A four-genes based diagnostic signature for osteoarthritis

Osteoarthritis (OA) is a primary leading cause of pain and disability. However, some cases are diagnosed at the later stage which delayed the timely treatment. This study aims to identify effective diagnostic signature for OA. The mRNA profile GSE48566 including 106 blood samples of OA patients and 33 blood samples of healthy individuals was downloaded from Gene Expression Omnibus (GEO) database. The potential OA-related genes were screened by weighted gene co-expression network analysis (WGCNA). Gene ontology (GO) term and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis were performed to reveal the functions or pathways of OA-related genes using the clusterProfiler function package of R software. Key genes significantly involved in OA progression were further screened by protein–protein interaction (PPI) network. The logistic regression model and the random forest model were conducted by bringing into optimal genes selected by stepwise regression analysis, and fivefold cross validation method was used to determine their reliability. A total of 146 genes, existed in three modules and might be associated with the occurrence of OA, were screened. 15 genes were screened from the PPI network and four genes, including CCR6, CLEC7A, IL18 and SRSF2, were further optimized. Finally, a logistic regression model and a random forest model were conducted by bringing into four optimal genes, and could reliably separate OA patients from healthy subjects. Our study established two effective diagnostic models based on CCR6, CLEC7A, IL18 and SRSF2, which could reliably separate OA patients from healthy subjects.