Circulating Tumor Cells in Patients with Testicular Germ Cell Tumors

Purpose: Germ cell tumors (GCTs) represent the most frequent malignancies among young men, but little is known about circulating tumor cells (CTCs) in these tumors. Considering their heterogeneity, CTCs were investigated using two independent assays targeting germ cell tumor and epithelial cell–specific markers, and results were correlated with disease stage, histology, and serum tumor markers. Experimental Design: CTCs were enriched from peripheral blood (n = 143 patients) and testicular vein blood (TVB, n = 19 patients) using Ficoll density gradient centrifugation. For CTC detection, a combination of germ cell tumor (anti-SALL4, anti-OCT3/4) and epithelial cell–specific (anti-keratin, anti-EpCAM) antibodies was used. In parallel, 122 corresponding peripheral blood samples were analyzed using the CellSearch system. Results: In total, CTCs were detected in 25 of 143 (17.5%) peripheral blood samples, whereas only 11.5% of patients were CTC-positive when considering exclusively the CellSearch assay. The presence of CTCs in peripheral blood correlated with clinical stage (P < 0.001) with 41% of CTC positivity in patients with metastasized tumors and 100% in patients with relapsed and chemotherapy-refractory disease. Histologically, CTC-positive patients suffered more frequently from nonseminomatous primary tumors (P < 0.001), with higher percentage of yolk sac (P < 0.001) and teratoma (P = 0.004) components. Furthermore, CTC detection was associated with elevated serum levels of α-fetoprotein (AFP; P = 0.025), β-human chorionic gonadotropin (βHCG; P = 0.002), and lactate dehydrogenase (LDH; P = 0.002). Incidence and numbers of CTCs in TVB were much higher than in peripheral blood. Conclusions: The inclusion of germ cell tumor–specific markers improves CTC detection in GCTs. CTCs occur frequently in patients with more aggressive disease, and there is a gradient of CTCs with decreasing numbers from the tumor-draining vein to the periphery. Clin Cancer Res; 20(14); 3830–41. ©2014 AACR.


Translational relevance
This study shows that CTCs can be detected in the peripheral blood of about 18% of patients diagnosed with germ cell tumors, both using the established CellSearch® system, as well as a custom made assay using CTC enrichment and a combination of immunocytochemical markers. The presence of CTCs is associated with histologically more aggressive nonseminomatous tumors, advanced clinical stages, increased serum concentrations of tumor markers (AFP, ȕHCG and LDH), and chemotherapy refractory relapse. The current study indicates the diagnostic potential of CTCs as prognostic biomarker and will stimulate future investigations, ideally with patients under both surveillance and chemotherapy and well documented follow-up to assess the clinical relevance of CTC analyses in GCTs. In addition, further characterization of CTCs in GCT patients will not only help to better understand the biology of the disease, but also offers the possibility to look for therapeutic targets in the era of targeted therapy.

