BRCA1 loss pre-existing in small subpopulations of prostate cancer is associated with advanced disease and metastatic spread to lymph nodes and peripheral blood

Patients

The patients with sporadic mono- and multifocal PCa were treated by radical prostatectomy in the Department of Urology of University of Muenster during 1998–2000. They did not receive any hormone therapy and did not manifest any metastasis at the time point of diagnosis and surgery. The tumors were characterized by different clinical parameters (Tab. 1). Their pathologic stages ranged from T2N0M0 to T3N1M0 and the final Gleason sums from 2 to 10. The mean patient age at the moment of surgery was 64 years old (range: 47 – 78). The concentration of serum PSA was measured after prostatectomy twice every year and occurrence of BR was defined as the end-point of disease-free survival. The follow-up was available for 186 patients with a median follow-up of 34 months (range: 1–114). Additionally, seven patients with metastatic PCa treated at the Department of Urology, University Medical Centre, Hamburg-Eppendorf, Germany were selected for CTCs analysis. Their short clinical description is embedded in Fig. 3B.

Clinical samples

Two tissue microarrays (TMA) containing 398 primary PCa and 196 LNM tissue cores (TCs), corresponding to 199 and 98 patients, respectively, were prepared as described (16). 53 (13%) TCs contained normal prostate tissue, further 337 (85%) tumor TCs belonged to 191 (96%) patients, 8 TCs were missing pathologic characteristics. All specimens were evaluated by a pathologist in the Institute of Pathology, University of Münster, Germany. The patients with multifocal PCa were represented on the TMA by TCs originated from two different tumor foci. From this cohort, 31 cases of primary tumors were selected for additional AI analysis and 11 paired PCa-LNM formalin-fixed, paraffin-embedded (FFPE) tissues were chosen for both FISH and AI analyses.

At the Department of Urology, University Medical Centre, Hamburg-Eppendorf, Germany, 20 ml of peripheral blood from seven unrelated patients with metastatic PCa and three healthy volunteers was collected into EDTA tubes and processed for CTC enrichment within three hours after phlebotomy as described (17).

FISH

FISH probes were prepared from DNA isolated from BAC clones RP11–242D8 and RP11-506G7 using a Large DNA Construct Isolation Kit (Qiagen, Germany) and the BioPrime® Total Genomic Labeling System (GibcoInvitrogen, UK) according to the manufacturers’ protocols. The deparaffinized 4-μm-thick TMA or FFPE sections were fixed in 2% formaldehyde in methanol for 10 min at −20°C and pretreated in Spot-Light Tissue Heat Pretreatment Buffer (Zymed) for 10 min at 97°C. The cells were digested for 10 min in 100 μl Enzyme-Buffer (Zymed) at 37°C and subsequently dehydrated in a graded alcohol series. One microliter of COT1™ Human DNA (Roche, Germany), 1 μl of CEP17 Spectrum Aqua® (Abbott Molecular) as a reference probe and 4 μl of Spectrum Orange®-labeled probe (Abbott Molecular) for a target gene suspended in 4 μl of hybridization buffer (20% dextran sulfate/50% formamide/4×SSC buffer) were applied on the top of specimens. For ERBB-2/HER-2 detection, 10 μl of commercial ready-to-use probe (PathVysionTM Her-2 DNA Probe Kit, Abbott Molecular) was used. The specimens were denatured for 3 min at 95°C and hybridized for at least 16 hours at 37°C. Afterwards, they were washed 2 min each in 2×SSC/NP-40 buffer at 72°C and at 21°C. The specimens were dehydrated in an alcohol series, counterstained with Vectashield® Mounting Medium (Vector Laboratories Inc., Burlingame, CA) containing 0.5 μg/ml 4′,6-diamidino-2-phenylindole (DAPI) and analyzed in a fluorescence microscope (Zeiss, Germany). For each tissue core, target and reference probe signals were counted in at least 20 cells showing a minimum of two signals for the reference probe. The gene dosage calculated as ratio of the number of target probe signal over reference probe signals was determined according to experimental cut-offs established on 20 normal prostate tissues for every probe combination.

