Introduction

Rare germline mutations in two breast cancer susceptibility genes, BRCA1 [1] and BRCA2 [2] account for 3–8% of all cases of breast cancer and 15–20% of familial cases. The penetrance of these mutations is highly variable between and within carrier families [3–5], suggesting that other environmental and genetic variants may act as modifying factors [6]. What is less clear is whether more common (presumably low penetrance) variants in BRCA1 or BRCA2 modify the risk of breast cancer in carriers of clearly deleterious mutations in BRCA1 or BRCA2 or are associated with risk of breast cancer in non-BRCA1 or BRCA2 carriers.

Results from studies of common genetic variants in BRCA1 and BRCA2 as potential risk factors for primary breast cancer have been largely inconsistent. Cox et al. used four haplotype tagging (ht)-single-nucleotide polymorphisms (SNPs) in BRCA1 and identified one haplotype associated with a significant increased risk in the Nurses’ Health Study [7], while Freedman et al. examined 9 tagSNPs in BRCA1 among women in the Multi-ethnic Cohort Study and found no significant effects [8]. For BRCA2, Freedman et al. also used 21 tagSNPs in BRCA2 and found one (rs206340) to be associated with an increase in risk for the homozygous genotype compared to the wildtype [9]. The British East Anglian SEARCH Study, which used controls from the European Prospective Investigation in Cancer and Nutrition (EPIC)-Norfolk Study evaluated five SNPs in BRCA1 and 16 SNPs in BRCA2 individually and in combination, finding no associations with breast cancer risk [10]. In addition, several studies have examined single missense polymorphisms in BRCA1 and BRCA2 in relation to breast cancer risk with conflicting evidence including BRCA2-N372H [11–15] and BRCA1-Q356R [7, 16, 17]. None of these studies accounted for BRCA1 or BRCA2 mutation status. Some evidence suggests that mutations in BRCA1 occur more often on specific haplotypes of BRCA1 [18], but the potential risk of common variants in these genes in carriers is not known.

In this study, we investigate the association between common SNPs in BRCA1 and BRCA2 and risk of developing asynchronous contralateral breast cancer (CBC) among women with a prior unilateral breast cancer (UBC) whose carrier status for deleterious mutations in BRCA1 and BRCA2 is known.

Materials and methods

Study population

The WECARE (Women’s Environmental Cancer and Radiation Epidemiology) Study has been previously described [19]. In brief, the WECARE Study is a multi-center, population-based case–control study of women with CBC (cases) and women with UBC only (controls). All participants were identified by five population-based cancer registries in the United States (Seattle, Iowa, Orange County/San Diego, and Los Angeles County) and Denmark (Danish Cancer Registry). CBC cases met the following eligibility requirements: (1) were diagnosed between 01/01/1985 and 12/31/2000 with a first primary invasive breast cancer that did not spread beyond the regional lymph nodes at diagnosis and a second primary in situ or invasive breast cancer diagnosed in the contralateral breast at least 1 year after the first breast cancer diagnosis (reference date); (2) resided in the same study reporting area for both diagnoses; (3) had no previous or intervening cancer diagnosis except squamous cell or basal cell skin cancer; (4) were able to provide informed consent to complete the interview; (5) provided a blood sample; and (6) were under age 55 years at the time of diagnosis of the first breast cancer. The “at-risk” time period was defined as the number of days between the case’s first and second breast cancer diagnoses.

Two control subjects were individually matched to each case on year of birth, year of diagnosis, registry region, and race, and were 1:2 counter-matched on registry-reported radiation exposure (for treatment of first primary breast cancer) so that each triplet consisted of one radiation-untreated and two radiation-treated subjects. Eligible UBC controls were (1) diagnosed between 01/01/1985 and 12/31/1999 with a first primary invasive breast cancer that did not spread beyond the regional lymph nodes at diagnosis; (2) resided in the same study reporting area as the case to which they were matched at diagnosis and at a reference date equal to the date of diagnosis plus the at-risk period; (3) had no diagnosis of any other cancer (other than squamous cell or basal cell skin carcinoma or cervical carcinoma in situ) prior to their breast cancer diagnosis or during the at-risk period; (4) were alive and able to provide informed consent to complete the interview; (5) provided a blood sample; and (6) had not had a prophylactic mastectomy of the contralateral breast prior to or within the at-risk period.

A total of 708 CBC cases participated in the study (of 998 eligible cases) and 1,399 women with UBC participated as controls (of 2,112 eligible women with UBC). Reasons for non-participation included: physician refusal, subject interview refusal, and subject blood draw refusal. For the current study, we restricted eligible women to Caucasians (649 eligible CBC cases and 1,288 UBC controls).

All participants were interviewed by telephone using a structured questionnaire that collected information on personal demographics, medical history, family history, reproductive history, body size, smoking, alcohol intake, oral contraceptive use, and post-menopausal hormone use. For all participants, medical records, pathology reports, and hospital charts, in addition to self-reported data, were used to collect detailed treatment information. Information on tumor characteristics of the first primary breast cancer was collected from medical records and cancer registry records.

The study protocol was approved by the Institutional Review Boards at each study site and by the ethical committee system in Denmark.

BRCA1 and BRCA2 mutation screening

Screening for mutations in BRCA1 and BRCA2 [20] was performed in all cases and controls, as previously described [21]. In brief, coding and flanking intronic regions were screened for mutations or polymorphic variants by denaturing high-performance liquid chromatography (DHPLC). All PCR-amplified fragments with variant heteroduplex output traces were sequenced to confirm the nucleotide variants. Three laboratories performed the screening using fixed sets of primers and DHPLC protocols. Consistency in screening between and within laboratories was assured via a rigorous laboratory quality control plan previously described [22]. We classified mutations as deleterious according to the guidelines provided by the Breast Cancer Information Core (http://research.nhgri.nih.gov/projects/bic/). Sequence variants classified as having a clearly deleterious effect included changes known or predicted to truncate protein production, splice site mutations located within 2 bp of an intron/exon boundary or shown to cause aberrant splicing, and missense changes with known deleterious functional effects. Of the eligible Caucasian women, four (three CBC cases and one UBC control) did not consent to screening for deleterious mutations in BRCA1 and BRCA2. Excluding these three cases and their respective matched controls (total n = 9) and the one non-consenting UCB control, 646 bilateral cases and 1,281 unilateral controls remained eligible.

Genotyping polymorphisms in BRCA1 and BRCA2

DNA samples were genotyped with Illumina’s HumanOmni1-Quad BeadChip. Samples with GenCall scores <0.15 at any locus were considered “no calls.” Each 96-well plate included one inter-plate positive quality control samples (NA06990, Coriell Cell Repositories). In addition, 38 blinded and 46 un-blinded quality control replicates from the study sample were genotyped. SNP data obtained from both the Coriell and study sample replicates showed a high concordance rate of called genotypes: >99.99% (for un-blinded and blinded study replicates that had call rates >90% for both replicates; inter-plate positive control—not determined). Default Omni1-Quad cluster definitions supplied by Illumina were used to call genotypes. The genotype clusters were visually inspected for all SNPs with significant associations.

Of the 1,927 eligible Caucasian women, four individuals were excluded because they did not consent to genome-wide association study (GWAS) genotyping beyond the initial BRCA1 and BRCA2 mutation screening. A series of quality control steps were applied leading to further subject exclusions: (1) women with SNP call rates <95% were excluded (n = 22); (2) population stratification was investigated using EIGENSTRAT [23]; using the first two principal components, 22 outliers (eight of which were CBC cases) were identified for exclusion; and (3) identity by descent was examined using PLINK [24] identifying three pairs of sisters, including one pair of identical twins. Following these exclusions, 26 additional participants were excluded due to incomplete matched sets. Analyses are based on the remaining 1,853 participants (632 CBC cases and 1,221 UBC controls).

In this report, we focus on common variants in BRCA1 and BRCA2. We defined the region for BRCA1 as base pairs 38444655–38545479 and for BRCA2 as base pairs 31782845–31871439 (Genome Build 36.2).

We performed additional quality control steps by comparing genotypes of common variants in BRCA1 and BRCA2 assayed on the Illumina Omni1-Quad platform with those we identified during DHPLC screening for deleterious mutations. Nine common (MAF > 0.05) SNPs in BRCA1 and six common SNPs in BRCA2 were identified. As homozygotes were not reliably determined in some instances from DHPLC screening, these common SNPS were also genotyped on a custom oligonucleotide probe panel using the Illumina GoldenGate™ assay on the Sentrix Array Matrix (Illumina, Inc., San Diego). A high concordance rate between results on the Illumina GoldenGate and Omni1-Quad (>96%) was observed for all SNPs.

SNPs were excluded from analysis if they had a MAF < 0.05 (n SNPs = 73). The call rate for all SNPs was >96% on the Omni1-Quad. In total, we included 1,853 samples and 52 SNPs (22 in BRCA1 and 30 in BRCA2) in the final analysis. The 22 SNPs in BRCA1 and 30 in BRCA2 captured 97 and 95% of all common variants in these genes, respectively, at r 2 > 0.80 using HapMap Phase II release 24.

Statistical analysis

We conducted single SNP and haplotype-based analyses. To investigate the association between each SNP or haplotype and the risk of CBC, we used conditional logistic regression analysis, including a log weight “covariate” in the model where the coefficient of this log weight was fixed at one (i.e., an “offset” in the model). The weights were created using the numbers of registry-reported radiation-treated and radiation-untreated women in the risk set to account for the counter-matched sampling design [19, 25]. All models were also adjusted for exact age at first breast cancer diagnosis (continuous). For individual SNP analysis, we considered both a log-additive and co-dominant model.

To investigate potential heterogeneity by BRCA1 and BRCA2 deleterious mutation carrier status, we included interaction terms between each SNP and mutation status indicator variables. To investigate heterogeneity in risk estimates across mutation carriers and non-carriers, we used a likelihood ratio test comparing the carrier status specific model to a model including only the main effect for the SNP.

For the haplotype-based analyses, we estimated the haplotype effects among all individuals irrespective of BRCA1 and BRCA2 mutation carrier status and among known non-BRCA1 and BRCA2 carriers. We were unable to estimate haplotype effects in BRCA1 and BRCA2 carriers due to small cell counts. Haplotype frequencies were estimated with the EM algorithm, as implemented in R version 2.10.1 (http://mayoresearch.mayo.edu/mayo/research/schaid_lab/software.cfm). The most common haplotype was used as the reference and rare haplotypes (combined frequency <0.5%) were combined into a single group.

When necessary, to account for missing information within a counter-matched set, a missing indicator variable was included in the conditional logistic regression models [26]. All statistical tests are two-sided. SAS release 9.2 (SAS Institute, Cary, NC) was used for the analyses.

Results

Among the WECARE Study Caucasian women, a total of 148 women carried a deleterious mutation in BRCA1 (n = 88) or BRCA2 (n = 60) (Table 1). BRCA1 and BRCA2 mutation carriers were more likely to be younger at first breast cancer diagnosis than non-carriers, but did not differ by time at risk. Women with CBC were more likely to have a family history of breast cancer and this proportion was higher among BRCA1 and BRCA2 carriers compared to non-carriers. BRCA1 carriers were younger at diagnosis of their first breast cancer and were more likely to be estrogen receptor-negative and progesterone receptor-negative compared to BRCA2 carriers and non-carriers.

Table 1 Selected characteristics of the BRCA1 and BRCA2 carriers and non-carriers in the WECARE Study
Full size table

Among all women, there were no significant associations with CBC risk for any single-SNP in BRCA1 using either the log-additive or co-dominant model (Table 2). In the log-additive model, rs1060915 was associated with a non-significant increased risk of CBC in BRCA1 carriers (per G allele, OR 1.96; 95% CI 0.91–4.26), no association in BRCA2 carriers and an inverse association in non-BRCA1 and BRCA2 carriers (OR 0.84; 95% CI 0.70–1.00). Similar results were observed for two additional SNPs, rs8176318 and rs16940, both in complete linkage disequilibrium (LD) with rs1060915 (r 2 = 1.0, r 2 = 1.0, respectively). No association with CBC was observed for any synonymous or non-synonymous amino acid coding SNPs including BRCA1-Q356R (rs1799950).

