Cancer Risks Associated With BRCA1 and BRCA2 Pathogenic Variants

1It is well established that pathogenic variants (PVs) in BRCA1 and BRCA2 (BRCA1/2) are associated with increased risks of breast and ovarian cancers in women for which reliable risk estimates are available. Accumulated evidence indicates that BRCA1/2 PVs are also associated with pancreatic cancer- and male breast cancer risks,,- and that BRCA2 PVs are associated with prostate cancer risk, particularly aggressive prostate cancer, whereas the association between BRCA1 PVs and prostate cancer risk is still debated.,,,,- Associations with risks for other cancers have also been suggested, including colorectal, liver, and stomach cancers for BRCA1/2 PVs; cervical, corpus uteri, kidney, and testis cancers for BRCA1 PVs;,,,,, and bone, brain, blood, and gallbladder cancers and malignant melanoma for BRCA2 PVs.,,,, However, these associations are based on studies with relatively small sample sizes, resulting in imprecise cancer risk estimates.

2Key Objective The associations of pathogenic variants (PVs) in BRCA1 and BRCA2 with cancers other than female breast and ovarian cancers remain uncertain. Precise risk estimates are required to inform effective cancer risk management. This study investigates the associations between the risks of 22 cancers and BRCA1/2 PVs using data from 5,341 families segregating BRCA1 or BRCA2 PVs. Knowledge Generated BRCA1 and BRCA2 PVs are associated with increased risks of male breast, pancreatic, and stomach cancers; male BRCA2 carriers are also at increased prostate cancer risk. No associations were found with risks of other cancers. The cumulative risks to age 80 years ranged from 0.4% for male breast cancer to approximately 2.5% for pancreatic cancer for BRCA1 carriers and from approximately 2.5% for pancreatic cancer to 27% for prostate cancer for BRCA2 carriers. Relevance The findings provide age-specific cancer risk estimates and will allow for improved cancer risk assessment of male and female carriers.

3Key Objective

4The associations of pathogenic variants (PVs) in BRCA1 and BRCA2 with cancers other than female breast and ovarian cancers remain uncertain. Precise risk estimates are required to inform effective cancer risk management. This study investigates the associations between the risks of 22 cancers and BRCA1/2 PVs using data from 5,341 families segregating BRCA1 or BRCA2 PVs.

5Knowledge Generated

6BRCA1 and BRCA2 PVs are associated with increased risks of male breast, pancreatic, and stomach cancers; male BRCA2 carriers are also at increased prostate cancer risk. No associations were found with risks of other cancers. The cumulative risks to age 80 years ranged from 0.4% for male breast cancer to approximately 2.5% for pancreatic cancer for BRCA1 carriers and from approximately 2.5% for pancreatic cancer to 27% for prostate cancer for BRCA2 carriers.

7Relevance

8The findings provide age-specific cancer risk estimates and will allow for improved cancer risk assessment of male and female carriers.

9The National Comprehensive Cancer Network and other guidelines recommend breast and ovarian cancer screening for BRCA1/2 carriers and prostate cancer screening particularly for BRCA2 carriers. Notably, National Comprehensive Cancer Network guidelines recently addressed testing and management for pancreatic cancer risk in BRCA1/2 carriers, but only in the presence of a positive family history of the disease., Overall, current guidelines suggest that men and women with BRCA1/2 PVs should consider participation in investigational screening studies and receive education regarding signs and symptoms of cancers possibly associated with BRCA1/2 PVs. The availability of more precise risk estimates will aid translation into evidence-based clinical guidelines for the cancer risk management in BRCA1/2 carriers and may guide treatment options for patients with cancer.

10To inform clinical management strategies and optimize guidelines for cancer risk management in female and male BRCA1/2 carriers, we comprehensively assess the associations of BRCA1/2 PVs with risks of 22 cancers, other than female breast and ovarian cancers.