Introduction
Germ cell tumors (GCTs) of testis are the most frequent malignancies in young men between the age of 20 and 40. Less than 10% of all GCTs arise in extragonadal sites (e.g. mediastinum), however their management follows that of testicular GCTs (1). The incidence of GCTs in Western countries is rising since decades (2). The available serum tumor markers -alpha-fetoprotein (AFP), ȕ-human chorionic gonadotropin (ȕHCG), and lactate dehydrogenase (LDH), used for diagnostics and follow up, are increased in up to 80% of patients with testicular GCTs (3). However, these markers are not elevated in patients with pure teratoma. The gold standard for detecting metastases is computed tomography of the chest, abdomen, and pelvis. However, approximately 10-20% and 30% of patients with stage I seminomatous and nonseminomatous testicular GCTs, respectively, have occult (invisible at the time of diagnosis) metastases, leading to relapse during surveillance (4,5). Therefore, additional markers would facilitate staging and clinical decision making.
Detection of circulating tumor cells (CTCs) has been associated with poor prognosis in carcinoma patients (6,7,8). However, little is known about the presence of CTCs in blood of GCT patients. Few studies have previously assessed presence of tumor-specific mRNA or whole cells in apheresis products of patients undergoing peripheral stem cell transplants (9,10,11). Additionally, in a small number of studies, putative CTCs in peripheral blood of testicular GCT patients were detected by reverse transcriptase chain reaction (RT-PCR) using alpha fetoprotein and human chorionic gonadotropin-specific mRNA as markers (12,13).
In the present study, we developed a new assay, using a label-free enrichment method based on physical properties of tumor cells (i. e., Ficoll density centrifugation)  (14).
Sal-like protein 4 (SALL4), a stem cell marker, has been reported as a novel sensitive and specific diagnostic marker that is present in all testicular GCTs types (15). OCT3/4 (POU5F1) is a transcription factor involved in regulation of pluripotency during normal development, and is a sensitive and specific marker for seminomas and embryonal carcinomas (16). Keratins 8 and 18 are the keratins of simple epithelia. These are the first two keratins expressed during mouse embryogenesis and might be expressed by seminomas and embryonal carcinomas (17,18,19,20).
Keratin 19 as a marker for epithelial cells was found to be present in some seminomas, and the majority of nonseminomas (10). Furthermore, epithelial cell adhesion molecule (EpCAM) is a homophilic, calcium-independent cell adhesion molecule and is uniquely expressed in germline and spermatogonial stem cells (21,22). Schoenberger et al. (2013) showed that EpCAM is highly expressed in malignant GCTs such as yolk sac tumors and choriocarcinomas (23).
In parallel, we applied the semi-automated CellSearch® system for CTC detection. This system is thus far the only assay cleared by the FDA for CTC detection, following the pivotal clinical studies in patients with metastatic breast, prostate, and colon cancer (6,24,25) (26,27). To the best of our knowledge, no data has so far been published on the use of the CellSearch® system in GCTs.
Thus, the aim of the present study was to determine the incidence of CTCs in GCT patients using two independent assays, and to correlate the findings to clinical parameters such as tumor histology, stage of disease, and tumor marker levels in blood serum.

Characteristics of patients
One-hundred forty one patients suffering from histologically proven testicular GCTs, and 2 patients with mediastinal GCTs were included in this study. Pathologists  individuals were included as control group, 9 suffering from non-germ cell tumor of the testis (Leydig cell tumor, Sertoli cell tumor), and 10 healthy male volunteers.
Detailed clinico-pathological parameters of all patients are summarized in Table 1.
The study was conducted according to REMARK study recommendations (28).

Control material
In order to select suitable markers for CTC detection, the expression of different

Enrichment of CTCs
Nine to seventeen mL of peripheral and 0.5 to 3.2 mL of testicular vein blood were collected into EDTA-tubes. All blood samples were processed within a maximum of 24 h after collection. Mononuclear cells (MNCs) were enriched using the Ficoll-Hypaque gradient centrifugation (29). The MNC fraction, preferably containing CTCs, was resuspended in 1 mL of 1xPBS and spinned down to prepare the microscopic slides. The slides were left overnight to air-dry at room temperature and

Detection of CTCs using the combination of GCT and epithelial markers
For each double immunocytochemical staining, cytospins containing 3x10 6 (for peripheral blood) or 1x10 6 (for testicular vein blood) MNCs were prepared as follows: The slides were fixed for 10 min. Finally, all the specimens were counterstained with DAPI VectaShield Medium (Vector Laboratories, Burlingame, CA, US) and covered with cover-slips. Slides were evaluated under the fluorescence microscope (Axioplan2, Zeiss, Germany).

Cell Search analysis of CTCs
In parallel, 122 blood samples (7.5 ml) were collected in CellSave tubes (Veridex, Raritan, NJ, USA). The CellSearch® system (Veridex) was used as previously described (8,27). Analyzer (Veridex). Results of the analyses were interpreted by researchers experienced with this system, and CTC assessment was performed on a cell by cell basis.

