Xinyi Liu,1 Siyao Wang,2 Xinru Du,1 Yulu Wang,3 Lingfei Mo,1 Hanchao Li,1 Zechao Qu,4 Xiaohao Wang,4 Jian Sun,5 Yuanyuan Li,1 Jing Wang1 

1Department of Rheumatology and Immunology, First Affiliated Hospital of Xi’an Jiaotong University, Xi’an, People’s Republic of China; 2Department of Gastroenterology, First Affiliated Hospital of Xi’an Jiaotong University, Xi’an, People’s Republic of China; 3Xi’an Jiaotong University College of Medicine, Xi’an, People’s Republic of China; 4Department of Spine Surgery, Hong Hui Hospital, Xi’an Jiaotong University, Xi’an, People’s Republic of China; 5Institute of Endemic Diseases, School of Public Health & Key Laboratory of Trace Elements and Endemic Diseases, Xi’an Jiaotong University, Xi’an, People’s Republic of China

Correspondence: Jing Wang; Yuanyuan Li, Department of Rheumatology and Immunology, First Affiliated Hospital of Xi’an Jiaotong University, Xi’an, 710061, People’s Republic of China, Email kidip@163.com; wudui220@163.com

Introduction: Cell death contributes to the pathogenesis of rheumatoid arthritis (RA) through various pathways. Disulfidptosis is a recently discovered novel form of cell death characterized by the abnormal accumulation of intracellular disulfide bonds. It remains unclear for the association between RA and disulfidptosis.
Methods: A comprehensive analysis of three GEO datasets was presented in this study. First, the analysis involved the use of weighted gene co-expression network analysis (WGCNA) and differential analysis and were employed to identify the key module genes related to RA and disulfidptosis-related genes. The machine learning algorithms were used to identify the hub genes. Second, a diagnostic model was constructed for RA based on the hub genes. The nomogram and receiver operating characteristic (ROC) curves were utilized to evaluate the diagnostic value of the model. Third, two RA subtypes were identified based on hub genes by using consensus clustering analysis. Then, the disease activity scores, clinical markers, and immune cells were compared between the two RA subgroups. Finally, the differential expression of hub genes was validated between healthy controls and RA patients by qPCR.
Results: Four hub genes (KLHL2, POLK, CLEC4D, NXT2) were identified. The expression of the four hub genes was verified to be significantly higher in RA patients compared with healthy controls. The superior diagnostic value of the model was validated, which demonstrated that the model outperforms each hub gene individually. Two subtypes of RA were determined. Patients in cluster A exhibited relatively lower levels of DAS28-CRP, DAS28-ESR, CDAI, SDAI, RF, CRP, and MMP3. In contrast, patients in cluster B had significantly higher levels of the above markers.
Conclusion: Four hub genes were identified to provide unique insights into the role of disulfidptosis in RA. Additionally, a promising diagnosis model and patient stratification were established based on the hub genes to assess the risk of RA onset and RA disease activity.

Keywords: rheumatoid arthritis, disulfidptosis, machine learning, bioinformatics, diagnostic model