Background
Non-Alcoholic Steatohepatitis (NASH) is a prevalent form of liver inflammation that can progress to fibrosis and even hepatocellular carcinoma. The purpose of this research is to explore the biomarkers for NASH-related fibrosis based on single cell sequencing and machine learning.
Methods
We retrieved three datasets from the GEO database (GSE228232, GSE162694, GSE130970). Within GSE228232, we conducted cell annotation, pseudotime analysis, cell communication, and high-dimensional Weighted Gene Co-expression Network Analysis (hdWGCNA). In GSE162694, differential analysis, immune cell infiltration, and enrichment analyses were performed to discern the gene differences between the NASH and NASH-F groups. Ultimately, multiple machine learning algorithms were employed to validate the biological markers of NASH-F.
Results
In the analysis of the GSE162694 dataset, immune infiltration studies revealed significant differences in various types of T cells between the NASH and NASH-F groups. Pseudotime analysis indicated a strong association between NASH-F and T cells with high expression of Cd8a/b, Cxcr6, and Pdcd-1. Through single-cell sequencing and transcriptome analysis, we have isolated a set of 15 genes conserved between mouse models and human cases of NASH. This conserved gene set includes BCL11B, CD2, CD3E, CD5, GLS, GZMK, ICOS, ITGAV, LEF1, NEURL3, NR4A3, PFKP, RGS1, THEMIS, and THY1. Subsequent machine learning models corroborated CD2 and ITGAV as biomarkers for NASH-F.
Conclusion
Leveraging single-cell sequencing and multiple machine learning, our study delves into the pathogenesis of T cells in NASH-associated fibrosis and identifies CD2 and ITGAV as biomarkers of NASH-F.