Table 1_Identify the potential target of efferocytosis in knee osteoarthritis synovial tissue: a bioinformatics and machine learning-based study.xlsx

IntroductionKnee osteoarthritis (KOA) is a degenerative joint disease characterized by the progressive deterioration of cartilage and synovial inflammation. A critical mechanism in the pathogenesis of KOA is impaired efferocytosis in synovial tissue. The present study aimed to identify and validate key efferocytosis-related genes (EFRGs) in KOA synovial tissue by using comprehensive bioinformatics and machine learning approaches.
MethodsWe integrated three datasets (GSE55235, GSE55457, and GSE12021) from the Gene Expression Omnibus database to screen differentially expressed genes (DEGs) associated with efferocytosis and performed weighted gene co-expression network analysis. Subsequently, we utilized univariate logistic regression analysis, least absolute shrinkage and selection operator regression, support vector machine, and random forest algorithms to further refine these genes. The results were then inputted into multivariate logistic regression analysis to construct a diagnostic nomogram. Public datasets and quantitative real-time PCR experiments were employed for validation. Additionally, immune infiltration analysis was conducted with CIBERSORT using the combined datasets.
ResultsAnalysis of the intersection between DEGs and EFRGs identified 12 KOA-related efferocytosis DEGs. Further refinement through machine learning algorithms and multivariate logistic regression revealed UCP2, CX3CR1, and CEBPB as hub genes. Immune infiltration analysis demonstrated significant correlations between immune cell components and the expression levels of these hub genes. Validation using independent datasets and experimental approaches confirmed the robustness of these findings.
ConclusionsThis study successfully identified three hub genes (UCP2, CX3CR1, and CEBPB) with significant expression alterations in KOA, demonstrating high diagnostic potential and close associations with impaired efferocytosis. These targets may modulate synovial efferocytosis-related immune processes, offering novel therapeutic avenues for KOA intervention.

引言
膝骨关节炎（Knee osteoarthritis, KOA）是一种退行性关节疾病，以软骨进行性退变与滑膜炎症为核心病理特征。膝骨关节炎发病机制中的关键环节为滑膜组织胞葬作用受损。本研究旨在通过综合生物信息学与机器学习方法，筛选并验证膝骨关节炎滑膜组织中关键的胞葬作用相关基因（efferocytosis-related genes, EFRGs）。

方法
本研究从基因表达综合数据库（Gene Expression Omnibus, GEO）中整合GSE55235、GSE55457与GSE12021三个数据集，以筛选与胞葬作用相关的差异表达基因（differentially expressed genes, DEGs），并开展加权基因共表达网络分析。随后，采用单变量logistic回归分析、最小绝对收缩与选择算子回归、支持向量机（Support Vector Machine）以及随机森林（Random Forest）算法对候选基因进行进一步筛选。将所得结果代入多变量logistic回归分析，构建诊断列线图。通过公共数据集与实时定量聚合酶链反应（quantitative real-time PCR, qRT-PCR）实验完成验证。此外，利用整合后的数据集，通过CIBERSORT工具开展免疫浸润分析。

结果
对差异表达基因与胞葬作用相关基因的交集进行分析，共获得12个与膝骨关节炎相关的胞葬作用差异表达基因。经机器学习算法与多变量logistic回归进一步筛选后，确定UCP2、CX3CR1与CEBPB为枢纽基因。免疫浸润分析结果显示，免疫细胞组分与上述枢纽基因的表达水平存在显著相关性。采用独立数据集与实验方法进行的验证证实了该研究结果的稳健性。

结论
本研究成功筛选出3个在膝骨关节炎中存在显著表达异常的枢纽基因（UCP2、CX3CR1与CEBPB），其不仅具备较高的诊断潜力，还与胞葬作用受损密切相关。这些靶点可调控滑膜组织中与胞葬作用相关的免疫过程，为膝骨关节炎的干预治疗提供了全新的潜在治疗方向。