Constructing and validating a diagnostic nomogram for multiple sclerosis via bioinformatic analysis

The purpose of this study was to identify biomarkers and construct a diagnostic prediction model for multiple sclerosis (MS). Microarray datasets in the Gene Expression Omnibus (GEO) were downloaded. Weighted gene coexpression analysis (WGCNA) was used to search for hub modules and biomarkers related to MS. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were used to roughly define their biological functions and pathways. Least absolute shrinkage and selection operator (LASSO) regression and multivariate logistic regression analysis were used to identify the diagnostic biomarkers and construct a nomogram. The calibration curve and receiver operating characteristic (ROC) curve were used to judge the diagnostic predictive ability. In addition, cell-type identification by estimating relative subsets of RNA transcripts (CIBERSORT) algorithm was used to calculate the proportion of 22 kinds of immune cells. GSE41850 was used as the training set, and GSE17048 was used as the test set. WGCNA revealed one hub module containing 165 hub genes. Most of their biological functions and pathways are related to cell metabolism and immune cell activation. The diagnostic nomogram contained ARPC5, ROD1, UBQLN2, ZNF281, ABCA1 and FAS. The ROC curve and the calibration curve of the training set and test set confirmed that the nomogram had great prediction ability. In addition, monocytes and M0 macrophages were significantly different between MS patients and healthy people. The expression of ARPC5, ZNF281 and ABCA1 is correlated with M0 macrophages. The nomogram provides new insights and contributes to the accurate diagnosis of MS.