Deleterious Germline Mutations in Patients With Apparently Sporadic Pancreatic Adenocarcinoma

Abstract

Purpose

Deleterious germline mutations contribute to pancreatic cancer susceptibility and are well documented in families in which multiple members have had pancreatic cancer.

Methods

To define the prevalence of these germline mutations in patients with apparently sporadic pancreatic cancer, we sequenced 32 genes, including known pancreatic cancer susceptibility genes, in DNA prepared from normal tissue obtained from 854 patients with pancreatic ductal adenocarcinoma, 288 patients with other pancreatic and periampullary neoplasms, and 51 patients with non-neoplastic diseases who underwent pancreatic resection at Johns Hopkins Hospital between 2000 and 2015.

Results

Thirty-three (3.9%; 95% CI, 3.0% to 5.8%) of 854 patients with pancreatic cancer had a deleterious germline mutation, 31 (3.5%) of which affected known familial pancreatic cancer susceptibility genes: BRCA2 (12 patients), ATM (10 patients), BRCA1 (3 patients), PALB2 (2 patients), MLH1 (2 patients), CDKN2A (1 patient), and TP53 (1 patient). Patients with these germline mutations were younger than those without (mean ± SD, 60.8 ± 10.6 v 65.1 ± 10.5 years; P = .03). Deleterious germline mutations were also found in BUB1B (1) and BUB3 (1). Only three of these 33 patients had reported a family history of pancreatic cancer, and most did not have a cancer family history to suggest an inherited cancer syndrome. Five (1.7%) of 288 patients with other periampullary neoplasms also had a deleterious germline mutation.

Conclusion

Germline mutations in pancreatic cancer susceptibility genes are commonly identified in patients with pancreatic cancer without a significant family history of cancer. These deleterious pancreatic cancer susceptibility gene mutations, some of which are therapeutically targetable, will be missed if current family history guidelines are the main criteria used to determine the appropriateness of gene testing.

INTRODUCTION

Pancreatic cancer is expected to be the second leading cause of cancer death in the United States by the year 2030.¹ Inherited gene mutations are known to contribute to pancreatic cancer in patients with familial pancreatic cancer (defined by the presence of two first-degree relatives with the disease),² but the extent to which deleterious gene mutations contribute to pancreatic cancer risk in individuals without a family history of pancreatic cancer is not well defined. Identifying inherited susceptibility gene mutations in an individual improves assessment and decisions regarding cancer screening for family members and can guide treatment of patients with pancreatic cancer.

The established familial pancreatic cancer susceptibility genes include the BRCA2, ATM, PALB2, CDKN2A, PRSS1, STK11, MLH1, and MSH2³ genes.^4-8 The results of whole-genome sequencing of more than 600 individuals with familial pancreatic cancer were recently reported, with analysis focused on the role of low-frequency truncating mutations.² Deleterious germline mutations in the BRCA2 gene account for the biggest fraction of known familial pancreatic cancer genes (found in approximately 5% to 10% of familial pancreatic cancer families^7,9-12), followed by ATM (deleterious mutations found in approximately 2% to 3%).^2,4 Deleterious germline mutations involving other genes are less common (each found in approximately ≤ 1% of affected individuals from familial pancreatic cancer kindred). These genes include CDKN2A (deleterious mutations cause familial atypical melanoma mole syndrome),^13-17 PALB2 (DNA mismatch–repair genes that cause Lynch syndrome),³ STK11 (Peutz-Jeghers syndrome), and PRSS1 (hereditary recurrent acute pancreatitis).^8,18-21 Germline BRCA1 mutations increase the overall risk of developing pancreatic cancer by approximately two- to -four-fold.^2,7,22,23 The role of other genes in pancreatic cancer susceptibility is still being evaluated.