FISH on CTCs

Isolated CTCs on slides were fixed in ice-cold 75% ethanol for 2 min then pretreated with 100 μg/ml RNase A for 40 min at 21°C, before cells were immersed in 1x citrate buffer (pH 6.0, Dako, Denmark) for 3 min at 120°C. The cells were postfixed in 1% formaldehyde in 1×PBS for 10 min, and dehydrated in an alcohol series. Air-dried cells were denatured in denaturation buffer (70% formamide, 0.6×SSC, pH 7.4) for 5 min at 73°C and dehydrated in alcohol. Hybridization was conducted as described above with minor modifications (2 μl of Spectrum Orange®-labelled RP11-242D8 probe, 0.5 μl CEP17, 6.5 μl hybridization buffer, Abbott Molecular). The next day, the slides were washed three times for 10 min in 50%formamide/2×SSC buffer, two times for 10 min. in 2×SSC buffer and once for 5 min in 0.1×SSC/0.1%Tween-20 buffer at 46°C. The cells were permeabilized in three changes of 1×Tris Buffered Saline/1% Tween-20 (TBST) for 3 min. each. The unspecific binding was reduced by a 20 min incubation in Blocking Serum (Dako). The primary mouse monoclonal antibodies against pan-cytokeratin (A45, Mikromet, Germany) or vimentin (RV202, DB Pharmingen™) diluted 1:100 or 1:200, respectively, in DakoREAL™ Antibody Diluent (Dako) were applied to the cells for 45 min at 21°C. The specimens were washed three times for 3 min with 1×TBST and incubated with a secondary 1:200-diluted Alexa-488-labeled anti-mouse rabbit antibody (MiBioTech) for 45 min at 21°C. Finally, the slides were washed twice for 3 min with 1×TBST and once for 3 min with 1×PBS, and counterstained with Vectashield® Mounting Medium containing DAPI. Cytokeratin- or vimentin-positive cells, showing no apoptotic or necrotic morphology, were analyzed for their BRCA1 gene dosage. Additionally, a total of 500 cells per sample were scored for BRCA1 losses and compared to the scores obtained in three healthy volunteers (false-hybridization index).

AI analysis

DNA isolation, microsatellite-sequence-based PCR and AI analysis of locus D17S855 (BRCA1) and D10S541 (PTEN) were performed as described (3).

Methylation assay

To investigate the BRCA1 promoter methylation status, a MLPA-assay, described in (18), was performed (Methylation Kit ME45, MRC-Holland, The Netherlands) according to the manufacturer’s protocol with minor modifications (initial denaturation for 20 min, 40 PCR cycles). Data were normalized by Methylation Status Analysis and further analyzed by the Coffalyser MLPA data analysis software.

Statistics

Kruskal-Wallis and Mann-Whitney tests were performed to assess genes dosages associations with clinical outcome. To estimate the hazard risk, Cox-Hazard-Potential regression analysis (CI 95%) was done. Disease free survival curves were generated by Kaplan-Meier test with log-rank univariate analysis.

Experimental strategy

To test the hypothesis that BRCA1 gene aberrations in PCa may be decisive events in primary PCa progression, our retrospective study investigated their frequency in primary PCa, LNM and CTCs isolated from peripheral blood of metastatic PCa patients. The samples flow is presented in Tab. S1. In order to assess specificity, efficiency and accuracy of the selected FISH probes, 20 normal prostate tissues and different prostate and breast cancer cell lines were tested. The experimental cut-offs for normal gene dosage were calculated as 0.71 – 1.29, 0.72 – 1.28 and 0.80 – 1.20 for BRCA1, 17q GRR and ERBB-2/HER-2, respectively.

BRCA1 gene dosage in primary prostate cancers

Two-hundred-thirteen TCs from 133 PCa patients were evaluable for FISH analysis of the BRCA1 gene (Fig. 1A–B). BRCA1 losses and gains occurred in 18 (14%) and 5 (7%) of patients, respectively. The BRCA1 gene dosage obtained by FISH agreed with BRCA1 AI status assessed in a subgroup of 31 selected tumors (R²= 0.614, p < 0.001, Fig. S1). Eighty-one tested patients (62%) were investigated in two TCs which enabled the assessment of heterogeneity of BRCA1 gene dosage within the analyzed PCa. Eighty-three percent of the BRCA1 aberrations were observed only in one TC which suggested the unique character of these alterations confined to the small population of tumor cells.

