肝细胞癌预后风险模型构建与免疫治疗疗效预测

Construction of hepatocellular carcinoma prognosis risk model and prediction of immunotherapy efficacy

目的:构建肝细胞癌(HCC)预后风险模型,阐明不同预后分层患者的免疫特征及免疫治疗响应。方法:下载肝细胞癌TCGA-LIHC、ICGC(LIRI-JP)的RNA-seq数据和GSE14520、GSE54236的基因芯片数据,以及对应样本的临床信息。首先筛选TCGA-LIHC、GSE14520、GSE54236中癌组织和癌旁组织样本的差异表达基因。对共有差异基因,利用TCGA-LIHC数据进行单因素cox回归分析得到HCC预后相关基因。随机选取5个基因为一组,采用Lasso-cox回归分析结合五折交叉验证方法在10 000个组合中筛选最优预后标志物组合。使用TCGA-LIHC数据作为训练集构建预后风险模型,ICGC数据作为验证集验证模型性能。肿瘤免疫功能障碍和排斥(TIDE)算法以及免疫表型评分(IPS)用于预测不同预后分层患者免疫治疗疗效。结果:与高风险组相比,低风险组HCC患者总生存期显著延长。高风险患者肿瘤增殖率、Treg和Th2细胞趋化、基质重构、促肿瘤细胞因子显著升高,而NK细胞、Th1细胞、效应细胞、内皮细胞在低风险患者中显著升高。免疫检查点分析表明,PDCD1、CTLA4和CD276在高风险患者中表达上调,PDCD1LG2在低风险患者中表达上调。TIDE评分和IPS结果预测低风险组患者免疫治疗疗效更优。结论:本研究构建了包含DNASE1L3、RDH16、DLGAP5 3个基因的预后风险模型,可有效预测HCC患者预后,辅助临床制定个体化免疫治疗决策。

Objective:To construct a prognostic risk model for hepatocellular carcinoma(HCC),and elucidate the immune characteristics and immunotherapy response in patients with different prognostic stratification.Methods:RNA-seq data of TCGA-LIHC and ICGC(LIRI-JP),and gene microarray data of GSE14520 and GSE54236 in hepatocellular carcinoma,as well as clinical information of the corresponding samples were downloaded.First,screening of differentially expressed genes in tumor and non-tumor tissue samples from TCGA-LIHC,GSE14520 and GSE54236.For the common differential genes,univariate cox regression analysis was performed using TCGA-LIHC data to obtain HCC prognosis-related genes.Five genes were randomly selected as a panel,and the optimal prognostic marker panel was screened among 10000 panels using Lasso-cox regression analysis combined with a five-fold cross-validation method.TCGA-LIHC data were used as training set to construct the prognostic risk model,and ICGC data were used as validation set to test the model performance.Tumor immune dysfunction and exclusion(TIDE)algorithm and Immunophenotypic score(IPS)were used to predict immunotherapy efficacy in patients in different prognostic groups.Results:Overall survival was significantly longer in low-risk group of HCC patients compared with high-risk group.Tumor proliferation rate,Treg and Th2 cell chemotaxis,stromal remodeling,and pro-tumor cytokines were significantly increased in high-risk patients,while NK cells,Th1 cells,effector cells and endothelial cells were significantly increased in low-risk patients.Immune checkpoint analysis showed that PDCD1,CTLA4 and CD276 were up-regulated in high-risk patients,while PDCD1LG2 was upregulated in low-risk patients.TIDE score and IPS results predicted that patients in low-risk group had better efficacy to immunotherapy.Conclusion:This study constructed a prognostic risk model containing three genes,DNASE1L3,RDH16 and DLGAP5,which can effectively predict the prognosis of HCC patients and assist in clinical decision making for individualized immunotherapy.