Pathogenic Variants in Cancer Susceptibility Genes Predispose to Ductal Carcinoma In Situ of the Breast

Ductal carcinoma in situ (DCIS) is a relatively common type of breast cancer and accounts for about 20% to 25% of all new breast cancer cases diagnosed in the United States (1, 2). DCIS is considered a non-obligate precursor to invasive ductal breast cancer (IDC), and understanding predisposing factors to DCIS could have implications for the early detection and prevention of breast cancer. Similarities in the risk factors for DCIS and IDC, including hormonal factors, mammographic density, and family history of breast cancer, have been identified (3–6). However, the contribution of rare germline pathogenic variants (PV) in cancer predisposition genes to the risk of DCIS is poorly understood. Prior studies of BRCA1 and BRCA2, which focused on those qualifying for germline genetic testing because of young age at diagnosis or family history of cancer, found that PV were less frequent in women with DCIS than those with invasive breast cancer (7–9) but were more frequent in DCIS relative to low-grade invasive disease (10). In contrast to BRCA1/2, the frequencies of PV in other breast cancer predisposition genes in DCIS and the risks of DCIS associated with these PV have not been established (7–11). Furthermore, an increased risk of contralateral breast cancer (CBC) has been established for carriers of PV in BRCA1, BRCA2, CHEK2, or PALB2 with breast cancer (12–15), but it is unknown whether these risk estimates apply to PV carriers with DCIS. In part due to a lack of understanding of genetic predisposition to DCIS, current guidelines on germline genetic testing and management of PV carriers do not distinguish between DCIS and IDC (16). A comprehensive assessment of genetic predisposition to DCIS is needed to personalize germline genetic testing in women with DCIS and identify appropriate management strategies for PV carriers with DCIS. Herein, we present the results of germline genetic testing among women with DCIS from clinical and population-based cohorts and identify the risk of DCIS and CBC among carriers of PV in cancer predisposition genes.

The study was approved by the Western Institutional Review Board-Copernicus Group, which exempted review of the clinical testing cohort, and by the Mayo Clinic Institutional Review Board. The “CAnceR RIsk Estimates Related to Susceptibility” (CARRIERS) consortium consisted of 12 population-based and family-based studies of breast cancer from the United States, including 4,628 women with DCIS, 18,193 women with IDC, and 25,007 unaffected women. The population-based cohort within CARRIERS included 3,876 women with DCIS, 13,701 women with IDC, and 22,186 unaffected women matched to breast cancer by age from nine prospective studies. Invasive or in situ CBC or second breast cancer (combined ipsilateral and CBC) risk was evaluated among 3,024 women with unilateral DCIS receiving ipsilateral surgery for initial breast cancer. DCIS with no invasive component was confirmed by the underlying cohorts within CARRIERS using data from medical records and cancer registries. Women diagnosed with bilateral breast cancer within 1 year of initial diagnosis, undergoing bilateral mastectomy for initial breast cancer, or those missing age at initial breast cancer or CBC status were excluded from the CBC evaluation. The details of the CARRIERS study have been previously published (15, 17–19) and are also provided in Supplementary Methods.

The clinical testing cohort consisted of a nationwide sample of 9,887 women with DCIS who underwent clinically indicated hereditary cancer multigene panel testing at Ambry Genetics from March 2012 to December 2016. Data on patient characteristics were collected from requisition forms at the time of testing and clinical notes provided by ordering clinicians. For comparison, information on 38,057 women with IDC who underwent germline genetic testing during the same time period at Ambry Genetics was extracted. The diagnoses and tumor pathology for 10% of the patients with DCIS and invasive disease were confirmed in medical records as part of a validation study. Women who had previously undergone testing for BRCA1 or BRCA2 prior to multigene panel testing were excluded.

The UK Biobank (UKB; RRID: SCR_012815) is a population-based prospective cohort study of more than 500,000 subjects (20). DCIS cases (n = 2,339) were defined by International Classification of Diseases-10 code D05, as determined by linkage to the UK National Cancer Registration and Analysis Service, or self-reported breast cancer. Both prevalent and incident cases were included. The control group consisted of females without breast cancer (n = 203,896).