Characterization of BRCA1 aberrations and their clinical relevance

All detected BRCA1 losses were monoallelic and appeared in majority of tumor cells localized in single TC (containing usually 20 to 1000 tumor cells). Promoter analysis of BRCA1 showed lack of methylation in all 31 PCa cases. Two other loci on chromosome 17q (17q GRR and ERBB-2/HER-2) were analyzed to investigate whether BRCA1 alterations are limited to BRCA1 or caused by larger alterations of chromosome 17q. The BRCA1 gains were found restricted to the gene locus. On the contrary, BRCA1 losses coexisted in 71% of the cases with the loss of 17q GRR containing genes already proposed to be involved in tumorigenesis, e.g. BECN1, EZH1 and VPS25 (Fig. 1C)(19–21). Losses of the entire region of chromosome 17 encompassing bands q12–21.1 were detected in 3% of all cases. Nevertheless, only BRCA1 alterations correlated significantly with clinical parameters (Tab. S2). BRCA1 losses were associated with more advanced T status (p=0.014) and higher Gleason score determined in TCs (p = 0.048). They occurred also more frequently in tumors with node-positive status (p=0.032), whereas BRCA1 gains were observed only in node-negative patients.

Prognostic value of BRCA1 aberrations

When compared to the BR data only BRCA1 losses classified according to the cut-off values ≤0.75 indicated the patients with the shorter disease-free survival (p = 0.0161, Fig. 2A). In addition, BRCA1 losses appeared to be prognostically significant in multivariate analysis (p = 0.048, Fig. 2B). However, the Cox Regression model was done on a relatively small group of patients and needs to be confirmed on a larger cohort.

BRCA1 aberrations in prostate cancer metastasis to pelvic lymph nodes and circulating tumor cells

In LNM, losses were found in 27% of 62 evaluable cases and were significantly increased in comparison to primary tumors (p=0.0072). BRCA1 gains were detected at a frequency similar to primary PCa (8%). The comparison of corresponding pairs of primary tumors and its metastases revealed that 8 of 11 pairs exhibited the matching BRCA1 gene status in LNM and at least one tumor focus (Tab. 2). In this analysis TCs were compared as well as the larger sections of tumor and metastasis specimens which helped to determine the additional foci of tumor cells bearing BRCA1 aberrations within primary tumors. It confirmed the suspected transmission of genetic abnormalities between these two sites suggesting that somatic BRCA1 mutations can predispose cells to dissemination. As a proof of principle, BRCA1 gene dosage was also assessed in tumor cells separated from peripheral blood of PCa patients suffering from metastasis. The BRCA1 losses were found in cytokeratin (CK+)-and of note vimentin (Vim+)-positive CTCs isolated from peripheral blood (Fig. 3A, B). Applying the strategy in which the frequency of cells with BRCA1 losses was counted in 500 randomly selected cells and compared to the indices of false-hybridization obtained in the normal healthy volunteers, significantly elevated number of cells with BRCA1 loss were found in 4 of 7 patients (p < 0.05). Of note is that both Vim staining and scoring method helped to point out the cells, particularly, when the pattern two BRCA1 copies : three CEP17 copies was observed (Fig. 3B). Such a genotype was rarely detectable in the blood of healthy volunteers. The clinically established parameters such as T and N status showed that the cells bearing BRCA1 losses in PCa seem to be efficient to penetrate LN, possibly survive the stress conditions in blood stream and consequently might invade distant organs.

Table 2.

The BRCA1 gene dosage and PTEN AI in 11 PCa – LNM pairs assessed in TMA and FFPE sections (WS).

Open in a new tab

Determination of PTEN gene dosage

Another tumor suppressor gene PTEN is reported to be lost in both primary PCa (22) and a majority of prostate CTCs (3) which is also believed to be one of the initiating neoplastic events during prostate tumorigenesis in mice (23). In the current study, PTEN gene dosage was analyzed in a fraction of PCa and LNM and compared with the status of BRCA1 gene. The AI of PTEN paralleled BRCA1 losses and gains, but was also found in cases with normal BRCA1 gene dosage both in 31 tested primary PCa and 11 corresponding pairs PCa-LNM (Tab. 2, S3). This observation leads to the conclusion that PTEN losses always underlie BRCA1 losses but not vice versa.