Recently, great insight into the genetic epidemiology of common cancers has been gleaned from genome-wide association studies (GWAS). This insight has the potential to lead to future predictive genetic testing for cancer risk and the discovery of new biological targets for therapeutic intervention. Difficulties in performing such studies include genotyping cost, problems in identifying controls in a hospital-based setting, and the large sample sizes required for such studies. To obviate these problems, the use of shared, common controls in GWAS for common diseases has been proposed. While not epidemiologically rigorous, such an approach could have several practical advantages. Here, the utility of this approach will be examined by rigorously testing the hypothesis that the use of shared controls can boost the power of genome-wide association studies in cancer genetic epidemiology. No matter whether findings from this research lead to acceptance or rejection of this hypothesis, the results will be useful for planning future genome-wide association studies in cancer genetic epidemiology. This hypothesis will be tested through achievement of the following aims: 1) compare the power of using shared controls with a traditional case-control study design;2) determine the optimal approach to account for and properly control for different population substructure in the case and control populations;and 3) test the ability of shared controls to identify a set of SNPs potentially associated with pancreatic cancer. Analytical power calculations and empirical simulations will be used to determine how well a shared control design will perform compared to a classical case-control study. The appropriate number of shared controls, the influence of environmental covariates, the use of imputation in the shared controls, and various methods for controlling population stratification will all be examined. An existing pancreatic cancer GWAS data set will be examined with shared controls from the CGEMS study to provide a real-world example of this approach.
Relevance Genome wide association studies are a powerful and expensive tool to understand the genetics of common diseases including cancer. Here, we will rigorously test an approach to improve these studies by combining data from healthy individuals in many such studies. This approach, which we will validate on a study of pancreatic cancer, could enable more powerful genome-wide association studies without a vast increase in cost.
| Mukherjee, Semanti; Simon, Jennifer; Bayuga, Sharon et al. (2011) Including additional controls from public databases improves the power of a genome-wide association study. Hum Hered 72:21-34 |