For the population-based cohort, germline DNA samples were subjected to multiplex amplicon–based analysis of 746 target regions covering all coding regions and consensus splice sites in cancer predisposition genes using a QIAseq custom panel (17). For the clinical testing cohort, testing of five to 49 genes, depending on the multigene panel ordered, was performed by targeted custom capture and sequencing of all coding domains and flanking 5′- and 3′-ends of all the introns and untranslated regions (21, 22). For both cohorts, germline genetic testing results from 18 cancer predisposition genes were assessed, and results for 10 breast cancer genes (ATM, BARD1, BRCA1, BRCA2, CDH1, CHEK2, PALB2, RAD51C, RAD51D, and TP53) were analyzed (Supplementary Table S1; refs. 16, 23, 24). For UKB, gene-specific sequencing data and called variants for these genes from whole-exome sequencing data were accessible through the UKB DNAnexus platform. Quality control metrics for genotype depth and quality were applied (25, 26). A five-tier system was used to classify variants using an American College of Medical Genetics framework (27). PV and likely PV were analyzed together. Known low-penetrance variants were excluded from all analyses (CDKN2A^I49T; CHEK2^H371Y, CHEK2^I157T, and CHEK2^S428F; TP53^R156H and TP53^R181H).

The frequency of PV in each gene was assessed among women with DCIS in the clinical testing and population-based cohorts and among subsets of women with DCIS based on estrogen receptor (ER) status, age, and family history of breast and ovarian cancers. To identify the genes associated with the increased risk of DCIS, the frequency of PV in each gene in women with DCIS in the clinical testing cohort was compared with the frequency from gnomAD 2.1 reference non-female, noncancer controls (http://gnomad.broadinstitute.org/, RRID: SCR_014964; ref. 28) with weighting for the relative frequency of different races and ethnicities utilizing a logistic regression model (17). Copy-number variations and gnomAD filter non-PASS variants were excluded (29). For the population-based cohort, case–control association testing for the risk of DCIS among PV carriers in each gene was conducted with logistic regression models adjusting for age at diagnosis, self-reported race and ethnicity, and study. For the UKB, DCIS case–control association analysis was performed using the Fisher exact test. Enrichment analysis comparing PV in DCIS and IDC in the clinical testing and population-based cohorts was conducted using logistic regression adjusting for age at diagnosis, family history of breast or ovarian cancer, and ER status. Among women with DCIS in the population-based cohort, the cumulative incidence of CBC for PV carriers in BRCA1 and BRCA2 combined, as well as in the three high-risk genes (BRCA1, BRCA2, and PALB2), was evaluated. This time-to-event analysis began at the time of the first breast cancer diagnosis and adjusted for the competing risk of death using the Fine and Gray method (30). Censoring occurred at the last follow-up or at contralateral prophylactic mastectomy, whichever came first. All tests were two-sided; P < 0.05 was considered statistically significant. All analyses were performed using R version 3.6 (RRID: SCR_001905).

All data presented in the article and supplementary materials are available directly from the authors. CARRIERS population-based data are available at dbGAP (phs002820).

The median age at diagnosis of DCIS was 55 years in the clinical testing cohort and 62 years in the population-based cohort. More than 81% of the women in the population-based cohort and 64% in the clinical cohort were non-Hispanic White, and >80% of the tumors with available information on receptor status were ER positive (Table 1).

The pooled frequency of PV in 10 known breast cancer predisposition genes among women with DCIS was 6.5% in the clinical testing cohort, whereas frequencies from nine genes (excluding TP53 because of the absence of DCIS cases) were 4.6% in the population-based cohort (Table 2; Supplementary Table S1). In both datasets, PV in CHEK2 and BRCA2 were the most common with frequencies >1%. PV in ATM had a mean frequency of 1%, and PALB2 PV ranged from 0.31% to 0.47% in the two cohorts. However, BRCA1 PV were relatively infrequent with a frequency range of 0.28% to 0.71% (Table 2).

Among women in the clinical cohort with first- or second-degree relatives with breast or ovarian cancer, the frequency of PV in 10 predisposition genes was 7.1%, whereas the frequency was 5.1% among those without a family history of these cancers (Supplementary Table S2). Similarly, in the population-based cohort, PV were detected in 7.7% of DCIS cases with first-degree relatives with breast or ovarian cancer and 3.8% of women without a family history (Supplementary Table S2). In both cohorts, PV in BRCA2 were significantly enriched (clinical P = 0.02; population-based P = 0.002) among women with a family history. PV in BRCA1 (P = 0.004) and PALB2 (P = 0.01) were significantly enriched in women with a first-degree relative with breast or ovarian cancer in the population-based cohort only. PV in CHEK2 were observed in >1% of women with DCIS without a family history of breast or ovarian cancer in both cohorts.

