Characterization of Statin Dose-response within Electronic Medical Records. W. Q. Wei1, Q. P. Feng2, L. Jiang3, M. S. Waitara2, O. F. Iwuchukwu2, D. M. Roden2,4,5,6, M. Jiang7, H. Xu7, R. M. Krauss8, J. I. Rotter9, D. A. Nickerson10, R. L. Davis11, R. L. Berg12, P. L. Peissig12, C. A. McCarty13, R. A. Wilke14, J. C. Denny1 1) Department of Biomedical Informatics, Vanderbilt University, Nashville, TN; 2) Department of Medicine, Vanderbilt University, Nashville, TN; 3) Department of Molecular Physiology and Biophysics, Center for Human Genetics Research, Vanderbilt University Medical Center, Nashville, TN; 4) Department of Pharmacology, Vanderbilt University School of Medicine, Nashville, TN; 5) Oates Institute for Experimental Therapeutics, Vanderbilt University School of Medicine, Nashville, TN; 6) Office of Personalized Medicine, Vanderbilt University School of Medicine, Nashville, TN; 7) Department of Biomedical Informatics, University of Texas, Houston, TX; 8) Children's Hospital Oakland Research Institute, Oakland, CA; 9) Medical Genetics Institute, Cedars-Sinai Medical Center, West Los Angeles, CA; 10) Department of Genome Sciences, University of Washington, Seattle, WA; 11) Kaiser Permanente Georgia, Center for Health Research Southeast, Atlanta, GA; 12) Biomedical Informatics Research Center, Marshfield Clinic Research Foundation, Marshfield, WI; 13) Essentia Institute of Rural Health, Duluth, MN; 14) Department of Internal Medicine, Sanford Healthcare, Fargo, ND.

   Background: Statin-mediated reduction in cardiovascular events represents one of the greatest advances of modern medicine. However, there is wide variability in the degree to which these drugs reduce low density lipoprotein (LDL) cholesterol within an individual. Efforts to define the genetic architecture underlying this variability have met with limited success, in part because genetic studies examining statin response have been limited to a single dose. By facilitating the reconstruction of full dose-response curves, electronic medical records (EMRs) offer a potential solution. In this study, we extract dose-response curves for simvastatin and atorvastatin, the two most commonly prescribed drugs in this class. We then leveraged these phenotypes to identify genetic predictors of statin potency and lipid lowering efficacy in the context of routine care. Methods: Two EMR-linked biobanks were utilized to construct dose-response curves for 2,026 subjects exposed to multiple doses of simvastatin and 2,252 subjects exposed to multiple doses of atorvastatin. We then fitted these data to a non-linear mixed effects model and extracted parameters representing potency (ED50), and maximal lipid lowering efficacy (Emax). We tested these parameters for association with 144 gene variants pre-selected based on (A) prior association with baseline lipids, (B) prior association with statin response, or (C) proven impact on statin pharmacokinetics. Findings: Atorvastatin was more efficacious, more potent, and demonstrated less inter-individual variability than simvastatin. A pharmacodynamic variant emerging from randomized trials (PRDM16) is associated with maximal efficacy (Emax) for both simvastatin (p = 0.04) and atorvastatin (p = 0.008). The impact of this variant on effect size was striking for atorvastatin (Emax = 51.7 mg/dl in subjects homozygous for the minor allele versus Emax = 75.0 mg/dl in subjects homozygous for the major allele). We also identified several loci associated with atorvastatin ED50, including SORT1, and several loci that were associated with simvastatin ED50, including SLCO1B1. Conclusion: Biobanks linked to EMRs improve our understanding of genetic factors contributing to drug response. The extraction of rigorously defined traits for pharmacogenetic association studies represents another promising approach to the meaningful use of EMRs.

You may contact the first author (during and after the meeting) at