Abstract
Objective: To investigate the clinical value of multiparameteric quantitative ultrasound combined with a non-invasive prediction model for assessing high-risk steatohepatitis. Methods: One hundred and ninety-four cases with metabolic-associated fatty liver disease (MAFLD) who underwent liver biopsy in Huashan Hospital, Fudan University, from June 2021 to September 2022 were selected. Shear wave elastography (SWE), shear wave dispersion (SWD) imaging, and attenuation imaging (ATI) examinations were conducted in all patients before biopsy. High-risk steatohepatitis was defined as a total activity score of ≥4 in patients with steatohepatitis, hepatocellular ballooning, and liver lobular inflammation based on pathological hepatic steatosis, inflammatory activity, and fibrosis scoring system (SAF), and fibrosis stage≥F2. Binary logistic regression analysis was used to identify the factors influencing high-risk steatohepatitis. A predictive model for diagnosing high-risk steatohepatitis was constructed using R language. The DeLong test was used to compare the area under the curve between groups. Measurement data was compared between groups using the t-test or rank-sum test, and count data were compared between groups using the χ(2) test. Results: There were 46 cases (23.7%) with high-risk steatohepatitis. The quantitative ultrasound parameters included elastic modulus (OR=2.958, 95%CI: 1.889-4.883, P<0.001), dispersion coefficient (OR=1.786, 95%CI: 1.424-2.292, P<0.001) and attenuation coefficient (OR=42.642, 95%CI: 3.463-640.451, P=0.004). Serological indexes of fasting blood glucose (OR=1.196, 95%CI: 1.048-1.392, P=0.011), alanine aminotransferase (OR=1.012, 95%CI: 1.006-1.019, P<0.001), aspartate aminotransferase (OR=1.027, 95%CI: 1.014-1.042, P<0.001), γ-glutamyl transferase (OR=1.008, 95%CI: 1.001-1.017, P=0.041) and HDL cholesterol (OR=0.087, 95%CI: 0.016-0.404, P=0.003) were the factors influencing its progression. The AUCs of elastic modulus, dispersion coefficient, attenuation coefficient, multiparametric ultrasound model, serological index model, and ultrasound combined with serology model for the diagnosis of high-risk steatohepatitis were 0.764, 0.758, 0.634, 0.786, 0.773 and 0.825, respectively. The results of the DeLong test showed that the ultrasound combined with the serological model was significantly better than the serological index model and the elastic modulus, dispersion coefficient, and attenuation coefficient alone (P=0.024, 0.027, 0.038 and <0.001). Conclusion: The combination of multiparametric quantitative ultrasound is helpful for the non-invasive diagnosis of high-risk steatohepatitis and possesses great clinical significance.