Identification of shared biomarkers in obesity and non-alcoholic fatty liver disease: a comprehensive analysis of Mendelian randomization and transcriptomic data

Introduction Obesity and non-alcoholic fatty liver disease (NAFLD) are known to be closely interlinked; however, the mechanism of their interaction is unclear. This study sought to explore the causal relationship and shared biomarkers in obesity and NAFLD by combining Mendelian randomization (MR) and transcriptomic data. Methods A bidirectional MR analysis was performed using summary statistics data from GWAS database to determine the causal relationship between obesity and NAFLD. Based on transcriptome data from GEO database, candidate genes were identified by combining differential expression analysis and expression level analysis. Subsequently, candidate genes were incorporated into least absolute shrinkage and selection operator (LASSO) and support vector machine-recursive feature elimination (SVM-RFE) algorithms to recognize shared biomarkers in obesity and NAFLD. The nomograms based on shared biomarkers were generated to predict the probability of developing obesity and NAFLD. Further, enrichment analysis and regulatory network construction were accomplished to excavate deeper into the molecular mechanism of obesity and NAFLD. Ultimately, expression level of shared biomarkers was validated through qRT-PCR. Rusults MR analysis indicated significant causal effect of obesity on NAFLD (IVW: P = 0.034, OR = 1.252), while no significant causal effect of NAFLD on obesity (IVW: P 0.05). Subsequently, based on the results of LASSO and SVM-RFE, we gained two shared biomarkers (NCAPH and IRS2) in obesity and NAFLD, and IRS2 was down-regulated and NCAPH was up-regulated in disease (obesity/NAFLD) samples. Receiver operating characteristic curves manifested that the efficacy of NCAPH and IRS2 in distinguishing between disease (obesity/NAFLD) and normal samples was all excellent. Besides, the nomograms constructed based on two shared biomarkers were reliable and had high reference value for predicting obesity and NAFLD. Enrichment analysis revealed that shared biomarkers could influence the development of both obesity and NAFLD via the “ECM-receptor interaction,” “Jak-STAT signaling pathway,” etc. In lncRNA-miRNA-mRNA network, lncRNAs (NEAT1, MALAT1, and FTX) could simultaneously regulate two shared biomarker via hsa-miR-493-5p. Ultimately, qRT-PCR results revealed that NCAPH was in line with the bioinformatics results. Conclusion This study discovered causal effect of obesity on NAFLD and identified two shared biomarkers (NCAPH and IRS2) in two diseases, providing new insights for further explore the pathogenesis of both diseases.