Abstract 4372707: Coronary Calcium Density Parameters Improve Cardiovascular Risk Prediction beyond Agatston Score in Adults Aged 75 and Older: The Atherosclerosis Risk in Communities (ARIC) Study

Abstract

Background: Coronary artery calcium (CAC) is usually quantified by Agatston score, integrating both density and plaque area. Recently, calcium density is shown to be inversely associated with cardiovascular disease (CVD) at a given CAC volume predominantly among middle-aged adults. However, its prognostic value has not been specifically assessed in adults aged ≥75 years. Also, most previous studies investigated average CAC density (total Agatston score divided by total CAC volume) whereas CAC density could be parameterized in various ways (e.g., the highest density across all plaques or average of the highest density across every plaque). Aim: To explore whether CAC density parameters can improve CVD risk prediction beyond Agatston score in the 75-and-older population. Methods: We analyzed 1,455 ARIC study participants (mean age 80.4 [SD 4.2] years, 58.2% female, 18.2% Black) with a CAC Agatston score >0 at visit 7 (2018–2019). We used elastic net regression to identify predictive density parameters out of 20 density parameters based on Hounsfield unit distribution of each calcified plaque across multiple plaques, if any, in every participant). Then, we evaluated improvement in risk prediction of CVD (coronary heart disease, stroke, heart failure) using Harrell's C-statistic and net reclassification improvement (NRI) based on Cox models. Results: Elastic net regression including log-transformed CAC volume selected 5 CAC density metrics (mean of density standard deviation, median of minimum density, maximum of minimum density, maximum of maximum density, minimum of maximum density across calcified plaques). Adding these 5 density parameters to traditional risk factors improved CVD risk prediction (ΔC-statistic 0.060 [95%CI 0.032, 0.093]) more than when we added Agatston score (0.046 [0.012, 0.093]) (top half of Table). These 5 CAC density parameters improved risk prediction even when Agatston score was included in the base model (the bottom row of Table). This result was similar for NRI. Conclusions: In the 75-and-older population, selected CAC density parameters significantly improved CVD risk prediction beyond and above Agatston score. Our findings underscore the potential of improving how to interpret and utilize CAC for clinical management in older populations.