Coronary atherosclerosis on AI-based plaque analysis in patients with chest pain and calcium score zero.
Level 3 - non-randomized controlled study
Non-randomized observational cohort study with longitudinal follow-up
PubMed 41549197 · doi:10.1007/s10554-026-03623-x
What was done
Researchers analyzed 606 symptomatic patients (156 men, 450 women; median age 54 vs. 61) with suspected coronary artery disease (CAD) and a coronary artery calcium score (CACS) of zero from a cohort of 2,216 patients evaluated between 2006 and 2016. Coronary computed tomography angiography (CTA) was evaluated by both standard clinical reading and artificial intelligence-based quantitative analysis (AI-QCT). Patients were followed for a median of 7.3 years for clinical outcomes (myocardial infarction, unstable angina, or death).
What was found
Obstructive CAD was present in ~5% of men and 2% of women with zero CACS on both clinical reading and AI-QCT. AI-QCT detected non-obstructive CAD in 88.5% of men and 85.6% of women, compared to 16.7% of men and 11.6% of women by clinical reading. Non-calcified plaques dominated, though small calcifications were detected. Over a median 7.3 years of follow-up, 3 men and 14 women experienced clinical events, yielding annual event rates of 0.26% in men and 0.43% in women.
Why it matters
AI-QCT reveals that non-obstructive plaque is present in the vast majority of symptomatic patients with zero calcium score, though their long-term clinical prognosis remains favorable.
Limits
Single-center retrospective study. The absolute number of clinical events was very low (17 events total across 606 patients over 7.3 years), precluding robust multivariable risk modeling.
Cited by
- supports Coronary artery calcium (CAC) scoring only detects calcified arterial plaque and cannot identify soft (non-calcified) plaque.