Pericoronary Adipose Tissue Imaging on Coronary CT Angiography: From Fat Attenuation Index to Radiomic Risk Phenotyping
Abstract
Coronary computed tomography angiography (CCTA) has evolved from an anatomical test into a platform for cardiovascular risk phenotyping. Pericoronary adipose tissue (PCAT), which is directly contiguous with the coronary adventitia, changes its composition in response to vascular inflammation and can be interrogated on routine CCTA. Importantly, PCAT attenuation, the fat attenuation index (FAI), and the standardized FAI score are related but nonequivalent measurements: Raw PCAT attenuation is a mean Hounsfield unit measurement within a defined PCAT region, whereas FAI denotes an attenuation‐based methodological construct and FAI score is a standardized, vessel‐specific output adjusted for technical, biological, and anatomical factors. Radiomics extends assessment beyond attenuation by extracting multivariable intensity, shape, and texture features. Selected radiotranscriptomically validated features within the fat radiomic profile (FRP) have been linked to fibrosis and vascularity, but this biological interpretation should not be generalized to all PCAT radiomic signatures. Outcome studies demonstrate clinically relevant associations for attenuation‐based and radiomic biomarkers, although effect sizes, incremental value, and validation differ across populations and analytic platforms. Translation is limited by retrospective evidence, acquisition and reconstruction heterogeneity, lack of universal raw‐attenuation cutoffs, dependence of some standardized outputs on proprietary software, and limited prospective outcome‐driven validation. This narrative review synthesizes the biological rationale, terminology, clinical evidence, technical reproducibility, and emerging technologies in PCAT imaging. We propose a vendor‐neutral reporting and validation framework and position PCAT biomarkers as complementary to coronary calcium, stenosis, plaque phenotype, CT‐derived physiology, and artificial intelligence–assisted risk stratification rather than as a hierarchy of prognostic tests.