# Validation of BMP8A fibrosis score to identify patients with metabolic dysfunction-associated steatohepatitis with advanced liver fibrosis

## TL;DR

This study validates a new non-invasive test called BFS to accurately identify patients with severe liver fibrosis in a liver disease called MASH.

## Contribution

The study introduces and validates the BMP8A Fibrosis Score (BFS) as a more accurate and reliable tool for detecting advanced liver fibrosis in MASH patients.

## Key findings

- BFS outperformed other fibrosis scores like FIB-4, NFS, APRI, and HFS in correctly classifying patients with advanced liver fibrosis.
- BFS uses a single cut-off value, reducing the number of patients with indeterminate results compared to other methods.
- BFS achieved an AUROC of 0.750 and correctly classified 70.9% of advanced fibrosis cases.

## Abstract

Liver fibrosis represents the main risk factor not only for liver-related but also for overall mortality in metabolic dysfunction-associated steatotic liver disease (MASLD) patients, being metabolic dysfunction-associated steatohepatitis (MASH) its more severe clinical form. We recently developed a non-invasive algorithm termed BMP8A Fibrosis Score (BFS) which is able to identify MASH patients with advanced liver fibrosis. The aim of this study was to validate the BFS comparing its diagnostic accuracy with that of other scoring systems developed to assess liver fibrosis in MASH patients. Serum BMP8A was measured in 302 patients with biopsy-proven MASH: 171 with non- or mild fibrosis (F0-F2) and 131 with advanced fibrosis (F3-F4) recruited from seven university hospitals located in different cities in Spain. BFS, Fibrosis-4 (FIB-4) Index, NAFLD Fibrosis Score (NFS), Hepamet Fibrosis Score (HFS), and AST-to-Platelet Ratio Index (APRI) were calculated for each patient. The diagnostic accuracy of the scoring systems was determined according to the area under the receiver operating characteristic (AUROC) curve, sensitivity, specificity, positive (PPV) and negative (NPV) predictive values, and likelihood ratios (LR). BFS showed higher overall accuracy than the other liver fibrosis algorithms calculated in the study cohort, presenting an AUROC of 0.750 for predicting advanced liver fibrosis (F3-F4), and correctly classifying 70.9% of F3-F4 patients with a sensitivity of 58.0%, a specificity of 80.7%, a 71.5% NPV, a 69.7% PPV, a 3.0 LR+, and a 0.5 LR-; the other predictive scores correctly classified a lower percentage of these patients (63.6% for FIB-4 ≥ 2.67, 63.2% for HFS ≥ 0.47, 57.3% for APRI ≥ 1.5 and 56.9% for NFS ≥ 0.675). BFS eliminates the grey area as it uses a single cut-off value (0.46), which is its key advantage over the others, reducing the number of patients with undetermined results (43.4% for FIB-4, 39.1% APRI, 37.4% for HFS, and 24.1% NFS). In sum, BFS properly classified more patients with advanced liver fibrosis (F3-F4) than the other scoring systems, eliminating indeterminate results and improving risk stratification.

The online version contains supplementary material available at 10.1186/s40364-025-00862-3.