Translational Informatics Approaches for Studying Cytochrome P450-Mediated Drug Interactions and Their Genetic Influences

Abstract

Drug-drug interactions (DDIs) involving cytochrome P450 (CYP) enzymes can alter drug metabolism, potentially leading to serious adverse drug events (SADEs). While pharmacokinetic mechanisms of CYP-mediated DDIs are relatively well understood, identifying clinically relevant ADE signals remains challenging, particularly given the limitations of pre-marketing clinical trials, spontaneous reporting biases, and the complexity of real-world clinical data. Furthermore, genetic variations in CYP enzymes significantly contribute to individual susceptibility to ADEs. Despite their clinical importance, drug-gene interactions (DGIs) remain understudied, leaving significant gaps in our ability to predict and prevent ADEs influenced by genetic factors. To address these gaps, this dissertation developed a comprehensive translational informatics framework integrating multiple data sources and advanced computational approaches: (1) Generating and prioritizing CYP-mediated DDI-SADE hypotheses through statistical disproportionality analyses and novel ranking algorithms applied to spontaneous reporting systems, followed by rigorous validation using electronic health record (EHR) databases. (2) Systematically mining biomedical literature to uncover previously unrecognized CYP-mediated DDI-SADE signals using advanced transformer-based natural language processing (NLP) models, and subsequently validating these signals through large-scale real-world EHR analyses. (3) Translating validated DDI-SADE signals into genetically informed DGI-SADE hypotheses by integrating comprehensive pharmacogenomic analyses leveraging genetic data linked to biobanks, systematically exploring CYP enzyme polymorphisms and their contributions to variability in ADE risks. Collectively, this dissertation enhances the current understanding of CYP-mediated drug interactions and genetic influences on drug safety, providing a broadly applicable computational framework for pharmacovigilance and personalized medication management through pharmacogenomic insights.