11Data on 7,618 families with at least one family member having a BRCA1 or BRCA2 PV were obtained from 26 study groups in the Consortium of Investigators of Modifiers of BRCA1/2 (Data Supplement, online only). Only families with a clearly PV identified were included. The majority of families (7,281) were ascertained through an index individual attending cancer family clinics, mainly because of having multiple affected relatives, and 337 families were ascertained through an index case with breast or ovarian cancer, unselected for family history. All index individuals were age ≥ 18 years. For each family member, data including familial relationship, BRCA1/2 PV status, sex, year of birth, and years or age at pedigree data collection, death, and cancer diagnoses were collected (Data Supplement). All participants provided written informed consent and participated in studies at the host institutions under ethically approved protocols.

12BRCA1 and BRCA2 families were analyzed separately. Complex segregation analysis, which considered the observed phenotype and observed or inferred genotype information of all family members, was used to estimate relative risks (RRs) for 22 first primary cancer sites, excluding female breast and ovarian cancers (Table 1). This involved comparing the observed cancer incidences for carriers with the age-, country- and birth cohort-specific population incidences (Cancer Incidence in Five Continents); thus, the estimated RRs were equivalent to standardized incidence ratios. Noncarriers were assumed to develop the cancers according to population incidences. Pedigree likelihoods were constructed and maximized using the pedigree analysis software MENDEL.

13Individuals were followed from birth until the age of the first primary cancer diagnosis, death, age at pedigree-data collection, risk-reducing mastectomy and/or salpingo-oophorectomy (if these occurred at least 1 year before breast or ovarian cancer diagnoses, respectively), or age 80 years, whichever occurred first. Missing year of birth and cancer diagnosis age were imputed (Data Supplement).

14Each individual was assumed to be at risk of developing the cancer of interest, as well as breast or ovarian cancer. The RRs for female breast and ovarian cancers were assumed to be equal to previous estimates; therefore, we only estimated the RR for the cancer of interest. We fitted models in which the RRs were assumed to be constant with age, birth cohort, sex, and study group and separate models with sex-specific RRs. For cancers with significant associations, we investigated whether the RRs varied by age. RRs from the best fitting models were used to estimate age-specific absolute risks on the basis of UK cancer incidences in year 2008-2012 (Data Supplement).

15Because family ascertainment varied across study groups, we adjusted for the ascertainment of each family separately using an ascertainment-assumption-free approach.- Pedigree likelihoods were computed conditional on any data that may be relevant to the ascertainment (Data Supplement). Noninformative families, in which no additional information beyond the data relevant to the ascertainment was available, were excluded from analysis. Since cancer family history was self-reported, we assessed the possibility of systematic under-reporting of specific cancers at the individual study group level and excluded any study groups in which under-reporting was likely relative to the population incidences (Data Supplement).

16Sensitivity analyses under alternative inclusion, censoring, or ascertainment assumptions were performed for cancers that demonstrated associations: (1) stratifying by geographical region (Asian countries v others); (2) including study groups with possible cancer under-reporting; (3) excluding individuals with missing age at diagnosis; (4) individuals with risk-reducing bilateral mastectomy and/or salpingo-oophorectomy were still considered to be at risk of developing the other cancers, except breast and ovarian cancers; and (5) assuming the data relevant to the ascertainment for clinic-based families do not include the family history of cancer of interest. To account for population differences in melanoma skin pigmentation, we also conducted sensitivity analyses for melanoma by using (1) only families from Australia, Northern Europe, and North America; (2) only families in which probands self-identified as White European; and (3) only the families satisfying both (1) and (2).

17All statistical tests were two-sided, and associations with a nominal P <.05 were considered statistically significant.

18After ascertainment adjustment, 3,184 BRCA1 families and 2,157 BRCA2 families were informative for inclusion in the analysis, including 14,979 carriers, 9,296 noncarriers, and 153,323 untested individuals (Data Supplement). 61.3% of probands had self-reported ethnicity data. Of those, 77.0%, 11.5%, 4.7%, 3.3%, and 1.2% self-identified as White European, Asian, Ashkenazi Jewish, Hispanic, and Black, respectively. Prostate, lung, colorectal, stomach, and pancreatic cancers were the most common cancers in the data set, aside from breast and ovarian (Table 1). The age at diagnosis for each cancer by PV status is shown in the Data Supplement. After excluding study groups in which there was potential cancer under-reporting (Data Supplement), the proportions of families included in the estimation of cancer-specific risks varied from approximately 15% for lymphoma and multiple myeloma to > 90% for pancreatic and male breast cancers (Data Supplement).

