Public Health Genomics: Exploiting SNP-NSAID interactions to prevent colorectal cancer

Abstract

Colorectal cancer (CRC) is a common disease with a high mortality rate. Non-steroidal anti-inflammatory drugs (NSAIDs) such as aspirin and ibuprofen have shown considerable promise as preventive agents for CRC. However, due to concerns over the balance between benefits and harms NSAIDs are not recommended as a means of preventing CRC in average risk groups. Investigation into genetic modifiers of NSAIDs' chemopreventive action may help to identify those for whom such drugs are beneficial or ineffective. This thesis explored genetic mediation of the effectiveness of NSAID prophylaxis for CRC. A review of basic CRC biology and methods for investigating interactions, a systematic review of the literature to identify candidate interactions, and a secondary analysis of a GWA case-control study were performed. Candidate SNPs were screened for potential interactions, and several possible interactions were identified. The joint effects were similar for aspirin and ibuprofen, but not acetaminophen, implying true biological effects. Potential interactions were investigated further using a stepwise model building procedure. This resulted in a model containing three SNPs, aspirin and/or Ibuprofen use, their interactions, sex and age. This model was better able to discriminate cases and controls, demonstrated better calibration, and had greater information content (by AIC) than models without the interaction terms. Finally, recommendations for further research were given.