AI plaque quantification in management of asymptomatic patients with coronary calcium: Results of the utility of plaque ANalysis for COronary vessel evaluation and risk-assessment (UNCOVER) study

Background Artificial Intelligence-enabled Coronary Plaque Analysis (AI-CPA, Heartflow) provides quantitative information from coronary computed tomographic angiography (CCTA) for total plaque and plaque volume subtypes. While AI-CPA has been well-studied in symptomatic patients, its clinical utility in asymptomatic patients with coronary calcium (CAC) has not been explored. Methods Three preventive cardiologists reviewed clinical summaries of 100 asymptomatic patients with positive CAC score or with incidental calcium noted on chest CT scans. An initial management plan (MP) for each case was established based on history, CAC score or incidental calcium, and CCTA report. AI-CPA was then provided for each patient, and a second MP was established. The primary endpoint was MP reclassification rate (RR) between treatment based on history, CAC findings, and CCTA vs with addition of AI-CPA. Results MP RR following AI-CPA review was 72/100 (72%, 95% CI 63.2%–80.8%). RR ranged from 57 to 80% across CAC score tiers, 54–79% across CAD-RADS categories, and 64–75% across LDL-C tiers. Hierarchical analysis showed medication escalation in 58% of patients. AI-CPA review was not associated with increased referral rates for coronary revascularization or non-invasive imaging. Physician confidence in selecting the optimal MP rose to 99.3%, with “very confident” ratings rising from 2 to 34% ( P < 0.0001) following AI-CPA review. Conclusion In asymptomatic patients with CAC who had already undergone CCTA, AI-CPA led to meaningful management changes and improved confidence in the clinical care plan. When added to CCTA, AI-CPA may provide a path towards personalized, preventive treatment in this asymptomatic at-risk patient population.