Sarah Rinehart, Ron Blankstein, Cian P. McCarthy, James L. Januzzi, Markus Scherer, Wesley T. O’Neal, Philip Green, Joseph Puma, Frank Corrigan, Jaydip Datta, Jens Eichhorn, Thomas Stuckey, Omar K. Khalique, Ziad Ali, Nicholas Ng, Whitney Huey, Sarah Mullen, Campbell Rogers, Leslee J. Shaw
BACKGROUND: Artificial intelligence-enabled coronary plaque analysis (AI-CPA) has been shown to improve cardiovascular risk prediction. However, little is known about how these measures influence management. OBJECTIVES: This study sought to define changes in management guided by AI-CPA as compared with management guided by measures of nonobstructive and obstructive stenosis on coronary computed tomographic angiography (CTA) alone. METHODS: The DECIDE (AI [Artificial Intelligence]-DErived Plaque Quantification: Coronary CTA and AI-QCPA [Artificial Intelligence-Derived Quantitative Coronary Plaque Analysis] for Determining Effective CAD [Coronary Artery Disease] Management) registry is a prospective, observational, pre-post interventional substudy. The substudy includes delayed release of AI-CPA findings to treating physicians until 90 days after the index coronary CTA, followed by an additional 90 days of follow-up. The primary outcome is change in management: modification of preventive/anti-ischemic therapies, new laboratory testing, referral to a specialist, or referral to stress testing/invasive coronary angiography post AI-CPA. RESULTS: A total of 972 symptomatic patients with atherosclerotic plaque were enrolled (median age 64 years [Q1-Q3: 56-72 years], and 50.2% were women). Changes in management following the availability of AI-CPA occurred in 51.3% (95% CI: 48.2%-54.5%) of participants and were more frequent in patients with more extensive plaque (up to 67.8% in the highest stage; P < 0.001). Intensifying medical therapy was the most common management change, occurring in 35.6% of participants. Participants with management changes realized greater reductions in low-density lipoprotein cholesterol than did patients without these changes (-11 mg/dL [Q1-Q3: -42.5 to 4 mg/dL] vs 1 mg/dL [Q1-Q3: -18 to 14 mg/dL]; P = 0.002). CONCLUSIONS: The DECIDE registry supports that AI-CPA was associated with preventive management changes, especially intensification of care for patients with more extensive plaque. Randomized trials to explore the utility of AI-CPA are warranted. (AI-DErived Plaque Quantification: Coronary CTA and AI-QCPA for Determining Effective CAD Management [DECIDE]; NCT06376851).