Every participant in the All of Us study carried a relevant variant, while 24.4% had a record of taking a drug affected by a variant they carried. The findings expand the data available to researchers but do not establish better treatment outcomes.
Researchers found 17,704 previously unreported variants in protein-coding regions of genes linked to drug response in a study of 98,590 participants. Every participant (100%) carried at least one relevant variant, but carrying one does not establish that a medicine helped or harmed them.
Most previous genetic research has focused on people of European ancestry. The study, published on 11 November in Science Translational Medicine, adds information on hundreds of drug-response variants in ancestry groups for which data had been incomplete.
The team reported how frequently hundreds of drug-response variants occurred in genetic ancestry groups for which data had previously been incomplete.
The team combined whole-genome sequencing, which reads a person’s genetic code, with electronic health records from the National Institutes of Health’s All of Us Research Program. It examined 16 pharmacogenes: genes that can affect how the body processes medicines, changing the dose a person needs, side effects or a drug’s effectiveness.
Around 99.5% of participants carried a variant for which the Food and Drug Administration or medical guidelines recommend a different drug or dose for particular medicines. A smaller share, 24.4%, had documented exposure to a medicine affected by a variant they carried. Common examples included the acid-reducing drugs omeprazole and pantoprazole and cholesterol-lowering statins.
Senior author Philip Empey, a University of Pittsburgh pharmacy professor who directs its Center for Clinical Pharmaceutical Sciences, said incomplete health records likely meant the study underestimated such exposures. He wrote in The Conversation that computer models predicted about a quarter of the newly identified variants might significantly affect drug processing. Those predictions are not experimentally demonstrated drug effects.
Empey cautioned that the study does not prove genetic testing would improve outcomes for each participant. Age, kidney and liver function, and other medicines also affect drug processing. Andrew Haddad and colleagues conducted the research.
| Ancestry | Participants |
|---|---|
| European | 48,112 |
| African | 23,179 |
| Americas | 15,133 |
| East Asian | 2,199 |
Empey said broader genetic testing may help clinicians choose medicines, but such testing is not routine. Many people have not had their genomes sequenced, and clinicians may not know when to order tests or how particular variants affect prescriptions.
Loren Saulsberry, a health policy researcher who was not involved in the study, said the diverse dataset could help limit disparities as findings move into clinical use. Empey disclosed funding from the National Institutes of Health, Veterans Administration, Shear Family Foundation and Myers Family Foundation.
The program’s future data collection faces funding pressure. Joshua Denny, who was All of Us CEO until the end of October, told STAT that cuts had sharply reduced recruitment and engagement at academic medical centers, federally qualified health centers and mobile units. The program has also stopped offering participants access to their genetic-testing results because of funding concerns.
All of Us receives annual congressional appropriations and funding through the 21st Century Cures Act, enacted in 2016. The latter stream is due to end in fiscal 2027.
Researchers published the finding on 11 November 2026, identifying adult and child prints on figurines from what is now Israel. They suggest adults taught children to work the clay, though the prints do not establish that.
2 min
Human Rights Watch says the plan announced on 5 November aims to expand electrification and renewable investment but sets no concrete target for phasing out fossil fuels.
3 min
Researchers using a different way to combine telescope measurements largely corroborate the original image of the shadow of M87*, a supermassive black hole.
2 min