19BRCA1 PVs were associated with male breast (RR = 4.30; 95% CI, 1.09 to 16.96), gallbladder (RR = 3.34; 95% CI, 1.34 to 8.28), pancreatic (RR = 2.36; 95% CI, 1.51 to 3.68), stomach (RR = 2.17; 95% CI, 1.25 to 3.77), and colorectal (RR = 1.48; 95% CI, 1.01 to 2.16) cancers (Table 2). No association was found for prostate cancer (RR = 0.82; 95% CI, 0.54 to 1.27). No difference in the RR estimates by sex was observed for any of the 17 non–sex-specific cancers (all P >.07; Table 3).

20A model with RRs stratified by age 65 years (Data Supplement) provided a significantly better fit for stomach cancer: RR = 3.50 (95% CI, 2.01 to 6.10) for age < 65 years and higher than 0.61 (95% CI, 0.16 to 2.30) for age ≥ 65 years (P-heterogeneity =.01). For male breast cancer, a model with RRs stratified by age decade provided a better fit than the model with an age-constant RR (P =.03), but this was mainly driven by the lack of cases in the age group of 50-59 years (Data Supplement).

21BRCA2 PVs were associated with increased risks of male breast (RR = 44.0; 95% CI, 21.3 to 90.9), stomach (RR = 3.69; 95% CI, 2.40 to 5.67), pancreatic (RR = 3.34; 95% CI, 2.21 to 5.06), and prostate (RR = 2.22; 95% CI, 1.63 to 3.03) cancers (Table 2). Female carriers had a higher risk of stomach cancer (RR = 6.89; 95% CI, 3.71 to 12.78) than male carriers (RR = 2.76; 95% CI, 1.59 to 4.80; P-heterogeneity =.04; Table 3).

22A model with RRs stratified by age 65 years (Data Supplement) provided a significantly better fit for pancreatic cancer: RR = 4.92 (95% CI, 2.96 to 7.80) for age < 65 years and higher than 1.77 (95% CI, 0.87 to 3.58) for age ≥ 65 years (P-heterogeneity =.03). There was a suggestion that the prostate cancer RR was greater for age < 65 years (RR = 3.10; 95% CI, 2.00 to 4.79) than age ≥ 65 years (RR = 1.69; 95% CI, 1.09 to 2.62), but this model did not fit significantly better than the model with an age-constant RR (P =.06).

23The results are described in detail in the Data Supplement. There was no significant difference in the RR estimates by geographical region. The observed cancer associations were robust to all sensitivity analyses, except for colorectal and gallbladder cancers. No association was found for melanoma even when analyses were restricted to families from Australia, Northern Europe, and North America or families in which probands self-identified as White European.

24RRs from the main analysis best-fitting models were used to calculate age-specific absolute cancer risks (Table 4 and Fig 1). By age 80 years, the male breast cancer risk for BRCA1 and BRCA2 carriers was 0.4% (95% CI, 0.1 to 1.5) and 3.8% (95% CI, 1.9 to 7.7), respectively; the pancreatic cancer risk varied between 2.3% and 3.0% for both male and female BRCA1 and BRCA2 carriers; the stomach cancer risks were 1.6% (95% CI, 0.7 to 4.0) for male and 0.7% (95% CI, 0.3 to 1.7) for female BRCA1 carriers and approximately 3.5% for both male and female BRCA2 carriers. The prostate cancer risk associated with BRCA2 PVs was 26.9% (95% CI, 20.5 to 34.7) by age 80 years and 33.1% (95% CI, 25.5 to 42.2) by age 85 years.

