Machine learning-based Algorithm Identifies Key Mitochondria-Related Genes in Non-Alcoholic Steatohepatitis

Abstract Background In hepatocytes, mitochondrial dysfunction drives aberrant fatty acid metabolism, oxidative stress, and cell apoptosis, promoting the occurrence and progression of NASH. Given the pivotal role of mitochondrial dysfunction in the advancement of NASH, the identification of mitochondrial core genes within NASH may offer potential targets for NASH treatment. Methods According to 101 machine learning algorithms assembled from 10 different machine learning algorithms, mitochondrial core genes were identified in NASH patients. The relationship between mitochondrial core genes and inflammation, lipid metabolism, liver fibrosis, and immune infiltration was investigated. Results AKR1B10, TYMS, and TREM2 were identified. A predictive model constructed using these three mitochondrial genes exhibited excellent diagnostic performance for NASH in the GEO cohorts. AKR1B10, TYMS, and TREM2 were significantly upregulated in NASH, F3-F4 stage liver fibrosis patients, and NAFLD-HCC patients. The expression levels of AKR1B10, TYMS, and TREM2 were positively correlated with pro-inflammatory genes, lipid synthesis genes, liver fibrosis genes, NAS score, pro-inflammatory immune signatures, and M1 macrophage content. Conversely, they were significantly negatively correlated with fatty acid oxidation genes and M2 macrophage content. Moreover, the biological and mitochondrial pathways enriched when AKR1B10, TYMS, and TREM2 were upregulated were related to NASH progression. NASH patients were further classified into Cluster 1 and Cluster 2. Pro-inflammatory genes, lipid synthesis genes, liver fibrosis genes, NAS score, pro-inflammatory immune signatures, and M1 macrophage content were significantly upregulated in Cluster 1. Conversely, fatty acid oxidation genes and M2 macrophage content were significantly downregulated in Cluster 1. Conclusion AKR1B10, TYMS, and TREM2 are associated with the severity of NASH. High expression of AKR1B10, TYMS, and TREM2 indicates a more severe condition in NASH patients.