Every One of 100,000 All of Us Participants Carries Drug-Response Gene Variants

One hundred percent of nearly 100,000 participants in NIH's All of Us Research Program carried at least one genetic variant that affects how their body processes medications — and 24.38% had already been prescribed a drug known to interact with a variant they carry. That's the headline result of a University of Pittsburgh-led study published in Science Translational Medicine, one of the largest and most diverse pharmacogenomic (PGx) datasets yet linked to real medication exposure records.

The team, led by senior author Philip E. Empey, PharmD, PhD, associate professor at Pitt's School of Pharmacy and associate director for pharmacogenomics at the Institute for Precision Medicine at Pitt and UPMC, evaluated 16 known drug-impacting variants across the cohort. The median participant carried nine such variants. Nearly all — 99.49% — had a predicted phenotype for a gene with a CPIC or FDA prescribing recommendation.

What did the prescription records show?

By matching genotype data to electronic health records, the researchers found that more than 24% of participants had prior exposure to at least one medication with pharmacogenomic prescribing guidance for a variant they carried. The most common implicated drugs were:

• Ondansetron (anti-nausea)
• Ibuprofen (anti-inflammatory)
• Omeprazole and pantoprazole (acid reflux and heartburn)
• Atorvastatin (cholesterol lowering)

These aren't rare specialty agents. They're among the most routinely prescribed drugs in routine care, which is exactly why the finding matters for health-system R&D and clinical decision-support budgets.

"Many people have experience with taking a medication that is not working well for them or that has a side effect," Empey said. "While many things impact how a patient responds to medication, our research tells us that genetics could be [a] factor for a large proportion of patients and it points to the value of precision medicine in prescribing."

Why the cohort's diversity matters

The study, titled "Frequency of pharmacogenomic variation and medication exposures among All of Us Participants," went beyond cataloging known variants. The team identified over 17,000 novel variants and reported medication-linked variant frequencies across multiple ancestry groups — including populations where those frequencies had never been estimated.

"Importantly, we report thousands of novel variants as well as hundreds of alleles and predicted phenotypes that deviate from known frequencies and/or were previously unreported, especially within Americas and African ancestry groups," the authors wrote.

Empey tied this directly to a methodological weakness in earlier work: "The limited diversity of racial and ethnic groups in early studies is a challenge in genetic research and skews risk estimates." The All of Us program — an NIH initiative targeting data from one million U.S. participants, with Pitt launching All of Us Pennsylvania in 2018 — was built to close that gap by enrolling people across a wide range of ages, backgrounds, and life experiences.

What are the implementation barriers?

More than 300 medications already carry label information on genetic variant interactions, Empey noted. One example: a colon cancer chemotherapy drug that can cause serious toxicity in up to 8% of patients carrying relevant variants, who are proactively tested before treatment. Yet broad PGx adoption remains limited, the authors wrote, "in part by the relative paucity of data on the frequency of variants, alleles, and predicted phenotypes across diverse biogeographical groups."

Empey's proposed model is panel-based screening — a one-time test for common variants affecting many medications, stored in the medical record "much like a test for kidney function." The approach is already being applied to patients on multiple concurrent medications or starting genetics-sensitive drugs such as clopidogrel, used after cardiac stenting.

He acknowledged the cost-benefit decision sits with policy makers and health insurers, calling it a very active research area. At UPMC, he is leading efforts to surface variant information in clinicians' prescribing workflows via EHR alerts, letting them consider an alternate drug or adjusted dose when needed.

"Just like with any new technology, it comes down to responsible implementation," Empey said. "Costs are coming down and the data is increasing, so it is important for patients and clinicians to be aware that the use of precision medicine in prescribing is becoming commonplace."