Effect of Germline Mutations in Homologous Recombination Repair Genes on Overall Survival of Patients with Pancreatic Adenocarcinoma

Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal malignancy with a 5-year survival of less than 10% (1, 2). There is an ongoing effort to improve the overall survival of PDAC by identifying potentially targetable somatic or germline mutations. In recent years, it has been demonstrated that germline mutations in several cancer predisposition genes may confer an increased risk of PDAC, and are detected in approximately 5%–15% of patients with PDAC (3–6). Some studies have also demonstrated that the biological characteristics of PDAC tumors in germline mutation carriers of homologous recombination repair (HRR) are different from sporadic PDACs (7, 8). In addition, germline mutations in HRR genes are known to confer higher sensitivity to DNA-damaging agents and/or ionizing radiation (4, 9–11). Furthermore, a new oral class of therapeutic agents called PARP inhibitors has been developed to take advantage of impaired HRR in tumor cells to induce synthetic lethality. Recently, maintenance therapy with a PARP inhibitor was shown to improve progression-free survival in patients with germline BRCA1 or BRCA2 (BRCA1/2 hereafter) mutations and metastatic PDAC compared with placebo (12).

Despite the improved understanding of differences in tumor biology and response to therapy in mutation carriers of HRR genes, the impact of mutation status on overall survival (OS) is still unclear. Prior studies on the impact of germline mutations on OS in PDAC had demonstrated mixed results, but these were mostly restricted to BRCA1/2 mutations and were limited by small sample sizes (13–16). Because several genes involved in HRR have similar biological functions, the differences in tumor biology and response to therapy may not be limited to BRCA1/2 mutation carriers. In this context, we evaluated the clinical characteristics of PDAC in germline mutation carriers of HRR genes and implications of these mutations on OS in large prospective hospital-based cohort of patients with PDAC.

The study sample was derived from the Mayo Clinic Biospecimen Resource for Pancreas Research, a prospective registry offering participation to patients evaluated at Mayo Clinic Rochester for initial diagnosis of pancreatic cancer. Patients enrolled in this registry between October 12, 2000 and October 6, 2017 with PDAC were included. Patients who were enrolled in the registry more than 3 months after their initial diagnosis and those who had no further follow-up after their diagnosis were excluded. While the results of germline sequencing have previously been reported (3), this analysis focused on OS and included 454 additional cases enrolled after the prior publication. This study was approved by the Institutional Review Board (IRB) at Mayo Clinic, and all aspects of the study were in accordance with the Declaration of Helsinki. All patients signed written, informed consent.

Patients consenting to the study completed a baseline questionnaire on personal and family history and provided a blood sample. In addition, patient demographics, tumor characteristics, family history, and clinical outcomes were abstracted from electronic medical records to verify existing information reported by patients. For patients who did not follow up at Mayo Clinic, medical records from local institutions were requested at 1 year postdiagnosis, 3 years postdiagnosis and at death. A medical oncologist or trained nurse abstractor reviewed each record for abstraction of information pertaining to any treatment the patient received. Furthermore, additional follow-up questionnaires were sent to patients to obtain information on treatment and outcomes while participating in the study. The abstraction of information on treatment was complete on approximately two-thirds (64.4%) of the study participants at the time of last follow up or death. For the purpose of this study, patients treated with cisplatin, oxaliplatin, or carboplatin at any point during the course of treatment were considered to have received platinum-based chemotherapy.

