London School of Hygiene & Tropical Medicine, United Kingdom
Background: Clinically significant drug-gene interactions are common but typically unavailable to prescribing physicians. Clinical Pharmacogenomics Implementation Consortium (CPIC) provides evidence-based, actionable guidelines on known drug-gene interactions to inform clinical care, but the real-world consequences of these interactions have not been well characterized.
Objectives: To characterize patterns of early discontinuation (no refill) among patients prescribed CPIC level A gene-interacting medications.
Methods: We extracted pharmacy dispensing and linked genetic data from the Veterans Aging Cohort Study (VACS)-Biomarker cohort, which includes 1730 patients with and 946 patients without HIV who consented to biomarker specimen collection in the Department of Veterans Affairs, the largest integrated healthcare system in the US. We selected the 10 most commonly prescribed medications in the cohort between 2000 and 2024 having actionable guidelines for prescribing based on the most recent CPIC level A criteria. Medications include amitriptyline, citalopram, codeine, efavirenz, nortriptyline, ondansetron, pravastatin, peginterferon, tacrolimus, and warfarin. We compared medication discontinuation within the first 30 days (no refill for a 30-day supply) among patients with and without the drug-gene interaction.
Results: Of 2242 patients with genetic data available, 1592 (71%) were of African genetic ancestry (genetic descent linked to Africa), 493 (22%) of European genetic ancestry (genetic descent linked to Europe), and 157 (7%) of mixed genetic ancestry. Nearly all (n=2197 98%) had received at least one medication with a CPIC level A drug-gene interaction. Comparing no refill within 30-days for a 30-day supply between patients with and without the relevant drug-gene interaction, the highest relative risk were for tacrolimus (7% vs. 4%, RR 1.83, CI 1.04-3.24), efavirenz (11% vs. 7%, RR 1.65, 95% CI 1.44-1.90) nortriptyline (33% vs. 24%, RR 1.39, 95% CI 0.85-2.29), peginterferon (24% vs. 19%, RR 1.26, CI 0.72-2.20), and amitriptyline (30% vs. 24%, RR 1.27, CI 1.05-1.54).
Conclusions: In real-world data, a higher proportion of people discontinued commonly prescribed medications for which they had a CPIC grade A drug-gene interaction in the first 30 days than those receiving the medication who did not have the interacting gene variant. Our findings highlight the need for clinical decision support tools and accessible pharmacogenomic data to optimize treatment selection, minimize early discontinuations, and reduce healthcare costs.
