AI-enhanced CTCA identified 16 out of 22 patients at moderate-to-high risk of cardiovascular events despite being classified as non-obstructive CAD by traditional methods.
Does AI-enhanced CTCA analysis impact clinical management and lipid modification prescribing in patients with 30-90% coronary stenosis?
While AI-enhanced CTCA can identify higher-risk patients within a non-obstructive CAD population, its implementation did not significantly alter lipid modification prescribing in this real-world service evaluation.
Absolute Event Rate: 0% vs 0%
Abstract Introduction Coronary artery disease (CAD) is a leading cause of premature death in the UK. Over the past six decades premature mortality rates have fallen dramatically. Recent trends, however, indicate a concerning reversal. Radiologically obstructive CAD accounts for only one third of cardiac deaths in the stable angina population with the majority of deaths occurring in patients with non-obstructive CAD. One method to risk stratify patients in the non-obstructive CAD cohort may be to apply artificial intelligence (AI) analysis of clinically indicated coronary computed tomography angiography (CTCA) studies. Several tools have shown early promise in identifying higher risk cohorts, however, real-world deployment of these tools remain limited. Purpose This evaluation investigates the real-world implementation and potential clinical impact of 3 AI-enhanced CTCA analyses, including fractional flow reserve computed tomography (FFR-CT), coronary plaque volume analysis, and a multi-factor tool including traditional and perivascular fat attenuation indices (CaRi-Heart), within an NHS district general hospital cardiothoracic centre. Methods A prospective service evaluation was conducted of 50 consecutive patients presenting to a Rapid Access Chest Pain Clinic in whom CTCA revealed 30-90% stenosis in a major epicardial vessel. Scans then underwent AI analyses from November 2024 to March 2025. Clinical outcomes were tracked via electronic patient records, examining adherence to existing and suggested clinical guidelines. Results Of the 50 patients evaluated (58% male, mean age 64 years). Amoung the 22 individuals with positive FFR-CT findings (≤0.80),11 had distal disease only, 3 required urgent angiography due to acute coronary syndrome (ACS), 3 underwent elective angiography, with 1 case requiring surgical intervention. Plaque volume analysis was completed in all 50 cases and indicated significant plaque volumes (≥565mm³) in 18 patients, 17 of whom were prescribed statins. Multi-factor scoring was completed in 22 patients and identified 16 patients at moderate-to-high risk (5% mortality over 8 years if no intervention provided) of whom 82% had non-obstructive CAD by conventional and FFR-CT assessments. Overall, 80% of patients were prescribed lipid modification agents with 8% citing intolerance. There was no significant difference in prescribing when stratified by AI risk tool. Conclusion AI-enhanced CTCA demonstrates the capacity to identify patients who may be at a higher cardiovascular risk within a non-obstructive CAD population, potentially guiding intensified preventive therapies. In this service evaluation, no difference was found in lipid modification between groups. Multimodal analysis is feasible within the NHS structure, however, this study does not delineate the clinical impact of the AI systems. Larger-scale, long term prospective studies are required to validate these findings and determine optimal intervention strategies.
Crichton et al. (Thu,) reported a other. AI-enhanced CTCA identified 16 out of 22 patients at moderate-to-high risk of cardiovascular events despite being classified as non-obstructive CAD by traditional methods.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: