Risk of Pancreatic Cancer Among Individuals With Pathogenic Variants in the ATM Gene

Key Points

Objective  To estimate the risk for pancreatic cancer among individuals with a pathogenic variant in the ATM gene.

Findings  This multicenter cohort study of 2227 members of 130 pancreatic cancer kindreds found that the risk of pancreatic cancer in individuals with a pathogenic variant in the ATM gene is 1.1% by age 50 years, 6.3% by age 70 years, and 9.5% by age 80 years.

Meaning  The findings of this cohort study suggest that individuals with a pathogenic variant in the ATM gene are at a high risk of pancreatic cancer and may benefit from enhanced early detection surveillance.

Importance  Pathogenic germline variants in the ATM gene have been associated with pancreatic cancer risk. Although genetic testing identifies these variants in approximately 1% to 3% of unselected patients with pancreatic cancer, the lifetime risk of pancreatic cancer among individuals with pathogenic ATM variants has not been well estimated.

Objective  To estimate age-specific penetrance of pancreatic cancer in individuals with a pathogenic variant in the ATM gene.

Design, Setting, and Participants  This was a multicenter cohort study of pancreatic cancer family registries in the US and Canada using pedigree data from 130 pancreatic cancer kindreds with a pathogenic germline ATM variant. Data analyses were performed from January 2020 to February 2021.

Main Outcomes and Measures  Observational age-specific risk of pancreatic cancer. Penetrance was estimated using modified segregation analysis.

Results  The study population of 130 families (123 [95%] White families) comprised 2227 family members (mean age [SD], 58 [22] years; 1096 [49%] women) with complete records (ie, including familial relationships, pancreatic cancer diagnosis, ATM status, proband status, and age), of which 155 individuals had positive results for an ATM pathogenic variant, 16 had a negative result, and the remainder did not have a test result. In these 130 families, 217 individuals had pancreatic cancer: 78 families had 1 such member; 34 families had 2 such members; and 18 families had 3 or more members with pancreatic cancer. The average (range) age at diagnosis was 64 (31-98) years. The cumulative risk of pancreatic cancer among individuals with a germline pathogenic ATM variant was estimated to be 1.1% (95% CI, 0.8%-1.3%) by age 50 years; 6.3% (95% CI, 3.9%-8.7%) by age 70 years; and 9.5% (95% CI, 5.0%-14.0%) by age 80 years. Overall, the relative risk of pancreatic cancer was 6.5 (95% CI, 4.5-9.5) in ATM variant carriers compared with noncarriers.

Conclusions and Relevance  This multicenter cohort study found that individuals with a germline pathogenic ATM variant were at an increased lifetime risk of pancreatic cancer. These risk estimates can help guide decision-making when evaluating the risks and benefits of enhanced early detection surveillance.

Pancreatic ductal adenocarcinoma is among the most lethal cancer types. In 2021, there will be an estimated 60 430 new pancreatic cancers diagnosed and 48 220 related deaths in the US.¹ Despite the near doubling of pancreatic cancer survival rates during the past decade, the 5-year survival of approximately 10% remains the lowest of any major tumor type.¹ Survival rates are higher for patients with early-stage disease,^1,2 highlighting the potential benefit of earlier detection.

Recently, germline genetic testing for patients with pancreatic cancer and their first-degree relatives has become the recommended standard of care³ because studies have demonstrated that pathogenic variants in the ATM, BRCA1, BRCA2, CDKN2A, PALB2, PRSS1 STK11, TP53, and Lynch syndrome collectively occur in 4% to 10% of patients.^4-7 While studies of the estimated risk of pancreatic cancer in carriers of a pathogenic variant in these genes are limited, BRCA2 carriers are estimated to have a 2- to 6-fold increased risk of developing pancreatic cancer, BRCA1 carriers a 1- to 2.5-fold increased risk, and CDKN2A carriers a higher 12-fold increased risk. Estimated risk for these and other genes is presented in eTable 1 in the Supplement.^6,8-11 Roberts and colleagues¹² first reported the association between pathogenic ATM variants and pancreatic cancer, observing pathogenic variants in 3% of patients with familial pancreatic cancer. Subsequent studies¹³ have replicated these findings and demonstrated that as many as 2.3% of patients with pancreatic cancer unselected for family history harbor germline pathogenic ATM variants with an odds ratio of 5.71 (95% CI, 4.38-7.33) in patients with pancreatic cancer compared with the control group of patients of European ancestry.⁶

The ATM gene product (OMIM 607585) is a serine/threonine kinase that plays essential roles in DNA double-strand break repair, cell-cycle control, and apoptosis. Identification of a germline ATM pathogenic variant in a patient with pancreatic cancer may have implications for future treatment decisions.^14,15 Furthermore, carriers of an ATM pathogenic variant without cancer should consider the risks and benefits of cancer screening. The risk of breast cancer among ATM pathogenic variant carriers has been determined to be moderately elevated.¹⁶ However, the lifetime risk of pancreatic cancer in carriers has not been well defined. Individuals with an inherited predisposition to pancreatic cancer, including those with a pathogenic ATM variant, may benefit from surveillance.¹⁷ Better risk estimates can help guide decisions regarding the role of pancreas surveillance in defined high-risk individuals. To this end, we sought to estimate age-specific penetrance of pancreatic cancer in kindreds with pathogenic ATM variants.