The prevalence of germline mutations in individual pancreatic cancer susceptibility genes in patients with apparently sporadic forms of the disease (ie, without a family history of pancreatic cancer) has also been studied.^24-27 These studies have primarily focused on BRCA genes. For example, germline BRCA2 mutations are found in a small percentage of patients with apparently sporadic pancreatic cancer,²⁸ with a higher prevalence found in populations with many individuals of Ashkenazi Jewish heritage because of the common 6174delT BRCA2 founder mutation in that population.^27,29-33 In one study, germline BRCA mutations were identified in 4.6% of an unselected series of 306 patients with pancreatic cancer from a single center.²⁴

The absence of a significant family history in patients with an established deleterious germline mutation is probably primarily as the result of incomplete penetrance, rather than de novo mutation in the germline. For example, the average lifetime risk of developing pancreatic cancer among BRCA2 gene mutation carriers is estimated to be approximately 5% to 10%.^34-36 Notably, these estimates have been determined primarily in families ascertained for breast and/or ovarian cancer and therefore may be an underestimate.

There is considerable potential clinical utility to identifying a germline susceptibility gene in a patient with pancreatic cancer. Mutation carriers with pancreatic cancer may have more options for personalized medicine directed against the genetic drivers of their cancer,³⁷ and their family members may benefit from cancer screening and cancer prevention strategies for pancreatic and extrapancreatic cancers.^38-42 Relatives of patients with apparently sporadic pancreatic cancer are at increased risk of mortality from other cancers.⁴³

In this study, we determined the prevalence of germline mutations in known and candidate pancreatic cancer susceptibility genes in a large hospital-based series of patients unselected for their family history of pancreatic cancer. We compared the prevalence of deleterious mutations in these patients with the prevalence in patients who underwent surgery for other periampullary cancers and diseases.

METHODS

Patients and Specimens

This study included 854 patients with pancreatic ductal adenocarcinoma who were evaluated and treated at the Johns Hopkins Hospital between 2000 and 2015. Patients were enrolled in the study during their preoperative evaluation or during their multidisciplinary clinic visit. Personal and family history information was obtained from the medical record and from the National Familial Pancreas Tumor Registry. To estimate the prevalence of deleterious germline mutations in patients with other periampullary/pancreatic diseases referred to the same clinical services, we included 339 patients who had undergone pancreatic resection for periampullary/biliary pathology other than pancreatic cancer, including 108 with other cancers (duodenal, biliary, gall bladder), 113 with other neoplasms (pancreatic neuroendocrine tumors, GI stromal tumors, carcinoid), 25 with precancerous neoplasms (duodenal, ampullary adenoma), 25 with serous cystadenoma, and 51 with non-neoplastic conditions including 37 with pancreatitis (Table 1); patient demographic data are listed in Table 2.

Table 1.

Pathology and Indications for Pancreatic Resection

Table 2.

Patient Characteristics

All elements of this study were approved by the Johns Hopkins Institutional Review Board, and written informed consent was obtained from all patients.

DNA Extraction

Genomic DNA was extracted from either frozen normal tissue from pancreatic resection specimens (duodenum, spleen, or pancreas) or peripheral blood mononuclear cells using QIAamp DNA Micro Kit (QIAGEN, Hilden, Germany), according to the manufacturer’s instructions. DNA samples were quantified using Quantifiler (Thermo Fisher Scientific, Waltham, MA).

