Yijun Sun, Guoqi Dong, Hui Gao, Yong Yao, Huayuan Yang

School of Acupuncture-Moxibustion and Tuina, Shanghai University of Traditional Chinese Medicine, Shanghai, 201203, People’s Republic of China

Correspondence: Huayuan Yang, School of Acupuncture-Moxibustion and Tuina, Shanghai University of Traditional Chinese Medicine, Shanghai, 201203, People’s Republic of China, Email yhy4@shutcm.edu.cn

Purpose: Rheumatoid arthritis (RA) is a persistent inflammatory condition, and electroacupuncture (EA) has been demonstrated to effectively reduce the symptoms associated with RA. However, the molecular mechanisms underlying the effects of EA in RA remained poorly understood. This study aimed to identify potential diagnostic biomarkers for RA and elucidated the molecular targets of EA by using bioinformatics analysis and machine learning algorithms in peripheral blood samples.
Methods: We obtained datasets from the Gene Expression Omnibus(GEO) database containing samples from RA patients (GSE15573) and from RA patients after EA treatment (GSE59526) for bioinformatics analysis. Diagnostic biomarkers were identified using three distinct machine learning algorithms (LASSO, Random Forest and SVM-REF). A rat model of RA was established using Complete Freund’s Adjuvant (CFA), and quantitative real-time PCR was performed to confirm the differential expression of identified diagnostic biomarkers and assess the modulatory impact of EA on these genes.
Results: Twenty-six genes were identified as differentially expressed following EA treatment. Three machine learning algorithms converged on ARHGAP17 and VEGFB as potential diagnostic biomarkers for RA, exhibiting robust diagnostic performance (AUC > 0.75) and consistent expression patterns across multiple RA cohorts (GSE17755, GSE205962 and GSE93272). Besides, EA treatment significantly increased the paw withdrawal threshold (PWT) and the peripheral blood expression of both ARHGAP17 and VEGFB in CFA rats.
Conclusion: This study employed three machine learning algorithms to identify potential diagnostic biomarkers for the alleviation of RA by EA. The biomarkers demonstrated robust diagnostic performance across multiple validation datasets. Furthermore, animal experiments confirmed that EA exerted a favorable regulatory effect on these diagnostic biomarkers. The findings of this study provided novel therapeutic targets for the EA treatment of RA.

Keywords: rheumatoid arthritis, machine learning, bioinformatics, diagnostic biomarkers, electroacupuncture