Location of Mutation in BRCA2 Gene and Survival in Patients with Ovarian Cancer

This article is featured in Highlights of This Issue, p. 251

Germline mutations in BRCA1 and BRCA2 genes have been identified as predisposing to hereditary breast and ovarian cancers (1). Up to 20% of high-grade serous ovarian carcinomas (HGSOC) show germline and/or somatic mutations of BRCA1/BRCA2 genes (2–5). Ovarian cancer patients with such mutations have better survival than noncarriers (2, 6, 7). Importantly, almost all the ovarian cancer patients in those studies received platinum-based chemotherapy, inducing interstrand crosslinks (8). Interstrand crosslinks formed by platinum lead to severe distortion of the DNA double helix, and consequently double-strand breaks (DSB; refs. 9, 10). Homologous recombination (HR), a major mechanism for protecting genome integrity in proliferating cells, is pivotal in repairing DSBs that arise during the processing of interstrand crosslinks.

From a biological point of view, BRCA1 and BRCA2 are both key players of DNA damage repair but have different functions (11). BRCA1 is a pleiotropic DNA damage response protein that functions in both DNA damage checkpoint activation and repair, including HR (11, 12). In contrast, the primary function of BRCA2 is HR. BRCA2 interacts directly with RAD51 and promotes its specific recruitment to DSBs sites where recombination is initiated (11). RAD51 is essential for HR. The improved survival of BRCA1/BRCA2 germline mutation carriers, especially BRCA2 carriers, compared with noncarriers (13, 14) has been linked to the role of these proteins in the HR pathway. However, there is no clear explanation of why BRCA2 carriers fare better than BRCA1 carriers.

BRCA2 is one of the largest proteins in the human body. The central portion of the protein contains 8 repeat sequences (called BRC repeats), which bind to RAD51, named RAD51-binding domain (RAD51-BD; refs. 11, 15). A second large segment of BRCA2 encompasses a DNA-binding domain (Fig. 1). Genetic and functional studies of BRCA2 revealed that mutations located in RAD51-BD impair the ability of BRCA2 to recruit RAD51, hampering HR (16, 17). In the current report, we investigated whether mutations in the RAD51-BD (exon 11) of the BRCA2 gene impact progression-free survival (PFS), platinum-free interval (PFI), and overall survival (OS) in ovarian cancer patients.

Clinical and genetic data were collected from a study cohort of ovarian cancer patients screened for germline mutations of BRCA1 and BRCA2 genes. We analyzed the curated dataset of HGSOC from The Cancer Genome Atlas (TCGA) as a validation cohort.

Study participants were women with confirmed invasive epithelial ovarian or fallopian tube or peritoneal carcinoma, who had been tested for germline BRCA1 or BRCA2 pathogenic mutations through blood tests between January 1995 and December 2015 and who received platinum-based chemotherapy. Ovarian cancer patients referred to the clinical genetics Units’ at Centre Leon Bérard (Lyon, France), Institut du Cancer Jean Mermoz (Lyon, France), and Hôpitaux Universitaires de Genève (Geneva, Switzerland) were included in the cohort. To increase the number of BRCA2 carriers in the study cohort, only BRCA2 carriers were included from Institut Curie (Paris, France). The study was conducted following ethical guidelines of the Declaration of Helsinki. The study was reviewed by the local Institutional Review Board in each hospital. Informed consent was obtained from all living patients in Geneva. All the French patients consented to the use of their data at the time of genetic analysis. Clinical and pathologic data were collected from medical records. These included patient demographics, tumor characteristics, surgical staging, macroscopic residual disease, platinum sensitivity, recurrence, and survival status. Surgical stage was classified according to the International Federation of Gynecology and Obstetrics (FIGO) at diagnosis. Information regarding residual disease following primary surgery was acquired from medical records. Pathology data, including histologic subtypes, tumor stages, and grades, were obtained from pathology reports.

