August 22, 2022JAMA Internal Medicine40 citations
View Full PaperIncremental Value of Polygenic Risk Scores in Primary Prevention of Coronary Heart Disease
Structured PICO
Does the addition of polygenic risk scores to traditional risk scores improve coronary heart disease risk prediction in primary prevention populations?
P
Population
Asymptomatic people in primary prevention populations, predominantly of European ancestry, from 5 highly cited studies (including ARIC, FINRISK, Framingham, MESA, and UK Biobank).
I
Intervention
Polygenic risk scores (PRSs) with >45,000 single-nucleotide variations added to traditional risk factor scores
C
Comparator
Traditional risk factor scores alone
O
Outcome
Incremental value in CHD risk prediction (measured by change in C statistic, net reclassification index, positive predictive value, and false-positive rate)surrogate
The addition of polygenic risk scores to traditional risk models provides negligible to modest improvements in CHD risk prediction and does not appear to meaningfully improve clinical decision-making in primary prevention.
Abstract
Importance: Risk prediction for coronary heart disease (CHD) is a cornerstone of primary prevention strategies. Polygenic risk scores (PRSs) have emerged as a new approach to predict risk in asymptomatic people. Polygenic risk scores for CHD have been studied in several populations, but there is lack of agreement about the incremental value of PRS beyond traditional risk factor scores in the primary prevention of CHD. Observations: This narrative review critically appraised the 5 most highly cited studies published through 2021 that also included a large number (>45 000) of single-nucleotide variations (formerly single-nucleotide polymorphisms) and evaluated the incremental value of PRS in CHD risk prediction according to published PRS reporting standards. The cohorts studied included the Atherosclerosis Risk in Communities Study, FINRISK, the Framingham Heart Study, the Multi-Ethnic Study of Atherosclerosis, and the UK Biobank. All of the studies focused predominantly on populations of European ancestry. The hazard ratio per standard deviation of PRS ranged from 1.24 (95% CI, 1.15-1.34) to 1.74 (95% CI, 1.61-1.86). The C statistic for PRS alone ranged from 0.549 to 0.623. The change in C statistic when PRS was added to a standard risk factor model ranged between -0.001 to +0.021. Net reclassification index was reported in 4 of the 5 studies and varied from 0.001 to 0.097. At a sensitivity (true-positive rate) of 90%, positive predictive values ranged from 1.8% to 16.6%, and false-positive rates ranged from 77.1% to 85.7%. Conclusions and Relevance: In this review, PRS was significantly associated with CHD risk in all studies. The degree of improvement in C statistic and the net reclassification indexes when PRS was added to traditional risk scores ranged from negligible to modest. Based on established metrics to assess risk prediction scores, the addition of PRS to traditional risk scores does not appear to provide meaningful improvements in clinical decision-making in primary prevention populations.