In genetic and genomic studies, it is important to study gene-environment interactions (GxE) in order to understand the etiology of complex diseases. Compared to marginal analysis, one difficulty with GxE arises from the fact that environmental exposures are often time-varying and measured with error. In this talk, we focus on testing GxE in the presence of measurement error and time-varying exposures. We first investigate the naïve test that ignores measurement error and show that it typically leads to biased estimates of the GxE effect and inflated type I errors under the null hypothesis of no interaction. This is different from the well-established result in testing main effects of E with measurement error, where the naive approach is valid for hypothesis testing purposes. We then obtain the analytic form of the bias term, and consider a regression calibration based approach for testing GxE when either validation data or replicates are available. Simulation studies are conducted to illustrate the performance of various tests in finite samples. The proposed methods are applied to study the gene-blood pressure interaction for coronary heart disease in an ancillary study of the Women's Health Initiative.
Available at: http://works.bepress.com/di/20/