BRCA1 and BRCA2 mutations were classified as truncating according to the ENIGMA BRCA1/2 Gene Variant Classification Criteria (http://www.enigmaconsortium.org/). Women with variants of uncertain significance were considered noncarriers. Blood samples for germline DNA testing were obtained when the patients were referred to clinical genetics Unit. All participants were screened for BRCA1 and BRCA2 mutations. The BRCA2 gene comprises 27 exons and encodes a 3418 amino-acid (AA) protein (Fig. 1). RAD51-BD corresponds to the region covering AA 1003-2082 of BRCA2 (exon 11). DNA-BD corresponds to AA 2481-3186 (http://www.ncbi.nlm.nih.gov/protein/NP_000050.2). The BRCA2 protein has a second binding site to RAD51 located in the carboxy-terminus (named RAD51-binding site or TR2), which we excluded from the analysis because it is dispensable for HR (11, 12).

We obtained the TCGA database of 316 HGSOC patients. Detailed patient information, including age at diagnosis, tumor stage and grade, and surgical outcome, has been described previously (13). Whole-exome sequencing of germline and somatic DNA was performed in all cases (2).

The primary endpoint was PFS. Secondary endpoints were PFI and OS. Date of first relapse was defined as the first instance of disease progression based on CT imaging by RECIST or clinical progression. PFS was defined as the interval between histologic diagnosis and first relapse, death, or the last follow-up (censored). OS was defined as the interval between histologic diagnosis and the date of death from any cause or last follow-up (censored). PFI was defined as the interval between the time of completion of platinum-based chemotherapy and the date of first relapse or death. Platinum-sensitive patients were defined as those having PFI >6 months.

Patient characteristics were compared using Pearson χ² or Fisher exact test for frequencies, and Kruskal–Wallis test for age distribution. Characteristics were compared pair by pair (BRCA2 carriers’ vs. noncarriers). The survival functions of the different subgroups were estimated using the Kaplan–Meier method and compared using a log-rank test. The relative hazards for each mutation group were estimated with a Cox proportional hazards model adjusted for tumor stage and macroscopic residual disease. The categorical covariates included in the model were FIGO stage (stage I–II vs. III–IV) and macroscopic residual disease (absent vs. present). P ≤ 0.05 was considered statistically significant; all tests were two-sided. Statistical analysis was carried out using R software.

Of the 353 women included in the cohort, 87 were excluded from the analysis: 74 women were found to carry BRCA1 mutation (they will be analyzed in another study), 2 patients harbored mutations in both BRCA1 and BRCA2 genes, 9 patients did not receive adjuvant chemotherapy (5 noncarriers, 1 BRCA1, and 3 BRCA2 carriers), and follow-up data were not available for 2 noncarriers (Supplementary Fig. S1). In total, 266 women (152 from Centre Leon Bérard, 40 from Hôpitaux Universitaires de Genève, 26 from Institut du Cancer Jean Mermoz, and 48 from Institut Curie) were analyzed for outcome: 188 were tested negative for germline mutations in BRCA1 and BRCA2 genes (hereafter “noncarriers”) and 78 were found to carry BRCA2 mutations (hereafter “BRCA2 carriers”). All mutations were germline. Patient demographics and clinical and treatment characteristics are summarized in Table 1.

The majority of patients had serous carcinomas and advanced stages (III/IV; 88%). Two thirds (67%) of the patients had no macroscopic residual disease and 85% were platinum sensitive. All patients received a platinum agent, and 95% of the patients received a combination with taxane.

There was no difference between the two groups regarding age of diagnosis (P = 0.814), histologic subtype (P = 0.451), FIGO staging (P = 0.655), and the presence of macroscopic residual disease (P = 0.15). Platinum sensitivity rates were not statistically different in the two groups of patients: 85% of noncarriers and 86% of BRCA2 carriers were sensitive (P = 0.722). However, BRCA2 carriers had longer PFI (P = 0.008).

Median follow-up of the cohort was 4 years. After adjustment for major prognostic factors (stage and macroscopic residual disease), PFS at 5 years was significantly higher in BRCA2 compared with noncarriers (P < 0.001; Fig. 2A; Supplementary Table S1). Regarding OS, BRCA2 carriers had significantly higher 5-year OS compared with noncarriers (P < 0.001; Supplementary Table S2).

