Ultra-high-resolution CT versus invasive angiography for obstructive coronary artery disease: real-world and core laboratory diagnostic performance in a multicenter study

Background Imaging studies interpretation in clinical practice may differ substantially from centralized Core Laboratory assessment, particularly in complex patients with severe calcification or stents. In the CORE PRECISION study, ultra–high-resolution computed tomography (UHR-CT) and invasive coronary angiography (ICA) demonstrated similar diagnostic performance under standardized Core Lab conditions, although statistical non-inferiority was not formally met. Whether real-world interpretation narrows or widens this performance gap remains clinically relevant. Purpose To compare the diagnostic performance of UHR-CT and ICA to detect hemodynamically significant stenosis according to local versus centralized core lab assessment. Methods In a prospective multicenter study conducted in Brazil, Japan, and the United States, 173 patients with a calcium score ≥400, known coronary stenosis ≥ 50% or stents underwent UHR-CT prior to an invasive coronary angiography (ICA) due to suspected obstructive coronary artery disease (CAD). UHR-CT and ICA images were analyzed locally at each site and by blinded and centralized Core Labs. Stenosis ≥70% on UHR-CT or ICA were deemed obstructive, and functional significance (reference standard) was defined by an abnormal quantitative flow ratio. The diagnostic accuracy of UHR-CT and ICA for detecting significant stenosis was calculated based on both Core Lab and site-level assessments. Results Among the 173 patients (mean age 65.2 years, 72% male), obstructive CAD was present in 115 (66%). The mean calcium score in patients without stents (n=115) was 1,004 (± 1,108). In real-world site-level readings, ICA accuracy decreased from 83.2% to 74.0% (p for difference=0.009), whereas UHR-CT showed a more modest decline, from 80.3% to 74.0% (p for difference=0.04). The reduction in performance was more pronounced for ICA specificity, resulting in convergence of diagnostic accuracy between modalities in routine practice (Fig 1A). Intercontinental variability further highlighted differences between controlled Core Lab conditions and real-world interpretation (Fig 1B). Conclusion In patients with severe coronary calcification or stents, UHR-CT offered similar diagnostic accuracy to ICA in detecting patients with significant CAD. The diagnostic performance gap between UHR-CT and ICA was narrower in real-world practice than under Core Lab adjudication, suggesting that variability in routine interpretation may influence comparative effectiveness analyses and the pragmatic performance of UHR-CT may be closer to ICA than suggested by controlled trial conditions, with important implications for real-world decision-making and study design.