Ethnic Differences in the Diagnostic Accuracy of a Quantitative Ultrasound-Derived Fat Fraction Estimator for Hepatic Steatosis: External Validation in Asian and Caucasian Cohorts.

Abstract

Purpose: Ultrasound fat fraction estimation (USFF) using quantitative ultrasonography (QUS) is a promising method for assessing liver fat in metabolic dysfunction-associated steatotic liver disease (MASLD). However, its performance across different ethnicities remains unclear. This study aimed to externally validate and compare USFF accuracy in Asian and Caucasian cohorts.

Methods: This retrospective analysis used QUS data from three institutions. Participants underwent QUS and magnetic resonance imaging of the proton density fat fraction (MRI-PDFF), the reference standard. A multivariable regression model was developed using two QUS parameters. Linear regression was used to estimate fat fraction, with adjusted R² values assessing model fit. The diagnostic performance of USFF for hepatic fat content thresholds (≥ 5%, ≥ 10%, ≥ 20%) was evaluated using the area under the receiver operating characteristic curve (AUC).

Results: The development set included 427 South Korean participants, while the validation set comprised 100 from Hungary and 79 from Italy. R² values were 0.689, 0.702, and 0.485 for the development, internal validation, and external validation sets, respectively. AUC for detecting hepatic steatosis (MRI-PDFF ≥ 5%) was 0.939 (development) and 0.863 (external validation). Using a cutoff > 0.513, USFF achieved 85.1% sensitivity and 87.4% specificity.

Conclusion: USFF is effective but shows variability across populations, highlighting the need for ethnicity-specific models.