Background: Non-invasive tools for the prediction of coronary artery disease (CAD) (coronary artery stenosis ≥70%) may improve identification of high-risk individuals who would benefit from further cardiac testing and closer follow-up.
Methods: We developed a risk estimation model for predicting significant CAD and adverse cardiovascular events from a prospective study of 407 consecutive patients undergoing elective coronary angiography for suspected CAD (“high-risk cohort” from MAGMA study, NCT01276678). Predictors of CAD were identified from candidate variables including demographics, clinical variables and biomarkers of thrombogenicity, inflammation and extended lipid panel. The model was validated for the prediction of coronary heart disease (CHD) (MI, coronary revascularization or CHD death) in the Jackson Heart Study (JHS), an ambulatory cohort of 4,485 African American participants (“low-risk cohort”). Comparisons were made with the predictive utility of the Framingham risk score (FHS).
Results: The risk estimation model was comprised of clinical variables (age, gender, history of CAD) and biomarkers (HDL3-C and lipoprotein(a)-C). In the model development cohort, the risk estimator had a c-statistic of 0.91 (95%CI 0.8-0.94, p<0.001) for predicting significant CAD (Figure a). In the “high-risk” cohort, the score had modest benefit in predicting incident CHD with c-statistic of 0.79 (95%CI 0.74- 0.82, p<0.001). The MAGMA risk score was comparable to FHS in the JHS cohort in predicting incident CHD with a c-statistic of 0.71 (95%CI 0.70-0.73, p<0.001) (Figure b).
Conclusions: Addition of lipoprotein sub-fractions enhances identification of anatomically severe CAD in a “high-risk” cohort (MAGMA). It has a modest utility in prediction of incident CHD in “high-risk” (MAGMA) and “low-risk” (JHS) patient populations comparable to conventional models. This is the first risk score to predict significant CAD and incident CHD in black patients.