摘要目的:为寻找肝细胞癌相关预后的铁死亡基因,寻找独立预后的因素,构建肝细胞癌患者铁死亡的预后模型。方法:通过TCGA数据库下载TCGA-LIHC的转录组数据和临床信息,使用Perl 语言对TCGA-LIHC的铁死亡相关基因的表达谱数据进行提取。使用R语言的limma包对铁死亡相关基因的表达谱数据进行差异分析。使用R语言的org.Hs.eg.db包对差异的铁死亡基因进行id注释,使用R语言的colorspace,stringi和ggplot2包对铁死亡基因进行GO和KEGG富集分析。使用Perl语言对铁死亡相关基因的表达谱信息和TCGA-LIHC患者的生存时间、生存状态进行整合,通过单因素cox分析,筛选出与LIHC患者生存预后显著相关的铁死亡基因,然后使用多因素cox分析对模型进行构建。使用R语言的survival包对肝细胞癌铁死亡预后模型进行分组,并且进行KM生存分析。使用R语言的pheatmap包对铁死亡预后模型进行风险曲线,生存状态图的绘制。使用Perl对患者的生存时间,生存状态,年龄,性别,TNM分期与铁死亡预后模型进行整合,进行单因素和多因素cox分析,寻找独立预后的因素,并使用了ROC曲线对模型的预后因素进行了评估。结果:TCGA-LIHC包含了个50正常组织,374个肝细胞癌组织。通过差异分析,共得到了TCGA-LIHC相关铁死亡差异基因83个,其中上调基因70个,下调基因13个。GO功能富集分析显示,response to oxidative stress和cellular response to oxidative stress是主要的铁死亡相关生物过程;KEGG功能富集分析显示,Ferroptosis和Central carbon metabolism in cancer是主要的生物通路。单因素cox分析筛选出了31个铁死亡预后相关基因,多因素cox分析构建了10个铁死亡相关基因的预后模型(FANCD2,ZEB1,BLOC1S5-TXNDC5,HMOX1,GABARAPL1,FLT3,IDH1,G6PD,VDAC2,MYB),多因素cox回归中的系数如下:铁死亡基因风险评分=(-0.3774×FANCD2的表达)+(-0.3774×ZEB1的表达)+(-0.3774×BLOC1S5-TXNDC5的表达)+(-0.3774×HMOX1的表达)+(-0.3774×GABARAPL1的表达)+(-0.3774×FLT3的表达)+(-0.3774×IDH1的表达)+(-0.3774×G6PD的表达)+(-0.3774×VDAC2的表达)+(-0.3774×MYB的表达)。KM生存分析显示,预后模型低表达的患者生存中更具备优势,P=1.992e-08。风险曲线,生存状态进一步验证了我们的结果。单因素和多因素cox分析分别显示了clinical stage的P<0.001和P=0.004,riskScore的P<0.001。ROC曲线评估了铁死亡模型的AUC为0.736,clinical stage临床分级的AUC为0.701,这证明了该模型的可靠性。结论:通过生物信息学分析,构建了10个肝细胞癌相关的铁死亡基因模型,可以用于患者预后的评估。
Abstract：Objective: To find the ferroptosis gene and independent prognostic factors, and to construct the prognostic model of ferroptosis in HCC patients. Methods: The transcriptomic data and clinical information of TCGA-LIHC were downloaded from the TCGA database, and the expression profiles of TCGA-LIHC genes were extracted using Perl language.The expression profiling data for ferroptosis-related genes were differentially analyzed using the limma package in R. Using the org.Hs.eg in the R language. Differential ferroptosis genes were id annotated using the org.Hs.eg.db package in R, and the GO and KEGG enrichment analysis of ferroptosis genes was performed using the colorspace, stringi and ggplot2 packages in R. The expression profile information of ferroptosis-related genes and the survival time and survival status of TCGA-LIHC patients were integrated using Perl language, and ferroptosis-related genes significantly associated with the survival prognosis of LIHC patients were screened by one-way cox analysis, and then the model was constructed using multi-factor cox analysis.Prognostic models of ferroptosis for hepatocellular carcinoma were grouped using the survival package in R and KM survival analysis was performed.Risk curves and survival status maps were plotted for the ferroptosis prognosis models using the pheatmap package in R. Patients’ survival time, survival status, age, sex, and TNM stage were integrated with an ferroptosis prognosis model using Perl to perform univariate and multifactor cox analyses to find factors that could be used as independent prognostic factors and ROC curves were used to assess the prognostic factors of the model.Results: The TCGA-LIHC contained 50 normal tissues and 374 hepatocellular carcinoma tissues.The differential analysis yielded 83 differential genes of TCGA-LIHC-related ferroptosis, including 