Polymorphisms in RAD51, XRCC2, and XRCC3 Are Not Related to Breast Cancer Risk

Highly penetrant, but rare, mutations in genes involved in double-strand break repair (i.e., BRCA1 and BRCA2) are associated with a risk for breast cancer of 40% to 65% by age 70 years (1, 2). Polymorphisms in other double-strand break repair genes are thought to contribute to the risk for the disease, either independently or through modifying the risk associated with rare mutations.

This study focuses on polymorphisms in three genes involved in the homologous recombination of double-strand breaks: RAD51 5′ untranslated region 135 G>C (rs1801320), X-ray repair cross-complementing group 2 (XRCC2) Arg¹⁸⁸His (rs3218536), and XRCC3 Thr²⁴¹Met (rs861539) in relation to breast cancer risk in the New York University Women's Health Study cohort.

The New York University Women's Health Study cohort collected questionnaires and blood samples from 14,274 healthy women ages 35 to 65 years in 1985 to 1991 (3). The current nested case-control study is matched for age and date at blood donation and includes incident cases of invasive breast cancer diagnosed before March 1998, with further methodologic details described by Shore et al. (4).

DNA was isolated using Qiagen QIAamp Blood Mini Kits (Qiagen, Inc.; ref. 4). Genotyping was done using PCR-RFLP methods described previously (ref. 4; see Appendix 1 for gene-specific PCR conditions and primer sequences). Blood clots and/or cell aggregates were available for 48% of the women. For the remaining women, serum specimens were used. Genotype results from clots/red cells and serum showed excellent concordance between repeated samples (n = 73) in pilot studies (97% for RAD51 135 G>C, 99% for XRCC2 Arg¹⁸⁸His, and 98% for XRCC3 Thr²⁴¹Met). Quality control duplicates showed 100% concordance for all three polymorphisms.

Deviation from Hardy-Weinberg equilibrium was assessed in controls using the χ² goodness-of-fit test. The relationship between genotype and breast cancer risk was evaluated using conditional logistic regression and the additive coding model. The dominant model was also assessed for RAD51 and XRCC2 because of the small number of individuals with the homozygous variant genotype. Tests for interaction between genotype and ethnicity, family history, body mass index, and smoking were planned a priori.

Given our sample size (612 cases and 612 controls) and the allelic frequencies in our population, we had sufficient power (99% for RAD51 135 G>C, 99% for XRCC2 Arg¹⁸⁸His, and 88% for XRCC3 Thr²⁴¹Met) to detect associations of the magnitude observed by Kadouri et al. (5) for RAD51 135 G>C and Kuschel et al. (6) for XRCC2 Arg¹⁸⁸His and XRCC3 Thr²⁴¹Met.

Genotype frequencies did not deviate from Hardy-Weinberg equilibrium (P > 0.5). Variant allele frequencies were comparable with those previously reported for populations of Caucasians of European descent for XRCC2 Arg¹⁸⁸His (8%; refs. 6-9) and XRCC3 Thr²⁴¹Met (36%; refs. 8-14), but the variant allele frequency for RAD51 135 G>C of 9% was somewhat lower than previous reports (5, 6, 9).

Table 1 describes study subject characteristics. As expected, significant differences in body mass index and parity/age at first full-term pregnancy were observed between cases and controls. However, these variables were not associated with genotype. Ethnicity was significantly associated with breast cancer risk and genotype. Asian and Hispanic women had a lower risk for breast cancer than non-Jewish White women (odds ratio, 0.49; 95% confidence interval, 0.29-0.81); this association is as expected (15). Ethnicity was significantly related to genotype for RAD51 GC/CC (P < 0.0001) and XRCC3 CT/TT (P < 0.0001) genotypes. Among Black women, 37.4% had at least one copy of the RAD51 135 G>C variant allele (non-Jewish White, 15.9%; Jewish White, 9.6%; others, 17.3%). The XRCC3 Thr²⁴¹Met variant was most common (67.3%) among Jewish White women (non-Jewish White, 60.3%; Black, 38.4%; others, 40.8%). XRCC2 Arg¹⁸⁸His variant was not significantly related to ethnicity.

Unadjusted and ethnicity-adjusted odds ratios and 95% confidence intervals are presented in Table 2. Although ethnicity was found to be related to genotype and risk, adjusting for ethnicity altered the odds ratios only slightly. In this population, none of the polymorphisms was found to influence breast cancer risk. The sum of variant alleles was also not related to risk (data not shown). Similar results were obtained when the analysis was restricted to Caucasians (data not shown). No significant interaction was found between genotype and ethnicity, body mass index, smoking, parity, or family history.

Genetic instability acquired through inefficient double-strand break repair is believed to be a component of breast cancer susceptibility. RAD51 plays a central role in homologous recombination, through direct interaction with XRCC2, XRCC3, BRCA1, BRCA2, etc., to form a complex essential for the repair of double-strand breaks and DNA cross-links (especially XRCC2 and XRCC3) and for the maintenance of chromosome stability (16).

Studies have suggested that RAD51 135 G>C modifies the breast cancer risk of women with a family history of breast cancer (17, 18) or carriers of BRCA2 mutations (5, 18-21). However, results have been inconsistent (22-24). Studies of non–BRCA2 mutation carriers or women without a family history have found no association between RAD51 135 G>C and breast cancer risk (5, 6).

Results for XRCC2 Arg¹⁸⁸His have been similarly mixed (6-8, 23). It is thought that this polymorphism has only a small effect on gene activity (7), although it may modify risk in those with low levels of plasma α-carotene (25) or plasma folate (26).

XRCC3 Thr²⁴¹Met has been found to be associated with increased DNA adducts (27), chromosomal deletions (28), and sensitivity to ionizing radiation and cross-linking agents (29, 30). Some (6, 17, 31) but not all (10, 23, 25, 32, 33) studies have found XRCC3 Thr²⁴¹Met to be related to an increased risk for breast cancer. Pooled analyses and meta-analyses show a small but significant increase in risk (8, 14, 22, 34).

Disruption of double-strand break repair is thought to contribute to carcinogenesis through the accumulation of genetic errors and genetic instability (35). However, in this study, the RAD51, XRCC2, and XRCC3 variants were found not to be associated with breast cancer risk. Unlike other reports, no relationship was found between RAD51 135 G>C and family history of breast cancer, perhaps because the participants in the study were not selected for having a family history of disease or being BRCA1/2 mutation carriers.