Evaluation of apoptosis
Apoptotic cells were assessed by characteristic morphology -presence of small pycnotic nuclei or apoptotic apoptotic bodies or speckled cytoplasmic staining seen in the CellSearch® images.

Fluorescence in situ hybridization analysis of CTCs
To confirm the germ-cell origin of CTCs, fluorescence in situ hybridization (FISH) was conducted using a probe derived from the Homo sapiens PAC clone 876C13 from

Statistical analysis
Statistical analyses were performed with the usage of SPSS software (Chicago, IL, US) ver. 21

Expression of selected markers in GCT cell lines and primary tumors
Strong expression of SALL4, OCT3/4, and keratins was found in more than 75% of the cells of all GCT cell lines. In some cells of the TCam-2 and NT2 cell lines, keratins were detected in a dot-like pattern ( Supplementary Fig. S1D). While EpCAM was strongly expressed in TCam-2 and 2102Ep cells, only weak or no expression was observed in NT2 and NCCIT cells (Supplementary Fig. S1A). Cells from all 4 cell lines were detected by the CellSearch® system ( Supplementary Fig. S1B). All 12 primary tumors showed strong SALL4 and OCT3/4 expression, whereas only 8

Analysis of recovery rates and assay specificity for GCT cell lines
Using Ficoll-Hypaque gradient centrifugation followed by ICC staining with selected markers, tumor cells spiked into blood of healthy donors were recovered in the range of 60-70% (data not shown). For the CellSearch® system the recovery rate was 80-100% (data not shown).
No positive cells for selected markers (SALL4, OCT3/4, Keratins and EpCAM) were found in the peripheral blood of 10 healthy volunteers or in non-germ cell tumor patients (data not shown).

CTCs in testicular vein blood
In total, 12 (63.2%) of 19 testicular vein blood samples were positive for tumor cells. The mean tumor cell number was 45 (range: 4-120; median: 16) per 1x10 6 of MNCs. In 7 (58.3%) patients, CTCs were forming clusters of 3 to 7 cells (Fig. 1B). In one patient, CTCs were found both in testicular vein and peripheral blood and no patient had CTCs in peripheral vein only. Due to the small number of samples, no further statistical analysis was done for tumor cells detected in testicular vein blood samples.

Detection of gains in 12p chromosomal region of CTCs
In 4 patients, CTCs with gains of the 12p chromosomal region were found in peripheral blood ( Fig. 2B; Table 3). In one tested sample from testicular vein, four tumor cells with 5-12 signals from the 12p region of interest and 5-7 centromere 12 signals were found which was similar to aberrations observed in the primary tumor tissue ( Fig. 2A; Table 3). Leukocytes present on slides showed 2 signals for 12p and centromere 12 each.

Association of CTCs with clinico-pathological parameters
In order to assess clinical characteristics associated with the detection of CTCs, correlations between the presence of CTCs and different clinico-pathological parameters of patients were analysed (

Preliminary follow up evaluation
The median follow up time was too short (mean: 13.7 months; range: 0.66-24.1; median: 14.9) for an in-depth analysis of the prognostic relevance of CTCs.
However, it is noteworthy that all six patients with treatment-refractory disease that were included in this analysis, were positive for CTCs. Of these patients, three died  Table S1). In PB of one patient, CTCs were detected using Ficoll-enrichment followed by ICC staining, the second patient showed CellSearch®-positivity, and the third patient was positive by both detection methods.