The study was reviewed and approved by the institutional review board of the Johns Hopkins Medical Institution. Written informed consent was obtained by each recruiting registry and only deidentified data were provided for the study analyses according to the Regulations for the Protection of Human Subjects (45 CFR §46). The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Included kindreds were enrolled in any of the family registries that participate in the Pancreatic Cancer Genetic Epidemiology Consortium: the National Familial Pancreas Tumor Registry (Johns Hopkins University); the Biospecimen Resource for Pancreas Research (Mayo Clinic); GI Cancer Genetics (Dana-Farber Cancer Institute); and the Ontario Pancreas Cancer Study (Mount Sinai Hospital, University of Toronto). Registry eligibility required that families have a least 1 member with pancreatic cancer. Families were eligible for inclusion if at least 1 family member was diagnosed with pancreatic cancer and at least 1 individual in the pedigree had a germline ATM pathogenic variant. Pathogenic ATM variants were identified in these individuals either through clinical sequencing or previously published research-based sequencing studies.^6,13,18,19 Information on familial relationships, pancreatic cancer diagnosis, ATM status (pathogenic variant carrier, noncarrier, or untyped), proband status, and age (at diagnosis of pancreatic cancer or at last contact or death if not diagnosed with pancreatic cancer) were obtained for pedigree members. Of the 130 families included in this study, 123 reported having European ancestry.

Using the modified segregation analysis approach available in MENDEL, version 16.0,²⁰ we estimated the cumulative risk of pancreatic cancer by age dependent on germline ATM variant status. In brief, using this approach, penetrance is estimated by the maximum likelihood of the observed genetic data given the phenotype of the observed pedigree members (age and affection status). The genetic model includes the mode of inheritance, disease allele frequency, and penetrance function. We assumed dominant inheritance and penetrance followed a proportional hazards model where the age-specific penetrance (cumulative risk, F[t]) was calculated by

cumulative incidence: Λ (t) = λ_0 (t) × e^ (β [t] g_i),

where g_i is an indicator for ATM carrier status. A population pathogenic ATM variant frequency of 0.2% was assumed (between estimate of 0.35% and 0.1% in prior studies^6,12) and sensitivity analysis was conducted. The baseline cumulative probability of pancreatic cancer was based on population cumulative probabilities from the Surveillance, Epidemiology, and End Results Program (US National Cancer Institute)²¹ for 2000 to 2017. Pedigree likelihoods were obtained using either the Elston-Stewart²² or Lander-Green algorithms depending on computational efficiency.²³ To account for the selection of families through a patient with pancreatic cancer and a known pathogenic ATM variant, analysis was conditioned on either a single proband (a patient with pancreatic cancer was initially found to have the ATM variant) or multiple probands (a family member without pancreatic cancer with an ATM pathogenic variant and their closest relative with pancreatic cancer was untyped for ATM). Age to pancreatic cancer was estimated using proportional hazards model, where the ATM variant carrier status was the predictor. We tested the hypothesis that penetrance was not dependent on ATM carrier status. Individuals without pancreatic cancer were censored at their age at last follow-up or age at death if deceased. Parameters were estimated using maximum-likelihood methods, such that the penetrance estimates that maximized the pedigree likelihoods were obtained, and variances were obtained from the observed information matrix.

Statistical tests were 2-tailed and P values < .05 were considered statistically significant. Data analyses were performed from January 2020 to February 2021 using MENDEL, version 16.0,²⁰ and R, version 4.0 (The R Foundation for Statistical Computing).

The study population of 130 families (123 [95%] White families) comprised 2227 family members (mean age [SD], 58 [22] years; 1096 [49%] women) with complete records (ie, including familial relationships, pancreatic cancer diagnosis, ATM status [pathogenic variant carrier, noncarrier, or untyped], proband status, and age; Table 1). Of these, 155 individuals had a test result that was positive for an ATM pathogenic variant, 16 had a negative result, and the remaining individuals did not have a test result. Because of the small sample size, race and ethnicity were categorized as White or non–White according to family ancestry. A higher prevalence of pancreatic cancer was observed in noncarriers compared with untyped individuals, but this was because of preferential genetic testing of patients with cancer over cancer-free individuals. Additional pedigree members with missing pancreatic cancer status and/or age were included to complete the family structure, but this had a limited influence on the risk estimates. In these 130 families, there were a total of 217 individuals with pancreatic cancer: 78 families with 1 family member with pancreatic cancer, 34 families with 2 members, and 18 families with 3 or more members. The average age (range) at pancreatic cancer diagnosis was 64 (31-98) years. Additional demographic information and pedigree details are provided in eTable 2 in the Supplement.