DNA Sequencing

Thirty-two genes (Appendix Table A1 [online only]) were sequenced using an AmpliSeq Custom Panel. Next-generation sequencing was performed using the Ion Proton system (Life Technologies [Life-Tech], Carlsbad, CA), according to the manufacturer’s protocols and as previously described.⁴⁴ These genes were either known pancreatic cancer susceptibility genes (BRCA2, ATM, PALB2, BRCA1, CDKN2A, MLH1, MSH2, PRSS1, STK11, and TP53), known cancer susceptibility genes (MSH6, PMS2, CDH1, RAD51C, RAD51D, BUB1B, and FANCJ), or candidate pancreatic cancer susceptibility genes (FANCA, FANCC, FANCG, FANCL, ARID1A, RECQL4, XRCC2, XRCC3, ERCC4, TERT, BAP1, BUB1, BUB3, and RNF43). Thus, 20 ng of DNA (10 ng per primer pool) were used for AmpliSeq polymerase chain reaction. After FuPa digestion, P1 adaptor/Xpress barcode ligation, and library clean up (Agencourt AMPure XP Reagent; Beckman Coulter, Brea, CA), libraries were eluted into low Tris-EDTA and quantified (Ion Quantitation Kit; Life-Tech). Libraries underwent emulsion polymerase chain reaction in an Ion OneTouch2 (Life-Tech) for 5 hours; Ion Sphere Particles were then cleaned and enriched in the OneTouch ES (Life-Tech). Enriched Ion Sphere Particles were loaded into P1v3 chips for sequencing (Ion Proton; Life-Tech). The postsequencing raw FASTQ files were launched in NextGENe (version 2.41; SoftGenetics, Chicago, IL) software for alignment to the hg19 human reference genome and single-nucleotide variant calling. Alignments were visually verified using Integrative Genomics Viewer (version 2.3; Broad Institute, Cambridge, MA) and NextGENe Viewer. The functional significance of variants was determined by interrogating ClinVar and PubMed. Variants of unknown significance are listed in Appendix Table A2 (online only). Variants identified as truncating (nonsense, frameshift, splice intervening sequence ± 1 or 2) and deleterious missense variants were validated by Sanger sequencing performed at The Johns Hopkins Synthesis & Sequencing Facility (Applied Biosystems DNA sequencers; Sanger validation primer sequences are listed in Appendix Table A3 [online only]). Deleterious variants were identified in several candidate pancreatic cancer susceptibility genes. To further evaluate these variants, primary pancreatic cancer tissue from three patients was laser-capture microdissected⁴⁴ and sequenced to evaluate for biallelic inactivation (tumor was not available from the patients with the RAD51D or BUB1B mutation). For genes with deleterious truncating variants (Table 3), we provide the truncating variant data in control subjects from the ExAC database (Broad Institute; Table 4).

Table 3.

Deleterious Mutations in Pancreatic Cancer Susceptibility Genes

Table 4.

Prevalence of Truncating Variants in Pancreatic Cancer Susceptibility Genes: ExAC Controls Versus PDAC Patient Cases

Statistical Analysis

The mean age of carriers of a deleterious germline variant was compared with noncarriers using the t test. SPSS software was used (version 22; IBM, Armonk, NY). A two-tailed P < .05 was considered statistically significant.

RESULTS

Characteristics of the study population are listed in Table 1 and Table 2. Thirty-three (3.9%) of the 854 patients with pancreatic ductal adenocarcinoma had an identifiable deleterious germline mutation, of which 28 were truncating (Table 3). Thirty-one (3.5%) of these patients had a deleterious mutation in a known pancreatic cancer susceptibility gene. This included 12 patients with deleterious germline BRCA2 mutations, five of whom carried the Ashkenazi Jewish founder BRCA2 mutation (6174delT); ten with ATM; three with BRCA1; two with PALB2; two with MLH1; and one each with a CDKN2A and TP53 mutation. Two patients had a mutation in a candidate pancreatic cancer susceptibility gene (BUB1B and BUB3). Patients with deleterious BUB1B mutations are prone to aneuploidy, chromosomal alterations, and cancer,⁴⁵ and BUB1B has been identified as a candidate pancreatic cancer susceptibility gene.² Evidence regarding the role of BUB3 as a cancer susceptibility gene is less clear^46,47; both BUB1B and BUB3 regulate mitotic checkpoints and, in mouse models, BUB3 heterozygotes are similarly prone to premature aging and chromosomal instability.⁴⁸ We sequenced the pancreatic cancer DNA from the patient with the deleterious germline BUB3 variant but did not find evidence of biallelic inactivation.

