Integrative Analysis of Hub Genes for Postoperative Cognitive Dysfunction Based on Bioinformatics Methods

Abstract

Abstract Background POCD is a frequent condition that originates after surgery and is characterized by impaired memory, inattention, and diminished learning or language abilities. It seriously threatens the quality of the patients’ life, especially in older adults. The identification of factors associated to POCD facilites the diagnosis and the prediction of clinical outcomes. The identification of biomarkers associated to POCD aids in the diagnosis and prediction of clinical outcomes. Therefore, this study aimed to identify hub genes and analyze their possible involvement in POCD pathogenesis. Methods The POCD expression profiles (GSE95070 and GSE95426) were obtained from the Gene Expression Omnibus (GEO) database, followed by the identification of differentially expressed miRNAs (DEmiRNAs) and mRNAs (DEmRNAs). Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were used to achieve functional enrichment. Subsequently, the target mRNAs of DEmiRNAs were identified, followed by the construction of the miRNA-mRNA network for POCD. The least absolute shrinkage and selection operator (LASSO) and support vector machine recursive feature elimination (SVM-RFE) algorithms were used to identify hub miRNAs. Finally, hub mRNAs were sorted by overlapping target mRNAs and DEmRNAs. A POCD mouse model was established and tested using the open field test (OFT) and Morris water maze (MWM). Finally, quantitative real-time polymerase chain reaction (qRT-PCR) was used for hub genes validation. Results Eleven DEmiRNAs were identified. Among them, mmu-miR-483-5p and mmu-miR-362-3p showed the highest positive correlation (R = 0.96), whereas mmu-miR-206-3p and mmu-miR-190a-3p showed the highest negative correlation (R = -0.88). GO analysis revealed that DEmiRNAs were mainly enriched in response to stress, anatomical structure development, and cellular nitrogen compound metabolic homeostatic processes. KEGG analysis identified pathways that were mainly enriched in regulating the actin cytoskeleton, sphingolipid signaling, axon guidance, and thyroid hormone signaling. Then, 7062 targeting mRNAs were identified with DEmiRNAs, and 53 DEmRNAs were identified. Finally, mmu-miR-3065-5p and mmu-miR-592-3p were identified as hub miRNAs by LASSO and SVM-RFE analyses, respectively. Finally, four mRNAs (Prdm14, Slc46a2, Nkx2-1, and Lrg1) were identified as hub genes and were validated by the POCD mouse model. Conclusions We identified two hub miRNAs and four hub mRNAs in the POCD model, all of which could serve as biomarkers and pharmacological targets in the treatment of POCD. Targeting hub genes and associated pathways using gene manipulation methods may help understand the mechanism and improve the outcome of POCD.