Genetics
Lp(a) and CAD polygenic risk scores do not improve coronary event prediction beyond the SMART score in post-stroke patients (J Am Heart Assoc 2026)
Original title: Lipoprotein(a) and Coronary Artery Disease Polygenic Risk Scores Do Not Improve Secondary Manifestations of Arterial Disease Score Prediction of Coronary Events After Ischemic Stroke
This prospective cohort analysis of 1202 UK Biobank participants with prior ischemic stroke and no baseline coronary artery disease evaluated whether baseline lipoprotein(a) and CAD polygenic risk scores add predictive value to the SMART secondary prevention score for incident coronary events. Over a median follow-up of 13.4 years, 155 participants (12.9%) developed coronary artery disease. The SMART score alone yielded a discrimination area under the receiver operating characteristic curve of 0.69. Adding lipoprotein(a), the polygenic risk score, or both produced no meaningful improvement in discrimination (change in area under the curve ≤0.002; all P>0.6) or reclassification. These null findings indicate that routine lipoprotein(a) testing does not refine coronary risk stratification in secondary prevention after stroke.
Original abstract
Background: Survivors of ischemic stroke are at high risk for coronary artery disease (CAD), but individual risk varies. Lipoprotein(a) [Lp(a)] and CAD polygenic risk scores (CAD-PRS) improve risk prediction in primary prevention; however, their prognostic value after stroke remains uncertain. We assessed whether Lp(a) and CAD-PRS add predictive value to the Secondary Manifestations of Arterial Disease (SMART) secondary prevention score.
Methods: We analyzed 1202 UK Biobank participants of European ancestry with prior ischemic stroke and no CAD at baseline (2006-2022). Baseline clinical characteristics, Lp(a), and CAD-PRS were recorded. The primary outcome was incident CAD, identified from hospital, primary care, and death records. Competing-risk models compared SMART alone with SMART plus Lp(a), CAD-PRS, or both. Ten-year model performance was assessed using the area under the receiver operating characteristic curve and Brier score. Prespecified subgroup analyses examined age, sex, diabetes, and low-density lipoprotein cholesterol.
Results: During a median follow-up of 13.4 years, 155 participants (12.9%) developed CAD. SMART alone provided moderate discrimination (area under the receiver operating characteristic curve, 0.69 [95% CI, 0.64-0.75]) and a Brier score of 0.08. Adding Lp(a), CAD-PRS, or both did not improve prediction (change in area under the receiver operating characteristic curve, ≤0.002; all P>0.6; continuous net reclassification improvement, ≤0.03; all P>0.20). Results were consistent across subgroups, with no significant interactions after correction for multiple comparisons.
Conclusions: Among survivors of ischemic stroke without CAD, neither Lp(a) nor CAD-PRS improved prediction of incident CAD beyond the SMART score. These findings do not support their incremental use for coronary risk stratification in this population.
epidemiologygeneticsrisk predictionstroketesting
Summary written by lp-a.org from the published abstract; figures as published. Page updated 31 October 2026. Methods.