Associations of AI-derived coronary CT angiography features with CZT-SPECT coronary flow reserve and FFR-CT

Objective: This study aims to explore the associations of artificial intelligence (AI)-derived coronary CT angiography (CCTA) features with coronary flow reserve (CFR) measured by cardiac-cadmium zinc-telluride single-photon emission computed tomography (CZT-SPECT) and CT-derived fractional flow reserve (FFR-CT), and to investigate their intrinsic relationships. Methods: This retrospective study included 251 patients (753 vessels) with suspected or known coronary artery disease (CAD), who underwent CZT-SPECT and concurrent CCTA. Myocardial ischemia was defined as CFR <2.0 or FFR-CT ≤0.8. Generalized estimating equations (GEE) were used to analyze the associations between CCTA coronary parameters and the two ischemia definitions. Results: < 0.05). Subgroup analysis revealed distinct plaque feature patterns among discordant CFR/FFR-CT statuses. The same set of coronary features achieved an adjusted AUC of 0.892 for FFR-CT-defined ischemia and 0.615 for CFR-defined ischemia. Conclusions: CFR and FFR-CT reflect different pathophysiological dimensions: CFR reduction is more associated with microvascular dysfunction in the context of inflammation and diffuse lesions, whereas FFR-CT mainly reflects focal, obstructive ischemia caused by high-risk plaques (such as low attenuation plaque). CCTA is an important tool for assessing obstructive coronary lesions, but coronary features alone are insufficient to predict whether CFR is abnormal or not. However, in the absence of invasive reference standards (invasive FFR and index of microcirculatory resistance), these findings should be considered hypothesis-generating and require confirmation in future studies incorporating invasive physiological assessment.