Among the 78 BRCA2 carriers of the study cohort, 42 had mutations located within the RAD51-BD (Fig. 1; Supplementary Table S3), including 5 carriers of the Ashkenazi founder mutation c.5946del/p.Ser1982Argfs*22. They had no macroscopic residual disease (P = 0.009) more frequently and had significantly prolonged PFI (P = 0.011), compared with noncarriers. There was no significant difference between the 36 other BRCA2 carriers and noncarriers, regarding any of the clinical, pathology, and treatment characteristics (Table 1).

After adjustment for stage and macroscopic residual disease, only BRCA2 carriers with mutations located in the RAD51-BD had significantly higher 5-year PFS than noncarriers (P = 0.001), whereas the other BRCA2 carriers did not have significant prolonged 5-year PFS (P = 0.094, Fig. 2B; Table 2). Regarding OS, events (deaths) occurred in 9 (11.5%) of BRCA2 carriers, insufficient for survival subgroup analyses.

We investigated whether mutations located in other domains of BRCA2 than the RAD51-BD impacted the outcomes of carriers. In our study cohort, 14 BRCA2 carriers harbored mutations located in the DNA-BD (AA 2481-3186; Fig. 1; Supplementary Table S3). Their 5-year PFS was not significantly longer than noncarriers [27%; HR = 0.73; 95% confidence interval (CI), 0.40–1.41; P = 0.38]. BRCA2 carriers with mutations located in exons 1–10 (19%; HR = 0.72; 95% CI, 0.36–1.42; P = 0.338) or exons 12–27 (32%; HR = 0.64; 95% CI, 0.35–1.16; P = 0.143) did not show higher 5-year PFS, compared with noncarriers.

To validate our observations, we explored the correlation between location of mutation within BRCA2 gene and survival in the TCGA cohort of HGSOC (n = 316). All patients received platinum-based chemotherapy. Among the 34 patients who had germline or somatic mutations of the BRCA2 gene (hereafter “BRCA2 carriers”), two of them harbored mutations in both BRCA1 and BRCA2 genes and were excluded from analysis. Three additional patients with missense mutations were excluded, resulting in 29 BRCA2 carriers being analyzed. We excluded patients with somatic or germline mutations of BRCA1. A total of 247 patients had no somatic or germline mutations of BRCA1 or BRCA2 (hereafter “noncarriers”).

Among the 243 patients for whom the information on residual disease was available, 20% had no macroscopic residual disease. After adjustment for stage and macroscopic residual disease, 5-year PFS was significantly higher in BRCA2 carriers with mutations within the RAD51-BD (P = 0.004), while BRCA2 carriers with mutations in other domains did not show better 5-year PFS than noncarriers (P = 0.639; Table 3; Fig. 3A). Five-year OS was significantly higher only in BRCA2 carriers with mutations within the RAD51-BD compared with noncarriers (P = 0.001). BRCA2 carriers whose mutations were not located in the RAD51-BD did not show any difference in OS compared with noncarriers (P = 0.594; Table 3; Fig. 3B).

In this study, we addressed the correlation between BRCA2 genotype and ovarian cancer patients’ survival, according to the functional domains of the protein. Data from two independent cohorts showed that ovarian cancer patients treated with platinum and harboring germline and/or somatic mutations located at the RAD51-BD (exon 11) of the BRCA2 gene have prolonged survival compared with noncarriers of BRCA1/BRCA2 mutations. Mutations occurring in the BRCA2 gene at other locations than the RAD51-BD did not impact the outcome, compared with noncarriers. Previous work investigated the correlation between type of mutation (base change, simple deletion, etc.) or location of the mutation across BRCA2 gene and patient outcome, and the authors did not report a correlation with outcome (3, 18). This can be explained by the absence of analysis of the location based on the different functional domains of the protein (3).

Our hypothesis is biology driven. Through HR, BRCA2 plays a central role in repairing DSBs induced by interstrand crosslinks (9). It coordinates the formation of RAD51 filaments at DSBs. BRCA2 interacts with monomeric RAD51 primarily via the highly conserved BRC repeats encoded by exon 11 (RAD51-BD). DNA DSBs induced by platinum can be considered as acting as a “targeted chemotherapy” in BRCA2-mutated tumors. We hypothesized that mutations at sites crucial for the association between RAD51 and BRC repeats could impair the ability of BRCA2 to recruit RAD51 to DNA breaks, hampering HR (Supplementary Fig. S2; refs. 16, 17) and impacting patients’ survival.

