Association of ultrasound-derived fat fraction with carotid intima-media thickness and carotid plaque in patients with type 2 diabetes: an exploratory cross-sectional study

Abstract

To investigate the associations of continuous ultrasound-derived fat fraction (UDFF) with carotid intima-media thickness (CIMT) and prevalent carotid plaque in patients with type 2 diabetes mellitus (T2DM), assess the incremental statistical information provided by UDFF in primary and broader traditional vascular-risk adjustment models, and compare its within-cohort discrimination with conventional qualitative liver ultrasonography. This single-center retrospective cross-sectional study included 153 patients with T2DM evaluated in the Department of Endocrinology, Affiliated Hospital of Jiangsu University, between September and December 2025. Continuous UDFF was the primary exposure. CIMT was analyzed using linear regression and prevalent carotid plaque using logistic regression. The primary multivariable model adjusted for age, sex, statin therapy, and low-density lipoprotein cholesterol (LDL-C). Each 1-percentage-point increase in UDFF was associated with a 0.0105-mm higher CIMT (95% CI, 0.0080–0.0131; P<0.001) and 10.9% higher odds of prevalent carotid plaque (OR = 1.109, 95% CI, 1.056–1.164; P<0.001) after multivariable adjustment. Associations remained materially similar in broader covariate models and after excluding 28 participants with prior atherosclerotic cardiovascular disease. Adding UDFF to the primary clinical adjustment model increased R² for CIMT from 0.274 to 0.498 and the C-statistic for carotid plaque from 0.633 to 0.747 (DeLong P = 0.018). In the broader traditional vascular-risk model, the C-statistic increased from 0.697 to 0.758 (ΔC-statistic=0.061; DeLong P = 0.045). In otherwise identical clinical models, discrimination for prevalent carotid plaque was greater with continuous UDFF than with the original four-category conventional ultrasonographic steatosis grade (0.747 vs 0.657; DeLong P = 0.008). Higher UDFF was associated with greater CIMT and prevalent carotid plaque in patients with T2DM. UDFF provided incremental statistical information within this cohort, although the magnitude of improvement in plaque discrimination was smaller in the broader traditional vascular-risk model than in the primary clinical adjustment model. Continuous UDFF also showed greater within-cohort discrimination than conventional four-category ultrasonographic steatosis grading. These cross-sectional findings do not establish causality or prospective cardiovascular risk prediction.