Potential Targets in Nonalcoholic Steatohepatitis Based on Bioinformatics Analysis and Machine Learning Strategies

NASH poses a significant threat to human health and is recognized as the leading contributor to HCC. In this study, we leveraged publicly accessible datasets to identify novel differentially expressed genes that may serve as potential targets in NASH or potentially NASH-induced HCC. The publicly available datasets were obtained from the GEO. Differential gene expression analysis and enrichment analysis was performed. Subsequently, WGCNA and PPI network were constructed. Lastly, machine learning was employed to identify key feature genes. Utilizing the integrated GEO database, we identified 446 genes exhibiting differential expression. Enrichment analysis indicated that these genes are predominantly associated with glucose and lipid metabolism and inflammatory processes. Through WGCNA, three modules were identified that demonstrated a significant correlation with NASH. Furthermore, core genes among the differentially expressed genes were extracted via protein and protein interaction analysis. Ultimately, machine learning techniques were employed, leading to the identification of three genes: FosB, Fos, and SOCS3. Notably, FosB exhibited consistent expression across various datasets, demonstrated strong predictive capabilities for NASH, and was associated with improved prognostic outcomes in hepatocellular carcinoma by data from TCGA. Additionally, in vitro immunohistochemistry experiments revealed significant reduction of FosB expression in NASH. Bioinformatics analyses conducted on various datasets, along with in vitro immunohistochemistry experiments, revealed significant downregulation of FosB in NASH. It indicates that FosB plays a critical role in the pathogenesis of NASH, and its expression is associated with the prognosis of patients with HCC. Further experimental studies are required to investigate the potential targeting of FosB in NASH and NASH-induced HCC.