25This study assessed the risks associated with BRCA1/2 PVs for 22 first primary cancers, other than female breast and ovarian cancers, and further clarified the cancer spectrum associated with BRCA1/2 PVs.

26The associations of BRCA1/2 PVs with the risks of male breast and pancreatic cancers were confirmed and refined, as well as the association of prostate cancer with BRCA2 PVs, regardless of age and aggressiveness.

27The lifetime male breast cancer risks were previously reported to be 2%-6% for BRCA1 and 7%-13% for BRCA2 carriers (Data Supplement).,,- We estimated these risks to be somewhat lower, 0.4% (95% CI, 0.1 to 1.5) and 3.8% (95% CI, 1.9 to 7.7), respectively. The pancreatic cancer associations were consistent with previously reported RRs of 2-3 and lifetime risks of 1%-4% for BRCA1 carriers,, and RRs of 3-6 and lifetime risks of 3%-5% for BRCA2 carriers (Data Supplement).,- Notably, the RR was higher for BRCA2 carriers age < 65 years.

28Previous retrospective studies reported prostate cancer RRs of 2-6 and absolute risks of 17%-31% by age 80 years for BRCA2 carriers (Data Supplement).,,,,- Our estimated absolute risk by age 85 years was 33%, lower than the recently reported prospective estimate of 60% by Nyberg et al. However, after adjusting for possible increased prostate-specific antigen screening effects in the prospective study, their estimate was 41% (95% CI, 22 to 59), consistent with our estimate. The present estimate is unlikely to be subject to increased screening biases since prostate cancer family history was retrospectively collected, and increased screening in relatives is unlikely to have taken place before the identification of BRCA2 PVs. The reported associations of BRCA1 PVs with prostate cancer risk are inconsistent, with RRs of 0.4-4, most not statistically significant.,,,,-,, This study confirms that BRCA1 PVs are not associated with overall prostate cancer risk.

29Among the suggested associations with other cancers, the association between BRCA1/2 PVs and stomach cancer is under considerable debate., This study validated and further elucidated this association: there were associations with both BRCA1 and BRCA2 PVs, with RRs of 2.17 (3.50 for age < 65 years) and 3.69, respectively. Our estimates better refined the previously reported RRs of 2-7 for BRCA1 carriers,, and approximately 2.6 for BRCA2 carriers (Data Supplement)., Notably, our findings showed that the stomach cancer RR for female BRCA2 carriers was higher than the estimate for male carriers although this translated in similar absolute risks, given the higher incidence of male stomach cancer in the general population. However, we cannot exclude the possibility that the higher female RR may be due to the misclassification of some ovarian cancers as stomach cancers.,

30Data in the current study come from either epidemiologic studies or families undergoing PV screening collected at genetics centers. Although individual studies and clinical genetic centers, where possible, confirmed reported cancer diagnoses in families through medical records or registries as part of standard clinical practice, cancer confirmation information is not available centrally and it was not feasible to collect this at such a large scale. However, a key advantage of the present study is the large sample size, which results in RR estimates with greater precision. Only a small number of family-based studies reported cancer confirmation rates.,,, Our RR estimates for stomach cancer, which may be susceptible to a greater degree of misclassification bias than other cancers,, are not significantly different from the estimates from studies that reported cancer confirmation. However, the present RRs have similar or greater precision than the published estimates from studies with high cancer confirmation rates (Data Supplement).

31In the present study, previously suggested associations of BRCA1/2 PVs with risks of other genitourinary cancers and melanoma, were not replicated. Although associations of BRCA1 PVs with colorectal and gallbladder cancers were observed, the results were not robust in the sensitivity analyses performed.

32Increased risks of bone and liver cancer have also been reported for BRCA1 or BRCA2 carriers.- However, liver and bone are common metastatic sites for breast, prostate, or pancreatic cancers and could be the presenting cancer. Since no pathology confirmation data were available, we did not examine these associations in the main analysis. If we assume that the reported bone and liver cancers in the data set are indeed first primaries, the data suggest no association with BRCA1 PVs, but that BRCA2 carriers are at seven-fold increased risk of bone cancer and five-fold increased risk of liver cancer without significant differences between males and females (Data Supplement). However, no conclusion for these associations can be drawn without pathology confirmation.

