Abstract
Genome-wide association studies (GWAS) are traditionally based on principles of medical genetics. This strategy is well adapted for Mendelian disorders. Genetics of phenotypes that leave human organisms vulnerable to diseases in late life (called age-related phenotypes) is, however, more complex. The fundamental complicating factor is the elusive role of evolution in fixing molecular mechanism of these phenotypes. This complexity implies a special type of an inherent genetic heterogeneity reflecting sensitivity of genetic associations with age-related phenotypes to the life course of individuals in different environments. Here we follow a two-stage genome-wide approach that leverages this heterogeneity. This approach is demonstrated by examining non-pleiotropic and pleiotropic genetic predisposition to 24 age-related phenotypes (16 biomarkers, 7 diseases, and death) in a modest sample (N=26,371) from five studies (ARIC, FHS, MESA, CHS, and CARDIA) from the Candidate Gene Association Resource. In Stage 1, we performed the traditional univariate GWAS for each of 24 phenotypes improved by leveraging information from longitudinal follow up in each study separately. In Stage 2, we used Fisher’s and two omnibus tests to combine statistics from Stage 1 leveraging different types of heterogeneity. Our analyses replicated 212 SNPs in 49 loci at genome-wide level and identified 53 novel or firstly attained genome-wide significance non-pleiotropic SNPs in 49 loci and 202 pleiotropic SNPs in 154 loci in a modest sample (all loci exclude the Major Histocompatibility Complex). Our findings demonstrate benefits of more comprehensive approaches than the currently prevailing ones to gain insights into the genetics of healthspan and lifespan.