Among women diagnosed with DCIS at ≤50 years of age, the frequency of PV was 6.9% in the clinical testing cohort and 6.3% in the population-based cohort (Supplementary Table S3). In comparison, the frequencies of PV among women diagnosed at >50 years were 5.9% and 4.1% in the clinical testing and population-based cohorts, respectively. Further analysis by ER status (Supplementary Table S4) demonstrated a comparable frequency of PV among women with ER-positive and ER-negative DCIS in the clinical testing (6.2% in ER-positive vs. 5.8% in ER-negative) and the population-based (4.8% in ER-positive vs. 4.6% in ER-negative) cohorts. However, in both cohorts, CHEK2 PV were more than twofold enriched, and the frequency of BRCA1 PV was substantially lower among women with ER-positive DCIS compared with those with ER-negative DCIS.

Case–control association analysis showed that PV in BRCA2 were significantly associated with a high risk (OR > 4) of DCIS, whereas PV in ATM, CHEK2, and PALB2 were associated with moderate risk (2 < OR < 4) in both the clinical testing and population-based cohorts (Fig. 1; Table 2). BRCA1 PV were associated with high risk [OR, 4.5, 95% (confidence interval) CI, 3.1–6.6, P < 0.001] of DCIS in the clinical testing cohort and moderate risk (OR, 2.1; 95% CI, 1.0–4.1; P = 0.047) in the population-based cohort. Interestingly, PV in MSH6 were associated with an increased risk of DCIS (OR, 2.6; 95% CI, 1.4–4.6; P = 0.002) in the clinical testing cohort and in the population-based cohort (OR, 2.1; 95% CI, 0.8–5.2; P = 0.11), although the latter association was not significant. Results from the UKB were consistent with the findings from the population-based study (Table 2). Further analyses of the clinical and population-based cohorts based on ER status of DCIS cases showed that PV in ATM, BRCA2, CHEK2, and PALB2 were consistently associated with increased risk of ER-positive DCIS (Supplementary Table S5). In contrast, only BRCA1 and BRCA2 PV were significantly associated with increased risk of ER-negative DCIS (Supplementary Table S5). Importantly, the ER-negative findings should be interpreted with caution because the numbers of ER-negative DCIS cases were limited.

The frequency of PV in the breast cancer predisposition genes was higher in IDC cases than DCIS cases overall in both the clinical (9.3% vs. 6.5%) and population-based cohorts (6.2% vs. 4.6%; Table 3). PV in BRCA1, BRCA2, and PALB2 were significantly enriched twofold or more in IDC cases compared with DCIS in both cohorts, whereas no difference in the frequency of ATM and CHEK2 PV was observed. Similarly, no difference in TP53 PV frequency was observed in the clinical cohort, even when restricted to DCIS diagnoses under 31 years of age (Table 3). However, consistent with recent findings (10), the frequencies of PV in high-risk genes (BRCA1, BRCA2, and PALB2; P = 0.0015) and CHEK2 (0.02) were significantly higher in grade 2/3 DCIS than in grade 1 IDC (Supplementary Table S6). Overall, DCIS was observed in 22.3% of women testing negative for PV in the breast cancer predisposition genes and in 20.8% of ATM and 27.8% of CHEK2 PV carriers but only in 5.9% of BRCA1, 14.2% of BRCA2, and 13.8% of PALB2 carriers in the population-based cohort, consistent with reported lower rates of DCIS in BRCA1 and BRCA2 PV carriers (9).

Among women with DCIS in the population-based cohort, the cumulative incidences of CBC in carriers of germline PV in BRCA1, BRCA2, or PALB2 were 11% (95% CI, 1%–20%) in 5 years, 17% (95% CI, 4%–28%) in 10 years, and 23% (95% CI, 6%–36%) in 15 years (Fig. 2). Restricting to BRCA1 and BRCA2 PV carriers yielded cumulative CBC incidences of 12% (95% CI, 1%–22%) in 5 years, 12% (95% CI, 1%–22%) in 10 years, and 20% (95% CI, 1%–36%) in 15 years. In contrast, the cumulative incidences of CBC among women with DCIS without germline PV in five common breast cancer predisposition genes were 3% (95% CI, 2%–4%) in 5 years, 5.5% (95% CI, 5%–7%) in 10 years, and 8% (95% CI, 6%–9%) in 15 years.

