Genetic risk scores can be less accurate for nonwhite patients

A Feb. 19, 2024 Nature Medicine study found that ten chronic disease polygenic risk scores still faced a major barrier to clinical use in diverse U.S. populations: their predictive performance dropped when the underlying genetic data did not reflect more than European ancestry. The result sharpened a problem that now sits at the center of precision medicine, where DNA-based scores are increasingly used to estimate risk for heart disease, diabetes and certain cancers.

The imbalance is built into the evidence base. A Nature Communications paper published July 25, 2019, on polygenic risk score usage and performance in diverse human populations showed that ancestry affects how well the tools travel across groups. More recent work in the European Journal of Human Genetics, published Jan. 5, 2024, found that European-centric cardiometabolic scores can behave differently in multi-ancestry populations. Other research has shown that scoring accuracy varies across the genetic ancestry continuum, making the same algorithm more reliable for some patients than for others.

That matters in the exam room. If a score underestimates a patient’s true risk, a doctor may delay screening, prevention or earlier treatment. If it overestimates risk, the patient may be pushed into extra testing or medication that was not warranted. The clinical stakes are highest for Black, Hispanic, Asian, Indigenous and mixed-ancestry patients because genetic studies have historically drawn heavily from people of European ancestry.

Researchers and public-health officials have been warning about the gap for years. At an ELSI Friday Forum held Jan. 12, 2024, Alicia Martin, PhD, discussed polygenic risk prediction in diverse populations and emphasized that genetic variation reflects human history, while national biobanks do not match global or even national diversity. The National Human Genome Research Institute has also published a public explainer on polygenic risk scores, and Centers for Disease Control and Prevention genomics staff asked in a July 5, 2022 blog post whether the tools are ready for clinical practice.

The policy pressure is rising as hospitals, researchers and payers move faster than the evidence can support. Carelon Medical Benefits Management issued Appropriate Use Criteria for Polygenic Risk Scores in Genetic Testing in 2024, while a Stanford-hosted paper warned that clinical use could widen health disparities. The next step is not just better algorithms, but more representative datasets, stronger validation and clearer limits before these scores are used at scale.