Biomarker Signatures in Time-Course Progression of Neuropathic Pain at Spinal Cord Level Based on Bioinformatics and Machine Learning Analysis.

Abstract

Neuropathic pain (NP) is a debilitating chronic pain condition with complex molecular mechanisms and inadequate therapeutic solutions. This study aims to identify temporal transcriptomic changes in NP using multiple bioinformatics and machine learning algorithms. A total of 10 mouse samples (5 per group) were harvested at each time point (day three, day seven, and day fourteen), following spared nerve injury and a sham operation. Differentially expressed gene (DEG) analysis and an intersection among the three time-point groups revealed 54 common DEGs. The GO and KEGG analyses mainly showed enrichment in terms of immune response, cell migration, and signal transduction functions. In addition, the interaction of the LASSO, RF, and SVM-RFE machine learning models on 54 DEGs resulted in Ngfr and Ankrd1. The cyan module in WGCNA was selected for a time-dependent upward trend in gene expression. Then, 172 genes with time-series signatures were integrated with 54 DEGs, resulting in 11 shared DEGs. Quantitative RT-PCR validated the temporal expressions of the above genes, most of which have not been reported yet. Additionally, immune infiltration analysis revealed significant positive correlations between monocyte abundance and the identified genes. The TF-mRNA-miRNA network and drug-target network revealed potential therapeutic drugs and posttranscriptional regulatory mechanisms. In conclusion, this study explores genes with time-series signatures as biomarkers in the development and maintenance of NP, potentially revealing novel targets for analgesics.