Germline DNA extracted from peripheral blood mononuclear cells was analyzed for germline pathogenic variants in the coding regions and consensus splice sites of 37 genes (Supplementary Information; Supplementary Table S1) using a custom amplicon-based QIAseq panel (QIAGEN). Pooled sample libraries from 768 samples were subjected to paired-end 150-base pair sequencing in each lane of a HiSeq4000 (Illumina) with a median coverage of 200×. Reads were trimmed with Cutadapt version 1.10 (17) and aligned with BWA-MEM (18). Sequence realignment, recalibration, haplotype calling, and depth of coverage were conducted using Genome Analysis Toolkit (GATK) version 3.4–46 (19). Large genomic rearrangements were detected with Pattern CNV v1.1.3 (20). Annotation of mutations was provided through the BioR toolkit (21) leveraging dbNSFP v3.0 (22), ClinVar (23), Clinical Annotation of Variants (24), and population frequencies from Genome Aggregation Database (gnomAD; ref. 11) and Exome Aggregation Consortium (ExAC; ref. 25). Pathogenic mutations were viewed with VCF-Miner (26). A five-tier system was used to classify mutations based on the American College of Medical Genetics and the Association for Molecular Pathology guidelines (27). Pathogenic and likely pathogenic variants were analyzed together as pathogenic variants.

Patients with a germline mutation in one of the eight genes (ATM, BARD1, BRCA1, BRCA2, BRIP1, PALB2, RAD51C, and RAD51D) involved in HRR were considered as carriers whereas those who tested negative for germline mutation in all 37 genes were considered noncarriers. These eight genes were selected based on prior studies demonstrating association with homologous recombination deficiency (28). Because other genes in the QIAseq panel may be associated with HRR, mutation carriers in genes other than the eight HRR genes were excluded from primary analysis to mitigate their effect. Baseline characteristics between carriers and noncarriers were compared using Fisher exact test for categorical variables and ANOVA for continuous variables. OS between carriers and noncarriers was compared in a multivariable Cox proportional hazard regression model including age, sex, and stage at diagnosis as covariates along with subset analysis by specific chemotherapy regimens, year of diagnosis, and specific gene mutations. Separate sensitivity analyses for differences in baseline characteristics and OS were performed in patients with germline mutations in the eight HRR genes (carriers) and in those who tested negative for mutations in the eight HRR genes (noncarriers). All statistical tests were performed using SAS 9.4 statistical software (SAS Institute).

Out of the 3,078 patients included in the final analysis, 175 (5.7%) were found to carry a mutation in one of the eight HRR genes: 67 (2.2%) in BRCA2, 65 (2.1%) in ATM, 20 (0.6%) in BRCA1, 12 (0.4%) in PALB2, four (0.1%) in BRIP1, four (0.1%) in RAD51C, two (0.1%) in BARD1, and one (0.03%) in RAD51D (Table 1). The noncarriers included 2,730 (88.7%) patients without mutations in any of the 37 genes evaluated. The median age at diagnosis of the cohort was 65.8 years with a median duration of follow-up of 9.9 years.

Baseline characteristics of 175 mutation carriers in eight HRR genes and 2,730 noncarriers are presented in Table 2. Mutation carriers were diagnosed with PDAC at a younger age (62.8 vs. 65.8 years, P < 0.001) and were more likely to present with metastatic disease at diagnosis (46.2% vs. 35.6%, P < 0.001) compared with noncarriers (Table 2). In addition, a higher proportion of mutation carriers had PDAC in the body or tail of the pancreas compared with noncarriers. Among patients with resectable PDAC, the rate of surgery was similar between the two groups. Furthermore, there were no statistically significant differences in the rates of chemotherapy between mutation carriers and noncarriers. The demographic and tumor characteristics of ATM carriers, which is one of the largest groups of mutation carriers in this cohort is presented in Supplementary Table S2.

The median OS for mutation carriers was 14.6 months [95% confidence interval (CI), 13.2–16.1 months] compared with 11.7 months (95% CI, 11.2–12.4 months) for noncarriers (Fig. 1). In univariate analysis, the difference in the OS between mutation carriers and noncarriers did not reach statistical significance (HR, 0.86; 95% CI, 0.73–1.01; P = 0.07). However, in multivariable analysis adjusting for age, sex, and stage at diagnosis, mutation carrier status was associated with a significantly longer OS compared with noncarriers (HR, 0.83; 95% CI, 0.70–0.97; P = 0.02; Table 3). Among surgically resectable patients, the median OS was 23.7 months in mutation carriers and 23.5 months in noncarriers (Supplementary Fig. S1), whereas among nonresectable/metastatic patients, the median OS was 11.5 months in mutation carriers and 9.6 months in noncarriers (Supplementary Fig. S2).

