Scott Buzby , 2025-04-18 13:51:00
April 18, 2025
3 min read
Key takeaways:
- AI-assisted coronary CT was more predictive of major CV events vs. standard risk scores for patients referred for coronary imaging.
- AI quantitative CT may help guide sex-specific treatment of symptomatic CAD.
CHICAGO — AI-assisted quantitative coronary computed tomography was more prognostic of major adverse CV events vs. standard risk scores among patients referred for imaging, especially for women with CAD, a speaker reported.
The results of the global, multicenter CONFIRM2 trial were presented at the American College of Cardiology Scientific Session and simultaneously published in Circulation: Cardiovascular Imaging.
AI-assisted coronary CT was more predictive of major CV events vs. standard risk scores for patients referred for coronary imaging. Image: Adobe Stock
“Women are underdiagnosed and undertreated for coronary artery disease. They have a distinct clinical presentation and suffer from adverse outcomes. Clinical risk scores have a lower accuracy and their performance in women is even weaker. Sex-specific differences in the adverse chronic plaque profile by CT have been reported, but data on their direct association with cardiovascular risk are limited,” Gudrun M. Feuchtner, MD, MBA, associate professor at Innsbruck Medical University in Innsbruck, Austria, said during a presentation. “AI-based quantitative coronary computed tomography (AI-QCT) is a novel technology … which allows for a quantification of the total plaque volume, as well as a subanalysis of the components of atherosclerosis. Noncalcified plaque can be distinguished from calcified and further from high-risk plaque features, among 16 other features of atherosclerosis. Because these features may enhance cardiovascular risk prediction, the purpose of this study was to define sex-specific patterns of atherosclerosis by AI-QCT for prediction of [major adverse CV events] compared to clinical risk scores.”
Their study included 3,551 patients with suspected CAD who were subsequently referred for coronary CT angiography (mean age, 59 years; 50% women).
The AI was utilized to measure plaque features including total plaque volume, noncalcified plaque volume, calcified plaque volume, low attenuation plaque volume, high-risk plaque, percentage atheroma volume and stenosis severity.
The primary endpoint was major adverse CV events including death, MI, late revascularization, cerebrovascular events, unstable angina and congestive HF.
During a mean follow-up of nearly 5 years, the primary endpoint occurred in 3.2% of women and 6.1% of men.
AI quantitative coronary CT-derived total plaque volume, noncalcified plaque, calcified plaque and percentage atheroma volume were associated with higher risk for the primary endpoint among women compared with men.
For every 50 mm3 increase, relative risk for the primary endpoint increased:
- 17.7% for total plaque volume (95% CI, 12%-24%) for women compared with 5.3% for men (95% CI, 3%-7%; P for interaction < .001);
- 27.1% for noncalcified plaque (95% CI, 17%-38%) for women compared with 11.6% for men (95% CI, 8%-15%; P for interaction = .0015); and
- 22.9% for calcified plaque (95% CI, 14%-33%) for women compared with 5.4% for men (95% CI, 1%-10%; P for interaction = .0012).
Feuchtner reported significant incremental improvement in the prediction of major adverse CV events for women when the AI model included evaluation of CV risk factors, total plaque volume and high-risk plaque (area under the curve = 0.791; 95% CI, 0.74-0.86; P vs. risk factors alone = .0046) as well as diameter stenosis (AUC = 0.797; 95% CI, 0.73; 0.85; P vs. risk factors alone = .0055) compared with risk factor evaluation alone (AUC = 0.668; 95% CI, 0.59-0.747).
Results were similar when AI quantitative coronary CT measures were added to Diamond-Forrester risk prediction of the primary endpoint compared with Diamond-Forrester alone, according to the presentation.
“Despite the total AI quantitative CT-derived plaque burden [being] higher in men, similar volumetric increments in AI quantitative CT-derived features confer a significantly higher relative risk [for major adverse CV events] in women, and this must be considered when using these features as imaging biomarkers, whereas the performance of clinical risk score was poor to moderate,” Feuchtner said during the presentation. “This approach using AI quantitative CT feature-based risk stratification, instead of relying on traditional scores, has the potential to enhance the precision of risk prediction, and as a result, also enables tailored personalized preventive interventions, for example, such as reinforced anti-atherosclerotic therapy, lower LDL target targets and other measures.”
Reference:
Sources/Disclosures
Collapse
Feuchtner GM, et al. Clinical and investigative horizons 1. Presented at: American College of Cardiology Scientific Session; March 29-31, 2025; Chicago (hybrid meeting).
Disclosures:
CONFIRM2 was funded by Cleerly Inc. Feuchtner reports no relevant financial disclosures.