Exploratory analysis based on small numbers of events estimated the rate of invasive CBC among BRCA1, BRCA2 or PALB2 PV carriers at 13% in 10 years and the rate of in situ CBC at 6% in 10 years. Restricting to BRCA1 and BRCA2 yielded rates of invasive CBC of 10% and in situ CBC of 3% in 10 years. Furthermore, combining ipsilateral and contralateral events demonstrated a cumulative incidence of 27% for second breast cancers in 10 years among carriers of PV in BRCA1, BRCA2, or PALB2, 22% in 10 years for BRCA1 and BRCA2 PV carriers only, and 10.9% in women without PV in the five predisposition genes. Further evaluation of these estimates in additional studies is needed.

In the largest study of genetic predisposition for DCIS to date, we present a comprehensive analysis of the frequency of germline PV in predisposition genes, the risks of DCIS associated with these PV, and the differences in gene-specific PV frequencies between IDC and DCIS. Thus, this study provides quantitative data that move beyond the assumptions currently in use regarding the relevance of predisposition genes to DCIS based on a series of small studies. These findings have implications for germline genetic testing, counseling of PV carriers for DCIS risk, and personalized management of DCIS among PV carriers. The consistency of the findings in unique clinical testing and population-based cohorts with distinct ascertainment, along with further validation in a UKB cohort, is a significant strength of the study.

The overall frequency of germline PV in BRCA2 and PALB2 and especially in BRCA1 was lower among women with DCIS compared with IDC, consistent with prior studies (31–33). In addition, among women with breast cancer in the population-based cohort, the proportion of DCIS was observed to be lower in BRCA1, BRCA2, and PALB2 PV carriers compared with noncarriers. These findings suggest that BRCA1, PALB2, and perhaps BRCA2 PV carriers may develop IDC without precursor DCIS or may progress rapidly from DCIS to IDC. Importantly, this observation did not hold for low-grade invasive disease, with grade 2/3 DCIS harboring substantially more PV than grade 1 IDC, as recently reported in another study (10). The difference in PV frequencies suggests that high-grade DCIS does not progress to form low-grade invasive breast cancers. In contrast, the frequencies of PV in CHEK2 and ATM were similar among women with IDC and DCIS. As identification of germline PV in ATM and CHEK2 in women with DCIS may have implications for early detection and prevention of second cancers (16), testing for all predisposition genes, including ATM and CHEK2 that predominantly predispose to ER-positive breast cancer, may be appropriate. However, given the concerns about overdiagnosis of DCIS (34), long-term follow-up studies are needed to understand whether the rates of overdiagnosis may differ between PV carriers and noncarriers.

The higher frequency of germline PV in predisposition genes among women with early-onset DCIS or a family history of breast or ovarian cancer is consistent with prior studies and the observations in IDC (17, 21, 22, 35). The current National Comprehensive Cancer Network (NCCN) and National Institute for Health and Care Excellence (NICE) guidelines rely on family history and age at diagnosis and receptor status of breast cancer for identifying women with breast cancer who may benefit from testing but do not distinguish between IDC and DCIS (16). In addition, there is an ongoing debate in terms of testing women with breast cancer based on age at diagnosis and family history versus testing all women with breast cancer (36–38). However, when considering the lower frequencies of germline PV in high-risk genes among women with DCIS compared with those with IDC and the lack of therapeutic indications with PARP inhibitors for DCIS even with BRCA1, BRCA2, or PALB2 PV (39–41) but the substantial risk of CBC and other cancers for PV carriers, perhaps germline genetic testing of women with DCIS is appropriate when based on family history, age at diagnosis, and receptor status.

This study provides insights into the magnitude of risk of DCIS in germline PV carriers. PV in BRCA2 were the only variants associated with a high risk of DCIS (OR > 4) in both the clinical testing and population-based cohorts, whereas PV in BRCA1 were associated with high risk of DCIS only in the clinical study. Importantly, in contrast to a high risk of IDC (17), PALB2 PV were associated with a moderate risk for DCIS in both cohorts. PV in ATM and CHEK2 were also consistently associated with moderate risk. The PV frequencies per gene were somewhat consistent with results from a study of 655 DCIS cases diagnosed <50 years of age (9), which had twofold higher rates of PV in BRCA2, PALB2, and TP53, likely because of the younger age at diagnosis, but similar rates of PV in BRCA1 and CHEK2.

