Abstract / Summary
Whether quantitative liver steatosis and stiffness provide vascular-risk information beyond conventional cardiometabolic factors remains uncertain. We evaluated their multivariable-adjusted associations and incremental utility for identifying subclinical carotid atherosclerosis. This retrospective cross-sectional study included 8948 adults undergoing routine health examination, carotid ultrasonography, and liver elastography at the Second Xiangya Hospital between 2022 and 2025. Controlled attenuation parameter (CAP) and liver stiffness measurement (LSM) were assessed simultaneously. Multivariable logistic regression, restricted cubic splines, threshold and joint-phenotype analyses, inverse-probability-of-selection weighting, repeated fivefold cross-validation, internal temporal validation, decision-curve analysis, and explainable machine-learning analyses were performed. Subclinical carotid atherosclerosis was present in 3997 participants (44.7%), and strict carotid plaque in 2750 (30.7%). After adjustment for the available cardiometabolic, renal, hepatic, and hematologic covariates, CAP was not associated with the primary outcome (OR per 40 dB/m, 1.02; 95% CI 0.94–1.09; P = 0.687), whereas LSM showed a weak residual association (OR per doubling, 1.15; 95% CI 1.00–1.31; P = 0.044). The LSM association was attenuated after multiple imputation and was not statistically significant for either strict carotid plaque or isolated carotid intima–media thickening. Selection-weighted estimates were essentially unchanged. Adding CAP and LSM to clinical variables produced negligible changes in discrimination, calibration, prediction error, or net benefit during repeated cross-validation and internal temporal validation, and limited hyperparameter tuning did not materially alter this finding. LSM showed only a weak and outcome-sensitive residual association with subclinical carotid disease, whereas the CAP association was largely accounted for by shared cardiometabolic factors. Neither measure provided meaningful incremental information beyond routinely available clinical variables for selecting asymptomatic adults for carotid ultrasonography in similar health-examination settings.