Associations of ATR and CHEK1 single nucleotide polymorphisms with breast cancer

Abstract

DNA damage and replication checkpoints mediated by the ATR-CHEK1 pathway are key to the maintenance of genome stability, and both ATR and CHEK1 have been proposed as potential breast cancer susceptibility genes. Many novel variants recently identified by the large resequencing projects have not yet been thoroughly tested in genome-wide association studies for breast cancer susceptibility. We therefore used a tagging SNP (tagSNP) approach based on recent SNP data available from the 1000 genomes projects, to investigate the roles of ATR and CHEK1 in breast cancer risk and survival. ATR and CHEK1 tagSNPs were genotyped in the Sheffield Breast Cancer Study (SBCS; 1011 cases and 1024 controls) using Illumina GoldenGate assays. Untyped SNPs were imputed using IMPUTE2, and associations between genotype and breast cancer risk and survival were evaluated using logistic and Cox proportional hazard regression models respectively on a per allele basis. Significant associations were further examined in a meta-analysis of published data or confirmed in the Utah Breast Cancer Study (UBCS). The most significant associations for breast cancer risk in SBCS came from rs6805118 in ATR (p=7.6 x 10(-5)) and rs2155388 in CHEK1 (p=3.1 x 10(-6)), but neither remained significant after meta-analysis with other studies. However, meta-analysis of published data revealed a weak association between the ATR SNP rs1802904 (minor allele frequency is 12%) and breast cancer risk, with a summary odds ratio (confidence interval) of 0.90 (0.83-0.98) [p=0.0185] for the minor allele. Further replication of this SNP in larger studies is warranted since it is located in the target region of 2 microRNAs. No evidence of any survival effects of ATR or CHEK1 SNPs were identified. We conclude that common alleles of ATR and CHEK1 are not implicated in breast cancer risk or survival, but we cannot exclude effects of rare alleles and of common alleles with very small effect sizes.

Figure 1. Association plots of the 32 typed SNPs and 454 imputed variants in the ATR region.

The most significant signal (rs6805118, p=7.6x10^-5) is labelled with the large blue circle. Circle symbols stand for the typed SNPs and diamond symbols represent the imputed variants. LD associations (r²) with rs6805118 were calculated in 85 CEU and 89 GBR (integrated call release as of 2010-11-23) in the 1000 genomes project. Gene transcripts are indicated by the dark green lines, with right arrowhead for the “+” strand and left arrowhead for the “-” strand. Recombination rate (blue line) was obtained from HapMap II.

Figure 2. Association plots of the 44 typed SNPs and 434 imputed variants in the CHEK1 region.

The most significant signal (rs2155388, p=3.1x10^-5) is labelled with the large blue circle. The typed SNPs and the imputed variants are shown in the circle and diamond symbols, respectively. Pair-wise r² with rs2155388 were calculated in 85 CEU and 89 GBR (integrated call release as of 2010-11-23) in the 1000 genomes project. Gene transcripts are indicated by the dark green lines, with right arrowhead for the “+” strand and left arrowhead for the “-” strand. Recombination rate (blue line) was obtained from HapMap II.

Figure 3. Forest plots of the most significant ATR and CHEK1 SNP associations.

(A) the ATR rs6805118 associations. SEARCH: Studies of Epidemiology and Risk Factors in Cancer Heredity [8]; NHS II: Nurses’ Health Study (premenopausal women) [9]; SBCS: Sheffield Breast Cancer Study (B) the CHEK1 rs2155388 associations. UBCS: Utah Breast Cancer Study. For each panel, fixed effect estimate (pooled OR) is shown, with p value for homogeneity (the Cochran’s Q test, p_het) and I-squared for the amount of heterogeneity in parenthesis. DL pooled OR stands for the random effect model derived from the DerSimonian-Laird estimator.

Figure 4. Meta-analysis of the association of rs1802904 (ATR region) with breast cancer risk.

Fixed effect estimate (pooled OR) is shown, with p value for homogeneity (the Cochran’s Q test, p_het) and I-squared for the amount of heterogeneity in parenthesis. DL pooled OR stands for the random effect model derived from the DerSimonian-Laird estimator. MEC: Multiethnic Cohort Study [7]; SEARCH: Studies of Epidemiology and Risk Factors in Cancer Heredity [8]; NHS II: Nurses’ Health Study (postmenopausal women) [22]; SBCS: Sheffield Breast Cancer Study.