Background: Although immune cells play a critical role in lipid metabolism and inflammation regulation in patients with non-alcoholic steatohepatitis (NASH), the specific immune cells involved and associated genes remain unclear.
Methods: We identified differential immune cell profiles between normal liver and NASH specimens using the CIBERSORT algorithm. Next, we conducted a weighted gene co-expression network analysis (WGCNA) to identify genes highly correlated with these immune cells in NASH. Subsequently, core genes of immune cells were identified using machine learning algorithms.
Results: The abundance of M1 macrophages significantly increased in patients with NASH. The Random Forest (RF) algorithm identified six M1 macrophage-related genes (COL10A1, FAP, IL32, STMN2, SUSD2, and THY1) crucial in NASH. These six genes positively correlated with five inflammatory genes (CCL2, IL1B, TNF, CSF1, and IL15), lipid synthesis gene (FAS), collagen synthesis genes (COL1A1 and COL3A1), liver fibrosis stage, NASH activity score (NAS), and aspartate aminotransferase (AST) levels. These were negatively correlated with the lipid transport gene (CD36), beta fatty acid oxidation gene (PPARA), and M2 macrophage abundance. Moreover, a predictive model based on these six genes achieved a C-index of 0.902 for diagnosing NASH across four cohorts. The expression of these six genes accurately stratified patients with NASH into low disease activity cluster 1 and high disease activity cluster 2.
Conclusion: These six core genes of M1 macrophages contribute to NASH progression by regulating inflammation, lipid metabolism, and liver fibrosis.
Keywords: M1 macrophages; NASH; cluster; diagnosis; machine learning.
© 2024 Xu et al.