【摘要】 目的:通过筛选与胃癌铁死亡相关的长链非编码RNA(LncRNA),构建预后风险模型，协助评估胃癌患者预后。方法:在癌症基因数据库(The Cancer Genome Atlas TCGA)下载胃癌的转录组和临床数据，从NCBI网站获得103个铁死亡相关基因。采用R软件进行Kaplan-Meier(KM)生存分析和单因素COX回归分析筛选出关键的LncRNA,将临床样本按7∶3的比例随机分为训练集和验证集，利用多因素COX回归分析建立胃癌铁死亡相关LncRNA的预后模型。两组分别进行生存分析、风险曲线分析、单因素和多因素独立预后分析、多指标受试者操作特征(ROC)曲线的绘制及列线图分析，最后对训练集进行GSEA的KEGG分析。结果:单因素Cox回归分析得到13个LncRNA(LINC00106、TNFRSF10A-AS1、AC244153.1、AC005586.1、AP001318.2、MAGI2-AS3、AP001528.2、MSC-AS1、PVT1、AC037198.1、AC010719.1、HAGLR),经多因素回归分析筛选出5个胃癌铁死亡相关的LncRNA(AC005586.1、AP001318.2、AP001528.2、AC010719.1、HAGLR),训练集和验证集构建的预后模型趋势一致，且ROC评估风险模型的预测能力大于其他临床指标。构建的5个LncRNA列线图，预测1年、2年、3年校准曲线也基本和45°的对角线重合，模型预测性能良好。结论:LncRNA AC005586.1、AP001318.2、AP001528.2、AC010719.1、HAGLR胃癌铁死亡相关LncRNA可有效预测胃癌患者的预后。更多还原

【Abstract】 Objective:The prognostic risk model was constructed to help evaluate the prognosis of patients with gastric cancer by screening long non-coding RNA(LncRNA) associated with ferroptosis in gastric cancer.Methods:The transcripts and clinical date of gastric cancer were downloaded from the Cancer Genome Atlas(TCGA) and 103 ferroptosis-related genes were downloaded from NCBI.The expression matrix of LncRNA associated with ferroptosis in gatric cancer was obtained through data processing by R and Perl software.Combined that with survival time and then the key LncRNAs were screened out by KM and univariate Cox regression an analysis using R software.Randomly divided clinical samples into training and validation sets in a 7∶3 ratio.The prognostic prediction models of the Train and Test group were established by multivariate Cox regression model.Survival analysis, risk cure analysis, single-factor and multi-factor independent prognostic analysis, multi-ROC curve drawing, etc.,were performed for the two groups respectively.Finally, line chart and KEGG analysis of GSEA were performed for the train.Results:13 LncRNAs(LINC00106,TNFRSF10A-AS1,AC244153.1,AC005586.1,AP001318.2,MAGI2-AS3,AP001528.2,MSC-AS1,PVT1,AC037198.1,AC010719.1,HAGLR) were obtained by univariate Cox regression analysis, and five LncRNAs(AC005586.1,AP001318.2,AP001528.2,AC010719.1,HAGLR) related to ferroptosis were screened out by multivariate regression analysis.5 LncRNAs related toferroptosis in gastric cancer were screened through multiple regression analysis(AC005586.1,AP001318.2,AP001528.2,AC010719.1,HAGLR),the trend of the prognostic models constructed from the training and validation sets was consistent.The predictive power of ROC assessment risk model was greater than that of other clinical indicators.The constructed 5 lncRNA histograms showed that the 1-year, 2-year and 3-year calibration curves coincided with the diagonal of 45°,indicating that the risk model had a good predictive ability.Conclusion: AC005586.1,AP001318.2,AP001528.2,AC010719.1 and HAGLR associated with iron death in gastric cancer can effectively predict the prognosis of gastric cancer patients.更多还原