# Identification of glycolysis-related clusters and immune cell infiltration in hepatic fibrosis progression using machine learning models and experimental validation

## TL;DR

This study identifies key glycolysis-related genes linked to liver fibrosis and immune cell infiltration, using machine learning and experiments to define subtypes and their immune profiles.

## Contribution

A novel glycolytic activity-based model for HF risk stratification and identification of six core glycolysis-related genes with distinct immune infiltration profiles.

## Key findings

- Six core glycolysis-related genes (B3GNT3, CHST4, DCN, GPC3, SOX9, VCAN) were identified as key drivers of hepatic fibrosis progression.
- Three molecular subtypes of HF were defined based on glycolysis-related gene expression, with Cluster C showing enhanced adaptive immune infiltration.
- Glycolytic activity was confirmed in fibrotic mice and cells, with consistent upregulation of core glycolysis-related genes.

## Abstract

Although glycolytic reprogramming constitutes a fundamental driver of hepatic fibrosis (HF), its precise mechanistic contributions remain incompletely characterized. This investigation systematically identified molecular signatures of glycolysis-related genes (GRGs) in HF. We further developed a glycolytic activity-based model for HF risk stratification.

Integrated analysis of GEO datasets (GSE276114, GSE84044, GSE49541) identified differentially expressed genes (DEGs) associated with HF progression. Integrated weighted gene co-expression network analysis (WGCNA) with six machine learning algorithms to identify core GRGs genes associated with HF progression, and systematically characterized their biological functions and immunoregulatory roles through immune infiltration assessment, functional enrichment, consensus clustering, and single-cell differential state analysis. Glycolytic activity was evaluated in CCl4-induced fibrotic mice and TGF-β-stimulated LX-2 cells. Additionally, the expression of core GRGs was validated using immunohistochemical staining and RT-qPCR.

Through the intersection of WGCNA, DEGs, and GRGs, machine learning identified six core GRGs: B3GNT3, CHST4, DCN, GPC3, SOX9, and VCAN. Based on the core GRGs, three GRG-based molecular subtypes were defined. Cluster C, with higher expression of the core GRGs, exhibited significantly enhanced immune infiltration, particularly of adaptive immune cells compared to Cluster A and B. Cluster C comprised a mixed landscape of T cells, mast cells, and pro-fibrogenic cells, distinct from the innate immune-dominant profiles of Clusters A and B. Both in vivo and in vitro analyses demonstrated enhanced glycolysis in fibrotic progression, accompanied by consistent upregulation of core GRGs.

Glycolytic reprogramming is a key pathogenic driver in HF progression and associated immune infiltration. Investigating this metabolic-immune dysregulation represents a promising therapeutic focus for progression of HF.