Radiomics-Based Precision Phenotyping Identifies Unstable Coronary Plaques From Computed Tomography Angiography

Area of Science:

• Cardiovascular Imaging
• Radiology
• Artificial Intelligence in Medicine

Background:

• The morphological characteristics of coronary atherosclerotic plaques that confer increased clinical event risk remain debated.
• Identifying specific vulnerable plaques within at-risk individuals is a significant challenge in managing coronary artery disease (CAD).

Purpose of the Study:

• To precisely phenotype culprit and nonculprit lesions in myocardial infarction (MI) and lesions in stable CAD using coronary computed tomography angiography (CTA)-based radiomic analysis.
• To evaluate the additive value of radiomic features for discriminating culprit lesions beyond traditional high-risk plaque (HRP) characteristics and plaque volumes.

Main Methods:

• Prospective coronary CTA in 60 acute MI patients matched with 60 stable CAD patients.
• Qualitative assessment of HRP characteristics, followed by semiautomated plaque quantification and extraction of 1,103 radiomic features.
• Machine learning models were developed to assess the incremental diagnostic value of radiomic features.

Main Results:

• Culprit lesions exhibited significantly higher volumes of noncalcified plaque (NCP) and low-density noncalcified plaque (LDNCP) compared to stable CAD lesions.
• A substantial proportion of radiomic features (14.9%) were associated with culprit lesions, and 9.7% with highest-grade stenosis nonculprit lesions, after adjusting for plaque volumes.
• Radiomic features significantly improved the discrimination of culprit lesions when added to a model with HRP and plaque volumes (AUC 0.86 vs 0.76, P=0.004).

Conclusions:

• Culprit lesions in MI demonstrate distinct radiomic signatures compared to lesions in stable CAD, suggesting potential for precision phenotyping.
• Coronary CTA-based radiomics offers a non-invasive method to identify unique plaque phenotypes, potentially pinpointing individual vulnerable plaques within high-risk patients.