In addition to the 33 patients with a deleterious mutation in a known or suspected pancreatic cancer susceptibility gene, three patients carried a deleterious mutation in another cancer susceptibility gene (one each involving CDH1, RAD51D, and RAD51B). The significance of these variants for pancreatic cancer susceptibility is not certain (they are listed in Appendix Table A2). RAD51D and RAD51C are ovarian cancer susceptibility genes.^49,50 Common variants in RAD51B are associated with increased breast cancer risk.⁵¹ Germline CDH1 mutations predispose to hereditary gastric cancer and lobular breast cancer. The CDH1 P373L mutation identified in one patient has been described in a hereditary gastric cancer family.⁵² We sequenced microdissected pancreatic cancer DNA from the patients with the germline RAD51B and CDH1 mutation, but we did not find evidence of biallelic inactivation of their germline mutated gene in their pancreatic cancer. Overall, there is not sufficient evidence to indicate that RAD51B and CDH1 germline mutations contribute to pancreatic cancer development.

Numerous variants of unknown significance were also identified (listed in Appendix Table A2). In addition to the 33 patients with deleterious mutations, two patients with pancreatic ductal adenocarcinoma carried the BRCA2 polymorphic stop variant p.K3326X. This variant was initially not thought to confer a cancer risk, but is now considered a modifier allele with evidence that carriers have a small increased risk of breast, pancreatic, and other cancers.^53-55

The majority (82%) of patients with deleterious germline mutations had a family history of other cancers, but only five mutation carriers (15%) had cancer family histories that suggested a familial cancer syndrome (Table 3). Eighteen of the 33 mutation carriers had a family history of breast cancer reported in a first- or second-degree relative, six had a family history of prostate cancer, and three had a family history of ovarian cancer. Only three (9%) of the 33 individuals with a germline deleterious mutation had a family history of pancreatic cancer. In comparison, among the 818 patients with pancreatic ductal adenocarcinoma without a germline mutation, 117 (14.3%) had a family history of pancreatic cancer identified; for 86 of these patients, it was in a first-degree relative.

Although there are no dedicated pancreatic cancer National Comprehensive Cancer Network (NCCN) guidelines for gene testing, the breast/ovarian cancer NCCN guidelines for gene testing⁵⁶ include recommendations for when to consider gene testing patients with pancreatic cancer. Candidates for gene testing include individuals 1with a close relative with pancreatic cancer, 2who are of Ashkenazi Jewish descent with pancreatic cancer, and 3with a close blood relative with ovarian cancer or young-onset breast cancer. On the basis of these criteria, five pancreatic cancer cases with the Ashkenazi BRCA2 founder mutation would be eligible for gene testing, as would nine others with a significant cancer family history (first-degree relative with pancreatic cancer, a close blood relative with ovarian cancer or young-onset breast cancer, or multiple close relatives with breast cancer).

The average age ± SD at diagnosis of patients with pancreatic ductal adenocarcinoma identified as having a germline mutation in a known (BRCA2, ATM, CDKN2A, PALB2, MLH1, BRCA1, and TP53) pancreatic cancer susceptibility gene was 60.8 ± 10.6 years, significantly lower than the average age of the patients without an identifiable susceptibility gene mutation (65.1 ± 10.1 years; P = .03).

A significantly smaller percentage (five of 339 patients [1.5%]; P = .02) of patients with diagnoses other than pancreatic ductal adenocarcinoma had an identifiable deleterious germline mutation. All of the nonpancreatic cancer cases with a deleterious germline mutation had another malignancy (five of 238 patients [2.1%]). These included two of the 47 patients with cholangiocarcinoma (one with an ATM mutation and another with a BRIP1 [FANCJ] mutation), one patient who had a renal cell carcinoma with pancreatic metastasis (ATM mutation), one patient with a duodenal GI stromal tumor (RAD51C mutation), and one patient who underwent pancreaticoduodenectomy for a recurrent bleeding duodenal ulcer after radiation therapy for lymphoma (BRCA2 mutation). The significance of the BRIP1 mutation is not clear. Evidence from large studies indicates that carriers of germline BRIP1 mutations are at moderately increased risk of developing ovarian cancer but not breast cancer.^57,58

No deleterious mutations were identified in any of the other disease control cases. Pancreatic cancer cases were more likely than disease controls without a malignant neoplasm to have a deleterious mutation (P = .035). We also compared the prevalence of truncating deleterious variants for each pancreatic cancer susceptibility gene in our pancreatic cancer cases with their prevalence in controls in the ExAC database. Germline truncating mutations involving BRCA2 and ATM were significantly more common in pancreatic cancer cases than in ExAC controls (Table 4).

