Generating genetic risk scores from intermediate phenotypes for use in association studies of clinically significant endpoints.

While previous results of genetic association studies for common, complex diseases (eg., coronary artery disease, CAD) have been disappointing, examination of multiple related genes within a physiologic pathway may provide improved resolution. This paper describes a method of calculating a genetic risk score (GRS) for a clinical endpoint by integrating data from many candidate genes and multiple intermediate phenotypes (IPs). First, the association of all single nucleotide polymorphisms (SNPs) to an IP is determined and regression beta-coefficients are used to calculate an IP-specific GRS for each individual, repeating this analysis for every IP. Next, the IPs are assessed by a second regression as predictors of the clinical endpoint. Each IP's individual GRS is then weighted by the regression beta-coefficients from the second step, creating a single, composite GRS. As an example, 3,172 patients undergoing coronary angiography were evaluated for 3 SNPs from the cholesterol metabolism pathway. Although these data provide only a preliminary example, the GRS method detected significant differences in CAD by GRS group, whereas separate genotypes did not. These results illustrate the potential of the GRS methodology for multigenic risk evaluation and suggest that such approaches deserve further examination in common, complex diseases such as CAD.