Current guidelines support MRI surveillance in addition to mammograms for women with PV in the high-risk BRCA1, BRCA2, and PALB2 genes and for ATM and CHEK2 PV carriers (16). The results provided here are consistent with this approach. Although the risk of DCIS may only be moderately elevated in BRCA1 or PALB2 PV carriers, prophylactic risk–reducing mastectomy is still appropriate in unaffected BRCA1 and PALB2 PV carriers because of the high risk of IDC (17). In addition, chemoprevention with endocrine agents may be a consideration, particularly for PV in genes such as ATM and CHEK2, which are primarily associated with an increased risk of ER-positive DCIS. However, the efficacy of endocrine agents for breast cancer prevention in germline PV carriers has not been thoroughly evaluated, with one prior study suggesting that tamoxifen decreased the risk of breast cancer in BRCA2 but not in BRCA1 PV carriers (42). As PV in CDH1, MSH6, and RAD51D were significantly associated with an increased risk of DCIS in the clinical cohort, in which ascertainment bias may play a role, these findings will need to be further evaluated in larger studies.

This study is the first to comprehensively assess the second breast cancer risk among women with DCIS from the general population who are carriers of PV in BRCA1, BRCA2, or PALB2. Carriers of germline PV in BRCA1 and BRCA2 with DCIS were noted to be at a substantially increased risk of CBC compared with noncarriers. Furthermore, the risk of CBC among PV carriers of these genes was observed to be similar regardless of whether the primary breast cancer is IDC or DCIS (15). Importantly, these findings suggest that contralateral risk–reducing mastectomy may be considered in women with DCIS who are carriers of PV in BRCA1 and BRCA2, similar to the strategy in PV carriers of these genes with IDC (16, 43). There were insufficient PALB2 carriers identified in this study to inform on the use of contralateral risk–reducing mastectomy. Furthermore, among carriers of PV in high-risk genes with DCIS who do not undergo contralateral prophylactic mastectomy, surveillance for CBC risk with supplemental MRI is warranted based on the >20% 15-year risk of CBC observed in this study (44, 45). Importantly, these surveillance and risk-reducing strategies are already available and recommended even for unaffected carriers of PV in these genes.

Although this is the largest reported study of genetic associations for DCIS, the sample size was still a limitation, in that PV in some of the predisposition genes resulted in associations with wide CI. In addition, the use of test requisition forms rather than medical records for ascertainment of clinical and phenotypic data for the clinical testing cohort and the lack of central pathologic validation of the diagnosis of DCIS across the entire dataset were limitations that may have introduced some uncertainty into the observed associations. Furthermore, this study established associations between PV in breast cancer predisposition genes and risk of DCIS, but prospective studies with long-term follow-up will be needed to estimate the absolute risk and/or RR of DCIS in PV carriers and the subsequent risk of progression to invasive breast cancer. The lack of long-term outcome data limits the demonstration of the clinical benefit of identifying PV carriers among women with DCIS beyond what would be expected for unaffected PV carriers.

PV in ATM, BRCA1, BRCA2, CHEK2, or PALB2 are associated with a moderate to high risk of DCIS. Among women with DCIS undergoing germline genetic testing, multigene panels may be appropriate because of high frequencies of PV in genes other than BRCA1 and BRCA2. Contralateral prophylactic mastectomy may be appropriate in women with DCIS who are carriers of PV in BRCA1, BRCA2, and possibly PALB2 because of the high risk of developing CBC.

R. Karam reports other support from Ambry Genetics outside the submitted work. A. Trentham-Dietz reports grants from NCI during the conduct of the study. A.H. Eliassen reports grants from NIH during the conduct of the study. J.R. Palmer reports grants from NIH during the conduct of the study. C.M. Vachon reports grants from NCI during the conduct of the study, as well as grants from GRAIL outside the submitted work. J.N. Weitzel reports grants from Breast Cancer Research Foundation during the conduct of the study, as well as personal fees from Natera, MyOme, CancerIQ, and AstraZeneca and other support from Natera outside the submitted work. P. Kraft reports grants from NIH during the conduct of the study. J.S. Dolinsky reports personal fees from Ambry Genetics outside the submitted work. M.E. Richardson reports personal fees from Ambry Genetics outside the submitted work. S. Yadav reports other support from AstraZeneca and Repare Therapeutics outside the submitted work. F.J. Couch reports grants from NIH and Breast Cancer Research Foundation during the conduct of the study, as well as grants from GRAIL and personal fees from AstraZeneca outside the submitted work. No disclosures were reported by the other authors.