The cumulative risk of pancreatic cancer in pathogenic ATM variant carriers and noncarriers is shown in the Figure and Table 2, and age and ATM variant status were associated with pancreatic cancer risk. Cumulative risk of pancreatic cancer in ATM variant carriers increased from 0.08% (95% CI, 0.07%-0.09%) by age 30 years to 9.5% (95% CI, 5.0%-14.0%) by age 80 years. To determine if there was evidence of penetrance differences between men and women, we allowed for sex-specific penetrance; however, it was not significant in the study model (β = 0.12; SE, 0.11; P = .29). Sensitivity analyses assuming a 10-fold increase and 5-fold decrease in ATM pathogenic variant frequency demonstrated consistent estimates of penetrance (data not shown).

This study presents what is, to our knowledge, the first quantification of the age-specific risk of pancreatic cancer. These models, which included proband correction, demonstrated the high lifetime risk of pancreatic cancer in ATM variant carriers. The risk was estimated to be 6.3% by 70 years of age and 9.5% by 80 years, findings that inform the potential benefit of early detection in these kindreds. Risk prior to age 50 years, even in this high-risk population, remained low (approximately 1.0%), supporting the current consensus that pancreatic cancer screening should generally begin at age 50 years among individuals with a germline pathogenic variant and a family history of pancreatic cancer.¹⁷ While there is high agreement that individuals with pathogenic variants in BRCA2 should be considered for early detection screening trials, there is less agreement for inclusions of ATM variant carriers in surveillance protocols because some previous studies^6,11,24 have suggested that the risk of pancreatic cancer in ATM variant carriers is lower. The relative risk estimate of 6.5 (95% CI, 4.5-9.5) in ATM variant carriers is substantial. Compared with studies that have estimated a 2.2- to 6.6-fold increased risk of pancreatic cancer in BRCA2 variant carriers,^6,11,24 the risk estimates of the present study strongly support inclusion of ATM variant carriers in screening trials.

Penetrance analysis in high-risk families typically yields higher-risk estimates when compared with alternative methods, even when proband corrections are implemented. However, this study’s risk estimate of an approximately 6.5-fold (95% CI, 4.5-9.5) increased risk of pancreatic cancer is consistent with estimates of prior studies⁶ that demonstrated a 5.7-fold (95% CI, 4.4-7.3) increased prevalence of ATM pathogenic variants in more than 3000 patients with pancreatic cancer unselected for family history vs general population controls.⁶ This concordance suggests that the present study’s penetrance estimates are not greatly inflated because of incomplete adjustment for ascertainment.

The study population was predominantly of European ancestry; thus, caution is needed when generalizing these results to individuals of different ancestry. Future studies to better understand the risk of germline ATM variants in non–European populations are needed and are underway.

This was multicenter cohort study found that the cumulative risk of pancreatic cancer in ATM pathogenic variant carriers was 9.5% by age 80 years. These findings underscore the need to develop appropriate surveillance and intervention strategies for these individuals at high-risk of developing pancreatic cancer.

Accepted for Publication: June 18, 2021.

Published Online: September 16, 2021. doi:10.1001/jamaoncol.2021.3701

Corresponding Author: Alison P. Klein, PhD, MHS, Department of Oncology, Pathology, and Medicine, Sidney Kimmel Comprehensive Cancer Center, 1550 Orleans St, Room 354, Baltimore, MD 21231 (aklein1@jhmi.edu).

Author Contributions: Dr Klein had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Hruban, Syngal, Petersen, Klein.

Acquisition, analysis, or interpretation of data: Hsu, Roberts, Childs, Porter, Rabe, Borgida, Ukaegbu, Goggins, Zogopoulos, Syngal, Gallinger, Petersen, Klein.

Drafting of the manuscript: Hsu, Klein.

Critical revision of the manuscript for important intellectual content: Roberts, Childs, Porter, Rabe, Borgida, Ukaegbu, Goggins, Hruban, Zogopoulos, Syngal, Gallinger, Petersen, Klein.

Statistical analysis: Hsu, Childs, Rabe, Klein.

Obtained funding: Gallinger, Petersen, Klein.

Administrative, technical, or material support: Porter, Ukaegbu, Goggins, Syngal, Gallinger, Petersen, Klein.

Supervision: Zogopoulos, Syngal, Gallinger, Klein.

Conflict of Interest Disclosures: Dr Hruban reported royalty distributions from a patent for GNAS, the technology used in this study, which is licensed under an agreement between Thrive Earlier Detection (a subsidiary of Exact Sciences Corporation) and Johns Hopkins University. Dr Syngal reported a patent for PREMM model licensed to Myriad Genetics during the conduct of the study. Dr Klein reported consulting fees from Merck outside the submitted work. No other disclosures were reported.

Funding/Support: This research was supported by grants from the US National Cancer Institute: P50CA62924 (Klein), P30CA006973 (Nelson), R01CA132829 (Syngal), R00CA190889 (Roberts), and R01CA097075 and P50CA102701 (Petersen); the Bowen-Chapman Family Research Fund (Syngal); Fonds de Recherche du Québec (Zogopoulos); the Lustgarten Foundation; the Rolfe Pancreatic Cancer Foundation (Roberts and Petersen); the Sol Goldman Pancreatic Cancer Research Center; and the Troper Wojcicki Foundation.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.