The polygenic risk score was not found to be superior to the clinical risk score in predicting the occurrence or risk stratification of incident coronary heart disease (CHD), according to study results published in JAMA.
Improving CHD risk stratification is critical in efforts to reduce over- or under-treatment of patients. In this retrospective cohort study, the predictive accuracy of a validated polygenic CHD risk score comprising > 6 million single nucleotide polymorphisms (SNPs) was compared with that of a clinical risk score based on the 2013 American College of Cardiology/American Heart Association (ACC/AHA) pooled cohort equations, using data from 2 trials.
The Atherosclerosis Risk in Communities (ARIC) study evaluated 4847 white individuals (mean age, 62.9±5.6 years; age range, 45-79; 56.4% women) enrolled between 1996 and 2015, and the Multi-Ethnic Study of Atherosclerosis (MESA) study examined 2390 participants (mean age, 61.8±9.6 years; 52.2% women). Participants from both studies were free of CHD at baseline.
The study’s primary outcome was the 10-year prediction of the occurrence of a first CHD event (ie, myocardial infarction, silent infarction, resuscitated cardiac arrest, revascularization, or coronary fatalities), assessed with model discrimination, calibration, and net reclassification improvement measures.
In the ARIC and MESA studies (median follow-up, 15.5 and 14.2 years, respectively), 696 and 227 participants, respectively (14.4% and 9.5%, respectively) had incident CHD events. There was a significant association between the polygenic risk score and 10-year CHD incidence in both the ARIC (hazard ratio [HR] per incremental standard deviation, 1.24; 95% CI, 1.15-1.34) and MESA (HR, 1.38; 95% CI, 1.21-1.58) cohorts.
When the polygenic risk score was added to the pooled risk equations, there were no significant increases in the C statistic in the ARIC (C statistic change, -0.001; 95% CI, -0.009 to 0.006) or MESA (C statistic change, 0.021; 95% CI, -0.0004 to 0.043) studies. Calibration in either cohort was not improved by the addition of the polygenic score to the clinical equations, with better calibration without the polygenic score in the ARIC study (P =.03). Using a ≤7.5% 10-year predicted risk threshold, the combination of the polygenic and clinical risk scores did not yield significant reclassification improvements in the ARIC (net reclassification improvement, 0.018; 95% CI, -0.012 to 0.036) or MESA (net reclassification improvement, 0.001; 95% CI, -0.038 to 0.076) cohorts.
Study limitations include the determination of allele weights from a genome-wide association study; the sole inclusion of participants of European ancestry, which restricted generalizability; a lack of optimization of SNP weighting for CHD subtypes; and possible overestimation of the polygenic risk score performance.
“These findings underscore the frequent discordance between statistical association and predictive performance,” noted the authors. “[A] polygenic risk score may not enhance risk prediction in a general, white middle-aged population.”
Conflicts of Interest Disclosures
Dr Gupta reported receiving grants from National Institutes of Health (NIH). Dr Psaty reported receiving grants from NIH and serving on the steering committee of the Yale Open Data Access Project funded by Johnson & Johnson. Dr Post reported receiving grants from NIH. Dr Rotter reported receiving grants from NIH. Dr Wang reported receiving personal fees from Novartis. No other disclosures were reported.
Mosley JD, Gupta DK, Tan J, et al. Predictive accuracy of a polygenic risk score compared with a clinical risk score for incident coronary heart disease. JAMA. 2020;323(7):627-635. doi: 10.1001/jama.2019.21782