Abstract Background Coronary computed tomography angiography (CCTA) is a non-invasive modality assessing lumen size and plaque characteristics. While CCTA reliably quantifies plaque burden and stenosis, its accuracy in determining distal reference diameters and its concordance with optical coherence tomography (OCT) remain uncertain. With the increasing use of AI-driven plaque quantification to support PCI, validation against OCT and identification of CCTA-derived predictors of suboptimal stent outcomes are needed. Methods The ISOS-PCI registry is a multicentre, multinational, retrospective registry of patients with chronic coronary syndromes (CCS) who underwent OCT-guided percutaneous coronary intervention (PCI) within three months after CCTA performed for suspected coronary artery disease (CAD). AI-driven quantification of plaque components (low-attenuation, fibrotic, calcified) was performed on both CCTA and OCT in anatomically matched segments identified by landmarks (Figure 1). The primary aim of this preliminary analysis was to compare the concordance between two AI-driven approaches for plaque characterization by CCTA and OCT. The overall objective of the registry is to evaluate the accuracy of blinded CCTA-based virtual stent planning in predicting the actual stent size and length selected with OCT guidance during PCI, and to identify CCTA-derived predictors of suboptimal stent outcomes (e.g., inadequate expansion). Results This preliminary ISOS-PCI analysis included 15 CCS patients (30 anatomically matched CCTA–OCT segments). Concordance between CCTA- and OCT-derived plaque composition varied by tissue type. Lipid volume (r=0.52, p=0.02) and fibrous tissue % (r=0.47, p=0.03) showed moderate correlations, whereas calcium volume (Pearson r=0.37, p=0.11) and % (Pearson r=0.29, p=0.22) did not, with CCTA consistently overestimating calcium (OCT 5% IQR 6 vs CCTA 42% IQR 29). Average CCTA-derived HU attenuation correlated with OCT calcium % (r=0.57, p=0.01) and volume (r=0.58, p=0.01) (Figure 2). Perivascular fat attenuation correlated negatively with maximal calcium arc (r=-0.51, p=0.02), while maximal HU attenuation of calcified components correlated positively with calcium thickness (r=0.47, p=0.02). Conclusions In this preliminary analysis of the ISOS-PCI registry, AI-driven plaque quantification by CCTA demonstrated variable concordance with OCT by plaque composition. Moderate correlations were observed for lipid volume and fibrous tissue percentage, whereas calcium quantification showed weaker agreement. This was largely explained by CCTA semi-automatic analysis extending beyond the inner vessel layers and incorporating peri-vascular calcification, leading to systematic overestimation of calcium compared with OCT. Importantly, additional CCTA-derived attenuation metrics correlated significantly with OCT measures of calcium burden and thickness, suggesting their potential role as surrogate markers for procedural planning.Case example showing the methodologyFor image description, please refer to the figure legend and surrounding text. Spearman correlationFor image description, please refer to the figure legend and surrounding text.
Bianchini et al. (Sun,) studied this question.