Empirical Drug Dosage Validates Pharmacogenomic Associations in All of Us
Clinical and Translational Science, 2025
The All of Us research program provides extensive medical history, drug dosage, and genomic data from a diverse U.S. population, creating an opportunity to uncover new links between genetic variation and medication response. Before pursuing novel discoveries, the dataset’s reliability must be assessed. Here, we test whether established drug–gene interactions can be reproduced using electronic health record data from All of Us. Focusing on the CYP450 enzyme family, which metabolizes roughly 90% of clinical drugs, we examine 61 CYP450-metabolized medications and assess dosage differences across metabolizer phenotypes. We replicate several known interactions involving CYP2D6, CYP2C19, CYP2C9, and CYP3A5, though not all expected associations emerge—likely due to data noise, limited dose adjustment in clinical practice, or phenoconversion. Overall, the results demonstrate that All of Us captures a subset of validated pharmacogenomic effects and has strong potential for future PGx discovery.