Candidate Pathway Gene Environment Interactions On Colon And Rectal Cancer Risk And Survival

Genetic association studies have adopted for a long time a traditional analytic approach that focuses on individual genetic markers, usually single nucleotide polymorphisms (SNPs), in association with disease or phenotype. A standard single-SNP analysis that ignores combined effects of multiple SNPs and furthermore their interactions with environmental exposures, explains a small portion of disease heritability: an often cited issue of 'missing heritability'. A comprehensive approach that accounts for these interactions carries the potential for identifying novel susceptibility loci and is more suited to decipher causal relationships and underlying molecular mechanisms of disease. The overall goal of this dissertation is to develop a methodologically sound framework that examines interactions in genetic association studies that is able to represent the biologic underpinnings of disease and yield interpretations that are statistically valid and of clinical and/or public health relevance. We first examined interactions between genetic variants at the gene level in genome-wide association study (GWAS) data of six common chronic diseases of the Wellcome-Trust-Case-Control-Consortium (WTCCC): bipolar disorder (BD); coronary artery disease (CAD); hypertension (HT); rheumatoid arthritis (RA); type 2 diabetes (T2D); and type 1 diabetes (T1D). We used logic regression to search for biologically plausible forms of SNP-set interactions within genes. Next, we extended our approach to test for gene-environment interaction (GEI) effects at the pathway level and applied it to the population-based case-control data of the Diet, Activity and Lifestyle as a Risk Factor for Colorectal Cancer Study. We focused on the candidate pathway of angiogenesis and three hypothesized environmental exposures: dietary protein intake; smoking; and alcohol consumption. Our approach consisted of 3-steps: the first two summarized the within gene effects and the full pathway effects; and the third step modelled the GEI effects on colon and rectal cancer risk and survival. Our interaction analysis was able to detect an appreciable number of susceptibility loci showing strong evidence of association with the six diseases in WTCCC, including novel signals supported by biologically plausible links to the diseases. The number of genes with strong evidence of association was: 13 for BD; 16 for CAD; 15 for HT; 72 for RA; 105 for T1D; and 19 for T2D. The top significant genes were: NFIA with BD, CDKN2B with CAD, COL4A4 with HT, BTNL2 with RA, and TCF7L2 with T2D. The majority of strong single-SNP signals of WTCCC and on average 46% of recent GWAS meta-analyses signals were confirmed in our analysis. The results of the GEI pathway analysis also yielded an appreciable number of significant and novel interactions. Overall the magnitudes of gene interaction odds and hazard ratios increased with increasing levels of the interacting environmental exposure. This observed positive gradient supported the plausibility of the interactions. We found five statistically significant GEIs associated with colon cancer risk and three GEIs with colon cancer survival involving all three environmental exposures. For rectal cancer, we found eight significant GEIs in association with risk involving six genes and five GEIs with survival. This dissertation showed how exploring interactions of all measured SNPs within each gene can identify appreciable numbers of novel susceptibility loci in GWAS. We also showed that GEI effects on colorectal cancer risk and survival can be identified by adopting a comprehensive candidate pathway approach that emphasizes the biologic hypothesis in the selection of the pathway genes and environmental exposures and carries that logic through the analysis.