Traditional Cardiovascular Disease Risk Scores Offer Similar Value as Polygenic Risk Score

Michael Pencina, PhD

New research from Duke Health’s AI Health Team suggests newer is not always better, at least when it comes to risk scores for predicting coronary heart disease in young and middle-aged adults.

An analysis comparing the predictive utility of a genome-wide polygenic comprising more than 6,000,000 genetic variants against a traditional 30-year risk factor score among patients from a pair of major studies, results demonstrate use of the polygenic risk score did not improve discrimination in the study’s primary prevention sample.

“While genetic tests use new technology, they can be high-priced,” said senior investigator Michael Pencina, PhD, vice dean for data science at Duke University School of Medicine and director of Duke AI Health, in a statement. “People should instead visit their doctor and have their actual, clinical factors measured, because this will do a much better job of determining their state of health. And for those who have a high risk of developing cardiovascular disease—especially young people—they should eat healthy foods, exercise and begin proper medications as warranted.”

As the rates of decline for cardiovascular disease have begun to plateau and the shortage of clinicians increases, a new emphasis has been placed on identifying means for improving primary prevention of cardiovascular disease for a resource-strapped health systems in the face of an aging US population. With continuous advances in technology, some have begun to argue for the adoption of strategies that incorporate use of polygenic risk scores in risk estimation for primary prevention populations. With this in mind, Pencina and a team of colleagues at Duke AI Health designed the current study to compare use of a previously validated polygenic risk score against a previously validated traditional 30-year risk model for in a cohort made up of patients from Framingham Offspring Study and the Atherosclerosis Risk in Communities study.

For the current study, investigators noted PRSice was used to derive the previously validated PRS by Khera et al. with more than 6,600,000 single-nucleotide polymorphisms based on previous genome-wide association studies from the Coronary Artery Disease Genetics consortium. In that study by Khera et al., use of the score identified 8.0, 6.1, and 3.5% of the population at greater than 3-fold increased risk for coronary artery disease, atrial fibrillation, and type 2 diabetes, respectively. The traditional risk factor model used in the study was validated in a previous study led by Pencina and published in 2009, which used data from the from the Framingham Heart Study and a modified Cox model that allows adjustment for competing risk of noncardiovascular death to construct a prediction algorithm for 30-year risk of hard cardiology. Specific factors included within the traditional risk factor model included age, sex, smoking, systolic blood pressure, antihypertensive treatment, diabetes, and total and HDL cholesterol.

Investigators noted the because the polygenic risk score was derived from simple of a predominantly European ancestry, their analyses were limited to those who self-reported White rase. The study sample was further stratified by age, with age categories defined as young (20-39 years), early midlife (40-59 years), and late midlife (45-59 years).

From the Framingham Offspring Study and the Atherosclerosis Risk in Communities study, investigators identified 9757 patients for inclusion in their analyses. Of these, 1863 were classified as young, 2154 were classified as being in early-midlife, and 5740 were classified as being in late-midlife. Upon analysis, results indicated both the traditional risk factor score (HR, 2.60 [95% CI, 2.08 to 3.27]; HR, 2.09 [95% CI, 1.83 to 2.40]; HR, 2.11 [95% CI, 1.96 to 2.28]) and the polygenic risk score (HR, 1.98 [95% CI, 1.70 to 2.30]; HR, 1.64 [95% CI, 1.47 to 1.84]; HR, 1.22 [95% CI, 1.15 to 1.30]) were significantly associated with incident coronary heart disease in young, early midlife, and late midlife, respectively.

Further analysis demonstrated discrimination was similar or better with the traditional risk factor score (C index, 0.74 [95% CI, 0.70 to 0.78]; C index, 0.70 [95% CI, 0.67 to 0.72]; C index, 0.72 [95% CI, 0.70 to 0.73]) compared with an age- and sex-adjusted polygenic risk score (C index, 0.73 [95% CI, 0.69 to 0.78]; C index, 0.66 [95% CI, 0.62 to 0.69]; C index, 0.66 [95% CI, 0.64 to 0.67]) in young, early-midlife, and late-midlife participants, respectively. Investigators also pointed out the C index when adding polygenic risk score to the traditional risk factor score was just 0.03 (95% CI, 0.001 to 0.05), 0.02 (95% CI, −0.002 to 0.037), and 0.002 (95% CI, −0.002 to 0.006) in young, early midlife, and late midlife, respectively.

“A lot of young people can be given a false sense of security if it looks like they have a low risk of inherited disease from their family,” Pencina added. “But in the nature vs. nurture battle, it’s nurture that is the stronger factor for cardiovascular disease: how a person lives throughout adulthood is a much bigger factor in the course of this disease.”

This study, “Predictive Utility of a Validated Polygenic Risk Score for Long-Term Risk of Coronary Heart Disease in Young and Middle-Aged Adults,” was published in Circulation.