Breast Cancer Risk Modification in Women with Pathogenic Variants in BRCA1, BRCA2, ATM, CHEK2, and PALB2 Open Access

Genetic testing for inherited cancer risk is indicated for most patients with breast cancer (1). Genetic testing results affect cancer treatment, secondary cancer risk reduction, and care of family members. Pathogenic variants (PV) are commonly found in BRCA1, BRCA2 (BRCA1/2), ATM, CHEK2, and PALB2 (2, 3). We and others have reported on the prevalence and penetrance (associated cancer risk) of PVs in these genes among women who were not selected based on strong family cancer history or early diagnosis age (4, 5).

For individuals who inherit PVs in breast cancer–associated genes, it is important to know whether potentially modifiable exposures, such as menopausal hormone therapy (MHT) or alcohol use, substantially increase breast cancer risk. Prior studies investigated risk factors, including smoking, alcohol consumption, parity, breastfeeding, physical activity, body mass index (BMI), diagnostic and therapeutic radiation exposure, oral contraceptives, and MHT, among carriers of BRCA1/2 PVs (6–21), but many were retrospective and few included ATM, CHEK2, or PALB2. We leveraged the Women’s Health Initiative (WHI), a prospective observational study and randomized clinical trials that enrolled postmenopausal women. We report on penetrance and breast cancer risk–modifying factors for PVs in BRCA1/2, ATM, CHEK2, and PALB2 among WHI participants.

The primary objective was to evaluate the association of established risk factors with breast cancer risk among carriers of inherited PVs. As a secondary objective, we tested whether established factors modify risk differently for PV carriers versus noncarriers.

The WHI (RRID: SCR_011789) is a prospective study of morbidity and mortality among postmenopausal women, including clinical trials and an observational cohort. Details of the WHI study design, conduct, and follow-up were published previously (22–24). WHI participants were of ages 50 to 79 years, postmenopausal, had life expectancy of at least 3 years, and were enrolled throughout the United States from September 1993 to December 1998.

This study combined two WHI ancillary studies (AS), AS508 and AS551 (AS551 is a component of the CARRIERS consortium). Details of AS508 and of the CARRIERS study were published previously (4, 5). Both ASs included women without history of breast cancer at WHI enrollment, who either were diagnosed with invasive breast cancer or were comparison women who remained cancer-free as of September 20, 2017. For both AS508 and AS551, women with incident breast cancer and cancer-free comparison women were randomly selected from all WHI participants with a banked DNA sample.

All incident breast cancers were centrally adjudicated, and participant demographic, clinical, and family history data were collected as previously described (22–24). WHI variables used in this study were age at enrollment, race, and ethnicity [according to previously published WHI categories (25)], BMI (underweight, <18.5 kg/m²; normal, 18.5–24.9 kg/m²; overweight, 25–29.9 kg/m²; and obese, ≥30 kg/m²), breast cancer diagnosis (yes/no), family history of breast cancer in first-degree relatives (yes/no), cigarette smoking (ever use: yes/no), alcohol consumption (ever use: yes/no), parity status (parous/nulliparous), breastfeeding history (ever/never), oophorectomy (yes/no), tubal ligation (yes/no), neighborhood socioeconomic status [NSES; a numeric variable used in prior WHI research (26)] and MHT use [ever use: yes/no, for estrogen and progestin (E + P) and estrogen only (E-only)]. For MHT use, we distinguished between participants in the WHI clinical trials versus observational study. If a participant was randomized to the placebo arm of either the E + P MHT clinical trial or the E-only MHT clinical trial, we used that participant’s self-reported data to categorize her MHT use for analyses.

