Abstract 4733: Gene-set analysis of colon cancer genome-wide association data: Identification of the TGF-beta pathway

First results from genome-wide association studies (GWAS) have demonstrated considerable success in identifying genetic variants associated with colorectal cancer (CRC). To date, 10 CRC susceptibility loci have been identified. However, these marginal single nucleotide polymorphism (SNP) associations explain only ∼6% of the heritable variation underlying CRC risk. We employed a pathway-based approach using our recently developed Gene-set Ridge Regression in Association Studies (GRASS) method to assess whether sets of functionally related genes are enriched for SNPs associated with colon cancer risk. We used GWAS data from three studies: The Women9s Health Initiative (WHI; 483 cases and 530 controls); the Prostate, Lung, Colon and Ovarian Cancer Screening Trial (PLCO; 546 cases and 1177 controls); and the Diet, Activity and Lifestyle population-based case-control study (DALS; 698 cases and 719 controls). Samples were genotyped on Illumina HumanHap 300K + 240K, 550K or 610K platforms. After applying stringent quality control criteria, our final analysis included 392,361 SNPs. We used genomic partition to assign SNPs to nearby genes and defined pathways using the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. We restricted our analysis to 170 pathways with 10 or more genes (range 10-253 genes). Our GRASS method uses eigenSNPs derived from principal components analysis within each gene. Regularized regression is performed for each pathway, using a novel group ridge penalty function. The penalty function results in selection of the most representative eigenSNPs within genes while assessing the association of all the genes, within a pathway, with disease risk. Permutations are used to standardize the test statistics and obtain p-values. The false discovery rate (FDR) was calculated using the Benjamini-Hochberg method. We analyzed WHI, PLCO and DALS separately, adjusting for age, sex, and genetic ancestry derived from principal components analysis, and performed a meta-analysis to combine the results. Five pathways were significant at a p-value cutoff of 0.05. The top pathway, the transforming growth factor beta (TGF-beta) signaling pathway (p-value=0.009), also met a FDR cutoff of 20%. This pathway remained significant (p=0.014) after excluding the known CRC susceptibility loci and all SNPs with linkage disequilibrium r2>0.2 with those 10 loci. The top genes in the pathway were TFGB1, SMAD4, FST, TFGB2, INHBA, PITX2, BMPR2 and PP2R2B (all p Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 4733.