Refining risk stratification in patients with a small burden of coronary atherosclerosis by AI-guided quantitative CCTA
Y Ding, P A Kamila, R Young, J J Bax, A van RosendaelAbstract
Background
Artificial intelligence–guided quantitative coronary computed tomography angiography (AI-QCT) enables comprehensive coronary plaque quantification and has demonstrated strong prognostic value in coronary artery disease (CAD). A validated CAD staging system shows stepwise risk increase with greater total plaque volume (TPV). Although patients with CAD stage 1 (TPV 1–250 mm³) are considered low risk, their event rates are approximately twofold compared with those without plaque. However, whether detailed AI-QCT plaque characteristics further stratify risk within this early-stage population remains insufficiently explored.
Purpose
To determine whether AI-QCT–derived plaque characteristics provide incremental prognostic stratification among patients at stage 1 CAD.
Methods
CONFIRM2 is a multicenter, international, observational cohort study of symptomatic patients without prior CAD undergoing clinically indicated CCTA with follow-up for major adverse cardiovascular events (MACE). Coronary plaque presence, extent, and composition were quantified using AI-QCT. The primary endpoint was MACE, defined as all-cause mortality, myocardial infarction, stroke, heart failure, late revascularization, or hospitalization for unstable angina. The present analysis focused on patients with TPV 1–250 mm³. Associations between plaque characteristics and MACE were evaluated using Cox proportional hazards models.
Results
Among 6,550 patients (58.6±12.1 years; 48.6% male; median follow-up 4.4 years), 4,901 had stage 1 CAD (TPV 1–250 mm³) with a 4-year MACE rate of 3.5%, compared with 1.7% in TPV 0 and 13.3% in TPV 250–750 mm³ (Table 1). Within stage 1, risk increased across TPV tertiles (2.2%, 2.7%, and 5.6%; HR tertile 3 vs 1: 2.44, 95% CI 1.64–3.63). Larger individual plaques were associated with higher risk, including any plaque >50 mm³ (HR 2.21, 95% CI 1.62–3.02) or >100 mm³ (HR 2.65, 95% CI 1.82–3.85). Obstructive stenosis (>50%) (HR 2.26, 95% CI 1.61–3.18), greater plaque number, larger individual plaque, longer plaque length, and increasing vessel diffuseness showed stepwise associations with MACE (Figure 1 and Table 1), with the highest diffuseness quartile conferring a 3.29-fold higher risk versus the lowest. However, adding diffuseness to TPV tertiles did not significantly improve discrimination in stage 1(Harrell’s C 0.625 vs 0.629).
Conclusion
Among patients with early-stage CAD, AI-QCT–derived measures of plaque diffuseness provide meaningful incremental prognostic stratification. Even within this conventionally low-risk group, greater plaque volume, larger individual plaques, and more diffuse disease are associated with substantially increased MACE risk. Comprehensive plaque characterization may therefore improve risk assessment and support earlier preventive strategies in early CAD.