欢迎访问《岭南现代临床外科》官方网站,今天是

岭南现代临床外科 ›› 2022, Vol. 22 ›› Issue (01): 14-23.DOI: 10.3969/j.issn.1009-976X.2022.01.003

• 论著与临床研究 • 上一篇    下一篇

胰腺癌预后相关铁死亡基因筛选及预后预测模型构建

何冲, 周姝睿, 李雅晴, 陈少杰, 练国达, 陈尚祥*, 黄开红*   

  1. 中山大学孙逸仙纪念医院消化内科,广州510120
  • 通讯作者: *陈尚祥,Email:chenshx67@mail.sysu.edu.cn;黄开红,Email:huangkh@sysu.edu.cn
  • 基金资助:
    国家自然科学基金面上项目(81874057); 国家自然科学基金青年项目(82103142); 国家自然科学基金青年项目(82003175); 国家自然科学基金青年项目(82002505); 中国博士后科学基金第70批面上项目(2021M703701); 广东省医学科学技术研究基金(A2016210); 广东省医学科学技术研究基金(A2019447); 广东省医学科学技术研究基金(A2021158); 广东省自然区域联合基金-青年基金项目(2020A1515110841); 广州市科技计划项目(202102020082); 广州市科技计划项目(202102020161)

Screening of ferroptosis-related genes and construction of prognostic prediction model in pancreatic cancer

HE Chong, ZHOU Shu-rui, LI Ya-qing, CHEN Shao-jie, LIAN Guo-da, CHEN Shang-xiang, HUANG Kai-hong   

  1. Department of Gastroenterology, Sun Yat-sen Memorial Hospital, Sun Yat-sen University, Guangzhou 510120, China
  • Received:2021-10-31 Online:2022-03-17 Published:2022-03-17
  • Contact: CHEN Shang-xiang, chenshx67@mail.sysu.edu.cn;HUANG Kai-hong, huangkh@sysu.edu.cn

摘要: 目的 探讨铁死亡相关基因对胰腺癌患者预后预测的价值,并进一步探索其分子机制。方法 从公共数据库中下载胰腺癌患者的基因表达及相应的临床病理数据,采用R软件“limma”包筛选胰腺癌组织中差异表达基因,使用单因素COX回归分析筛选预后相关的铁死亡相关基因,利用LASSO COX回归分析构建预后模型。通过Kaplan-Meier生存分析检验该模型在预测胰腺癌预后中的意义,用ROC曲线评估模型预测胰腺癌预后的准确性。在ICGC验证队列中通过Kaplan-Meier生存分析、ROC曲线评估该模型在外部队列中的预测价值。依照风险值中位数,将TCGA训练队列分为两组,比较其基因表达差异、差异基因富集通路差异,评估免疫细胞浸润丰度。结果 通过单因素COX回归分析,37个铁死亡相关基因被鉴定为预后基因(P<0.05)。利用LASSO COX回归分析,构建了一个基于15个铁死亡相关基因的预后预测模型。根据模型计算将队列中所有患者分为高风险组和低风险组,Kaplan-Meier生存分析表明高风险组的胰腺癌患者较低风险组生存时间短(HR=2.16,P<0.05)。ROC曲线分析证明了预测模型在胰腺癌预后预测中的准确性:ROC曲线下面积值分别为0.74(1年)、0.82(3年)、0.88(5年)。对高低风险组差异表达基因进行功能富集分析,发现两组胰腺癌患者免疫微环境之间存在差异,高风险组中幼稚B细胞、浆细胞、CD8+T细胞浸润程度较低风险组低,但M0和M1巨噬细胞的浸润程度高(P<0.05)。此外,在PD-1和CTLA4在TCGA-PAAD队列低危组和高危组中的表达中,TCGA-PAAD队列低风险组患者中PD-1和CTLA4的表达水平较高风险组患者更高(P<0.05)。结论 本研究构建了胰腺癌中铁死亡相关基因预后预测模型,具有预测胰腺癌患者预后的作用,并发现TCGA队列中高低风险组免疫微环境存在差异,可为免疫治疗提供参考。

关键词: 胰腺癌, 铁死亡, 预后

Abstract: Objective To explore the value of ferroptosis-related genes in predicting the prognosis of patients with pancreatic cancer, and further explore the potential molecular mechanism. Methods Gene expression and corresponding clinicopathological data of pancreatic cancer patients were downloaded from the public database. The differentially expressed genes in pancreatic cancer tissues were screened by R software “limma” package. Differentially expressed genes in pancreatic cancer tissues were verified by univariate COX regression analysis, and a prognostic model was established by LASSO COX regression analysis. We performed Kaplan-Meier survival analysis to examine the significance of the model in the prognosis of prediction pancreatic cancer. The accuracy of the model in the prognosis of prediction pancreatic cancer was analyzed by the ROC curve. In the ICGC verification queue, the predictive value of the model was evaluated by Kaplan-Meier survival analysis and ROC curve. Then we divided the TCGA-PAAD cohort into two groups according to the risk value, compared the differences in gene expression and gene enrichment pathways, and evaluated the abundance of immune cells in the tumor microenvironment. Results Thirty-seven ferroptosis related genes were identified as prognostic genes by univariate COX regression analysis. We constructed a prognostic model based on 15 iron death-related genes using LASSO COX regression analysis. According to the model calculation, all patients in the cohort were divided into the high-risk group and the low-risk group. Kaplan-Meier survival analysis suggested that the survival time of pancreatic cancer patients in the high-risk group was shorter than that in the low-risk group (HR=2.16, P<0.05). ROC curve analysis proved the accuracy of the predictive model in predicting the prognosis of pancreatic cancer (the area under the ROC curve were 0.74 (1 year), 0.82 (3 years), and 0.88 (5 years), respectively). Functional enrichment analysis of differentially expressed genes in the high-risk and low-risk groups showed that there were significant differences in immune microenvironment between the two groups. The infiltration degrees of B naive cells, plasma cells, and CD8+T cells in the high-risk group were lower than that in the low-risk group, but the infiltration degrees of M0 and M1 macrophages were higher in the high-risk group than in the low-risk group(P<0.05). In addition, we also studied the expression of PD-1 and CTLA4 in the low-risk group and high-risk group of the TCGA-PAAD cohort and found that the expression levels of PD-1 and CTLA4 in the low-risk group of the TCGA-PAAD cohort were higher than those in the high-risk group. Conclusion We have constructed a prognostic model of ferroptosis-related genes in pancreatic cancer, which can predict the prognosis of pancreatic cancer patients, and found that there are differences in immune microenvironment between the high-risk group and the low-risk group in the TCGA cohort, which can provide a reference for immunotherapy.

Key words: pancreatic cancer, ferroptosis, prognosis

中图分类号: