Germ-line mutations in the BRCA1 and BRCA2 genes confer a high lifetime risk of developing breast and other cancers. Estimates of the cumulative risk of breast cancer to age 70 years vary from 40% to 85% (Easton et al, 1995; Ford et al, 1998; Antoniou et al, 2003; Chen et al, 2006; Milne et al, 2008). Other genetic and/or environmental factors (modifiers) are likely to explain these differences, at least in part. Because of the large sample size required to identify such effects, few reliable associations have been reported to date, all coming from the Consortium of Investigators of Modifiers of BRCA1/2 (CIMBA) initiative, which was set up to provide large samples of BRCA1 and BRCA2 mutation carriers to reliably assess even modest associations with single nucleotide polymorphisms (SNPs) (Chenevix-Trench et al, 2007). Recently, CIMBA assessed three SNPs in the FGFR2, TNRC9 and MAP3K1 genes that had been previously found by a genome-wide association study to be associated with increased breast cancer risk for women in the general population (Easton et al, 2007). Consistent associations were found for BRCA1 and/or BRCA2 mutation carriers (Antoniou et al, 2008), indicating that SNPs involved in the susceptibility to develop breast cancer in the general population are good candidates to be tested as potential modifiers in BRCA1 and BRCA2 mutation carriers.

The ERCC4 gene is involved in the nucleotide excision repair (NER) pathway, which has led to the investigation of its role in the susceptibility to develop different types of cancer including breast cancer (Garcia-Closas et al, 2006; Mechanic et al, 2006; Moreno et al, 2006; Kiyohara and Yoshimasu, 2007; Hooker et al, 2008; Smith et al, 2008). We previously reported that the minor G allele in a SNP on intron 1 of ERCC4 (rs744154) was associated with protection from breast cancer in the general population (OR under a recessive model 0.61; P=0.0002) (Milne et al, 2006). On the basis of this finding, we aimed to assess ERCC4-rs744154 as a breast cancer risk modifier in BRCA1 and BRCA2 mutation carriers. The study was performed in two stages, the first comprising 837 mutation carriers from three CIMBA centres, and the second comprising 15 040 mutation carriers from all the CIMBA studies, including those used in stage I.

Materials and methods


Eligible subjects were female carriers of deleterious mutations in BRCA1 and BRCA2 aged 18 years or older. Further details regarding eligibility and the information collected from subjects are described elsewhere (Antoniou et al, 2008). Subjects who reported having ethnicity other than White European were excluded. This gave a total of 15 040 female mutation carriers (9408 with mutations in BRCA1 and 5632 with mutations in BRCA2), 8088 of whom had been diagnosed with breast cancer (4956 and 3132 with mutations in BRCA1 and BRCA2, respectively). All carriers participated in clinical or research studies at the host institution under ethically approved protocols.

A total of 34 collaborating CIMBA studies, carried out in 18 countries, contributed genotype data for ERCC4-rs744154 to this study. Details of each study along with the number of samples included from each are provided in Table 1. Seven studies (CBCS, GOG, ILUH, MSKCC, NNPIO, OSUCCG and IOVHBOCS) had not participated in previous CIMBA collaborations (Antoniou et al, 2008). Subjects from the CNIO, HEBCS and MBCSG studies were used in the first stage and comprised 837 mutation carriers (469 in BRCA1 and 368 in BRCA2). All 9408 BRCA1 and 5632 BRCA2 mutation carriers CIMBA subjects were included in the second stage.

Table 1 Number of BRCA1 and BRCA2 mutation carriers by study


The genotyping platform used by each study is detailed in Table 1. For 11 studies, matrix-assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF MS) was applied to determine allele-specific primer extension products using Sequenom's MassARRAY system and iPLEX technology (Sequenom, San Diego, CA, USA). The design of oligonucleotides was carried out according to the guidelines of Sequenom and performed using MassARRAY Assay Design software (version 3.1). One study determined genotypes by direct sequencing. Genotyping was carried out for the remaining studies by nuclease assay (Taqman). Taqman genotyping reagents were designed by Applied Biosystems (Foster City, CA, USA) ( as Assays-by-Design. Genotyping was performed using the ABI PRISM 7900HT, 7700 or 7500 Sequence Detection Systems according to the manufacturer's instructions. All studies complied with CIMBA genotyping quality control (QC) standards (Antoniou et al, 2008).

Statistical analysis

To test for departure from Hardy–Weinberg equilibrium, a single subject was randomly selected from each family and Pearson's χ2 Test (1 d.f.) was applied to genotypes from this sample set. The association of ERCC4-rs744154 with breast cancer risk was assessed by estimating hazard ratios (HR) and their corresponding 95% confidence intervals (CI) using weighted multivariable Cox proportional hazards regression with robust estimates of variance (Antoniou et al, 2005). For each mutation carrier, we modelled the time to diagnosis of breast cancer from birth, censoring at the first of the following events: bilateral prophylactic mastectomy, breast cancer diagnosis, ovarian cancer diagnosis, death and last date known to be alive. Subjects were considered affected if they were censored at breast cancer diagnosis and unaffected otherwise. The weighted cohort approach involves assigning weights separately to affected and unaffected individuals such that the weighted observed incidences in the sample agree with established estimates for mutation carriers (Antoniou et al, 2003). This approach has been shown to adjust for the bias in the HR estimates that is a consequence of the ascertainment criteria used (Antoniou et al, 2005), which leads to an over-sampling of affected women. Weights were assigned separately for carriers of mutations in BRCA1 and BRCA2 and by age interval (<25, 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70).

We considered log-additive and codominant genetic models and tested these by applying in the first case a Wald test based on the log-HR estimate per allele and its standard error, and in the second the χ2 equivalent of a Wald test (on 2 degrees of freedom [d.f.]) based on the log-HR estimates for heterozygotes (CG) and minor-allele-homozygotes (GG) vs common homozygotes (CC) and the corresponding variance–covariance matrix. Additional independent variables included in all analyses were year of birth (<1930, 1930–1939, 1940–1949, 1950–1959, 1960–1969, 1970), study centre and country. Heterogeneity in HRs by study centre was assessed by the χ2 test described above, but applied to interaction terms for the per-allele effect by centre (on 33 d.f.). A number of sensitivity analyses were applied, including censoring at bilateral prophylactic oophorectomy (BPO), adjusting for BPO (as a time-varying covariate), excluding data used in the initial analysis (from CNIO, HEBCS and MBCSG) and excluding prevalent cases, defined as those diagnosed >3 years before interview or DNA extraction. In addition, an alternative analysis based on a previously described retrospective likelihood approach (Chenevix-Trench et al, 2007) was also applied using the pedigree analysis software MENDEL (Lange et al, 1988).

All statistical analyses were carried out using Stata: Release 10 (StataCorp. 2007. Stata Statistical Software: Release 10.0. College Station, TX: Stata Corporation LP) unless otherwise stated. Robust estimates of variance were calculated using the cluster subcommand, applied to an identifier variable unique to each family.

Results and discussion

It is thought that any SNP involved in the susceptibility to develop breast cancer in the general population could also be a phenotypic modifier in carriers of mutations in the high-risk susceptibility genes BRCA1 and BRCA2. This has been confirmed in a recent report from CIMBA in which minor alleles in three SNPs in the FGFR2, TNRC9 and MAP3K1 genes, previously found to be associated with increased breast cancer risk in the general population (Easton et al, 2007), were found to increase breast cancer risk in BRCA1 and/or BRCA2 mutation carriers as well (Antoniou et al, 2008). We therefore aimed to investigate the role of the rs744154 SNP in ERCC4 as a potential BRCA1/BRCA2 risk modifier, based on our earlier finding that the minor G allele was associated with breast cancer protection in the general population (OR under a recessive model 0.61; P=0.0002) (Milne et al, 2006).

This study was performed in two stages, the first analysing the SNP in 837 carriers of mutations (469 in BRCA1 and 368 in BRCA2) from three CIMBA studies (CNIO, HEBCS and MBCSG). Results of the first stage are summarized in Table 2. We observed a marginally significant association of the G allele in ERCC4-rs744154 with breast cancer risk for both BRCA1 (HR: 0.78, 95% CI: 0.60–1.00, P=0.05) and BRCA2 (HR: 0.68, 95% CI: 0.45–1.02, P=0.06) mutation carriers. As this result was consistent with our previously reported protective association in the general population (Milne et al, 2006), we genotyped this SNP in subjects from the remaining CIMBA studies and repeated the analysis in the combined series of 9408 BRCA1 and 5632 BRCA2 mutation carriers (see Table 2). However, when the whole CIMBA series was analysed (stage II), there was no longer evidence of an association with breast cancer risk for either BRCA1 (HR: 0.98, 95% CI: 0.93–1.04, P=0.5) or BRCA2 (HR: 0.97, 95% CI: 0.89–1.06, P=0.5) mutation carriers. Several sensitivity analyses were performed (see Materials and Methods) but results did not change substantially and so those from the main analysis are presented in this report.

Table 2 Genotype frequencies of ERCC4-rs744154 by mutation and disease status and hazard ratio estimates from stages I and II

Subsequent to the initiation of this study, a study from the Breast Cancer Association Consortium (BCAC), performed in >30 000 breast cancer cases and 30 000 controls has now found no evidence for an association of rs744154 with breast cancer risk in the general population (Gaudet et al, 2009). This is consistent with the lack of association found for BRCA1 and BRCA2 mutation carriers. These findings highlight the necessity of very large collaborative efforts to obtain reliable conclusions in genetic association studies. On the basis of the combined results obtained from the BCAC analysis of sporadic breast cancer and the CIMBA analysis of BRCA-related breast cancer – each the largest study of its kind – there no longer seems to be convincing evidence that ERCC4-rs744154 is associated with breast cancer risk.