Sequencing for AS508 was performed by Myriad Genetics and for AS551 by the Couch Laboratory at Mayo Clinic, as previously reported (4, 5). For AS508, next-generation sequencing (NGS) was performed with sequence and large rearrangement (LR) analysis for all genes. LR testing for deletions and duplications was successfully performed by NGS dosage analysis on most samples. Portions of noncoding intronic regions surrounding the exons were analyzed by NGS. Long-range PCR was used for initial amplification of CHEK2 gene regions to avoid well-characterized pseudogenes. Samples were pooled and loaded onto massively parallel NextGen sequencers for 2 × 150 base paired-end reads. The minimum coverage depth used for sequence determination by NGS was 50× per base. AS551 sequencing was performed using a custom amplicon-based QIAseq panel. Target regions were exons <150 bp with 10 bp of flanking intronic sequence. Samples were pooled in batches of 768 and loaded onto massively parallel NextGen sequencers for 2 × 150 base paired-end reads. The minimum coverage depth used for sequence determination was 20× per base with a minimum alternative allele frequency of 0.20. LRs were identified by PatternCNV (27).

Variant classification followed criteria of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (28). Variants were classified as PV or likely PV if (i) they conferred a truncation, loss of select initiation codon, or splice donor/acceptor effect; (ii) functional data demonstrated an effect on protein function relevant to disease phenotype; or (iii) pathogenicity was otherwise demonstrated in the literature. If no significant functional data were available, then missense, silent, and intronic variants were classified as variants of uncertain significance (VUS), likely benign or benign based on modified American College of Medical Genetics and Genomics guidelines and according to standard laboratory procedures. The CHEK2 I157T variant was not classified as pathogenic, consistent with practice guidelines.

All analyses were conducted using R version 4.1.3 and independently reproduced using R version 4.3.2. We assessed the penetrance of PVs (defined as the magnitude of breast cancer risk associated with PV carriage) in each gene and breast cancer risks associated with clinical factors in terms of ORs, two-sided 95% confidence intervals (CI), and two-sided P values from multivariable logistic regression models with the Firth penalized likelihood. Additional details regarding the coding of variables are provided in Supplementary Methods S1.

To evaluate gene penetrance, we constructed a separate logistic regression model for each gene. Breast cancer status was the dependent variable. Independent variables included PV status, age at WHI enrollment, and AS (AS508 or AS551). Women with PVs in breast cancer–associated genes other than the gene of interest were excluded, as were women with VUS in the gene of interest, unless the VUS co-occurred with a PV in the same gene.

For the primary analysis, we evaluated breast cancer risks associated with established risk factors in terms of odds ratios (ORs) and two-sided 95% CIs from logistic regression models constructed separately in subcohorts of women who tested negative for PVs (noncarriers) and in women with PVs in each gene. For these models, breast cancer status was the dependent variable, and independent variables included the risk factor of interest, age at WHI enrollment, and AS.

Reasoning that associations of larger magnitude could have more significance in clinical decision-making, we sought to rule out twofold or greater increases in risk with established risk factors. Accordingly, we calculated one-sided upper 95% CIs to rule out twofold or greater increases in risk for PV carriers due to clinical factors expected to be associated with higher breast cancer risk (BMI, family history of breast cancer, smoking, alcohol consumption, NSES, and MHT), and one-sided lower 95% CIs for clinical factors expected to be associated with lower risk (parity, breastfeeding, oophorectomy, and tubal ligation).

For the secondary analysis, we tested whether established factors modify breast cancer risk differently for PV carriers versus noncarriers by fitting interaction terms in logistic regression models. Models were constructed separately in subcohorts comprised of noncarriers and carriers of PVs in the gene of interest. For each model, breast cancer status was the dependent variable. Independent variables included the risk factor of interest, PV status for the gene of interest, age at WHI enrollment, study, and a term for “risk factor × PV” interaction.

In an exploratory analysis, we reasoned that the effects of risk factors might differ by the estrogen responsiveness of breast cancers to which PV carriers are predisposed. Thus, we repeated the analyses described above by groups of genes according to association with estrogen receptor (ER)-positive breast cancer (ER-positive genes: ATM and CHEK2, which are not significantly associated with ER-negative disease) and ER-negative breast cancer (ER-negative genes: BRCA1 and PALB2, with ORs for ER-negative disease at least double those for ER-positive disease). These associations were based on results of the large CARRIERS and BRIDGES studies (4, 29), not on the ER status of tumors that developed in participants of AS508 or AS551.

