Abstract PL08-02: Fine mapping of genetic associations in African American colorectal cancer

Abstract

Abstract Genome-wide association studies (GWAS) of colorectal cancer (CRC) have identified risk variants in 14 genomic regions through the study of Europeans with CRC. In genotype analysis of a European American CRC case-control group, effects sizes and directions of association for 24 SNPs tested, which localize to 10 of the 14 previously identified regions, were consistent with GWAS results; in contrast, in an African American CRC case-control group, the associations ran in the opposite direction for 9 of 22 SNPs tested, indicating significant genetic heterogeneity in CRC associations in African Americans compared to European Americans. We conclude that genetic markers that identify CRC risk in persons of European ancestry do not identify CRC risk in African Americans. Two hypotheses may explain the genetic heterogeneity between European and African Americans: (i) The risk alleles in both populations are the same, however, the genetic architecture differs such that other SNPs that are linked to these same risk alleles will make better markers for African American CRC risk. (ii) The underlying risk alleles are different between the two populations. A corollary of this hypothesis is that the allele frequencies of the risk alleles must be low (<5%), because most the frequent genetic polymorphisms are shared between European and African populations. Distinguishing these hypotheses is important because genetic risk associated with rare alleles could have larger effect sizes (ORs>2, for example), which could make genetic risk modeling more determinative. Furthermore, identification of multiple rare alleles, each with a moderate risk effect, could explain the missing heritability problem. To address these hypotheses, we have drilled down on each of the genetic regions identified in the European GWAS using systematic genotyping and deep re-sequencing in African American CRC cases and population-matched controls. Analysis of tagSNPs in two genes so far analyzed, GREM1 and BMP4, have demonstrated multiple, independent SNP associations, which are not correlated with SNPs previously identified in persons of European descent. Because novel SNP associations are detected in each of these genes, we suggest that rare alleles underlie these genetic risk associations. We are presently using the deep re-sequencing analysis to identify candidate risk variants, which we will then test in functional studies. Ultimately, the analysis will permit better quantification of CRC risk, and it may lead to risk assessment models that can be used clinically for screening and surveillance. Citation Information: Cancer Epidemiol Biomarkers Prev 2011;20(10 Suppl):PL08-02.