Discussion
This is the first study demonstrating the presence of CTCs in a large number of GCT patients including all clinical stages and all histological subtypes. We were able to detect CTCs in approximately 18% of GCT patients using immunocytochemical staining with germ cell tumor-and epithelial cell-specific markers in two independent CTC assays based on different physical capture technologies. Detection of CTCs was correlated to higher tumor stages, more aggressive tumor histology, increased tumor markers in serum, and early relapses, suggesting that CTCs detected by our assays might either contribute to or at least indicate disease progression in GCTs.
Interestingly, there was little overlap between our new assay and the FDA- Using FISH, the detected CTCs showed gain of the 12p chromosomal region which is a cytogenetic hallmark of GCTs, present in about 80% of tumors (30,31).
This finding additionally confirms the germ-cell origin of CTCs and thus specificity of our assays.
To investigate whether the detection of CTCs in GCT patients is associated with an increased risk of metastatic relapse and progression, as shown for other tumor entities (32,33), future long term follow up evaluations are required.
Interestingly, we could already show that all patients with treatment refractory disease included in this analysis were positive for CTCs in the present study.
Additionally, a strong correlation was found between the presence of CTCs and more advanced clinical stages of GCTs. In a much smaller cohort study, Hautkappe et al., Nonseminomatous tumors tend to be more aggressive and are more prone to metastasize than seminomatous tumors. In this study, CTCs were more frequently found in patients with nonseminomatous than in seminomatous tumors, especially in those with higher percentages of yolk sac and teratoma components. Teratomas and yolk sac tumors have been reported as the most common types of tumors observed in patients with late relapses (34,35,36). Teratomas are not producing common tumor markers such as AFP, ȕHCG or LDH. Therefore, evaluation of CTCs might be of special value to detect metastasis or relapse in patients suffering from teratomas.
In several studies, the presence of predominantly embryonal carcinoma components has been reported as a factor for poor prognosis (35,37,38,39). Hautkappe et al. (2000) found AFP-and/or ȕHCG-mRNAs mostly in patients with embryonal carcinomas (13). In the current study, a strong correlation between the presence of CTCs and a higher percentage (>50%) of embryonal carcinoma component within primary tumors was only found in blood samples analysed with the CellSearch® system, suggesting that this approach might be particularly useful to detect CTCs derived from embryonal carcinomas expressing both EpCAM as well as one of the detected keratins. diagnosis (40). In the present study, CTCs were significantly associated with elevated serum concentrations of AFP, ȕHCG, and LDH. Higher levels of serum tumor markers after orchiectomy are associated with worse outcome in metastasized nonseminoma (41). An association between elevated serum tumor markers and the presence of CTCs might indicate a prognostic significance of CTCs. However, CTCs were found also in 4 marker-negative patients, suggesting that determination of CTCs might help to minimize the diagnostic gap of conventional tumor markers.
To the best of our knowledge, this is the first study investigating intraoperatively collected blood from the testicular vein of patients with GCTs. The testicular vein carries deoxygenated blood from testis to the inferior vena cava or one of its tributaries and might be the first path of haematogenous tumor-cell spread in GCT (42). Our results seem to support this hypothesis. Of note, we observed very high numbers of tumor cells of up to 120 per 1x10 6 MNCs in TVB. In the testicular vein, the CTC yield was higher in comparison to PB, which is similar to the gradients observed in breast and colorectal cancers, where significantly more CTCs could be detected in the central venous blood or the mesenteric vein, respectively (43,44). In more than half of GCT patients, clusters of 3-7 tumor cells were observed, which was much more frequent than in PB (20 %). These findings suggest that a high number of isolated and clustered CTCs is shed from the primary tumor into the local blood stream, and during the circulation CTCs may undergo anoikis or they might be filtered in the lungs (or other organs), which may cause the observed gradient between TVB and PB. metastasis formation (45,46) and may serve as a "liquid biopsy" assessing potential targets for therapy [e. g., CD30 or glypican-3 (47,48,49)] or gene mutations relevant to targeted therapy [e.g., c-KIT or BRAF (50)]. Sequential blood analyses during therapy may also hold the promise to gain insights into drug resistance in individual patients. Thus, the present work opens a new avenue to personalized medicine in GCT patients.