Analysis of risk factors and establishment of a predictive model for liver disease severity in metabolic dysfunction–associated fatty liver disease

This is the authors' abstract. We don't add key points for this paper.

Objective To identify risk factors associated with liver disease severity in patients with metabolic dysfunction–associated fatty liver disease (MAFLD) and develop a risk prediction model. Methods Clinical data were collected from MAFLD patients diagnosed via liver biopsy at the Second Department of Hepatology, Beijing Ditan Hospital, Capital Medical University, from January 2018 to December 2022. Patients were initially grouped by ALT levels to analyze its limitations in MAFLD diagnosis. Pathological reports were used to classify patients into non-significant (S/F/G < 2) and significant (S/F/G ≥ 2) groups based on fibrosis (S), inflammation (G), and steatosis (F) severity. Logistic regression identified independent risk factors for liver disease severity, and a predictive model was established. Model performance was validated using the area under the curve (AUC). Results Significant differences were observed between ALT-normal and ALT-abnormal groups in AST, GGT, ALP, BMI, hyperlipidemia prevalence, age, and diabetes prevalence ( p < 0.05). Univariate analysis revealed age, BMI, AST, ALB, ALP, TG, and high-calorie diet as significant variables ( p < 0.05) across inflammation, steatosis, and fibrosis subgroups. When stratified by inflammation severity, multivariate analysis identified sex, age, BMI, ALB, and PcIII as independent predictors of significant inflammation in MAFLD. High BMI and hyperlipidemia were risk factors for MAFLD steatosis. When grouped by fibrosis severity, age, height, BMI, ALB, and PcIII emerged as independent predictors of pathologically significant fibrosis in MAFLD. Patients with G ≥ 2 and/or S ≥ 2 were classified as the significant pathology group, while others formed the non-significant group. A logistic regression model was constructed using regression coefficients and constants, p = 1/(1 + e -Y ), Y = −12.486 + 0.913 × Sex +0.442 × BMI + 0.04 × PcIII-0.085 × ALB. The area under the ROC curve was 0.833 (95% CI: 0.797–0.869), with a maximum Youden index of 0.456, corresponding to sensitivity and specificity of 67.3 and 83.3%, respectively. Conclusion Sex, BMI, PcIII, and ALB are closely associated with MAFLD severity. The logistic regression model demonstrates potential clinical utility for predicting significant liver pathology.