DISCUSSION

We found a significant yield of deleterious germline mutations in pancreatic cancer susceptibility genes in patients with pancreatic cancer without a pancreatic cancer family history. These patients with apparently sporadic pancreatic cancer also often do not have an extensive family history of pancreatic or other cancers that would trigger consideration for germline gene testing. The prevalence of deleterious germline mutations in patients with apparently sporadic pancreatic cancer is approximately half of that reported to date in patients with familial forms of pancreatic cancer.^7,27

Family history remains one of the best predictors of future pancreatic cancer risk.^59-61 For common cancers such as breast/ovarian and colorectal cancer (where risk assessment, genetic counseling, and gene testing are well established), current guidelines recommend using family history to risk stratify family members to identify those individuals most likely to benefit from gene testing.⁶² In contrast, our results for pancreatic cancer highlight the limitations of relying solely on current NCCN family history guidelines to determine which patients are most likely to carry a deleterious pancreatic cancer susceptibility gene. Indeed, most deleterious germline mutations found in patients with pancreatic cancer are in those who do not meet the familial criteria for gene testing.

Although the evidence is mostly anecdotal in the context of pancreatic cancer, not only can affected relatives have the opportunity to undertake cancer screening and prevention strategies, but it also would have an impact on the patient’s treatment. Identifying BRCA mutations in patients with pancreatic cancer would provide the opportunity to have personalized therapy with poly (ADP-ribose) polymerase inhibitors or platinums,⁶³ and Lynch syndrome carriers with pancreatic cancer would have the potential to benefit from immunotherapy.⁶⁴ It is suspected that patients with ATM germline mutations would benefit from radiotherapy to control local disease (assuming that the gene is biallelically inactivated in their cancer)⁶⁵ and may also be more sensitive to certain chemotherapeutics.

Although a large gene panel was evaluated for this study, our results indicate that gene testing patients with pancreatic cancer should be limited to established pancreatic cancer susceptibility genes (BRCA2, ATM, PALB2, CDKN2A, and BRCA1, and mismatch-repair genes), with testing for PRSS1 and STK11 mutations for families suspected of having the corresponding clinical syndromes. Although using large gene panels to identify cancer susceptibility genes as part of research studies can be informative, early experience with using large gene panels to perform germline gene testing in clinical settings has shown that such panels pose challenges.⁶⁶ Because of the greater chance of identifying variants of unknown significance, many patients receive inconclusive results, and the potential for misunderstanding and anxiety is significant.

The potential to benefit the few individuals with actionable gene mutations would seem to justify the effort to routinely offer gene testing to all patients with pancreatic ductal adenocarcinoma to identify such cases.^39,67 However, offering widespread genetic testing for patients with pancreatic cancer has significant challenges, not the least being that patients should undergo genetic counseling before and after such testing to provide understanding and reassurance and to avoid harm.⁶⁸ Unfortunately, there are not enough genetic counselors to provide this service. This shortage of genetic counselors applies to other cancers and, as a result, most patients, even those who undergo gene testing for BRCA mutations, do so without genetic counseling.⁶⁹ The demand for BRCA gene testing has led to alternative approaches. One approach taken in some centers that are sequencing cancer samples is to provide an opt-out option for patients who do not want to know about germline information. Another approach in situations where there is considerable demand, such as for patients with ovarian cancer, is to provide genetic counseling where clinicians (including nurses trained in genetic counseling) provide the service.⁷⁰ Because of the potential for adverse events when gene testing is performed without adequate genetic counseling, most experts recommend that appropriate counseling⁷¹ and testing should only be undertaken by those with the expertise.⁷² Genetic counseling for patients with pancreatic cancer poses additional challenges. Because pancreatic cancer often progresses rapidly, patients can greatly benefit from optimal selection of their first-line therapy; thus, ideally, counseling and testing would be incorporated into routine patient care so that it can be performed rapidly. Relatives of patients with pancreatic cancer who are also mutation carriers can also benefit from gene testing. Here, the need for genetic counseling is perhaps more important, because the lifetime estimates of developing pancreatic and other cancers for carriers of deleterious mutations need to be better defined, and the benefits of pancreatic screening are still being established.^38,41 Because cancer genetics risk assessment is not a routine component of pancreatic cancer care, it would be valuable to undertake studies to determine the benefits and challenges of incorporating risk assessment and gene testing into routine pancreatic cancer practice. Additional studies are also needed to determine how cancer family history and other risk factor information can help refine pancreatic cancer risk in mutation carriers. Furthermore, the pancreatic cancer risk associated with mutations in some pancreatic cancer susceptibility genes (such as ATM and PALB2) is not well defined, and the effectiveness of screening these mutation carriers is not established.