Further exploratory analyses were performed by stage of diagnosis, year of diagnosis, and type of chemotherapy (Table 3). Among patients with metastatic or nonresectable PDAC, mutation carriers were observed to have a significantly longer OS compared with noncarriers (HR, 0.81; 95% CI, 0.67–0.97; P = 0.02). In addition, mutation carrier status was also significantly associated with longer OS among patients diagnosed after May 2011 (when FOLFIRINOX use started) compared with noncarriers diagnosed during the same time period (HR, 0.75; 95% CI, 0.56–1.00, P = 0.04). However, specific chemotherapy regimens including platinum agents and FOLFIRINOX were not associated with OS. Similarly, on separate analysis of metastatic/nonresectable and surgically resectable PDAC cases receiving platinum-based chemotherapy, mutation carriers were not noted have a statistically significant difference in OS compared with noncarriers (Table 3). Furthermore, no difference in OS by mutation carrier status was noted among patients not receiving platinum-based chemotherapy. However, restricting the analysis to mutation carriers with metastatic or nonresectable disease, patients who received platinum-based chemotherapy had better OS compared with mutation carriers who did not receive platinum-based chemotherapy (HR, 0.46; 95% CI, 0.30–0.68; P < 0.001; Supplementary Fig. S3).

Further gene-level analysis demonstrated that germline ATM mutation carriers had longer OS compared with patients without germline mutations in any of the 37 HRR genes (HR, 0.72; 95% CI, 0.55–0.94; P = 0.01). Significant differences in OS were not noted for BRCA1, BRCA2, or PALB2 carriers compared with noncarriers (Table 3; Supplementary Figs. S4 and S5).

In sensitivity analysis including 175 mutation carriers in eight HRR genes compared with the remaining 2,903 patients negative only for a mutation in these eight genes, the overall differences in baseline characteristics and OS were similar to the primary analysis (Supplementary Tables S3 and S4).

In a large prospective registry of patients with PDAC, we observed an association between germline HRR mutations and improved OS. Prior studies primarily focused on OS in PDAC with BRCA1/2 mutations and exposure to platinum based chemotherapy (13, 14, 29–33). These studies have demonstrated mixed results, with some studies demonstrating longer OS (29, 33) while others demonstrated shorter or similar OS (13, 14). Fewer studies have evaluated the role of other DNA-damage repair or HRR genes (15, 33–36). However, the majority of the prior studies were limited by their retrospective nature and small sample. In contrast, this study is the largest to date, to the best of our knowledge, for carriers of HRR genes in patients with PDAC. In addition, this study included a large number of noncarriers. Furthermore, the prospective collection of clinical outcomes is a significant strength of this study. The findings of our study have significant implications for future research, primarily in understanding the differences in tumor biology and response to standard treatments in germline mutation carriers with PDAC. In addition, these findings will aid in appropriate counseling of prognosis in mutation carriers with PDAC.

The difference in OS observed in this study could be due to differences in tumor biology and/or responses to standard treatments in germline mutation carriers with PDAC. Differences in OS due to unique tumor biology would be suggestive of prognostic nature of HRR mutations while OS difference due to response to therapy, for example, platinum-based chemotherapy, would point toward predictive nature of HRR alterations. The prognostic versus predictive nature of HRR defect deserves special attention. Molecular studies involving sporadic PDAC have identified a complex mutational landscape with a high frequency of somatic mutations in KRAS, TP53, SMAD4, CDKN2A, and in genes involved in cell-cycle regulation, TGFβ signaling, HRR, chromatin regulation, and axonal guidance pathways (7, 8, 37–39). It is not entirely clear whether germline mutation–associated PDACs have similar landscapes of somatic mutations and mutational signatures as sporadic PDAC, but differences in tumor biology have been reported from smaller series. For example, significant enrichment for germline mutations in HRR genes in KRAS wild-type PDAC (8) and among patients with the mutational signature associated with high genomic instability (7) have been observed. While this study did not evaluate somatic mutations in PDAC, it does establish a foundation for future studies to evaluate if differences in tumor biology may explain the association with improved survival observed in this study.

