Identification of ferroptosis-related biomarkers in depression using a bioinformatics approach

Abstract

Abstract Background Depression is a common psychiatric disorder. Many studies have highlighted the involvement of ferroptosis in the pathological process of depression.Methods All datasets on depression: GSE98793、GSE201332、GSE76826、GSE54564、GSE44593、GSE38206 and GSE54570 were collected from the Gene Expression Omnibus (GEO) database and categorized into the test and validation sets, respectively. In addition, depression-associated module genes were detected using the weighted gene co-expression network analysis (WGCNA), based on the expression data from the GSE98793 test set. Afterward, Ferroptosis-Related Genes (FRGs) were extracted from the FerrDb database. Furthermore, pyroptosis-related genes (PRGs) were extracted from the MSigDB and GeneCard databases as controls. Subsequently, ferroptosis-related and pyroptosis-related potential biomarkers were screened by machine learning (ML) algorithms. Nomograms were constructed separately based on the above genes to predict disease occurrence. The reliability of the nomograms was assessed after analyzing the receiver operating characteristic (ROC) curve. The clinical predictive effects of the two cell death modalities were compared to highlight the specificity of ferroptosis in the pathological process of depression. Moreover, Next, we evaluated the expression levels and clinical predictive power of ferroptosis-related biomarkers in the samples in the GSE201332, GSE76826, GSE54564, GSE44593, GSE38206 and GSE54570 validation sets. Gene Set Enrichment Analysis (GSEA) and immune function analysis were performed for ferroptosis-related biomarkers. Finally, miRNAs and drugs associated with ferroptosis-related biomarkers were predicted.Results In total, 27 FRGs and 19 PRGs were identified. Ferroptosis-related potential biomarkers (AKR1C3, IDO1, LCN2, PANX2, and PEX12) and pyroptosis-related potential biomarkers (GZMA, ELANE, CD274, TUBB6, and CD14) were screened by ML algorithms. Subsequently, nomograms incorporating these biomarkers were constructed separately. The area under the ROC curve (AUC) values for the FRGs (0.689) were seen to be greater than the values for the PRGs (0.619), suggesting that ferroptosis is more specific in the pathogenesis of depression compared to pyroptosis. Most of the ferroptosis-related biomarkers were significantly expressed in the validation set, and all biomarkers could distinguish disease samples from normal samples. GSEA suggested that immune-related pathways such as primary immunodeficiency had been significantly enriched, in addition to the ferroptosis-related pathway. Subsequently, five ferroptosis-related biomarkers were seen to be significantly related to NK cells resting, T cells CD4 memory activated, and T cells regulatory (Tregs). Finally, 55 miRNAs and 10 key drugs were predicted.Conclusions Ferroptosis is more specific in the pathological mechanisms of depression compared to pyroptosis. In addition, AKR1C3, IDO1, LCN2, PANX2, and PEX12 are ferroptosis-related potential biomarkers in depression.