Technical Acoustic Measurements Combined with Clinical Parameters for the Differential Diagnosis of Nonalcoholic Steatohepatitis

Background and aim: Diagnosing nonalcoholic steatohepatitis (NASH) is challenging. This study intended to explore the diagnostic value of multiple technical acoustic measurements in the diagnosis of NASH, and to establish a diagnostic model combining technical acoustic measurements with clinical parameters to improve the diagnostic efficacy of NASH. Methods: We consecutively enrolled 75 patients with clinically suspected nonalcoholic fatty liver disease (NAFLD) who underwent percutaneous liver biopsy in our hospital from June 2020 to December 2021. All cases underwent multiple advanced acoustic measurements for liver such as shear wave dispersion (SWD), shear wave speed (SWS), attenuation imaging (ATI), normalized local variance (NLV), and liver–kidney intensity ratio (Ratio) examination before liver biopsies. A nomogram prediction model combining the technical acoustic measurements and clinical parameters was established and the model is proposed to improve the diagnostic performance of NASH. Results: A total of 75 cases were included in this study. The classification of pathological grade for NASH was as follows: normal liver, (n = 15, 20%), nonalcoholic fatty liver (NAFL), (n = 44, 58.7%), and NASH, (n = 16, 21.3%). There were statistically significant differences in SWS (p = 0.002), acoustic coefficient (AC) (p = 0.018), NLV (p = 0.033), age (p = 0.013) and fasting blood glucose (Glu) (p = 0.049) between NASH and non-NASH. A nomogram model which includes SWS, AC, NLV, age and Glu was built to predict NASH, and the calibration curves showed good calibrations in both training and validation sets. The AUCs of the combined nomogram model for the training set and validation set were 0.8597 and 0.7794, respectively. Conclusion: There were statistically significant differences in SWS, AC, NLV, age and Glu between NASH and non-NASH. A nomogram model which includes SWS, AC, NLV, age and Glu was built to predict NASH. The predictive model has a higher diagnostic performance than a single factor model in the diagnosis of NASH and has good clinical application prospects.