H. Huang: Data curation, formal analysis, investigation, writing–original draft, writing–review, and editing. R.E. Couch: Data curation, formal analysis, investigation, writing–original draft, writing–review and editing. R. Karam: Data curation, validation, investigation, writing–original draft, writing–review and editing. C.A. Haiman: Data curation, formal analysis, validation, investigation, writing–original draft, writing–review and editing. N. Boddicker: Formal analysis, writing–original draft, writing–review and editing. E.C. Polley: Formal analysis, writing–original draft, writing–review and editing. J. Na: Data curation, formal analysis, methodology, writing–original draft, writing–review and editing. C.B. Ambrosone: Resources, validation, writing–original draft, writing–review and editing. S. Yao: Resources, data curation, formal analysis, writing–original draft, writing–review and editing. A. Trentham-Dietz: Resources, data curation, writing–original draft, writing–review and editing. A.H. Eliassen: Data curation, writing–original draft, writing–review and editing. K. Penney: Data curation, investigation, writing–original draft, writing–review and editing. K. Brantley: Data curation, investigation, writing–original draft, writing–review and editing. C. Bodelon: Data curation, investigation, writing–original draft, writing–review and editing. L.R. Teras: Data curation, investigation, writing–original draft, writing–review and editing. J. Hodge: Data curation, investigation, writing–original draft, writing–review and editing. A. Patel: Data curation, investigation, writing–original draft, writing–review and editing. C.A. Haiman: Data curation, investigation, writing–original draft, writing–review and editing. E.M. John: Data curation, investigation, writing–original draft, writing–review and editing. S.L. Neuhausen: Data curation, investigation, writing–original draft, writing–review and editing. E. Martinez: Data curation, investigation, writing–original draft, writing–review and editing. J.V. Lacey: Data curation, investigation, writing–original draft, writing–review and editing. K.M. O’Brien: Data curation, investigation, writing–original draft, writing–review and editing. D.P. Sandler: Data curation, investigation, writing–original draft, writing–review and editing. C.R. Weinberg: Data curation, investigation, writing–original draft, writing–review and editing. J.R. Palmer: Data curation, investigation, writing–original draft, writing–review and editing. K.A. Bertrand: Data curation, investigation, writing–original draft, writing–review and editing. C.M. Vachon: Data curation, investigation, writing–original draft, writing–review and editing. J.E. Olson: Data curation, investigation, writing–original draft, writing–review and editing. K.E. Ruddy: Data curation, investigation, writing–original draft, writing–review and editing. H. Anton-Culver: Data curation, investigation, writing–original draft, writing–review and editing. A. Ziogas: Data curation, investigation, writing–original draft, writing–review and editing. D.E. Goldgar: Data curation, investigation, writing–original draft, writing–review and editing. K.L. Nathanson: Data curation, investigation, writing–original draft, writing–review and editing. S.M. Domchek: Resources, data curation, funding acquisition, writing–original draft, writing–review and editing. J.N. Weitzel: Data curation, investigation, writing–original draft, writing–review and editing. P. Kraft: Data curation, formal analysis, investigation, writing–original draft, writing–review and editing. J.S. Dolinsky: Data curation, investigation, writing–original draft, writing–review and editing. T. Pesaran: Data curation, investigation, writing–original draft, writing–review and editing. M.E. Richardson: Data curation, investigation, writing–original draft, writing–review and editing. S. Yadav: Resources, data curation, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. F.J. Couch: Conceptualization, resources, data curation, formal analysis, funding acquisition, investigation, writing–original draft, project administration, writing–review and editing.

This study was supported by NIH grants R35CA253187 and R01CA225662; an NIH Specialized Program of Research Excellence (SPORE) in Breast Cancer (P50CA116201) to Mayo Clinic; the Paul Calabresi Program in Clinical/Translational Research at Mayo Clinic (2K12CA090628); and the Breast Cancer Research Foundation (F.J. Couch, S.M. Domchek, K.L. Nathanson, J.N. Weitzel, and C.B. Ambrosone).