Semi-quantitative coronary computed tomography angiography plaque scores predict adverse cardiovascular outcomes: a systematic review and meta-analysis

Background Semi-quantitative Coronary Computed Tomography Angiography (CCTA) scoring systems - including the Segment Involvement Score (SIS), Segment Stenosis Score (SSS), CT-adapted Leaman score (CT-LeSc), Leiden score, and the Coronary Artery Disease-Reporting and Data System (CAD-RADS) are used to quantify atherosclerotic burden, however their relative performance and suitability for standardised clinical application remain uncertain. Purpose To evaluate and compare the associations of commonly used semi-quantitative CCTA scoring systems for major adverse adverse cardiovascular outcomes (MACE). Methods We conducted a systematic review and meta-analysis for studies reporting associations between semi-quantitative CCTA scores and MACE. Random effects models were used to derive pooled hazard ratios (HR) for SIS, SSS, CT-LeSc and Leiden scores, and a risk ratios (RR) for CAD-RADS. Results Thirty-seven studies, examining 54,719 patients, were analysed, and all scoring systems were significantly associated with MACE. SIS showed the largest per-unit association (HR 1.25, p<0.001), while SSS showed a similar association (HR 1.18, p<0.001). The CT-LeSc (HR 1.09, p<0.001) and Leiden scores (HR 1.11, p<0.001) demonstrated comparable associations, while CAD-RADS showed a stepwise gradient (RR 1.71 per category, p<0.001). Associations remained directionally consistent in analyses restricted to hard MACE. To enable cross-score comparison, HRs were rescaled to represent MACE risk per 1-standard deviation increase in each score. CT-LeSc and Leiden scores showed the most consistent associations with low heterogeneity (both I²=0%), whereas SIS, SSS and CAD-RADS retained moderate-to-high heterogeneity. Conclusion Overall, semi-quantitative CCTA scores were associated with subsequent cardiovascular events, and anatomically weighted scores (CT-LeSc and Leiden) showed more consistent associations, while simpler metrics such as SIS, SSS and CAD-RADS supported broad risk stratification. By providing a reproducible estimate of plaque burden, these scores have potential utility for longitudinal monitoring of disease progression and treatment response on follow-up imaging. Further within-cohort, head-to-head comparisons and evaluation of clinically actionable thresholds are warranted.