Diagnostic Performance of Transabdominal Ultrasonography for Pancreatic Fat in Adults with Metabolic Risk Factors for NAFPD: Comparison with a Study-Defined Two-Echo MRI Index.

Background/

Objectives

To evaluate an exploratory ordinal transabdominal ultrasonography score against a study-defined semiquantitative two-echo chemical-shift MRI pancreatic fat index ≥ 5%.

Methods

This prospective paired diagnostic accuracy study included 150 adults aged 30-60 years with metabolic abnormalities or risk factors and technically adequate ultrasonography. A score ≥ 1 was the primary positive index-test result. Diagnostic accuracy, ROC AUC, and ordinal correspondence were evaluated.

Results

After nine clinical exclusions, pancreatic ultrasonography was technically evaluable in 150 of 170 participants (88.2%; 95% CI, 82.5-92.3%). For the study-defined MRI index ≥ 5%, sensitivity was 76.8% (95% CI, 66.6-84.6%), specificity was 77.9% (95% CI, 66.7-86.2%), accuracy was 77.3% (95% CI, 70.0-83.3%), and AUC was 0.812 (95% CI, 0.748-0.876). Exact four-category agreement was 45.3%, and quadratically weighted kappa was 0.504 (95% CI, 0.387-0.612).

Conclusions

These estimates describe agreement and discrimination relative to the study-defined two-echo MRI index in selected adults with metabolic risk and technically adequate ultrasonography. They should not be interpreted as diagnostic accuracy for histologically confirmed pancreatic steatosis or standardized MRI-PDFF-defined disease. The exploratory score is not a quantitative substitute for MRI and requires reproducibility assessment and external validation.