P467 - LBA_ECE_1195 - Improved detection of advanced fibrosis using a metabolically informed model in the MASLD era: comparison with FIB-4
TL;DR
The proposed model demonstrated superior discriminative ability compared to FIB-4 and improved risk stratification, with a lower prevalence of advanced fibrosis in the low-risk group and better separation across intermediate and high-risk categories.
Abstract
The transition from nonalcoholic fatty liver disease (NAFLD) to metabolic dysfunction–associated steatotic liver disease (MASLD) reflects a paradigm shift emphasizing the central role of metabolic dysfunction in liver disease progression. While non-invasive scores such as FIB-4 are widely used for risk stratification, their performance may be suboptimal in metabolically high-risk populations. We aimed to develop and evaluate a metabolically informed model for predicting advanced fibrosis and to compare its performance with FIB-4. In this cross-sectional study, we included 346 adults with body mass index (BMI) ≥25 kg/m2 who underwent liver stiffness assessment using transient elastography. Advanced fibrosis was defined according to established liver stiffness measurement thresholds. Multivariable logistic regression analysis was performed to identify independent predictors of advanced fibrosis, including liver enzymes, fasting glucose, and prediabetes/diabetes status. Model performance was assessed using receiver operating characteristic (ROC) analysis and compared with FIB-4. Risk stratification performance was evaluated across predefined low-, intermediate-, and high-risk categories. Advanced fibrosis was identified in a subset of patients characterized by a high metabolic burden. In multivariable analysis, prediabetes/diabetes emerged as the strongest independent predictor of advanced fibrosis (OR 4.18, 95% CI 1.87-9.36, P < .001). Additional independent predictors included AST (OR 1.09), GGT (OR 1.01), and fasting glucose (OR 1.01), while ALT showed an inverse association (OR 0.96). Notably, BMI was not associated with advanced fibrosis (P = .355). The proposed model demonstrated superior discriminative ability compared to FIB-4 (AUC 0.816 vs 0.711). Furthermore, the model improved risk stratification, with a lower prevalence of advanced fibrosis in the low-risk group (3.8% vs 15.3%) and better separation across intermediate and high-risk categories. In patients with overweight or obesity, advanced fibrosis is primarily driven by metabolic dysfunction rather than adiposity itself. A metabolically informed model incorporating glycemic status and liver enzymes outperforms FIB-4 in both discrimination and risk stratification. These findings support the conceptual shift from NAFLD to MASLD and highlight the importance of integrating metabolic parameters into fibrosis risk assessment.Figure 1Model performance and predicatiors of advanced fibrosis in MASLD. (A) ROC curve comparison swhowing improved discrimination of the proposed model compared to FIB-4. (B) Prevalence of advances fibrosis across risk categories demonstrating improved stratification by the proposed model. (C) Forest plot of multivariable logistic regression identifying independent predictors of advanced fibrosis