33Overall, the estimated age-specific relative and absolute risks suggest that, in addition to breast and ovarian cancers, the clinical management of BRCA1/2 carriers should focus on cancer sites, which now show robust associations, such as prostate (BRCA2 carriers only), pancreatic, and possibly stomach cancers. Notably, although rare, pancreatic and stomach cancers are associated with poor prognosis and their incidences have been rising over time, and thus, our results highlight the importance of screening for upper gastrointestinal tract malignancies for BRCA1 and BRCA2 carriers, particularly for age < 65 years. On the other hand, some cancers previously taken into consideration for screening for BRCA1/2 carriers, like melanoma, may be reconsidered, to further optimize cancer prevention screening strategies and eventually reduce carriers' distress. Given that the cancer risk associations were found for both male and female carriers, the results also suggest that male relatives of known BRCA1/2 carriers should be informed about their individual cancer risk and encouraged to be tested., It has been shown that knowing the germline BRCA1/2 PV status can influence treatment options for patients with cancer, leading to improved prognosis. For example, poly (ADP-ribose) polymerase inhibitor therapies that have been used successfully in the treatment of BRCA-related breast and ovarian cancers are now beginning to be used for pancreatic and prostate cancers,, and in the near future, they might also be used for stomach cancer.

34To avoid biases in the risk estimates related to the ascertainment of clinic-based families, on the basis of multiple affected family members, we used a conservative ascertainment adjustment approach by conditioning on the family histories of cancers of breast and ovary and the cancer site under investigation. When only family history of female breast and ovarian cancers was considered in the ascertainment, the RR estimates were somewhat higher for most cancers but with narrower CIs (Data Supplement). Therefore, conditioning on the family history of the cancer of interest is unlikely to have led to substantial underestimation of risk. A notable exception was male breast cancer, where much higher RR estimates were obtained. However, this estimate is most likely biased because male breast cancer family history has been an important factor in considering BRCA1/2 germline genetic testing since the discovery of BRCA1/2.

35This study has several limitations. First, this is a retrospective family-based study, with self-reported cancer family history, which may be inaccurate., Second, 7%-40% of reported cancer cases had missing age at diagnosis, with stomach cancer having the largest proportion. To minimize these potential biases, we performed sensitivity analyses excluding any study groups in which under-reporting was likely and any cases with missing age at diagnosis, and conclusions remained similar for most cancers. Third, we presented our results without any multiple testing adjustment. However, even using a false discovery rate adjustment, all the observed associations for BRCA2 carriers and the pancreatic cancer association for BRCA1 carriers had false discovery rates < 0.05. Fourth, the ethnicity of the family proband was not systematically collected by all studies because of variations in local data collection protocols. Among those with recorded ethnicity, in Asia-based studies, 97.7% of probands were Asian and in the rest of the studies 86.1%, 5.2%, 3.7%, 1.3%, and 1.1% of probands were White European, Ashkenazi, Hispanic, Black, and Asian, respectively. Therefore, the power to investigate the associations by all ethnic groups was limited. However, we did not find evidence of heterogeneity in the RRs by geographical region (Asia v others). Whether our risk estimates are applicable to non-European populations requires further investigation. Fifth, we did not have data on other genetic and environmental factors, so we were unable to investigate the modification effects of these factors; therefore, our risk estimates should be interpreted as the average risks across all potential genetic and environmental modifiers.

36In conclusion, this study confirms that, aside from female breast and ovarian cancers, BRCA1/2 PVs are associated with increased risks of breast cancer in men, and pancreatic and stomach cancers in both sexes, and that only BRCA2 carriers are at elevated prostate cancer risk. BRCA1/2 PVs were not associated with the risks of any other cancers previously suggested. The association results and estimated age-specific risks will improve the cancer risk management for men and women with BRCA1/2 PVs.