NIH 'All of Us' Study Finds Everyone Carries Drug-Response Gene Variants

Every person. Not most people, not a majority. Every single one of the nearly 100,000 participants examined.

That's the headline finding from a University of Pittsburgh-led study published November 11 in Science Translational Medicine, using whole-genome sequencing data from the NIH's All of Us Research Program. The paper, by Andrew Haddad and colleagues, characterizes the frequency of variants across 16 well-established pharmacogenes, the subset of genes that govern how the body processes medications.

<cite index="13-4">Every participant carried at least one genetic variant known to affect how the body processes medications, with a median of nine variants per person.</cite> That's not a fringe finding tucked into the supplement; it's the central result, and it has real clinical weight.

<cite index="13-5">The study found 24.4% of participants had been prescribed a medication known to interact with a genetic variant they carried, most commonly ondansetron, ibuprofen, omeprazole, pantoprazole, and atorvastatin.</cite> These aren't exotic drugs. They're some of the most commonly dispensed medications in the country, which makes the mismatch between genetic profile and prescription practice worth taking seriously.

The diversity angle is where this paper earns its place in the literature. <cite index="11-3,11-4">Pharmacogenomics promises improved outcomes through individualized prescribing, but the lack of diversity in pharmacogenomics studies has historically impeded clinical translation and the application of precision medicine across populations.</cite> Most of the foundational datasets skew heavily toward people of European ancestry, which means the variant frequencies used to build prescribing guidelines don't generalize well.

<cite index="10-4,10-5">The observed frequencies in pharmacogenes such as SLCO1B1 in the African ancestry group deviated from previous reports used to develop pharmacogenomics guidelines</cite>, according to the EurekAlert release from the American Association for the Advancement of Science. That deviation matters because SLCO1B1 variants affect how the liver transports statins, and miscalibrated guidance for a drug taken by tens of millions of people compounds quickly into population-scale harm.

<cite index="16-5,16-6">The team characterized pharmacogenomic variant frequencies across 16 well-established pharmacogenes using whole-genome sequencing data from nearly 100,000 All of Us participants, representing one of the largest and most diverse pharmacogenomics datasets with linked medication exposure records to date.</cite>

The study design has limitations worth naming. This is an observational analysis of existing genomic and electronic health record data, not a prospective trial measuring patient outcomes. Knowing that a quarter of participants carry a variant-drug mismatch tells you something about exposure, but it doesn't, by itself, tell you how many of those people experienced an adverse event or a subtherapeutic response as a result. That's the next clinical question, and it'll require a different study architecture to answer.

There's also the question of what clinicians are supposed to do with this. <cite index="12-5">The paper points to the need to give more attention to the role genetics plays in the clinic</cite>, according to reporting by STAT News, which covered the paper on the day of publication. That's a reasonable conclusion, but the infrastructure for routine pharmacogenomic testing at the point of prescribing is still uneven across health systems and far from universal.

<cite index="12-6">The All of Us Research Program has the goal of collecting genetic data from 1 million Americans from diverse backgrounds</cite>, which means this dataset will only get larger and, in principle, more representative. For now, a cohort of nearly 100,000 is a credible foundation.

The underlying paper is indexed at Science Translational Medicine under DOI 10.1126/scitranslmed.adt8040. Read the methods section before you cite the median.

Sources cited:
- Science Translational Medicine (Haddad et al., 2026) (https://www.science.org/doi/10.1126/scitranslmed.adt8040)
- STAT News (https://www.statnews.com/2026/11/11/all-of-us-study-finds-gene-variants-drug-responses/)
- EurekAlert / AAAS (https://www.eurekalert.org/news-releases/1151027)
- MedicalXpress / University of Pittsburgh (https://medicalxpress.com/news/2026-11-genes-affect-medications-highlights-high.html)