Genetic Association to adverse drug events in the eMERGE Pharmacogenomics cohort

Abstract

National Human Genome Research Institute (NHGRI) Electronic Medical Records and Genomics (eMERGE) network was created to combine DNA biorepositories with electronic medical (EMR) data to support genomic research. The eMERGE Pharmacogenomics (PGx) project is a partnership between eMERGE[1,2] and the Pharmacogenomics Research Network (PGRN)[3], collecting clinical data and pharmacogenetic variance data. The eMERGE PGx project aims to connect clinical data from electronic medical records to targeted pharmacogenetic information[4]. First phenotypes are extracted from the clinical data. This dissertation proposes a simple automated approach to identify potential adverse drug events (ADE) in the eMERGE PGx cohort. Following the identification of potential ADEs, I utilized them as phenotype outcomes for genetic associations tests at the single variant, gene, and gene-set level.I examined data from the EMR of the study participants through the lens of a database of known adverse drug events, the Drug Evidence Base[5]. Diagnosis codes that were known to be adverse events and appeared in a participant’s medical record following a medication order were labeled as a potential ADE. This analysis resulted in 1731 participants out of 6379 (~27%) potentially having experienced at least one ADE, and 372 different phenotypes identified. Each phenotype is identified using a drug-diagnosis pair. I used the more common of these phenotypes in genetic association tests. First, I used logistic regression to evaluate association to single nucleotide variants (SNV). I found two phenotypes with a statistically significant association to one or more variants. Clotrimazole with edema was associated with variants rs143661234 and rs 1800822 on gene FM03 (p-value needed). The second was Venlafaxine with Hyponatremia associated with variant rs28416595 on gene CYP4F12. I continued with Sequence Kernel Association Tests (SKAT)[6] to assess gene-level associations to the same phenotypes. I found ten different phenotypes with a significant association to one or more genes. Finally, for the ten phenotypes, which had a significant finding using SKAT, I used Gene Set Enrichment Analysis (GSEA)[7] to identify any gene set level (pathway) associations. The numeric SKAT score was used in place of an enrichment value. One phenotype, hydrochlorothiazide with pulmonary edema, was identified as having both a significant adjusted p-value as well as an acceptable false discovery rate (below .25) for the gene set Positive Regulation of Cellular Component Biogenesis.