|Home | About | Journals | Submit | Contact Us | Français|
Recent genome-wide association studies (GWAS) have identified multiple common genetic risk variants for breast cancer among women of Asian and European ancestry. Investigating these genetic susceptibility loci in other populations would be helpful to evaluate the generalizability of the findings and identify causal variants for breast cancer. We evaluated 11 GWAS-identified genetic susceptibility loci for breast cancer in a study including 2,594 African-American women (810 cases and 1,784 controls). Two single-nucleotide polymorphisms (SNPs), rs13387042 (2q35) and rs1219648 (FGFR2 gene), were found to be associated with breast cancer risk. Risk increased nearly linearly with number of affected risk alleles, with a two-fold elevated risk for women homozygous for the risk alleles in both SNPs. No additional significant associations, however, were identified for the other 9 loci evaluated in the study. The results from this study extend some of the recent GWAS findings to African Americans and may guide future efforts to identify causal variants for breast cancer.
Recent genome-wide association studies (GWAS) conducted among Chinese women (1) or women of European ancestry have identified multiple single-nucleotide polymorphisms (SNPs) associated with breast cancer risk (2–6). It is unclear, however, whether these genetic variants may also be associated with breast cancer risk among African-American women whose genetic architecture differs considerably from other ethnic groups. In this paper we evaluated 11 GWAS-identified loci in relation to breast cancer risk in African-American women using data from two ongoing studies, the Southern Community Cohort Study (SCCS) and the Nashville Breast Health Study (NBHS).
The SCCS is a prospective cohort study initiated in 2002 focusing on investigating racial disparities in cancer risk (7). Included in the current project were 527 women who reported having a prior breast cancer diagnosis and provided a blood or buccal cell sample at the baseline interview. Controls (n =1,054) were selected randomly from those who were cancer-free at baseline and frequency-matched to cases in a 2 to 1 ratio on age at enrollment (+/− 1 year), recruitment method, and sample type (blood/buccal cell). The NBHS, a population-based case-control study (1), provided additional subjects to the current project. Incident breast cancer cases were identified through the Tennessee State Cancer Registry and a network of major hospitals that provide medical care for breast cancer patients. Controls were identified via random digit dialing of households in the same geographic area as cases and frequency-matched to cases on age (5-year group). Included in the current project were 291 cases and 178 controls who provided buccal cell samples. To increase the statistical power of the study, additional controls (n=564) were randomly selected from cancer-free SCCS participants and frequency-matched to NBHS cases by age (± 1 year), family income, and education. In total, 810 cases and 1,784 controls from the SCCS and NBHS were included in the current study after excluding subjects whose DNA samples were limited.
In addition to the 12 SNPs initially reported from GWAS (Table 1), 20 additional SNPs were selected to tag all common SNPs (minor allele frequency, MAF≥5%) in a ±50kb region flanking each of the initially-reported SNPs. These common SNPs were identified from the HapMap data (release 24) that are in high linkage disequilibrium (LD, r2≥0.8) with each of the initially-reported SNPs in Chinese (for rs2046210) or European descendants (for all other SNPs) but not in Africans (r2<0.8). For the FGFR2 gene, two additional SNPs (rs2981578 and rs7895676) were considered, as these SNPs were reported to be potentially functional (8). SNP rs7895676, however, was found to be non-polymorphic and was excluded from the analysis. No tagging SNPs were found for 5p12/MRPS30, 11p15.5/LSP1, and 16q12.1/TOX3 (1) loci that meet the above-mentioned criteria. All 33 SNPs were successfully genotyped.
To adjust for population stratification, we selected the top 30 SNPs from 276 ancestry-informative markers (AIM) that were previously genotyped in 2,552 SCCS participants. These 30 SNPs show the largest difference in allele frequency between European and African descendants and correctly classified 98.7% of the participants in the previous study (data not yet published). Twenty-eight of these SNPs were successfully genotyped, and population structure was estimated using the principal component method (9).
With the exception of five SNPs (rs2981578, rs6929137, rs851974, rs8051542, and rs7895676) that were genotyped using TaqMan assays, all other SNPs were genotyped using Sequenom. Quality control samples were included and showed 99.5 to 100% concordance rates.