Prior studies have demonstrated increased responsiveness to radiation and chemotherapy, particularly platinum-based chemotherapies, in PDAC associated with germline mutations (9, 14, 36, 40, 41). In addition, among patients with advanced PDAC treated with platinum-based chemotherapy, OS has been shown to be longer for patients with HRR defect (33, 42, 43), which is suggestive of a predictive effect of HRR mutation on OS with platinum-based chemotherapy. While we observed that mutation carriers with advanced (nonresectable or metastatic) PDAC who received platinum-based chemotherapy had better OS compared with mutation carriers who did not (Supplementary Fig. S1), we did not observe an association between OS and mutation carrier status among patients receiving platinum-based chemotherapy (Table 3). These findings highlight the importance of platinum-based chemotherapies in mutation carriers but do not confirm the predictive effect of HRR mutations on OS with platinum-based chemotherapy. Similarly, the association with improved OS in mutation carriers after platinum-based therapy became standard of care in May 2011 and in nonresectable/metastatic cases (in whom platinum-based chemotherapy is more likely to be administered) suggest, but do not confirm, that HRR defect may be predictive of differential response to platinum-based therapies. In contrast, the observation of no significant difference in OS between mutation carriers and noncarriers with specific chemotherapy regimens, such as FOLFIRINOX, argues against predictive effect of platinum-based chemotherapy. However, the lack of predictive effect of HRR mutation on OS with platinum-based chemotherapy in this study may be due to smaller numbers of patients in the analyzed subsets, and does not rule out this possibility.

Ultimately, this study is not adequately powered to delineate whether the improvement in OS in germline mutation carriers is due to higher response to chemotherapy or inherent differences in tumor biology or a combination of both. However, similar to other studies (41–43), this study suggests that platinum-based therapies are important in treatment of mutation carriers with advanced PDAC. The current National Comprehensive Cancer Network guidelines on PDAC supports the use of cisplatin in BRCA1/2 mutation carriers, but the use of these agents in PDAC associated with other HRR genes has not been well-studied. Our study suggests that the sensitizing effects of platinum agents may extend beyond BRCA1/2 to other HRR genes. Larger randomized studies are needed to identify whether mutation carriers in HRR genes other than BRCA1/2 demonstrate significant response to particular chemotherapy regimens. Biomarkers predictive of differential therapeutic response are also urgently needed. In addition, with the availability of PARP inhibitors, future clinical trials could include rational combinations of PARP inhibitors with chemotherapeutic agents, immunotherapy, and/or radiotherapy to leverage the germline mutation status to further improve OS (10).

This study also demonstrated significant difference in OS between ATM mutation carriers and noncarriers. This is in contrast to a prior study that demonstrated poor survival with tumoral loss of ATM by IHC in PDAC (44). The biological characteristics of PDAC tumors associated with germline ATM carriers are understudied. In a small study of 24 breast cancer tumors from germline ATM carriers, none of the tumors displayed high activity of mutational signature 3 associated with HRR defect (45). In addition, it has been noted that germline ATM mutations and TP53 somatic mutations may be epistatic (41, 45). Somatic mutation in TP53 is known to be associated with poor prognosis in multiple malignancies including PDAC (46–48). These findings raise the possibility that germline ATM associated PDAC might be a distinct entity compared with BRCA1/2-associated PDAC or sporadic PDAC. Further evaluation of tumor biology of ATM associated PDAC tumors through DNA sequencing and RNA expression studies are needed to fully understand the unique nature of these tumors and implications on OS. In contrast, this study did not observe a difference in OS for BRCA1 or BRCA2. These analyses at gene-level may be underpowered to detect a difference in OS. However, further evaluation of differences in tumor biology and response to therapy between ATM, BRCA1/2 and PALB2 carriers should also be evaluated in larger studies in the future.

