TOPLINE

A post hoc analysis of the SCOT-HEART trial demonstrated that AI-derived left ventricular myocardial radiomic phenotypes from cardiac CT angiography (CTA) improved the long-term prediction of fatal or non-fatal myocardial infarction (MI) beyond traditional cardiovascular risk factors and coronary artery assessments among patients with stable chest pain.

METHODOLOGY

• Researchers conducted a post hoc analysis of the SCOT-HEART trial — a multicentre, randomised controlled trial — to evaluate AI-derived myocardial radiomic phenotypes from cardiac CTA.
• A total of 1736 participants with stable chest pain (mean age, 57.7 years; 56.6% men) were included in the analysis, with a median follow-up duration of 8.6 years.
• The researchers used TotalSegmentator — an open-source deep learning-based tool — to segment the left ventricular myocardium and extracted 105 radiomic features using PyRadiomics, which were grouped into six clusters: global variance, local dimness, global intensity, morphologic structure, bright region, and local uniformity.
• Participants were divided into two cohorts: Cohort A (n = 109) included individuals free from coronary artery disease (CAD) and cardiovascular risk factors, and cohort B (n = 1627) included participants with evidence of CAD (a coronary artery calcium [CAC] score above 0 or coronary stenosis > 10%) or at least one cardiovascular risk factor.
• The researchers determined the prognostic value of myocardial radiomic phenotypes in predicting fatal or non-fatal MI beyond that of a baseline clinical model incorporating the ASSIGN score — which assesses the cardiovascular risk according to the Scottish Intercollegiate Guidelines Network recommendations — the CAC score, and obstructive CAD.

TAKEAWAY

• A total of 82 participants developed a fatal or non-fatal MI during follow-up.
• In the univariable analysis, male sex (hazard ratio [HR], 2.75; P = .0001), diabetes (HR, 2.09; P = .0068), hypertension (HR, 1.90; P = .0037), the ASSIGN score (HR, 1.26; P < .0001), and the CAC score (HR, 1.21; P = .0009) were linked to an increased risk for MI.
• In the multivariable analysis, male sex and hypertension were each linked to increased myocardial size and signal uniformity, older age was linked to a smaller myocardium, and CAC scores were linked to changes in the myocardial texture rather than structure.
• Adding myocardial radiomic phenotypes to a clinical model based on the cardiovascular risk score, CAC score, and obstructive CAD improved discrimination for future MI events, raising the cross-validated concordance index from 0.693 to 0.722 and significantly improving model fit (likelihood ratio test: P = .009).

IN PRACTICE

"[C]ardiac CTA-derived myocardial radiomic phenotypes may hold promise as novel biomarkers for long-term prediction of fatal and non-fatal myocardial infarction. This could offer a potential avenue to refine cardiovascular risk stratification beyond traditional coronary assessment," the authors wrote.

SOURCE

The study was led by Jian Chen, MSc, University of Cambridge, Cambridge, England. It was published online on September 10, 2026, in European Heart Journal - Cardiovascular Imaging.

LIMITATIONS

The study was restricted by the relatively small number of MI events, the homogeneous Scottish cohort, and the limited number of participants without traditional risk factors.

DISCLOSURES

The SCOT-HEART trial was supported by grants from the Chief Scientist Office of the Scottish Government Health and Social Care Directorates, with supplementary awards from the Edinburgh and Lothians Health Foundation Trust and the Heart Diseases Research Fund. Several authors reported receiving support through the Cambridge Trust Scholarship and the Royal Brompton and Harefield Hospitals Charity Fellowship, as well as support from the National Institute for Health and Care Research Cambridge Biomedical Research Centre and the British Heart Foundation Centre of Research Excellence.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.