Prediction of major adverse cardiovascular events (MACE) from disease progression in low-cost (no-cost) screening CT calcium score images

Coronary artery disease is a leading cause of death. Agatston score, which non-linearly sums coronary artery calcifications (CACs) in CT calcium score (CTCS) exams, is the best risk predictor for major adverse cardiovascular events (MACE). CTCS follow-up exams are recommended but over a broad time range (2-10 years). We have identified 3390 patients with both baseline and follow-up exams from University Hospitals of Cleveland’s no-cost CTCS (CLARIFY registry) study. This data provides an opportunity to assess the rate-of-change of disease progression (∆Ag/∆t) on prediction of MACE. We designed time-to-event Cox models using Agatston scores for baseline (model1) and follow-up (model2). Alternatively, we designed our progression model using baseline Agatston score, rate-of-change, and clinical features (model3). Our progression model gave improved performance (c-index: 79.4) as compared to baseline (68.8) and follow-up (66.8). The rate-of-change in progression model showed a risk effect with HR of 2.51[CI±95:1.68, 3.76; p<0.0001]. Hence, as compared to no-progression, an increase of 100-Ag/year increases MACE risk by 69%. Diabetic patients were at even greater risk with progression (HR=4.19). Smoking and female were non-significant features. Kaplan-Meier plots showed the importance of progression model over baseline Agatston model, with successful separation of high and low-risk patients. Our model correctly classified 67.3% of cases with net-reclassification-index, NRI = 0.149. Progression model results emphasize the importance of timely follow-up CTCS exams, particularly for patients at high risk, thereby improving early intervention and treatment strategies, something that conventional models solely based on initial characteristics fail to achieve.