The data generated in this study are available upon request from the corresponding author. Sequencing data used in this study are available upon request to the WHI Coordinating Center, in accordance with the WHI policy for accessing genetic data (available at https://www.whi.org/md/gwas).

A total of 12,957 WHI participants were included: 4,517 comprising AS508 and 8,440 comprising AS551 (Table 1). The median age at WHI enrollment was 63 years. The racial distribution was 80.6% White in AS508 and 99.0% White in AS551, for an overall distribution of 92.6% White, 4.0% Black, 1.4% Asian, 0.2% American Indian or Alaskan Native, <0.1% Native Hawaiian or Pacific Islander, 1.1% with two or more races, and 0.7% with unknown or not reported race. The ethnic distribution was 5.4% Spanish/Hispanic/Latina in AS508 and 0.5% Spanish/Hispanic/Latina in AS551, for an overall distribution of 2.2% Spanish/Hispanic/Latina, 97.7% not Spanish/Hispanic/Latina, and 0.2% missing data. The BMI was 29.2% obese and 34.6% overweight; 16.6% had a first-degree relative with breast cancer. Approximately half of participants ever smoked, most consumed some alcohol and were parous, approximately half had breastfed, and 17% to 18% had oophorectomy or tubal ligation. The mean NSES was 76.8 (SD 7.65), similar to that in other WHI studies (26). Slightly less than half (43.8%) of participants never used MHT; 32.6% used E + P MHT (685 in the WHI E + P clinical trial and 3,536 by self-report in the observational study), and 23.6% used E-only MHT (297 in the WHI E-only clinical trial and 2,761 by self-report in the observational study; Table 1).

PV prevalence was as follows: BRCA1, n = 34, 0.26%; BRCA2, n = 62, 0.48%; ATM, n = 65, 0.50%; CHEK2, n = 93, 0.72%; and PALB2, n = 33, 0.25%. Participant characteristics according to PV status are shown in Supplementary Table S1. Penetrance was as follows: BRCA1, OR = 4.87 (95% CI, 2.11–13.60); BRCA2, OR = 4.71 (95% CI, 2.57–9.41); ATM, OR = 2.24 (95% CI, 1.34–3.87); CHEK2, OR = 2.05 (95% CI, 1.34–3.21); and PALB2, OR = 6.94 (95% CI, 2.83–21.78; Table 2).

Table 3 shows results from multivariable models of the increase in the odds of developing breast cancer associated with each risk factor among noncarriers and carriers of PVs. We observed modest evidence of reduced breast cancer risk associated with breastfeeding for carriers of PVs in PALB2 (OR = 0.08; 95% CI, 0.00–0.92; P value = 0.042). When examining individual genes, statistical power was limited due to the small number of PV carriers. However, power was sufficient to rule out a twofold or greater increase in the odds of developing breast cancer based on one-sided 95% CIs for the following risk factors: obesity for ATM or BRCA2 carriers; smoking and alcohol for CHEK2 carriers; not breastfeeding for ATM carriers; not having oophorectomy for BRCA2 or CHEK2 carriers; not having tubal ligation for CHEK2 carriers; and NSES for carriers of PVs in any gene.

Table 4 shows results from multivariable models of the exploratory analysis of the increase in the odds of developing breast cancer associated with each risk factor among women who carry PVs in ER-positive breast cancer risk genes (ATM and CHEK2) and ER-negative breast cancer risk genes (BRCA1 and PALB2). The results suggest that E + P MHT use may be associated with high breast cancer risk in women who carry PVs in ER-positive genes (OR = 7.31; 95% CI, 1.14–64.20; P value = 0.036). Power was sufficient to exclude a twofold or greater increase in the odds of developing breast cancer due to smoking, alcohol consumption, and not having an oophorectomy or tubal ligation for ER-positive PV carriers and due to NSES for both ER-negative and ER-positive gene PV carriers.

At the 5% significance level, we found no evidence supporting an interaction of clinical risk factors with PVs in any gene or group of genes (all P values for interaction ≥0.05). The strongest results from interaction testing suggested a possibly greater risk reduction due to breastfeeding for PALB2 carriers versus noncarriers (P value for interaction = 0.053) and a potentially greater increase in risk due to E + P MHT for ER-positive gene carriers versus noncarriers (P value for interaction = 0.094).

We characterized breast cancer risk associated with established risk factors among WHI participants who carry PVs in BRCA1/2, ATM, CHEK2, and PALB2. There was evidence of reduced breast cancer risk with breastfeeding in PALB2 PV carriers, which may offer a practical option for risk modification. Whereas statistical power was limited, it was sufficient to rule out a twofold or greater increase in breast cancer risk for PV carriers due to obesity, alcohol consumption, tobacco use, and NSES. These results should in no way discourage public health messages to modify these risk factors but may offer reassurance that exposure does not substantially increase breast cancer risk (by twofold or higher) among PV carriers; however, more modest increases could not be ruled out. Moreover, in an exploratory analysis of grouping genes that were previously reported to predispose toward ER-positive disease, use of E + P MHT was modestly associated with high breast cancer risk in women with ATM or CHEK2 PVs.

Prior studies have focused on BRCA1/2 PV carriers and generally shown little or no increase in breast cancer risk associated with alcohol consumption, tobacco use, or obesity (7). Such prior findings contrast with those for the general population, in which alcohol consumption has been associated with relative risks in range of 1.1 to 1.2 (for one drink per day), tobacco use in range of 1.2 to 1.5, and postmenopausal obesity in range of 1.2 (30, 31). High parity has been associated with a lower risk of breast cancer (in range of 0.7) among BRCA1/2 PV carriers, and breastfeeding with a lower risk (in range of 0.6) among BRCA1 PVs carriers (32, 33), similar to the lower risks associated with high parity (in range of 0.7) and breastfeeding (in range of 0.8) in the general population (30, 31). Premenopausal salpingo-oophorectomy, recommended for BRCA1/2 PV carriers given their high ovarian cancer risk, has been variably associated with lower breast cancer risk (in range of 0.2–0.9) depending on the affected gene and age at surgery (34) and aligns with general population findings. MHT after premenopausal salpingo-oophorectomy in BRCA1/2 PV carriers has generally not been associated with higher breast cancer risk (35). Although this study lacked power to detect risk increases larger than twofold, its findings are largely consistent with prior reports, suggesting that most established breast cancer risk factors do relatively little to modify breast cancer risk in BRCA1/2 PV carriers.

Whereas studies have found little modification of BRCA1/2-associated risks, there has been evidence of greater risk modification with moderate penetrance genes, notably CHEK2 (36). Given the small number of PVs, we jointly analyzed two moderate penetrance genes, ATM and CHEK2, due to their shared association with ER-positive breast cancer in prior studies (4, 29); the result of this exploratory analysis was an increase in breast cancer risk (OR = 7.3) associated with E + P MHT. These findings should be considered hypothesis-generating, given their wide CI. However, they concur with the WHI randomized clinical trial results that showed higher breast cancer risk with E + P but not E-only MHT (37). Unlike BRCA1/2, PVs in ATM and CHEK2 do not confer sufficient ovarian cancer risk to warrant recommendation of risk-reducing salpingo-oophorectomy. Thus, MHT after natural menopause (as was studied in WHI) may be a clinically relevant option for ATM or CHEK2 PV carriers, and its potential impact on breast cancer risk should be further studied. Notably, however, WHI used a single dose of combined E + P for all clinical trial participants, which may not reflect the needs of women in different stages of menopause and differs from regimens typically used nowadays (38). Research on the potential interaction of modern MHT regimens with breast cancer risk due to PVs in ATM, CHEK2, and other genes should be prioritized.

