[Preliminary investigation into the mechanisms of thalidomide in inhibiting the malignant transformation of oral leukoplakia].

内容摘要

This study aims to integrate network pharmacology, machine learning, and survival analysis to preliminarily explore the molecular mechanisms underlying the inhibitory effects of thalidomide on the malignant transformation of oral leukoplakia (OLK). First, multisource databases and an OLK transcriptomic cohort (GSE26549) were integrated to identify the intersecting feature genes between thalidomide and OLK with epithelial dysplasia, followed by pathway enrichment analysis. Subsequently, a protein-protein interaction (PPI) network was constructed, and ensemble machine learning algorithms were applied to screen for core biomarkers. Kaplan-Meier curves and multivariate Cox proportional hazard regression models were utilized to evaluate the clinical prognostic efficacy of these core biomarkers. Finally, molecular docking was employed to validate the physical binding potential between thalidomide and the core biomarkers of OLK. A total of 16 intersecting target genes between thalidomide and OLK with epithelial dysplasia were identified. These genes were primarily enriched in the signaling pathways of the cell cycle, phosphatidylinositol 3-kinase/protein kinase B (PI3K-Akt), and microRNAs in cancer. PPI network analysis combined with ensemble algorithms ultimately identified four core biomarkers: MET proto-oncogene, receptor tyrosine kinase (MET), Aurora kinase A (AURKA), DNA methyltransferase 1 (DNMT1), and poly(ADP-ribose) polymerase 1 (PARP1). Survival analysis revealed that high expression levels of MET (P=0.004) and AURKA (P=0.029) increased the risk of malignant transformation in OLK. Furthermore, MET exhibited crucial independent prognostic value in the multivariate Cox proportional hazard regression model (hazard ratio=2.58, P=0.051). Molecular docking demonstrated that thalidomide could stably bind to the MET receptor (binding energy=-38.178 kJ/mol). Thalidomide may exert its inhibitory effects against the malignant progression of OLK by primarily targeting MET and synergistically ac-ting on key biomarkers, including AURKA, DNMT1, and PARP1, thereby suppressing downstream PI3K-Akt and cell cycle signaling pathways. This study not only provides robust computational biology evidence for the "drug repurpo-sing" of thalidomide but also highlights a promising pharmacological option for the clinical intervention of OLK's malignant transformation. 目的: 本研究旨在整合网络药理学、机器学习与生存分析，初步解析沙利度胺抑制口腔白斑病（OLK）恶变的分子机制。方法: 首先联合多源数据库与OLK转录组队列（GSE26549），获得沙利度胺与OLK伴有上皮异常增生的交集特征基因集并进行通路富集分析;继而构建蛋白质-蛋白质相互作用（PPI）网络，联合集成机器学习算法筛选核心标志物;然后利用Kaplan-Meier曲线及多因素Cox比例风险回归模型评估核心标志物的临床预后效能;最后借助分子对接技术验证沙利度胺与OLK核心标志物的物理结合潜能。结果: 共获得16个沙利度胺与OLK伴有上皮异常增生的交集靶点基因，主要富集于细胞周期信号通路、磷脂酰肌醇-3-激酶/蛋白激酶B信号通路（PI3K-Akt）和癌症中的微小RNA信号通路。拓扑网络分析联合机器学习算法最终筛选出MET原癌基因（MET）、极光激酶A（AURKA）、DNA甲基转移酶 1（DNMT1）及聚（ADP-核糖）聚合酶1（PARP1）为四大核心标志物。生存分析显示，MET（P=0.004）和AURKA（P=0.029）的高表达可增加OLK恶变风险，且MET在多因素Cox比例风险回归模型中表现出关键的独立预后潜能（风险比=2.58，P=0.051）。分子对接显示:沙利度胺能够与MET受体稳定结合（结合能为 -38.178 kJ/mol）。结论: 沙利度胺可能以MET为核心干预靶点，协同AURKA、DNMT1及PARP1等关键分子，通过抑制下游PI3K-Akt及细胞周期等信号通路，进而抑制OLK的恶性转变进程。本研究不仅为沙利度胺的“老药新用”提供坚实的计算生物学依据，也为OLK的临床恶变干预提供具有潜力的药物选择。.