Improving Cardiovascular Risk Assessment by Abdominal Adiposity Phenotyping on Non-contrast Computed Tomography

Abstract

Background/Objectives: Abdominal obesity represents a risk factor for cardiovascular disease and subcutaneous abdominal fat (SAF) and visceral abdominal fat (VAF) have distinct metabolic profiles. First, we aimed to develop a comprehensive model for cardiovascular risk assessment through phenotyping abdominal obesity on computed tomography (CT). Then, we identified the radiomic features of SAF and VAF that best discriminate patients with coronary calcium score (CCS) >0 from patients with CCS=0 and built a model to assess individual cardiovascular risk. Subjects/Methods: We performed a cross-sectional study of 1736 patients (57±10yo, 57% male) with no prior cardiovascular disease who had undergone a non-contrast CT for CCS and abdominal single slice acquisition between L4/L5-S1. Then we performed a texture analysis of SAF and VAF. Radiomic features were extracted from structures with mean attenuation values ranging from -150 to -50 HU. Results: Men were younger and had lower body mass index, but higher CCS values and higher waist circumference than women. Men accumulated more VAF, while women accumulated more SAF. Indexed SAF area added to a traditional risk model was the best performing model to detect CCS. The model adding VAFi/SAFi ratio to the traditional risk factors was the best to rule-out coronary calcification. In non-contrast CT, SAF has a brighter, heterogeneous and coarser texture than VAF. A radiomic based-model including the best SAF and VAF texture features provided the best AUC [0.74 (95% CI: 0.71-0.77)] to distinguish asymptomatic patients with CCS>0 from CCS=0 in cardiovascular risk assessment. Conclusions: SAF and VAFi/SAFi ratio significantly improved cardiovascular risk assessment. Low SAF expandability explains the greater accumulation of VAF and increased cardiovascular risk in men. The texture analysis of the SAF and VAF demonstrated the heterogeneity of the two compartments and improved the assessment of cardiovascular risk better than the SAF and VAF areas.