PHILADELPHIA- Polygenic risk scores (PRSs) are a cutting-edge tool in genetics, combining information from genetic markers across the genome to estimate a person's risk of developing certain diseases, such as coronary artery disease (CAD). By analyzing a person's DNA, PRSs offer insights into an individual's genetic predisposition for conditions like heart disease, potentially informing a more personalized approach to healthcare. But there can be significant variability across currently available PRSs, which may limit their reliability for individual predictions, according to new research from the Perelman School of Medicine at the University of Pennsylvania published this week in JAMA and presented at the American Heart Association's Scientific Sessions in Chicago.
The researchers analyzed data from more than 260,000 participants from diverse backgrounds and found that although most PRSs performed similarly when predicting CAD risk across populations, individual-level predictions varied widely. Many participants were placed in both high and low-risk categories by different PRSs, suggesting that patients could receive conflicting advice based on which score is used.
"Polygenic risk scores represent an exciting frontier in personalized medicine that has been gaining traction in clinics and as commercial health tests, but our findings suggest that they need to be used carefully," said co-lead author Michael G. Levin, MD, an assistant professor of Cardiovascular Medicine and cardiologist at Penn and the Corporal Michael Crescenz VA Medical Center (CMCVAMC). "At the individual level, these scores can vary quite a bit, which means that the same patient could receive dramatically different risk assessments that impact how doctors make decisions about prevention and treatment."
The research, conducted with data from the National Institute of Health's All of Us Research Program, Penn Medicine Biobank, and UCLA ATLAS Precision Health Biobank, compared 48 different CAD PRSs using health and genetic data. While 46 of the scores provided similar population-level predictions, 20% of participants had at least one score placing them in both the highest and lowest 5% of risk, depending on which score was used.
"The goal of PRSs is to help identify people at higher genetic risk for diseases like heart disease," explained Scott M. Damrauer, MD, Vice Chair for Clinical Research in Penn's Department of Surgery and a vascular surgeon at Penn and the CMCVAMC. "But for clinical use, it's important that the results are consistent and reliable, especially when decisions about someone's health is on the line."
"Our research underscores a critical gap in our understanding of PRSs, which has implications for their use in personalized medicine," said the study's lead author Sarah Abramowitz, BA, a medical student at the Zucker School of Medicine at Hofstra/Northwell and a Sarnoff Cardiovascular Research Fellow at the Perelman School of Medicine. "While these scores show promise for population-level CAD risk assessment, we need more robust methods to quantify and communicate the uncertainty of individual-level predictions."
The study's findings highlight the need for more refinement before PRSs can be widely adopted in healthcare to guide an individual's CAD risk assessment. Researchers recommend that clinicians consider potential inconsistencies and use these scores as part of a broader risk assessment strategy that considers clinical and lifestyle factors, among others.
This research was supported by the National Institutes of Health (HL169458, T32HL007843), Sarnoff Cardiovascular Research Foundation, and US Department of Veterans Affairs Biomedical Research and Development Award (IK2-BX006551).