Table 2 Association between SNPs in BRCA1 and contralateral breast cancer risk by BRCA1 and BRCA2 mutation status
Full size table

For BRCA2-rs11571686, we found that the variant homozygous genotype (C/C) was significantly associated with an increased risk of CBC when compared to the wildtype homozygous genotype (OR 2.75; 95% CI 1.17–6.48, Table 3). When stratified by deleterious mutation status, we observed that the increased risk associated with this SNP was restricted to non-carriers (OR 1.32, 95% CI 1.04–1.68); there was no evidence of an association in BRCA1 carriers (OR 0.60; 95% CI 0.18–2.18) or BRCA2 carriers (OR 0.27; 95% CI 0.04–1.87).

Table 3 Association between SNPs in BRCA2 and contralateral breast cancer risk by BRCA1 and BRCA2 mutation status
Full size table

Among BRCA1 mutation carriers, we found evidence that two SNPs in BRCA2 were significantly associated with the risk of CBC (Table 3). BRCA2-rs206115 was inversely associated with risk (OR 0.41; 95% CI 0.19–0.87) and BRCA2-rs206117 was associated with an increased risk (OR 2.46; 95% CI 1.10–5.48). These two SNPs are in modest LD (r 2 = 0.61). We found no evidence of an association between BRCA2-N372H (rs144848), V1269V (rs543304) or S2414S (rs17999555) and risk of CBC.

Combining CBC and UBC women, we estimated six common haplotypes with frequency ≥0.05 in BRCA1 and six in BRCA2 (Table 4). We found no association between these BRCA1 or BRCA2 haplotypes and risk of CBC among all women (Table 5). When we excluded carriers of deleterious BRCA1 or BRCA2 mutations, results suggested a lower risk of CBC among women who carried haplotype_3.

Table 4 Common haplotypes among all women in the WECARE Study
Full size table
Table 5 Association between haplotypes in BRCA1 and BRCA2 and contralateral breast cancer risk
Full size table

Discussion

In this large case–control study designed to determine genetic factors that increase risk of CBC, we observed that the association of common variants in BRCA1 and BRCA2 with CBC risk may be modified by BRCA1 and BRCA2 deleterious mutation carrier status. Selected SNPs in BRCA1, rs8176318, rs1060915 and rs16940, appeared to confer an increased risk of CBC among women who carried known deleterious mutations in BRCA1, but were associated with a lower risk in non-BRCA1 and BRCA2 carriers. BRCA2-rs11571686 conferred an increase risk among non-BRCA1 and BRCA2 carriers, and two other SNPs in BRCA2, rs206115 and rs206117, were associated with risk only in BRCA1-mutation carriers. We found no evidence that common haplotypes of BRCA2 were significantly associated with risk of CBC among all women or among non-carriers. However, we identified one common BRCA1 haplotype that was inversely associated with risk, but only among non-BRCA1 and BRCA2 carriers.

BRCA1 and BRCA2 are involved in the detection and repair of DNA double-strand breaks in response to DNA damage [27]. Selected mutations in these genes have been classified as deleterious, but the role of common variants is less clear and to date has only been examined in relation to the risk of first primary breast cancer. Our findings suggest that selected common variants in BRCA1 may contribute to the risk of CBC among BRCA1 carriers, although our results did not reach statistical significance. The risk estimates for CBC associated with being a deleterious BRCA1 mutation carrier are significantly higher than those among non-carriers [28] and the penetrance of selected BRCA1 deleterious mutations may in part be modified by common allelic variants.

