| 陈松鹤,陈烨,虞听,陈亚栋.基于TCGA数据库的胰腺癌临床预后风险模型的构建与药物筛选[J].浙江中西医结合杂志,2023,33(8): |
| 基于TCGA数据库的胰腺癌临床预后风险模型的构建与药物筛选 |
| Construction and drug screening of pancreatic cancer clinical prognosis risk model based on TCGA database |
| 投稿时间：2022-04-22 修订日期：2023-05-22 |
| DOI： |
| 中文关键词: TCGA 胰腺癌 预后风险模型 药物筛选 |
| 英文关键词:TCGA database pancreatic cancer prognostic risk model drug screening |
| 基金项目: |
|
| 摘要点击次数: 637 |
| 全文下载次数: 6 |
| 中文摘要: |
| 目的：基于基于癌症基因组图谱(The Cancer Genome Atlas,TCGA)数据库构建胰腺癌预后风险模型和筛选候选药物。方法：通过TCGA数据库下载胰腺癌转录组和临床数据，通过R软件进行加权基因共表达网络分析(weighted correlation network analysis,WGCNA)、预后风险模型的构建、差异分析、Kaplan-Meier法生存分析、风险分析和临床相关性分析。通过受试者工作特征曲线(receiver operating characteristic curve，ROC曲线)的曲线下面积(area under curve，AUC)和多因素cox回归分析判断模型准确性和独立性，最后运用CMAP平台进行胰腺癌药物的筛选。结果：下载得到胰腺癌转录组数据182例和临床数据185例，数据合并交集后纳入病例样本177例。WGCNA分析筛选模块基因为“MEgreen（P=0.04<0.05）”；通过单因素cox回归分析得到28个预后相关免疫基因和70个预后相关的lncRNA，进一步用Lasso回归和多因素cox回归分析进行预后风险模型构建。模型评价显示该模型可区分高低风险组的患者，高风险的患者较低风险预后较差（P<0.05）。训练组和验证组ROC曲线下面积（AUC）显示该模型具有一定的准确性。多因素cox回归分析显示风险得分可作为胰腺癌独立预后因子（P<0.05）。临床相关性分析显示年龄、性别、Grade、Stage、M和N期等临床性状在高低风险组之间无明显差异（P>0.05），T期在高低风险组之间具有明显差异（P=0.0215<0.05）。利用CMAP平台筛选出前5抑制胰腺癌高风险基因表达的药物中，HDAC抑制剂2种，FLT3抑制剂、种VEGFR抑制剂与细菌细胞壁合成抑制剂各1种。结论：根据TCGA患者生存数据和表达谱，结合基因库获得胰腺癌预后相关免疫基因和lncRNA构建的预后风险模型能作为胰腺癌预后判断的新指标。 |
| 英文摘要: |
| Objective: To construct a pancreatic cancer prognostic risk model and screen candidate drugs based on The Cancer Genome Atlas TCGA database.Methods: Transcriptome and clinical data of pancreatic cancer were downloaded from TCGA database, and WGCNA module analysis, construction of prognostic risk model, differential analysis, Kaplan-Meier method survival analysis, risk analysis and clinical correlation analysis were performed by R software.The accuracy and independence of the model were judged by the area under the ROC curve (AUC) and multivariate cox regression analysis. Finally, the CMAP platform was used to screen pancreatic cancer drugs.Results: Transcriptome data of 178 cases and clinical data of 185 cases of pancreatic cancer were downloaded.The WGCNA analysis screening module gene is "MEgrey (P=0.04<0.05)".Twenty-eight prognosis-related immune genes and 70 prognosis-related lncRNAs were obtained by univariate cox regression analysis, and further prognostic risk models were constructed by Lasso regression and multivariate cox regression analysis.Model evaluation showed that the model could distinguish patients in high and low risk groups, and high risk patients with lower risk had poor prognosis (P<0.05).The area under the ROC curve (AUC) of the training and test groups showed that the model had some accuracy.Multivariate cox regression analysis showed that risk score could be used as an independent prognostic factor for pancreatic cancer (P<0.05).Clinical correlation analysis showed that there was no significant difference in age, gender, Grade, Stage, M and N stages and other clinical traits between high and low risk groups (P>0.05), while T stage had significant differences between high and low risk groups (P =0.0215<0.05).The CMAP platform was used to screen out the top 10 drugs that inhibit the expression of high-risk genes in pancreatic cancer, including 6 HDAC inhibitors, 2 FLT3 inhibitors, 1 VEGFR inhibitor and 1 bacterial cell wall synthesis inhibitor.Conclusion: According to the survival data and expression profiles of TCGA patients, the prognostic risk model constructed by combining the gene bank to obtain the prognosis-related immune genes and lncRNA of pancreatic cancer can be used as a new indicator for the prognosis of pancreatic cancer. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
