Construction and validation of m6A-related diagnostic model for psoriasis

Abstract

Background Psoriasis is a chronic immune-mediated inflammatory disease. N6-methyladenosine (m6A) is involved in numerous biological processes in both normal and diseased states. Herein, we aimed to explore the potential role of m6A regulators in the diagnosis of psoriasis and predict molecular mechanisms by which m6A regulators impact psoriasis. Methods GSE30999 (170 human skin tissue samples) and GSE13355 (180 human skin tissue samples) were downloaded as the training analysis dataset and validation dataset respectively. M6A-related genes were obtained from the literature and their expression levels in GSE30999 samples were measured to identify M6A-related DEGs between psoriasis lesions (LS) and non-lesional lesions (NL). We identified m6A-related DEGs using differential expression analysis and assessed their interactions through correlation analysis and network construction. A logistic regression analysis followed by LASSO optimization was employed to select m6A-related DEGs for the construction of a diagnostic model. The performance of the model was validated using support vector machine (SVM) methodology with sigmoid kernel function and extensive cross-validation. Additionally, the correlation between m6A-related DEGs and immune cell infiltration was analyzed, as well as the association of these DEGs with psoriasis subtypes. Functional analysis of the m6A-related DEGs included the construction of regulatory networks involving miRNAs, transcription factors (TFs), and small-molecule drugs. The m6A modification patterns were also explored by examining the gene expression differences between psoriasis subtypes and their enriched biological pathways. Finally, the expression of significant m6A regulators involved in the diagnostic model was examined by RT-qPCR. Results In this study, ten optimal m6A-related DEGs were identified, including FTO, IGF2BP2, METTL3, YTHDC1, ZC3H13, HNRNPC, IGF2BP3, LRPPRC, YTHDC2, and HNRNPA2B1. A diagnostic model based on these m6A-related DEGs was constructed, demonstrating high diagnostic accuracy with an area under the curve (AUC) in GSE30999 and GSE13355 of 0.974 and 0.730, respectively. Meanwhile, the expression level of m6A regulators verified by RT-qPCR was consistent with the results in GSE30999. The infiltration of activated mast cells and NK cells was significantly associated with all ten m6A-related DEGs in psoriasis. Among them, YTHDC1, HNRNPC, and FTO were targeted by most miRNAs and were regulated by nine related TFs. Therefore, patients may benefit from dorsomorphin and cyclosporine therapy. Between the two subgroups, 1,592 DEGs were identified, including LRPPRC and METTL3. These DEGs were predicted to be involved in neutrophil activation, cytokine-cytokine receptor interactions, and chemokine signaling pathways. Conclusions A diagnostic model based on ten m6A-related DEGs in patients with psoriasis was constructed, which may provide early diagnostic biomarkers and therapeutic targets for psoriasis.

Keywords

• Psoriasis
• m6A
• Diagnostic model
• Bioinformatics analysis