Biomarkers Identification of Early Rheumatoid Arthritis via Bioinformatics Approach and Experimental Verification

Objective: To screen and identify potential biomarker for early rheumatoid arthritis (RA) by bioinformatic analysis and experimental investigation.

Methods: Transcriptome data of RA synovium was downloaded from GEO. Differentially expressed genes (DEGs), gene ontology (GO) functional annotation, and KEGG pathway were analyzed to inspect significative target genes. The protein-protein interaction was constructed using STRING database and Cytoscape to screen hub genes with least absolute shrinkage and selection operator (LASSO). The diagnostic effectivity of screened hub genes was analyzed with receiver operating characteristic (ROC). RA synovial fibroblast (SF) was treated with TNF-α (20ng/mL for 24h). RT-qPCR and Western blotting were used to measure mRNA and protein for screened hub gene.

Results: A total of 271 DEGs were found in GEO dataset. GO analysis indicated that DEGs mainly involved in phagocytosis, recognition and complement activation, etc. KEGG analysis suggested that DEGs were mostly enriched in the cytokine-cytokine receptor interaction, regulation of lipolysis in adipocytes, PPAR signaling pathway. LASSO regression and ROC curve indicated that ADIPOQ, CIDEA, FABP4, AQP7, LOC102723407, PLIN4, LIPE, CIDEC, PLIN1, and LEP had excellent diagnostic value. The area under ROC was 0.734. The level of ADIPOQ, LEP, LIPE, PLIN1, and PLIN4 were lower in RA group rather than that of control group (p<0.01). The higher expressions of CIDEC and FABP4 were found in RA group comparing to control group (p<0.001).

Conclusions: Identified hub genes might be valuable biomarkers for early RA diagnosis to promote precise and personal therapy.