Identification and Experimental Validation of Key Lipid Metabolism-Related Genes in Intracerebral Hemorrhage Based on Machine Learning

Background: The relationship between lipid metabolism and intracerebral hemorrhage (ICH) is not fully understood, particularly regarding its potential contribution to disease initiation and progression. Accordingly, this study employed machine learning methods to identify and validate pivotal lipid metabolism-related genes in ICH. Methods: We integrated our mouse mRNA sequencing data with the ICH-associated datasets GSE216607 and GSE200575. Lipid metabolism related genes were obtained from the MSigDB database. The analysis of differentially expressed genes (DEGs), Weighted correlation network analysis (WGCNA) and intersecting with lipid metabolism-related gene sets to screen the DEG of lipid metabolism-related in ICH. Multiple machine learning (ML) algorithms and a protein protein interaction network were then used to further identify key genes, which were subsequently validated in an independent dataset. Enrichment analysis, immune infiltration analysis, and clustering analysis were conducted to explore their potential biological functions. A mouse ICH model was then established and evaluated by MRI, HE staining, and Nissl staining. Quantitative PCR, immunofluorescence, and Western blotting were performed for preliminary experimental validation of the key genes and pathways. Results: In ICH, 84 differentially expressed lipid metabolism-related genes were identified, with enrichment mainly observed in lipid metabolism and the PPAR signaling pathway. By integrating multiple ML algorithms with protein protein interaction network analysis, Plin2, Cd36, and Abca1 were ultimately identified as three key genes. Validation in an independent dataset showed that all three genes were significantly upregulated in the ICH group. In the mouse ICH model, the mRNA expression levels of these genes were also markedly increased, consistent with the bioinformatics results. In addition, the PPAR signaling pathway was implicated by enrichment analysis and preliminary validation. Conclusion: Abnormal lipid metabolism may be involved in the pathogenesis and progression of ICH, and Plin2, Cd36, and Abca1 may serve as potential biomarkers associated with ICH-related lipid metabolic changes.