AI-Driven Quantification of Coronary Plaque Outperforms Standard Care in Suspected CAD

Artificial intelligence (AI)-based quantitative coronary topography (CT) scans in patients with suspected coronary artery disease (CAD) may improve diagnostic certainty and expedite necessary treatment.¹

These data were presented at the American Heart Association (AHA) Scientific Sessions 2026 in Chicago, Illinois, by Todd Villines, MD, Donald C. Barnes Professor of Cardiology in Internal Medicine at the University of Virginia School of Medicine, executive editor of JACC Advances, and vice chief of clinical services in the cardiovascular division at UVA Health.¹

“While patients oftentimes do get definitive answers with stress testing, we know that many times it leaves them still questioning whether they have plaque,” Villines told HCPLive in an exclusive interview. “That diagnostic certainty can be important, because when providers order a test, they want a definitive answer.”

PARAMOUNT was a prospective randomized open-label trial evaluating the hypothesis that a personalized management strategy for patients with suspected CAD using an AI-powered, CT-based coronary atherosclerotic plaque assessment improves diagnostic certainty, control of risk factors, and efficiency of invasive coronary angiography (ICA) referral. Patients were eligible if they were >18 years old and were symptomatic with suspicion of CAD. Those with LDL <100 mg/dL, those who were currently or previously treated beyond primary prevention guidelines, and those with suspected acute coronary syndrome, among other criteria, were excluded.²

Patients were randomly assigned to receive either coronary plaque-based care or usual care – the former saw patients undergo CCTA with an AI-enabled quantitative assessment of morphology and composition of atherosclerotic plaque, while the latter saw patients referred to their usual care providers for standard-of-care management post-stress test findings. The primary endpoint was whether a management strategy improves efficiency and effectiveness – this was divided into multiple elements, including improved diagnostic certainty, reductions in LDL, freedom from ICA without obstructive CAD, and freedom from PCI without findings of significant stenosis.²

Villines and colleagues randomized a total of 530 patients between January 2025 and February 2026 across 21 participating sites. Among these patients, median age was 57 years, 57% were women, and median baseline LDL-C was 128 mg/dL. By 180 days, plaque-based care was shown to be favorable to standard-of-care (win ratio, 2; 95% CI, 1.37-2.56; P <.0001). Diagnostic certainty regarding the presence of CAD was the largest contributor to the primary endpoint, with a win ratio of 16.6%.³

Ultimately, the team concluded that AI-QCT plaque assessments may be able to give physicians greater certainty when evaluating chest pain. However, these data were still preliminary – the full results are reportedly expected in early 2027.³

“The ability to send these data sets and rapidly analyze them and get a standardized report back is only going to get quicker and faster and more widely utilized with time,” Villines said. “We think that, with the growing number of coronary CT scans, that AI will help reading physicians be more efficient: they’ll help with pre-reads, help with plaque staging, things that will hopefully make their job easier.”

Editors’ Note: Villines reports being a consultant for Cleerly, HeartFlow, and Milestone.

References

1. Villines T, Foldyna B, Aquino M, et al. A Randomized Comparison of Personalized Management for Medical and Interventional Therapy Based on Coronary Atherosclerotic Plaque versus Usual Care for Symptomatic Patients With Suspicion of Coronary Artery Disease: Primary Results of the PARAMOUNT Trial. Presented at the American Heart Association (AHA) Scientific Sessions 2026, Chicago, IL. November 6-9, 2026.

2. Cleerly. A Randomized Comparison of Personalized Therapy Mgmt Based On Coronary Atherosclerotic Plaque Vs. Usual Care for Symptomatic Patients With Suspicion of CAD (PARAMOUNT). ClinicalTrials.gov Identifier: NCT06713239. Updated June 5, 2026. Accessed November 7, 2026. https://clinicaltrials.gov/study/NCT06713239?term=NCT06713239&rank=1

3. PARAMOUNT, the First Randomized Trial of AI-Enabled Plaque-Based Care, Meets Primary Endpoint in Late-Breaking Results at AHA 2026. BusinessWire. November 7, 2026. Accessed November 7, 2026. https://www.businesswire.com/news/home/20261107769280/en/PARAMOUNT-the-First-Randomized-Trial-of-AI-Enabled-Plaque-Based-Care-Meets-Primary-Endpoint-in-Late-Breaking-Results-at-AHA-2026