Development of a Bayesian Joint Logistic Model (BJLM) to Better Study the Association Between Haplotypes and Disease

• Activity Number: 526
• Type: Contributed
• Date/Time: Wednesday, August 1, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Anthony D'Amelio*+ and Carol Etzel
• Companies: MD Anderson Cancer Center and MD Anderson Cancer Center
• Keywords: Bayesian ; Epidemiology ; Methods ; Glioma ; Hodgkin ; Lung Cancer

Abstract

With the ever-increasing applications of cancer genetics in epidemiology, there is potential to identify genetic risk factors that would help identify individuals with increased cancer susceptibility. In this project, we use a Bayesian framework that directly incorporates information from HapMap to elucidate prior information for each haplotype block, and also non-genetic information to develop risk models. Models based on haplotypes using prior genetic information generally show increased and more precise estimates of risks for Hodgkin, Glioma, and Lung Cancer especially for rare and recessive genetic trait haplotypes. For example, the BJLM inferred haplotype TC had a substantially higher estimated odds ratio (OR=12.16, 95% Credible Interval = 2.47-90.1 vs. 9.24, 95% Confidence Interval = 1.81-47.2) and more significant p-value using WinBUGS (0.00044 vs. 0.008) for Hodgkin Disease compared to a traditional logistic regression approach. Full genetic models developed with the BJLM result in significantly higher discriminatory power and a significantly higher Net Reclassification Index compared to a traditional logistic regression approach for all cancers studied.