Identification and immunological characterization of lipid metabolism-related molecular clusters in nonalcoholic fatty liver disease

Abstract

Abstract Background Nonalcoholic fatty liver disease (NAFLD) is now the major contributor to chronic liver disease. Disorders of lipid metabolism are a major element in the emergence of NAFLD. This research intended to explore lipid metabolism-related clusters in NAFLD and establish a prediction biomarker. Methods The expression mode of lipid metabolism-related genes (LMRGs) and immune characteristics in NAFLD were examined. The “ConsensusClusterPlus” package was utilized to investigate the lipid metabolism-related subgroup. The WGCNA was utilized to determine hub genes and perform functional enrichment analysis. After that, a model was constructed by machine learning techniques. To validate the predictive effectiveness, receiver operating characteristic curves, nomograms, decision curve analysis (DCA), and test sets were used. Lastly, gene set variation analysis (GSVA) was utilized to investigate the biological role of biomarkers in NAFLD. Results Dysregulated LMRGs and immunological responses were identified between NAFLD and normal samples. Two LMRG-related clusters were identified in NAFLD. Immune infiltration analysis revealed that C2 had much more immune infiltration. GSVA also showed that these two subtypes have distinctly different biological features. Thirty cluster-specific genes were identified by two WGCNAs. Functional enrichment analysis indicated that cluster-specific genes are primarily engaged in adipogenesis, signalling by interleukins, and the JAK-STAT signalling pathway. Comparing several models, the random forest model exhibited good discrimination performance. Importantly, the final five-gene random forest model showed excellent predictive power in two test sets. In addition, the nomogram and DCA confirmed the precision of the model for NAFLD prediction. GSVA revealed that model genes were down-regulated in several immune and inflammatory-related routes. This suggests that these genes may inhibit the progression of NAFLD by inhibiting these pathways. Conclusions This research thoroughly emphasized the complex relationship between LMRGs and NAFLD and established a five-gene biomarker to evaluate the risk of the lipid metabolism phenotype and the pathologic results of NAFLD.

Keywords

• Nonalcoholic fatty liver disease
• Lipid metabolism
• Immune infiltration
• Machine learning
• Biomarkers
• Molecular clusters