In addition to the survival differences, this study also showed that germline mutation carriers with PDAC are more likely to be younger and present with metastatic disease at diagnosis. Germline mutation carriers of HRR genes are diagnosed with breast and ovarian cancers at an earlier age (49) and are also more likely to have metastatic rather than localized prostate cancer compared with the general population (50). Similar associations in PDAC, although suspected, had not been observed previously and are novel findings of this study.

Because this study was performed using a prospective registry from a single institution, it has several strengths compared with similar prior studies. However, a number of limitations exist. Because we only analyzed patients who consented to participate in the registries, a selection bias cannot be ruled out. We evaluated combined outcomes for eight clinically relevant HRR genes, but it is possible that not all of the HRR genes may have similar biological properties. In addition, we did not take into account somatic mutations in tumors, which can independently affect outcomes as patients with somatic HRR defect are known to respond to platinum based chemotherapy and PARP inhibitors (41, 51, 52). Furthermore, the lack of ethnic/racial diversity and incomplete information on treatment is an important limitation of this study. Finally, even though we evaluated a large number of patients, the total number of germline mutation carriers was relatively small, and we may not have had power to detect differences in subgroup analyses.

This study demonstrates that germline mutation carriers with PDAC have a longer OS compared with noncarriers after adjusting for age, sex, and stage at diagnosis. Further research into tumor biology and therapeutic response to chemotherapy in PDAC associated with germline mutations in HRR genes are needed to identify not only prognostic but also predictive biomarkers to develop personalized treatment options for this unique patient population.

S. Yadav reports grants from Conquer Cancer Foundation "Young Investigator Award" during the conduct of the study. P.M. Kasi reports other from Foundation Medicine (consultancy/advisory board), Natera (consultancy/advisory board), Taiho (consultancy/advisory board), Ipsen (consultancy/advisory board), and from AstraZeneca (travel grant) outside the submitted work. F.J. Couch reports personal fees from QIAGEN, AstraZeneca, and Ambry Genetics outside the submitted work. R.R. McWilliams reports grants from NCI during the conduct of the study; grants from GSK, BMS, and grants from Merck outside the submitted work; and personal fees from NewLink Genetics. No potential conflicts of interest were disclosed by the other authors.

S. Yadav: Formal analysis, writing-original draft, writing-review and editing. P. Kasi: Investigation, writing-review and editing. W. Bamlet: Data curation, formal analysis, writing-review and editing. T. Ho: Data curation, writing-review and editing. E. Polley: Conceptualization, data curation, methodology, writing-review and editing. C. Hu: Formal analysis, writing-review and editing. S. Hart: Formal analysis, methodology, writing-review and editing. K. Rabe: Data curation, formal analysis, writing-review and editing. N. Boddicker: Investigation, writing-review and editing. R. Gnanaolivu: Investigation, writing-review and editing. K. Lee: Investigation, writing-review and editing. T. Lindstrom: Investigation, writing-review and editing. G. Petersen: Supervision, writing-review and editing. F. Couch: Conceptualization, resources, formal analysis, supervision, investigation, methodology, writing-review and editing. R. McWilliams: Conceptualization, resources, data curation, funding acquisition, investigation, methodology, writing-review and editing.

This study was supported in part by NIH Specialized Program of Research Excellence in Pancreatic Cancer (CA102701), NIH Specialized Program of Research Excellence in Breast Cancer (CA116201), NIH grants CA116167, CA176785, CA192393, and CA225662, the Breast Cancer Research Foundation and the Conquer Cancer Foundation Young Investigator Award.

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.