Tag-SNP analyses of activin receptors with risk of breast cancer and breast density.

A191

Introduction : The TGF-B family of polypeptide growth factors plays an important role in normal mammary gland development and function. Activin and its receptors are part of this pathway. We examined the association of nonsynonymous coding SNPs and tagSNPS in four activin receptors (ACVRL1, ACVR1, ACVR1B and ACVR2) with breast cancer (BC) as well as with breast density in a case-control study. Methods : Incident cases of invasive BC were identified at the Mayo Clinic in Rochester, MN from 2001 through 2005. Caucasian controls (N=843) were frequency matched to cases (N=798) on age and region of residence. Percent density (PD) was estimated using a thresholding algorithm on those controls for whom mammograms were available (N=686). Genotype data were downloaded from public databases. LDSelect (Carlson, 2004) was utilized with parameters of MAF>0.05 and minimumr²=0.8 to select tagSNPs. Genotyping was performed on an Illumina BeadLab using the Illumina GoldenGate genotyping assay. Logistic regression was used to calculate odds ratios and 95% confidence intervals (CI) for BC. Linear regression was used to estimate mean and 95% CI for PD for each copy of the minor allele. All models were adjusted for age, residence, and confounding factors. A global test of significance for the entire gene was performed for both BC and PD using multiple regression (logistic/linear, respectively) or haplotype analyses. Results : The mean age of cases and controls was 56.6 and 57.6, respectively. 483 (64%) cases and 579 (72.5%) controls were postmenopasual at the time of questionnaire. Two (of 18) SNPS in ACVR1 were associated with risk of BC in a dominant manner. The first, located in intron 3 was associated with decreased risk of BC (OR=0.75; 0.58-0.99); the second, located in intron 1, was associated with increased risk of BC (OR=1.24; 1.00-1.53). No SNPs in the other three genes were associated with risk of BC. Variation in none of the four genes were significant when tested at the gene level. Five SNPs in ACVR1 (not overlapping with those for breast cancer above) were associated with increased PD, however there was no association at the gene level. ACVRL1 was significantly associated with PD at the gene level. Two variants in this gene were significantly associated with PD. Both displayed a U-shaped association, with the highest breast density found in women with 1 copy of the common allele. One SNP in the 5'UTR of ACVR1B was significantly associated with PD (mean PD=21.2, 20.0-22.3; 25.3, 22.5-28.1; and 22.4, 11.1- 33.7 for 0, 1 and 2 copies of the variant allele, respectively.) This gene also was not associated with PD when tested at the gene level. Conclusion : These analyses suggest that SNPs in ACVRL1, which is known to cause increased endothelial cell proliferation, are associated with PD. In addition, there is a possible association with ACVR1 SNPs and risk of BC.