There are some limitations to our study. First, this was a retrospective study where we relied on self-reported family history of pancreatic and other cancers obtained from the medical record, and patient reporting of their familial cancer history is often incomplete. We were also not able to determine if detecting these mutations resulted in clinical benefit to the patients or their families.

In summary, we found that there is a significant yield of deleterious germline mutations in pancreatic cancer susceptibility genes in unselected patients with apparently sporadic pancreatic cancer. Routine gene testing of patients with newly diagnosed pancreatic cancer and their families may yield significant clinical benefits.

Appendix

Table A1.

Targeted Regions of the 32 Genes*

Table A2.

Variants of Uncertain Significance

Table A3.

Sanger Sequencing Validation PrimersGene

AUTHOR CONTRIBUTIONS

Conception and design: Jun Yu, Alison P. Klein, Michael Goggins

Collection and assembly of data: Koji Shindo, Jun Yu, Shahriar Fesharakizadeh, Christy Cho, Anne Macgregor-Das, Abdulrehman Siddiqui, P. Dane Witmer, Koji Tamura, Tae Jun Song, Jose Alejandro Navarro Almario, Aaron Brant, Michael Borges, Madeline Ford, Thomas Barkley, Jin He, Matthew J. Weiss, Christopher L. Wolfgang, Ralph H. Hruban, Alison P. Klein, Michael Goggins

Data analysis and interpretation: Koji Shindo, Jun Yu, Masaya Suenaga, Anne Macgregor-Das, Koji Tamura, Nicholas J. Roberts, Alison P. Klein, Michael Goggins

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS’ DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Deleterious Germline Mutations in Patients With Apparently Sporadic Pancreatic Adenocarcinoma

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO’s conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/site/ifc.

Koji Shindo

No relationship to disclose

Jun Yu

No relationship to disclose

Masaya Suenaga

No relationship to disclose

Shahriar Fesharakizadeh

No relationship to disclose

Christy Cho

No relationship to disclose

Anne Macgregor-Das

No relationship to disclose

Abdulrehman Siddiqui

Stock or Other Ownership: 22nd Century Group

P. Dane Witmer

No relationship to disclose

Koji Tamura

No relationship to disclose

Tae Jun Song

No relationship to disclose

Jose Alejandro Navarro Almario

No relationship to disclose

Aaron Brant

No relationship to disclose

Michael Borges

No relationship to disclose

Madeline Ford

No relationship to disclose

Thomas Barkley

No relationship to disclose

Jin He

No relationship to disclose

Matthew J. Weiss

No relationship to disclose

Christopher L. Wolfgang

No relationship to disclose

Nicholas J. Roberts

No relationship to disclose

Ralph H. Hruban

Leadership: miDIAGNOSTICS

Patents, Royalties, Other Intellectual Property: Myriad Genetics

Alison P. Klein

Patents, Royalties, Other Intellectual Property: Myriad Genetics

Michael Goggins

Patents, Royalties, Other Intellectual Property: Myriad Genetics