A few studies have used LD patterns to define haplotype blocks that describe the majority of common variation in BRCA1 in Caucasian women. Cox et al. genotyped four haplotype (ht)-SNPs in BRCA1 and identified one haplotype associated with an increased risk of breast cancer among Caucasians in the Nurses’ Health Study (OR 1.18; 95% CI 1.02–1.37) [7]. Freedman et al. examined 9 tagSNPs in both coding and non-coding regions of BRCA1 across different ethnic/racial groups in the Multiethnic Cohort Study and found no evidence of association with the risk of a first primary invasive breast cancer [8]. Furthermore, results from the British East Anglian SEARCH Study of Caucasian women found no association between any individual or combination of the 5 tagSNPs in BRCA1 and breast cancer risk [10]. Unlike these previous studies, we focused on CBC in the WECARE Study and accounted for BRCA1 and BRCA2 mutation carrier status. We identified one common haplotype that was associated with a lower risk of CBC in non-BRCA1/BRCA2 deleterious mutation carriers that requires further confirmation in other studies.

Several studies have examined the role of BRCA1-Q356R in the risk of developing breast cancer. Johnson [17] found that this SNP was associated with an increased risk of first primary (OR 1.31; 95% CI 1.14–1.51) and second primary breast cancer (OR 1.72, P = 0.0002). In a review by Dunning et al., this SNP had previously been described as being inversely associated with breast cancer risk [16], but this was not replicated in the Nurses’ Health Study [7] or in the current analysis of CBC in the WECARE Study.

For BRCA2, Freedman et al. examined 21 tagSNPs in BRCA2 in a breast cancer case–control study nested within the Multiethnic Cohort Study and found one (rs206340) to be associated with an increase in risk for the homozygous genotype compared to the wildtype (OR 1.59; 95% CI 1.18–2.16) [9]. In our study, BRCA2-rs543304 was not associated with CBC. In addition, a few studies have suggested that BRCA2-1342A>C/N372H (rs144848) is associated with an increased risk of breast cancer [12, 13], but not other cancers [11, 14, 15]. Our results also do not support an association between BRCA2-N372H and risk of CBC. We did not find that any of the common haplotypes of BRCA2 were associated with risk, in agreement with the British East Anglian SEARCH Study of Caucasian women [10]. However, when we stratified by BRCA1 and BRCA2 carrier status, we observed that BRCA2-rs11571686 conferred an increase risk among non-BRCA1 and BRCA2 carriers, and two other SNPs in BRCA2, rs206115 and rs206117, were significant risk alleles in BRCA1-mutation carriers.

Recent results from genome-wide association studies suggest that identified susceptibility loci may have different effects in BRCA1 and BRCA2 carriers compared to non-carriers [29]. These results may partially explain the variability in risk among carriers [6]; however, published studies examining haplotypes in BRCA1 and BRCA2 did not examine potential heterogeneity by carrier status.

In spite of the strengths of this study, there were some limitations. We considered only common variants occurring at a frequency of at least 5% in the population and we may have missed low-frequency haplotypes that are subtypes of the major haplotypes. The relevance of rare alleles may be critical and merits further investigation. Further, we are limited by the small number of BRCA1 and BRCA2 deleterious mutation carriers, which precluded more detailed analyses with carrier status. Our statistical power to detect modest SNP effects among BRCA1 and BRCA2 carriers is limited.

This is the first study to investigate the role of common variants in BRCA1 and BRCA2 in the development of CBC by BRCA1 and BRCA2 mutation carrier status. We have used both a comprehensive tagging and haplotype-tagging approach to account for the majority of common variants in BRCA1 and BRCA2. We have also implemented strong quality control procedures to ensure accurate genotyping of BRCA1 and BRCA2. We have limited the potential for population stratification by restricting our analysis to Caucasians and matching by center/country of origin. Further, by ascertaining cases and controls through population-based cancer registries, our study avoided biases associated with use of high-risk or selected populations thereby increasing the generalizability of the findings.

In conclusion, we provide evidence that selected common alleles of BRCA1 and BRCA2 may modify risk of CBC conditional on BRCA1 and BRCA2 mutation carrier status. These results add to a growing literature suggesting that risk may be influenced by multiple loci with modest effects and may help explain the variability in penetrance estimates for BRCA1 and BRCA2 mutations in carrier families [3–5].