Of course, this study has several limitations. First, the cohort is derived from a retrospective study that only included patients screened for germline mutations. The criteria for genetic analysis of BRCA1/BRCA2 genes have evolved over time. Systematic screening for all ovarian cancer patients has recently been implemented. Thus, our cohort study included patients selected for their young age at diagnosis and/or significant family history of cancer and does not reflect the general population of ovarian cancer patients. Second, our noncarrier group probably included some patients with somatic mutations of BRCA1, BRCA2, and other HR genes (2). The enrichment in BRCA2 carriers, who were mainly recruited at Institut Curie could also be considered as a limitation. On the other hand, compared with the literature, our study cohort is representative of BRCA2 carriers in terms of clinical characteristics (age of diagnosis, histologic subtype, stage, etc.) and clinical behavior with BRCA2 carriers showing significantly prolonged survival compared with noncarriers (4, 13).

Two thirds of the patients included in the study cohort had no macroscopic residual disease, explaining high survival rates, as described previously (19). The low number of events in the BRCA2 carriers group allowed subgroup investigation only for PFS, not OS. OS data for BRCA2 patients will require longer follow-up before subgroup analysis. It should be noted that PFS has been shown to be a surrogate marker for OS in BRCA carriers in the TCGA cohort (13) and the GOG clinical trials 218 and 262 (4). Moreover, we observed in the TCGA cohort that OS was dramatically prolonged in these patients. One major difference between our study cohort and the TCGA cohort is the percentage of patients with no macroscopic residual disease, which was lower in the TCGA cohort (66% vs. 20%, respectively).

Our study and the TCGA cohort showed that only mutations in the RAD51-BD of BRCA2 gene lead to prolonged PFS, PFI, and OS in ovarian cancer patients, compared with noncarriers. As BRCA2 is a very large protein, it is possible that mutations in other domains of the protein could also impact clinical outcome of ovarian cancer patients. In our study cohort, we did not observe that mutations located in the DNA-BD impact PFS.

A very large proportion of pathogenic BRCA2 mutations are truncating deletions, meaning that cells containing early mutations (exons 1–10) are likely to produce no BRCA2 protein. In an exploratory analysis, we compared BRCA2 carriers with mutations located in either exons 1–10 or exons 12–27, and we did not observe a significant difference in the outcome of both groups. Our analysis is limited by the number of cases in each subgroup and needs to be confirmed on larger cohorts before deriving any conclusion. From a biological point of view, little is known about the functions of the large part of BRCA2 protein corresponding to exons 1–10 (AA 1-1002), except PALB2-BD (AA 10-40). This is an important question that needs to be addressed through preclinical functional studies.

Our work highlights the importance of RAD51 in the context of ovarian cancer. Germline mutations in RAD51 paralogs (20, 21), in particular RAD51C and RAD51D genes, increase the risk for ovarian cancer (21) and lead to high genomic instability and sensitivity to PARP inhibitors (22). Secondary gene reversion mutations restoring the open reading frame of RAD51C or RAD51D is a mechanism of acquired resistance to PARP inhibitors (23), as it was previously shown for BRCA1 and BRCA2 genes (24). Edwards and colleagues studied 12 independently derived PARP inhibitor–resistant CAPAN1 tumor cell lines (BRCA2 mutation p.Ser1982Argfs*22 located in the RAD51-BD) and found that none of them showed a secondary mutation event restoring the wild-type allele, but all exhibited BRCA2 DNA deletion events that restored the open reading frame and encoded the RAD51-BD (24, 25). It is perhaps ironic that the secondary mutations in BRCA2 or RAD51 leading to resistance may also arise because of this same HR deficiency (25).

