Coronary Computed Tomography Angiography (CCTA), currently the preferred non-invasive imaging modality for clinical diagnosis of coronary artery disease (CAD), not only enables intuitive assessment of coronary stenosis severity but also allows quantitative and qualitative analysis of coronary atherosclerotic plaque burden and phenotype. Quantitative CCTA integrated with artificial intelligence, particularly convolutional neural network algorithms, has achieved rapid and accurate quantitative diagnosis of atherosclerotic plaques. A large body of research has demonstrated that high-risk plaque features identified by CCTA (such as low-attenuation plaques, positive remodeling, spotty calcification, and the napkin-ring sign) are closely associated with the occurrence of acute coronary syndrome and can be used to predict major adverse cardiovascular events (MACEs). With advancements in image post-processing techniques, CCTA can also compute CT Fractional Flow Reserve (CT-FFR) via dedicated software, enabling non-invasive assessment of whether a lesion causes hemodynamic abnormalities and myocardial ischemia, thereby partially overcoming the limitations of purely anatomical evaluation. Notably, CCTA not only focuses on intravascular and vascular wall changes but also allows quantitative characterization of phenotypic changes in Perivascular Adipose Tissue (PVAT). By calculating the Perivascular Fat Attenuation Index (pFAI), it can reflect alterations in PVAT density and composition, indirectly indicating the inflammatory activity status of the local coronary artery wall, thus providing a novel imaging perspective for understanding CAD from the "inflammation and metabolism" viewpoint. Furthermore, CCTA can characterize the lesion patterns of coronary atherosclerosis based on physiological and geometric distribution characteristics of the vasculature. For instance, the Pullback Pressure Gradient (PPG) derived from CCTA allows comprehensive quantitative assessment of the longitudinal distribution of CAD vasculature, enabling the classification of CAD phenotypes as predominantly focal or diffuse. The CT-FFR gradient, defined as the peak instantaneous CT-FFR gradient per unit length of the vessel, has also been recently confirmed to possess incremental predictive ability for future coronary events, thereby more comprehensively describing coronary atherosclerosis in terms of both "lesion distribution pattern" and "local functional severity".
Coronary Angiography (CAG), the gold standard for diagnosing CAD, has traditionally relied on stenosis severity and SYNTAX score to assess the anatomical complexity of lesions and the difficulty of interventional treatment. However, anatomical information alone remains limited for predicting plaque stability and event risk. In recent years, emerging parameters such as Radial Wall Strain (RWS) extracted from single or multiple angiographic views have provided new tools for identifying high-risk plaques from a vascular mechanics perspective. Studies have shown that RWS can reflect the radial deformation characteristics of the arterial wall during the cardiac cycle and exhibits good correlation with vulnerable plaques and the risk of plaque rupture, holding promise as a supplementary indicator for risk stratification beyond traditional angiographic parameters.
Meanwhile, intravascular imaging techniques such as Intravascular Ultrasound (IVUS) and Optical Coherence Tomography (OCT) can visualize plaque composition, burden, and fibrous cap morphology with high resolution, more closely approximating the pathological "vulnerable plaque" phenotype. Coronary functional assessments, such as FFR, Quantitative Flow Ratio (QFR), Murray Law-Based QFR (μFR), and Coronary Angiography-derived FFR (caFFR), evaluate from a hemodynamic perspective whether a stenosis causes functional myocardial ischemia. The continuous development of these technologies has addressed the limitation of traditional CAG, which only assesses anatomical stenosis and cannot accurately distinguish between "ischemia-causing" and "non-ischemia-causing" lesions, thereby reducing the risks of overtreatment and missed lesions. Beyond epicardial large vessel disease, Coronary Microvascular Dysfunction (CMVD) is also a significant mechanism underlying clinically common myocardial ischemia. For assessing microcirculatory function, various non-invasive or minimally invasive functional indices have emerged in recent years, such as the Coronary Angiography-derived Index of Microcirculatory Resistance (caIMR), which comprehensively reflects myocardial ischemic burden from the perspectives of microcirculatory resistance and perfusion efficiency. This provides a novel solution to the clinical conundrum of "presence of ischemia and symptoms without significant angiographic stenosis".
In addition to imaging and functional information, routine clinical metabolic and inflammatory markers also play a crucial role in the development, progression, and prognostic evaluation of CAD. Examples include low-density lipoprotein cholesterol, triglyceride-glucose index, lipoprotein(a), glycated hemoglobin, high-sensitivity C-reactive protein, homocysteine, fasting blood glucose, and N-terminal pro-B-type natriuretic peptide. These indices reflect the body's overall metabolic status and cardiovascular risk from different dimensions, including lipid metabolism, glycolipid metabolic disorders, chronic inflammation, endothelial function, myocardial stress, and cardiac function. Studies have suggested that the combined application of metabolic and inflammatory markers with coronary imaging features facilitates more refined individualized risk stratification and prognostic prediction.
In summary, the current assessment of CAD is gradually shifting from a single focus on "anatomical stenosis" to a multidimensional comprehensive characterization encompassing "anatomical structure + functional evaluation + plaque phenotype + perivascular environment + systemic metabolic status". However, there remains a lack of systematic research on how to integrate the aforementioned multi-source information (such as CCTA plaque and functional parameters, CAG and intravascular imaging information, microcirculatory function parameters, and clinical metabolic indicators) into a unified risk prediction model, construct the model through retrospective analysis, and subsequently validate its predictive value and clinical feasibility in a prospective cohort. Therefore, this project aims to integrate coronary imaging features (including CCTA and CAG-related parameters), functional indices, and metabolic markers to construct a multimodal CAD risk prediction model. The model will be developed using a retrospective cohort and its performance validated using a prospective cohort, exploring its application value in clinical risk stratification and decision-making.
This study is led by the National Clinical Research Center for Cardiovascular Diseases and invited to be carried out by several Chinese centers including Beijing Anzhen Hospital.
