Are artificial intelligence derived coronary atherosclerotic characteristics associated with subsequent development of acute coronary syndrome (ACS)

Abstract Funding Acknowledgements Type of funding sources: Public hospital(s). Main funding source(s): Scripps Health Krueger Wyeth Grant Background ACS results from unstable atherosclerotic plaques. Current measures to assess risk do not consider the patient’s coronary atherosclerotic composition. This was a case control study using a prospective single-center registry of patients undergoing CCTA (N=2,700) from 2017–2020 to assess the association between plaque characterization and the occurrence of ACS. Methods ACS cases were adjudicated independently by two cardiologists using data from the EMR. Patients with prior ACS prior to CCTA, were excluded. Controls were selected 1:1 by nearest neighbor propensity matching (R. v. 4.0.3.) based on age, sex, smoking, HTN, HLD, and DM. Plaque burden and stenosis features were calculated by a commercially available artificial intelligence (AI) coronary analysis (Figure 1). Results There were 21 cases and 21 controls. Mean age of 68, 43% women, 5% Hispanic, 29% smoking hx, 76% HTN, 81% HLD, 24% DM, CAD-RADS score 2.3, and CAC of 487. Median time from CT to ACS for cases was 329 days. Presence of total non-calcified plaque, calcified plaque, fibro-fatty plaque, maximum stenosis diameter percentage and total plaque were each associated with greater risk of ACS (Table 1). Conclusion AI assessment of CCTA plaque characteristics were associated with ACS. This study shows the association between AI derived coronary atherosclerotic features and development of ACS in patients undergoing CCTA and highlights the potential utility of using AI-driven plaque characterization to predict ACS.