Incremental Predictive Value of Deep Learning–Quantified Coronary Atherosclerotic Volume: A SCAPIS Cohort Analysis
Authors' conclusion
Automated deep learning-quantified TPV improved prediction of first coronary events beyond SCORE2, coronary artery calcium score, and manually derived coronary CTA measurements, and identified even small plaque volumes associated with increased risk in a general population without established atherosclerotic cardiovascular disease.
Study context
- Population
- Randomly selected men and women 50–64 years from a population-based cohort, without prior atherosclerotic CVD, with SCORE2 data and coronary computed tomography angiography (CTA) of acceptable quality287 coronary events
- Participants
- 23,314 participants
- Exposure
- Deep learning-quantified total plaque volume (TPV) added to SCORE2 and segment involvement score vs SCORE2 and segment involvement score alone
- Outcome
- CHD death or MI
- Follow-up
- Median 7.8 years
Methodological detailspopulation-based
- Population-based
Main finding
As reported in the abstract
C-statistic, SCORE2 + segment involvement score + TPV vs without TPV
0.798 vs 0.783; P = 0.022
Net reclassification index with TPV added
0.093
95% CI 0.041–0.146
HR, highest TPV category
6.36
95% CI 3.00–13.52
Observational data: associations, not effects.
Further findings
Adding stenosis ≥50% or any segment with only noncalcified plaque
- Did not materially improve the C-statistic
Noncalcified plaque volume (NCPV) instead of TPV
- Similar results
Original publication
Summarised from the published abstract; every number is checked against it