2C):1883-1890. final manuscript. 20. Stathopoulou A, Angelopoulou K, Perraki M, Georgoulias V, Malamos N, Lianidou E: Quantitative RT-PCR luminometric hybridization assay with Competing interests an RNA internal standard for cytokeratin-19 mRNA in peripheral blood ZR, having ZR, DT and EO as inventors, filed a patent application based of patients with breast cancer. Clin Biochem 2001, 34(8):651-659. upon this manuscript. 21. Roncella S, Ferro P, Bacigalupo B, Pronzato P, Tognoni A, Falco E, Gianquinto D, Ansaldo V, Dessanti P, Fais F, et al: Human mammaglobin Received: 22 February 2010 Accepted: 3 December 2010 mRNA is a reliable molecular marker for detecting occult breast cancer Published: 3 December 2010 cells in peripheral blood. J Exp Clin Cancer Res 2005, 24(2):265-271. 22. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, Ellis B, References Gautier L, Ge Y, Gentry J, et al: Bioconductor: open software development 1. GLOBOCAN 2002: Cancer Incidence, Mortality and Prevalence for computational biology and bioinformatics. Genome Biol 2004, 5(10): Worldwide. [http://www-dep.iarc.fr]. R80. 2. Ashworth T: A case of cancer in which cells similar to those in the 23. Ge YDS, Speed TP: Resampling-based multiple testing for microarray data tumours were seen in the blood after death. Aus Med J 1869, 14:146-149. analysis. Test 2003, 12:1-44. 3. Paterlini-Brechot P, Benali NL: Circulating tumor cells (CTC) detection: 24. Gleiss A, Sanchez-Cabo F, Perco P, Tong D, Heinze G: Adaptive trimmed t- clinical impact and future directions. Cancer Lett 2007, 253(2):180-204. statistics for identifying predominantly high expression in a microarray 4. Smerage JB, Hayes DF: The measurement and therapeutic implications of experiment. Stat Med . circulating tumour cells in breast cancer. Br J Cancer 2006, 94(1):8-12. Obermayr et al. BMC Cancer 2010, 10:666 Page 12 of 12 http://www.biomedcentral.com/1471-2407/10/666 25. Klein CA, Seidl S, Petat-Dutter K, Offner S, Geigl JB, Schmidt-Kittler O, Wendler N, Passlick B, Huber RM, Schlimok G, et al: Combined transcriptome and genome analysis of single micrometastatic cells. Nat Biotechnol 2002, 20(4):387-392. 26. Mikhitarian K, Martin RH, Ruppel MB, Gillanders WE, Hoda R, Schutte del H, Callahan K, Mitas M, Cole DJ: Detection of mammaglobin mRNA in peripheral blood is associated with high grade breast cancer: interim results of a prospective cohort study. BMC Cancer 2008, 8:55. 27. Zeillinger R, Fabjani G: Method and kit for diagnosis of a cancerous disease, method for determining the reaction of a patient to the treatment for a cancerous disease, medicament for the prophylaxis or treatment of a cancerous disease. European Patent No. 1786929 2006. 28. Gervasoni A, Monasterio Munoz RM, Wengler GS, Rizzi A, Zaniboni A, Parolini O: Molecular signature detection of circulating tumor cells using a panel of selected genes. Cancer Lett 2008, 263(2):267-279. 29. Konigsberg R, Gneist M, Jahn-Kuch D, Pfeiler G, Hager G, Hudec M, Dittrich C, Zeillinger R: Circulating tumor cells in metastatic colorectal cancer: efficacy and feasibility of different enrichment methods. Cancer Lett 293(1):117-123. 30. Lauper N, Beck AR, Cariou S, Richman L, Hofmann K, Reith W, Slingerland JM, Amati B: Cyclin E2: a novel CDK2 partner in the late G1 and S phases of the mammalian cell cycle. Oncogene 1998, 17(20):2637-2643. 31. Muller V, Stahmann N, Riethdorf S, Rau T, Zabel T, Goetz A, Janicke F, Pantel K: Circulating tumor cells in breast cancer: correlation to bone marrow micrometastases, heterogeneous response to systemic therapy and low proliferative activity. Clin Cancer Res 2005, 11(10):3678-3685. 32. Forozan F, Karhu R, Kononen J, Kallioniemi A, Kallioniemi OP: Genome screening by comparative genomic hybridization. Trends Genet 1997, 13(10):405-409. 33. Knuutila S, Bjorkqvist AM, Autio K, Tarkkanen M, Wolf M, Monni O, Szymanska J, Larramendy ML, Tapper J, Pere H, et al: DNA copy number amplifications in human neoplasms: review of comparative genomic hybridization studies. Am J Pathol 1998, 152(5):1107-1123. Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2407/10/666/prepub doi:10.1186/1471-2407-10-666 Cite this article as: Obermayr et al.: Assessment of a six gene panel for the molecular detection of circulating tumor cells in the blood of female cancer patients. BMC Cancer 2010 10:666. Submit your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit

Journal

BMC CancerSpringer Journals

Published: Dec 3, 2010

There are no references for this article.