This study has some limitations. As noted previously, the number of PV carriers was small, limiting statistical power. The WHI enrolled only women who had not developed breast cancer before menopause: thus, the findings may not apply to younger PV carriers. The sample had limited racial and ethnic diversity and, given its enrollment from 1993 to 1998, may not reflect the exposures, reproductive experiences, or, as noted above, MHT regimens of more recent birth cohorts. Study strengths include a large, well-characterized sample within the prospective WHI; more than 20 years of follow-up time; and confirmation of all cancer diagnoses, with careful ascertainment of risk factors, including type of MHT.

In summary, postmenopausal carriers of PVs in BRCA1/2, ATM, CHEK2, and PALB2 are unlikely to have substantial (twofold or greater) elevations in the risk of developing breast cancer due to reproductive and lifestyle exposures. However, additional research is warranted to determine whether MHT modifies the risk conferred by PVs in breast cancer susceptibility genes.

A.W. Kurian reports grants from Myriad Genetics during the conduct of the study, and grants from the National Institutes of Health, Breast Cancer Research Foundation and Susan G. Komen Foundaton outside the submitted work. E. Hughes reports other support from Myriad Genetics outside the submitted work. R. Bernhisel reports other support from Myriad Genetics outside the submitted work. E. Hu reports other support from Myriad Genetics outside the submitted work. S. Yadav reports other support from AstraZeneca and grants from Repare Therapeutics, Exelexis, and 858 Therapeutics outside the submitted work. J.L. Caswell-Jin reports grants from QED Therapeutics, Effector Therapeutics, and Novartis Therapeutics outside the submitted work. R. Chlebowski reports personal fees from UpToDate and Pfizer during the conduct of the study. P. Kraft reports grants from the NIH during the conduct of the study. F.J. Couch reports grants from the NIH and grants from 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.

A.W. Kurian: Conceptualization, resources, supervision, funding acquisition, investigation, methodology, writing–original draft. E. Hughes: Conceptualization, formal analysis, methodology, writing–original draft. R. Bernhisel: Formal analysis, methodology, writing–original draft. E. Hu: Formal analysis, methodology, writing–review and editing. E.C. Polley: Resources, investigation, writing–review and editing. S. Yadav: Investigation, writing–review and editing. C. Hu: Resources, investigation, writing–review and editing. J.L. Caswell-Jin: Investigation, writing–review and editing. E.M. John: Investigation, writing–review and editing. A.H. Shadyab: Resources, investigation, writing–review and editing. R. Chlebowski: Resources, investigation, writing–review and editing. R. Nassir: Resources, investigation, writing–review and editing. P. Kraft: Resources, investigation, writing–review and editing. M.L. Stefanick: Resources, investigation, writing–review and editing. F.J. Couch: Conceptualization, resources, formal analysis, funding acquisition, investigation, methodology, writing–original draft.

Myriad Genetics, Inc. (A.W. Kurian, E. Hughes, R. Bernhisel, E. Hu); Suzanne Pride Bryan Fund for Breast Cancer Research (A.W. Kurian); Jan Weimer Faculty Chair in Breast Oncology (A.W. Kurian); BRCA Foundation (A.W. Kurian); G. Willard Miller Foundation (A.W. Kurian); Carole and Alan Kushnir Fund (A.W. Kurian); and Nancy McDaniel PlayForHer Fund (A.W. Kurian). WHI: NHLBI, NIH, and the U.S. Department of Health and Human Services HHSN268201600018C, HHSN268201600001C, HHSN268201600002C, HHSN268201600003C, and HHSN268201600004C (A.W. Kurian, A.H. Shadyab, R. Chlebowski, R. Nassir, and M.L. Stefanick).