Although mutations of RAD51-BD concerns a minority of ovarian cancer, this has to be put in the perspective of other cancers, such as breast (26), pancreatic (27), or castration-resistant prostate cancers (28) where BRCA2 mutations occur in 5% to 7% of the patients. This work, which needs further confirmation in larger cohorts (29), could have consequences for the clinical management of ovarian cancer patients. On the basis of the location of mutation in the BRCA2 gene, we can predict whether there will be a survival advantage and prolonged PFI in BRCA2 carriers compared with noncarriers. These patients may benefit more than other patients from reintroduction of platinum. This observation also questioned the impact of such alteration on selecting patients for PARP inhibitors. PARP inhibitors showed promising results in preclinical studies (30), but the response rate varies from one study to another (31, 32). PARP inhibitors lead to longer PFS especially in patients selected on the basis of their platinum sensitivity (31–35). Thus, platinum sensitivity probably acts as an “in vivo functional test” for HR defects (36), selecting BRCA carriers who will be more likely to benefit from PARP inhibitors. A recent report, showing an enrichment of BRCA2 carriers with mutations located in the RAD51-BD among patients who are long responding (>2 years) to olaparib as a maintenance therapy, reinforces our observations (37). We are currently investigating the PFS and OS of BRCA2 carriers treated with PARP inhibitors based on their genotype.

In conclusion, our data suggest that PFI and survival in BRCA2 carriers are prolonged specifically when the mutations occur in the RAD-51 BD (exon 11) of the gene. These results confirmed the new paradigm of personalized therapy in BRCA carriers.

S.I. Labidi-Galy, M. Rodrigues, and A. Bodmer are consultant/advisory board members for AstraZeneca. O. Tredan is a consultant/advisory board member for AstraZeneca, Lilly, Novartis, Pfizer, and Roche. M-H. Stern has ownership interests (including patents) at Myriad Genetics. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S.I. Labidi-Galy, T. Olivier, O. Tredan, P.O. Chappuis, I. Ray-Coquard

Development of methodology: S.I. Labidi-Galy, T. Olivier, P.O. Chappuis, A. Buisson, I. Ray-Coquard

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.I. Labidi-Galy, T. Olivier, M. Rodrigues, D. Ferraioli, O. Derbel, A. Bodmer, N. Chopin, O. Tredan, P.-E. Heudel, S. Stuckelberger, P. Meeus, V. Viassolo, A. Ayme, P.O. Chappuis, C. Houdayer, D. Stoppa-Lyonnet, L. Golmard, V. Bonadona, I. Ray-Coquard

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.I. Labidi-Galy, T. Olivier, M. Rodrigues, P. Petignat, P. Meraldi, P.O. Chappuis, M.-H. Stern, A. Buisson, L. Golmard, I. Ray-Coquard

Writing, review, and/or revision of the manuscript: S.I. Labidi-Galy, T. Olivier, M. Rodrigues, D. Ferraioli, O. Derbel, A. Bodmer, P. Petignat, B. Rak, O. Tredan, P.-E. Heudel, S. Stuckelberger, V. Viassolo, A. Ayme, P.O. Chappuis, C. Houdayer, D. Stoppa-Lyonnet, A. Buisson, L. Golmard, V. Bonadona, I. Ray-Coquard

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.I. Labidi-Galy, T. Olivier, M. Rodrigues, N. Chopin, C. Houdayer, L. Golmard

Study supervision: S.I. Labidi-Galy, T. Olivier, I. Ray-Coquard

The authors thank Laurence Zulianello for iconographic support, Gaelle Pot-Benais for technical support, Dr. Therese Wilhelm for critical review of the manuscript, and Dr. Sven Rottenberg for instructive discussions. We thank Drs. Michel Forni, Anne Hugli, and Bernard Exquis for providing clinical data. S.I. Labidi-Galy is supported by a grant from Fondation pour la lutte contre le cancer et pour des recherches médico-biologiques. M. Rodrigues received financial support from ITMO Cancer AVIESAN (Alliance Nationale pour les Sciences de la Vie et de la Santé, National Alliance for Life Sciences & Health) within the framework of the Cancer Plan.

The results shown here are based in part on data generated by TCGA Research Network: http://cancergenome.nih.gov.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.