Related Experiment Video
Updated: Jan 16, 2026
Author Spotlight: Fu's Subcutaneous Needling for Knee Osteoarthritis Pain
Published on: March 24, 2023
Unveiling CDKN1A and RELA: Machine learning-driven aging biomarkers for precision diagnosis and therapy in knee
Zehao Hu1,2, Zixuan Xu3, Tao Xiao3
1Shantou Hospital of Traditional Chinese Medicine, Guangzhou University of Traditional Chinese Medicine, Shantou, 515000, China.
Background:
Knee Osteoarthritis (KOA) is a degenerative joint disease marked by progressive cartilage deterioration, closely tied to cellular senescence. Despite advances in understanding KOA mechanisms, systematically identifying aging-related biomarkers remains challenging.
Methods:
KOA gene expression datasets (GSE12021, GSE169077) were sourced from the GEO database, and aging-related genes from GenAge and CellAge databases. Differentially expressed aging-related genes (ARDEGs) were identified using the limma package. Functional enrichment involved Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Gene Set Enrichment Analysis (GSEA). Core biomarkers were screened via LASSO regression, SVM-RFE, and Random Forest, with cross-tabulation using Venn software. Immune cell infiltration correlations were assessed using CIBERSORT and Pearson analysis. Single-cell RNA sequencing (GSE176308) was analyzed with Seurat and UMAP. Molecular docking with AutoDockTools evaluated curcumin interactions, and microRNA-mRNA networks were built using STRING and Cytoscape. In vivo validation in a mouse KOA model (DMM-induced) confirmed gene expression via qRT-PCR and Western blotting.
Results:
Integration of KOA datasets identified 543 (GSE12021) and 1909 (GSE169077) differentially expressed genes (DEGs), intersecting with 517 aging-related genes to yield 90 ARDEGs. Enrichment linked ARDEGs to senescence, metabolism, and signaling. Machine learning pinpointed CDKN1A and RELA as core biomarkers. Single-gene GSEA tied CDKN1A to sphingolipid/glucose metabolism and RELA to FGF/CXCR4 pathways. CIBERSORT revealed correlations with immune cells (CDKN1A: Tregs/NK cells; RELA: plasma cells/M1 macrophages). Single-cell analysis showed high expression in progenitor chondrocytes. Molecular docking confirmed curcumin binding, and microRNA-mRNA networks predicted upstream regulators. Multi-dataset validation supported CDKN1A's diagnostic robustness, with RELA showing cohort-specific significance. Experimental validation confirmed downregulated CDKN1A and RELA in KOA synovial tissue.
Conclusion:
Integrated transcriptomic and machine learning analyses identified CDKN1A and RELA as downregulated biomarkers in KOA, validated experimentally. CDKN1A deficiency impairs cell cycle regulation, and RELA disruption affects NF-κB-mediated inflammation, driving cartilage degeneration. Their immune-metabolic correlations and curcumin affinity highlight their potential as diagnostic and therapeutic targets for precision KOA intervention.