Logistic regression models were used to estimate odds ratios (OR) and 95% confidence intervals (CI) associated with each of the 33 SNPs in the 11 previously-reported loci adjusting for age, education, study (SCCS or NBHS), and the first 4 principal components defined by the 28 AIMs.
Of the 33 SNPs, only two, rs13281615 (8q24.21) and rs12598982 (16q12.1), were found to deviate from the Hardy-Weinberg equilibrium at p≤0.05. Significant associations (p<0.05) with breast cancer risk were found for rs13387042 (2q35, p=0.02) and rs1219648 (FGFR2 gene, p=0.004) (Table 2). The pattern of the association for these two SNPs tended to be consistent in the SCCS and NBHS, although some of the study-specific point estimates were not statistically significant (Table 3). To evaluate the combined effect of these two SNPs, a genetic score was constructed by counting the number of risk alleles each woman carries. A dose-response association was observed between the number of risk alleles and risk of breast cancer (p for trend, 0.004), with a 1.2-fold increase per risk allele and a two-fold elevated risk observed for women homozygous for the risk alleles in both SNPs (Table 3). Having one or more risk alleles of these two SNPs was associated with a population attributable risk (PAR) of 37.9%.
To our knowledge, this is the first study in African Americans that has systematically evaluated breast cancer susceptibility loci recently-identified through GWAS. Results from our study are helpful to assess the generalizability of previously-identified associations and guide fine-mapping efforts to search for causal variants. Of the 33 SNPs evaluated, 12 SNPs reported from previous GWA studies and 21 tagging SNPs, only two, in two different loci, were found to be significantly associated with breast cancer risk. This finding perhaps is not surprising, given the large difference in genetic architecture between African Americans and women of Chinese or European descent.
Similar to our study, a significant association between SNPs in the FGFR2 gene and breast cancer risk was also reported from a recent study conducted in 1,250 cases and 1,245 controls of African descent (10). On the other hand, rs3803662 at 16q12.1 was not found to be associated with breast cancer risk in our study, while a significant association was identified in the study by Stacey et al. involving 422 cases and 447 controls of African American descent (6). Reasons for this inconsistency are unknown, and the sample size for both studies is small.
In an attempt to explore the region surrounding previously-reported susceptibility loci, for each locus we evaluated all SNPs in a 100 kb region that are in strong LD with the initially-reported SNPs in Chinese (for rs2046210) or European descents (for all other SNPs) but not in Africans. None of these SNPs, however, were found to be associated with breast cancer in African-American women, indicating that these SNPs are neither causal variants nor in LD with causal variants in African Americans. Additional fine-mapping work in a larger region with more extensive coverage is needed to identify SNPs in these loci for breast cancer risk in African-American women.
The sample size of the study was not large. With the exception of a few SNPs, most of the SNPs evaluated in the study had a minor allele frequency (MAF) of 0.2 or higher. It is estimated that the study had 80% statistical power to identify SNPs with a MAF of ≥ 20% and an allelic OR of ≥ 1.22 at a significance level of 0.05. Therefore, it is possible that some of the null associations observed in this study could be due to the inadequate statistical power to identify a weak association. Another possible limitation of the study is that prevalent cancer patients in the SCCS were included in the study. As reported recently, however, these variants were not associated with breast cancer survival (11), and thus including prevalent cases is unlikely to introduce bias. Indeed, the results from the SCCS (prevalent cases) and NBHS (incident cases) were generally consistent.
The results from this study demonstrate the complexity of uniformly applying GWAS findings across ancestral groups. Large-scale studies are needed to identify genetic risk variants for breast cancer in this under-studied African-American population.
This research was supported in part by US National Institutes of Health grants R01CA92447, R01CA100374, and U54CA091405. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The authors wish to thank study participants and research staff for their contributions and commitment to this project, Regina Courtney and Qing Wang for DNA preparation, and Brandy Venuti for clerical support in the preparation of this manuscript. Genotyping assays using Sequenom platform were conducted at Proactive Genomics. Sample preparation and part of the genotyping assays were conducted at the Survey and Biospecimen Core that is supported in part by the Vanderbilt-Ingram Cancer Center (P30 CA68485).
This research was supported in part by US National Institutes of Health grants R01CA92447, R01CA100374, and U54CA091405