70 upregulated genes and 13 downregulated genes.GO functional enrichment analysis showed that response to oxidative stress and cellular response to oxidative stress were the major ferroptosis-related biological processes; KEGG functional enrichment analysis revealed that Ferroptosis and Central carbon metabolism in cancer were the major biological pathways. Univariate cox analysis screened out 31 genes associated with ferroptosis prognosis, and multivariate cox analysis constructed a prognostic model of 10 ferroptosis-related genes (FANCD2, ZEB1, BLOC1S5-TXNDC5, HMOX1, GABARAPL1, FLT3, IDH1, G6PD, VDAC2, MYB). The coefficients in the multivariate cox regression were as follows: ferroptosis gene risk score = (-0.3774 FANCD2 expression) + (-0.3774 ZEB1 expression) + (-0.3774 BLOC1S5-TXNDC5 expression) + (-0.3774 HMOX1 expression) + (-0.3774 GABARAPL1 expression) + (-0.3774 FLT3 expression) + (-0.3774 IDH1 expression) + (-0.3774 G6PD expression) + (-0. 3774 VDAC2 expression) + (-0.3774 MYB expression).KM survival analysis showed showed a greater advantage in survival for patients with low expression of the prognostic model, P=1.992e 08. Risk curve and survival status further validated our results. Univariate and multivariate cox analyses showed P<0.001 and P=0.004 for clinical stage, and P<0.001 for riskScore, respectively. ROC curve evaluated the AUC of 0.736 for the ferroptosis model and the AUC of 0.701 for the clinical stage, which demonstrateed the reliability of the model.Conclusion: Through bioinformatics analysis, 10 ferroptosis gene models can be used to evaluate patient prognosis.
[1] Yuan K,et al.TXNDC12 promotes EMT and metastasis of hepatocellular carcinoma cells via activation of β-catenin[J].Cell Death & Differentiation,2020,27(4):1355~1368. [2] Lee A,F Lee.Medical oncology management of advanced hepatocellular carcinoma 2019:a reality check[J].Frontiers of Medicine,2020,14(3):273~283. [3] 张正阳,等.肿瘤铁死亡调控机制的新进展[J].江苏大学学报(医学版),2020,30(2):121~124. [4] Merkofer M,et al.Fenton chemistry and iron chelation under physiologically relevant conditions:electrochemistry and kinetics[J].Chemical Research in Toxicology,2006,19(10):1263~1269. [5] 娄丽丽,王中峰.金属硫蛋白1G通过抑制铁死亡促进肝癌细胞索拉非尼耐药[J].临床肝胆病杂志,2016,32(10):1887. [6] Hisateru Komatusu,T M T I,et al.Clinical significance of FANCD2 gene expression and its association with tumor progression in hepatocellular carcinoma[J].Anticancer Research,2017,37(3):1083~1090. [7] Shen Y,et al.Iron metabolism gene expression and prognostic features of hepatocellular carcinoma[J].Journal of Cellular Biochemistry,2018,119(11):9178~9204. [8] Yang Y,et al.Nedd4 ubiquitylates VDAC2/3 to suppress erastin-induced ferroptosis in melanoma[J].Nat Commun,2020,11(1):433. [9] LIU C,et al.Low expression of GABARAPL1 is associated with a poor outcome for patients with hepatocellular carcinoma[J].Oncology Reports,2014,31(5):2043~2048. [10] Ally A,et al.Comprehensive and integrative genomic characterization of hepatocellular carcinoma[J].Cell,2017,169(7):1327~1341. [11] Hong X,et al.PTEN antagonises Tcl1/hnRNPK-mediated G6PD pre-mRNA splicing which contributes to hepatocarcinogenesis[J].Gut,2014,63(10):1635~1647.
2022, Vol. 28 
冀公网安备13080202000677号