Diet, inflammation, and genetic predictors of tissue-specific gene expression: A functionally-informed gene-environment interaction analysis for risk of colorectal cancer.

Colorectal cancer (CRC) is the second most commonly diagnosed cancer in both sexes combined worldwide, as well as the second leading cause of cancer deaths. Pathogenesis of CRC can be attributed to both genetic (G) and environmental (E) factors. Among environmental risk factors, dietary factors contribute significantly to the etiology of CRC. Extensive research has focused on relationships between dietary factors, food constituents, and dietary patterns to CRC risk. There is compelling evidence from observational studies which supports the carcinogenicity of increased consumption of processed meat, and red meat. Evidence for protective effects of fruit, vegetable, whole grains, or fiber intake is less certain. Dietary factors may also have a direct effect on CRC risk by modifying the effect of genetic predisposition for CRC. Furthermore, evidence for a relationship between inflammation and cancer has also been established by previous epidemiological studies as well as through basic research, yet mechanisms that underlie the association between inflammation and CRC remain uncertain. C-reactive protein (CRP) is a positive acute-phase protein produced primarily in the liver as a response to infection, tissue injury, and systemic inflammation. Circulating levels of CRP are commonly used as a measure of chronic inflammation, and pre-diagnostic serum CRP has been associated with increased risk of CRC in previous studies. This dissertation research project utilized self-reported dietary histories and measures of circulating C-reactive protein to examine the interaction between these risk factors and genetic variation on the risk of colorectal cancer. Utilizing genotype array data from both population-based case-control studies and nested case-control studies in an international collaboration of well-characterized studies, the Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO), we employed a novel statistical approach to reduce multiple testing burden and increase power with a priori functional information from predicted tissue-specific gene expression. Our GxE analysis for dietary factors had a study sample comprised of 22 studies with 20,236 cases and 19,838 controls. In our dietary GxE analysis we detected a significant interaction (P = 2.30 x10-7) between processed meat intake and variants predictive of RAC Family Small GTPase 1 (RAC1) gene expression in the colon and risk of CRC. The most significant interaction of a single variant within RAC1 was rs2346263 (βGxE = -0.127; P = 1.32 x 10-4). In our CRP GxE analysis of 5 studies comprising 5,423 individuals, we observed a 4 % increase of risk of CRC from a two-fold rise in CRP (mg/L) levels [OR:1.04 (95% CI:1.00, 1.08)]. There was significant heterogeneity in study specific effect sizes for the circulating CRP [P = 7.6 x 10-3]. At an FDR < 0.20 we detected no significant interactions between circulating CRP and variants predictive of gene expression in the colon. Our results suggest that the RAC1 gene may be involved in the ROS and inflammatory pathways that lead to CRC from intake of processed meat. Furthermore, our findings also suggest that CRP may not be associated with CRC and we did not find evidence that genetically determined gene expression influences the association between circulating CRP levels and CRC risk.