Diagnostic performance of ultrasound-derived fat fraction for detecting hepatic steatosis in patients with tumor-related liver disease

Abstract

Accurate assessment of hepatic steatosis is crucial for preoperative risk stratification in patients with liver tumors. However, the diagnostic performance of the noninvasive ultrasound-derived fat fraction (UDFF) remains uncertain, where tissue heterogeneity can impair measurement reliability. Using histopathology as the reference standard, this study aimed to evaluate the ability of UDFF to detect and grade background hepatic steatosis. In this prospective study, 64 adults scheduled for hepatic surgery were enrolled. All patients underwent preoperative UDFF assessment, and postoperative histopathology of non-tumorous liver tissue served as the reference. Diagnostic performance, incremental predictive value, and clinical utility were evaluated using ROC analysis (Youden index for cutoff determination), multivariable regression (ΔAUC assessment), and decision curve analysis. UDFF values increased progressively with histological steatosis grades ( p < 0.0001). The AUCs for detecting S ≥ 1 and S ≥ 2 steatosis were 0.885 and 0.967, with corresponding optimal cutoffs of 4.75% and 8.50%. UDFF was an independent predictor of steatosis (OR = 1.95 for S ≥ 1; OR = 2.58 for S ≥ 2) and significantly improved model discrimination (ΔAUC = 0.161 and 0.256). Decision curve analysis demonstrated a consistent net clinical benefit across clinically relevant threshold probabilities. UDFF provides accurate, independent, and clinically meaningful quantification of hepatic steatosis in tumor-related liver disease, and the population-specific cutoff values identified in this study may support improved preoperative evaluation and treatment planning.