Abstract 9971: Distribution of Coronary Artery Calcium and Risk for Cardiovascular Disease: The Multi-Ethnic Study of Atherosclerosis

Abstract

Introduction: We tested the hypothesis that specific arteries with coronary artery calcium (CAC) would be associated with higher risk for incident cardiovascular disease (CVD) outcomes, even after accounting for traditional risk factors, and that adding this metric to current risk scores would improve event prediction. Methods: We analyzed data from 3,255 MESA participants with baseline CAC and event follow-up. Cox proportional hazards models assessed the independent associations between CAC presence in each of the main coronary arteries and incident myocardial infarction (MI), hard coronary artery disease (CHD), and hard CVD, adjusting for the ASCVD risk score and CAC volume and density. Area under the receiver operating characteristic curve (AUC) analyses assessed for risk discrimination by adding CAC distribution to existing risk models. Results: At baseline [mean age 66 years (SD 9.6); 1,893 female (58%)], 2,909 (89%) participants had CAC in the left anterior descending (LAD), 1,871 (58%) in the left circumflex (LCx), 1,691 (52%) in the right coronary artery (RCA), and 693 (21%) in the left main (LM). Over a median of 15.5 years, 275 developed MI, 408 CHD, and 598 CVD. After adjustment for the ASCVD risk score and CAC volume and density, only the RCA remained significant for MI and CHD, and was borderline for CVD (Table). AUC increased after adding RCA CAC presence to the ASCVD risk score for MI (0.5992 to 0.6660; 95% CI 0.031-0.102) and CHD (0.6483 to 0.6818; 95% CI, 0.003-0.064), but not for CVD, and there was no significant increase for any endpoints when added to models accounting for CAC volume and density. Conclusion: Prevalent CAC was most common in the LAD, followed by the LCx and RCA. Accounting for risk factors and CAC measures, the RCA was the only artery independently associated with additional MI and CHD risk. Moreover, MI and CHD risk prediction were improved with the addition of CAC distribution to the ASCVD score but not when also accounting for CAC volume and density.