Statins among the most widely prescribed agents worldwide for prevention of cardiovascular diseases, produce substantial pleiotropic effects. Pleiotropic effects are unanticipated outcomes other than those for which the drug was originally developed, either therapeutic (beneficial) or detrimental (adverse drug reactions). Statin pleiotropic effects are unanticipatedly broad, including increasing the risk of developing type 2 diabetes mellitus and cataract, decreasing cancer-related mortality, and reducing dementia. Many effects are still not determined. In addition, individual responses to statins are highly variable. Genetics studies have identified loci that are significantly associated with statin response. However, it is unclear if either of the genetic variants within these regions is also associated with statin pleiotropic effects. We propose to investigate statin pleiotropic effects using whole de-identified electronic health records (EHRs) of >2.5 million individuals at Vanderbilt, including >110,000 statin exposure individuals. By linking this cohort to BioVU, the Vanderbilt de-identified DNA biobank, >10,000 of these statin exposure individuals have extant genome-wide genotyping. We argue that 1) previous inconclusive results are largely caused by inconsistent phenotype definitions, and 2) using the EHR to develop a novel, drug-based phenome-wide association studies (PheWAS) provides an ideal approach to discover unknown statin effects. The still-growing Vanderbilt de-identified EHRs allow large amounts of individuals' clinical data shared to support validation of known pleiotropic effects and to enable novel discoveries. Our previous work demonstrated our ability to develop consistent EHR-based phenotype definitions that can be deployed across multiple EHRs and institutions. We have expertise leveraging state-of-the-art informatics techniques, including natural language processing and ontologies, for pharmacogenetic studies, including for statins. We first described the PheWAS approach to not only replicate genetic associations but also discover novel, pleiotropic associations. Our informatics expertise combined with an ideal EHR/DNA population, will enable us to validate and discover statin pleiotropic effects. Accordingly, we propose the following three aims: 1. develop and test EHR-based phenotype algorithms for four controversial statin pleiotropic effects, 2. conduct a PheWAS to discover unknown statin pleiotropic effects, and 3. evaluate and discover genetic predictors of statin pleiotropic effects.
Statins (HMG-CoA reductase inhibitors), among the most widely prescribed agents worldwide for prevention of cardiovascular diseases, produce substantial pleiotropic effects. Statin pleiotropic effects are unanticipatedly broad and the genetic variants associated with statin pleiotropic effects remain unclear. We propose to investigate statin pleiotropic effects using whole de-identified electronic health records (EHRs) of >2 million patients at Vanderbilt (including 10,000 genotyped statin recipients) to answer these questions.
