Individualized prediction model for coronary artery disease integrating traditional risk factors and novel biomarkers in Korean population

Abstract

Background: Investigators have evaluated biomarkers beyond traditional risk factors to predict coronary artery disease (CAD). However, the incremental usefulness of adding multiple biomarkers remains to be clarified. Methods: We collected clinical and biomarker data from 1006 subjects (503 CAD patients and 503 age and sex-matched control subjects). The case group included relatively young patients (male < 55 years, female < 60 years) with angiographically documented multi-vessel CAD. We recorded clinical risk factors (hypertension (HTN), diabetes mellitus (DM), smoking, body mass index (BMI)) and assessed total cholesterol (T-chol), triglyceride (TG),high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), fasting glucose, and insulin level. We also checked other biomarkers (hs-CRP, IL-6, adiponectin, Lp-PLA2, RAGE and RANTES) thathad previously been demonstrated to be associated with CAD by our group. We developed a prediction model for CAD using multiple logistic regression analysis. Results: In multiple logistic regression analysis, the following variables were identified as independent determinants of CAD (each variable is followed by the adjusted odds ratio): HTN (36.47), DM (3.36), BMI (0.98), T-chol (1.0), TG (1.0), HDL-C (0.94), glucose (1.04), insulin (1.12), hs-CRP (1.18), IL-6 (1.01), adiponectin (1.04), and RAGE (1.001). By using our analyzed data, we developed a final prediction model for CAD. C statistics for models of CAD was 0.941 with the final prediction model, whereas it was 0.924 without novelbiomarkers (hs-CRP, IL-6, adiponectin, and RAGE). We added one of the biomarkers that showed a statistically significant difference in the multivariate analysis in addition to major traditional risk factors: HTN and DM. In this with or without HTN and DM plus one of the biomarkers prediction model, only hs-CRP and IL-6 resulted a statistically significant increase in prediction. Conclusions: Our results indicate that incorporation of novel biomarkers may modestly improve risk stratification for CAD beyond the model based on only the traditional risk factors. We confirmed an improvement in prediction of CAD when specific biomarkers (hs-CRP, IL-6) were added even in the absence of the major traditional risk factors (HTN and DM).