目的 通过生物信息学技术筛选胰腺癌（pancreatic cancer）与慢性胰腺炎（chronic pancreatitis，CP）组织的差异基因，预测能够干预胰腺癌“炎-癌”转化进程的中药及其潜在治疗靶点及机制。方法 从基因表达数据库（Gene Expression Omnibus，GEO）获取GSE151945、GSE30134基因芯片，应用R软件进行数据标准化、差异表达基因（differentially expressed genes，DEGs）筛选，并进行基因本体论（gene ontology，GO）和京都基因与基因组百科全书（Kyoto encyclopedia of genes and genomes，KEGG）富集分析；通过STRING数据库构建蛋白-蛋白互作网络，应用Cytoscape构建蛋白互作网络图，应用CytoHubba插件筛选关键基因；使用R软件对关键基因进行生存分析，筛选显著影响胰腺癌预后的基因，通过Kaplan-Meier曲线进行可视化展示；将上述基因与Coremine Medical数据库相互映射，预测潜在治疗作用的中药。从TCMSP和TCMID数据库获取中药化学成分，利用Cytoscape构建“中药-成分-靶点”网络图，并使用CytoHubba插件筛选关键靶点。结果 共筛选出178个DEGs，其中88个上调基因，90个下调基因；DEGs主要参与病毒生命周期、细胞外结构组织、细胞外基质组织、染色质的共价修饰等生物功能；KEGG通路分析显示DEGs主要富集在人乳头瘤病毒感染、志贺菌病、亨廷顿病、癌症蛋白多糖和丝裂原活化蛋白激酶（mitogen-activated protein kinase，MAPK）信号通路；共得出20个关键基因，生存分析提示黏附连接相关蛋白纽蛋白（vinculin，VCL）、核不均一性核糖核蛋白L（heterogeneous nuclear ribonucleoprotein L，HNRNPL）、小泛素样修饰物3（small ubiquitin like modifier 3，SUMO3）、整合素α3（integrin subunit alpha 3，ITGA3）、整合素β5（integrin subunit beta 5，ITGB5）、黏结蛋白聚糖1（syndecan 1，SDC1）和神经细胞黏附分子1（neural cell adhesion molecule 1，NCAM1）显著影响胰腺癌预后；筛选得到干预胰腺癌“炎-癌”转化进程的潜在中药有夏枯草、金钱草、紫苏、生地、赤芍、菟丝子等。结论 胰腺癌“炎-癌”转化机制复杂，中药可通过多靶点干预胰腺癌“炎-癌”转化，该研究将为胰腺癌发生机制和治疗药物的研究提供参考方向。

Objective To screen the differentially expressed genes (DEGs) between pancreatic cancer (PC) and chronic pancreatitis (CP) tissues through bioinformatics techniques, in order to predict the traditional Chinese medicine (TCM) that can interfere with the "inflammation-cancer" transformation process and their the potential therapeutic targets and mechanisms. Methods GSE151945 and GSE30134 gene chips were obtained from Gene Expression Omnibus (GEO), and R software was applied to normalize the data, screen DEGs, and perform gene ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis. The protein-protein interaction network was constructed by STRING database, the protein-protein interaction network map was created by Cytoscape, and the hub genes were screened by CytoHubba plug-in. R software was used to analyze the effect of the hub gene on the survival analysis of pancreatic cancer and screen for genes that significantly affect the prognosis of pancreatic cancer, visualized by Kaplan-Meier curves. The above genes were mapped with Coremine Medical database to predict potential therapeutic TCM. The chemical components of the TCM were obtained from TCMSP and TCMID databases. The network diagram of "TCM-component-target" was constructed by Cytoscape, and the key targets were screened by CytoHubba plug-in. Results A total of 178 DEGs were screened, including 88 up-regulated and 90 down-regulated genes. DEGs were mainly involved in biological functions such as virus life cycle, extracellular structure, extracellular matrix organization, and covalent modification of chromatin. KEGG pathway analysis showed that DEGs were mainly enriched in human papillomavirus infection, shigellosis, Huntington's disease, cancer proteoglycan, and mitogen-activated protein kinase (MAPK) signaling pathway. A total of 20 hub genes were obtained, and survival analysis showed that adhesion-linkage-associated protein nucleoprotein vinculin (VCL), heterogeneous nuclear ribonucleoprotein L (HNRNPL), small ubiquitin like modifier 3 (SUMO3), integrin subunit alpha 3 (ITGA3), integrin subunit beta 5 (ITGB5), syndecan 1(SDC1), and neural cell adhesion molecule 1 (NCAM1) significantly affected the prognosis of PC. The potential TCMs that can interfere with the "inflammation-cancer" transformation process of PC include Xiakucao (Prunellae Spica), Jinqiancao (Lysimachiae Herba), Zisu (Perilla frutescens), Shengdi (Rehmanniae Radix), Chishao (Paeoniae Radix Rubra), Tusizi (Cuscutae Semen). Conclusion The transformation mechanism of "inflammation-cancer" of PC is complex, and TCM can interfere with the "inflammation-cancer" transformation process of PC through multiple targets. This study will provide a reference direction for the study of the pathogenesis and therapeutic agents of PC.

R285

成都中医药大学大学生科研实践创新课题(ky-2021049,ky-2022004)