Vessel-level patterns and predictors of discordance between computed tomography coronary angiography and invasive coronary angiography

BACKGROUNDCoronary calcification is a recognized source of computed tomography coronary angiography (CTCA) stenosis overestimation.We evaluated whether discordance between CTCA and invasive coronary angiography (ICA) also shows clinically relevant vessel-territory patterns in a selected real-world paired-imaging cohort. AIMTo evaluate patient-and vessel-level CTCA diagnostic performance, quantify territoryspecific discordance, assess three-category stenosis agreement, and identify predictors of discordance in a selected paired-imaging cohort. METHODSWe systematically screened the hospital CTCA database for October 2019-December 2024 and included consecutive adults with paired CTCA and ICA within 120 days, subject to the study exclusions.After reviewing interval revascularisation, we excluded one patient in whom percutaneous coronary intervention (PCI) occurred after diagnostic ICA but before the subsequent CTCA, leaving 403 patients and 1612 major epicardial vessel observations.Severe/occlusive disease was the positive binary endpoint, with a left main coronary artery (LMCA)-specific severe threshold of ≥ 50% in both modalities.Patient-level predictors were assessed with logistic regression and vessel-level predictors with mixed-effects logistic regression using a patient random intercept. RESULTSICA demonstrated severe/occlusive disease in 204/403 patients (50.6%).Patient-level sensitivity, specificity, positive predictive value (PPV), and negative predictive value 4 / 41 (NPV) were 91.7% (95%CI: 87.0%-95.1%),35.2% (95%CI: 28.6%-42.2%),59.2% (95%CI: 53.5%-64.6%),and 80.5% (95%CI: 70.6%-88.2%),respectively.Across 1612 vessels, sensitivity was 80.8% (cluster-bootstrap 95%CI: 76.3%-85.1%),specificity 81.7% (95%CI: 79.5%-83.7%),PPV 50.4% (95%CI: 45.4%-55.2%),NPV 94.9% (95%CI: 93.5%-96.1%),and discordance 18.5% (95%CI: 16.6%-20.3%).At least one discordant vessel was present in 229 patients (56.8%).Discordance was highest in the left anterior descending artery (LAD) (31.8%) and lowest in the LMCA (2.5%).Three-category agreement was moderate (weighted kappa 0.500, 95%CI: 0.459-0.539).Calcium burden remained independently associated with discordance at the patient level [odds ratio (OR) = 1.66 per category, 95%CI: 1.03-2.73;P = 0.042] and vessel level (OR = 1.40, 95%CI: 1.04-1.87;P = 0.025).The LAD had higher adjusted odds of discordance than the right coronary artery (OR = 1.82, 95%CI: 1.32-2.53;Holm-adjusted P = 0.0003) and left circumflex artery (OR = 2.01, 95%CI: 1.44-2.80;Holm-adjusted P < 0.001). CONCLUSIONIn this selected cohort of patients who underwent both CTCA and ICA, greater visual coronary calcium burden was associated with anatomical discordance, and discordance varied by epicardial territory, being highest in the LAD.Because referral for ICA was selective, study readings were not independently blinded, and ICA stenosis was visually graded, the observed diagnostic metrics may not generalise to unselected CTCA populations.No downstream testing pathway, invasive-referral decision, or patient outcome was evaluated; the findings are descriptive and hypothesis-generating.