Identification of Chondrocyte Stemness-Associated Biomarkers in Osteoarthritis Using Bioinformatics and Machine Learning Approaches

Siyi Xie,1,2 Meiling Liu,1,2 Yuzhong Wang,1 Shuxing Cao,1 Yiming Yang,2 Ruixue Chen,2 Tianlin Gao,1,2 Jun Ma,2,3 Zhe Lv,4 Yongzhou Song1– 3 1Department of Orthopedics, The Second Hospital of Hebei Medical University, Shijiazhuang, Hebei Province, People’s Republic of China; 2Hebei Medical University-National University of Ireland Galway Stem Cell Research Center, Hebei Medical University, Shijiazhuang, Hebei Province, People’s Republic of China; 3Hebei Technology Innovation Center for Stem Cell and Regenerative Medicine, Shijiazhuang, Hebei Province, People’s Republic of China; 4The Second Hospital of Hebei Medical University, Shijiazhuang, Hebei Province, People’s Republic of ChinaCorrespondence: Yongzhou Song; Zhe Lv, The Second Hospital of Hebei Medical University, Shijiazhuang, Hebei Province, 050017, People’s Republic of China, Email [email protected]; [email protected]: Osteoarthritis (OA) represents a prevalent degenerative joint condition, in which chondrocyte dysfunction plays a key role in disease progression. Although accumulating evidence underscores the importance of cellular stemness regulation in OA development, systematic screening of related biomarkers has been insufficient. The current study sought to discover and validate potential biomarkers through bioinformatics and machine learning (ML), offering novel perspectives for early detection and therapeutic intervention in OA.Methods: The present study examined six OA-related transcriptomic profiles from the Gene Expression Omnibus (GEO) to discover and validate stemness-associated biomarkers. Differentially expressed genes (DEGs) were selected and analyzed for enriched biological functions. OA-related modules were determined via weighted gene coexpression network analysis (WGCNA). Key stemness-related genes were selected using ML algorithms, including support vector machine (SVM), random forest (RF), extreme gradient boosting (XGBoost), and the least absolute shrinkage and selection operator (LASSO) regression. Receiver operating characteristic (ROC) analysis was implemented to determine diagnostic accuracy. Utilizing single-sample gene set enrichment analysis (ssGSEA), the link with immune cell infiltration was examined. Ultimately, immunohistochemistry was employed for experimental validation.Results: Intersection analysis identified 56 stemness-related DEGs in OA cartilage. WGCNA analysis yielded 7 modules significantly associated with stemness genes, and a combined screening approach identified 60 candidate genes. Using four machine learning algorithms—SVM, LASSO, XGBoost, and RF—four feature genes were ultimately determined (WWP2, CDKN1A, IL11, and CRTAC1), among which WWP2, CDKN1A, and CRTAC1 showed significant differential expression between OA and normal samples and demonstrated good diagnostic performance in both the training and validation cohorts (AUC > 0.7). ssGSEA analysis revealed that the expression of these three genes was significantly correlated with specific immune cell subpopulations. Immunohistochemistry further confirmed that WWP2 and CDKN1A were downregulated in OA tissues, whereas CRTAC1 was upregulated.Conclusion: Through bioinformatics analysis and IHC validation, we identified three stemness-associated biomarker genes (WWP2, CDKN1A, CRTAC1) in OA. These findings may provide meaningful implications for future clinical assessment, treatment, and research on OA.Keywords: osteoarthritis, cellular stemness, machine learning, biomarkers, chondrocytes