肺鳞状细胞癌自噬相关基因预后风险评分模型构建及分析

Construction and analysis on prognostic risk score model of lung squamous cell carcinoma:based on autophagy-associated genes

目的探索构建基于自噬相关基因(ARGs)肺鳞状细胞癌(LUSC)预后风险评分模型并分析。方法通过GeneCards数据库获得4418个ARGs。从TCGA数据库收集了551例LUSC患者的基因表达谱及临床数据,提取所有ARGs的表达数据,利用R软件筛选差异表达的ARGs。对差异表达的ARGs进行富集分析。利用Cox回归模型构建ARGs的预后风险评分模型。根据风险评分计算公式计算出每个样本的风险评分,以中位数为cut-off值,将患者分为高风险评分组和低风险评分组。绘制多指标受试者工作特征曲线并计算风险评分评估模型性能。最后利用单因素和多因素Cox回归分析评价模型是否具有独立预后价值,并分析其临床相关性。利用外部数据集对其验证。结果初步筛选了50个有预后价值的差异表达的ARGs,以此为基础,利用Cox回归分析构建了由5个ARGs(LAMP2、TUSC1、CDKN1A、ITGB1、RGS19)组成的LUSC预后风险评分模型。该模型中,低风险评分组与高风险评分组的生存时间比较差异有统计学意义(3.032年比2.275年,t=3.23,P<0.001)。风险评分在单因素和多因素Cox回归分析中与LUSC患者预后比较差异均有统计学意义(P值均<0.001),提示风险评分可作为LUSC潜在的独立预后因素。并且,与外部数据集交叉验证仍有良好预测效果。临床特征相关性分析表明高风险评分与年龄、性别和发生不良预后密切相关。结论构建了一个由5个ARGs组成的LUSC风险评分模型,该模型可为预测LUSC患者预后提供参考,未来或可与恶性肿瘤分期联合应用于LUSC患者的预后预测。

Objective To construct and analyze on a prognostic risk score model of lung squamous cell carcinoma(LUSC)based on autophagy-associated genes(ARGs).Methods Using R software to screen differentially expression of obtained ARGs,which 4418 ARGs data were from the GeneCards database and the other ARGs data were from Gene expression profiles and clinical data of 551 LUSC patients collected from TCGA database,and the enrichment analysis was performed to assess the differentially expression of ARGs.The prognostic risk scoring model of ARGs were conducted by the Cox regression model,and then each sample risk score was obtained by calculation formula of this model,the patients were assigned into high-risk group and low-risk group according to the critical of the median score as cut-off value.Multivariate receiver operator characteristic curve were plotted to calculate risk scores and to assess model performance.Finally,univariate and multivariate Cox regression analysis were used to evaluate independent prognostic value of the model and to analyze its clinical correlation,with verification by external dataset.Results On the basis of primary screening 50 differentially expressed ARGs with prognostic value,a LUSC prognostic risk score model including five ARGs(LAMP2,TUSC1,CDKN1A,ITGB1,RGS19)was constructed using Cox regression analysis.In this model,the effect in the low-risk group was significantly superior to the high-risk group in mean survival time(3.032 years vs 2.275 years,t=3.23,P<0.001).The risk score was significantly associated with the prognosis of LUSC patients in univariate and multivariate Cox regression analyses(all P<0.001),suggesting the risk score as a potential independent prognostic factor for LUSC,meantime,a good prediction effect found in the cross validation with external dataset.Clinical correlation analysis indicated that high-risk scores were closely associated with age,gender,and poor prognosis.Conclusions The LUSC risk scoring model comprised from 5 ARGs can provide a reference for predicting prognosis of LUSC patients.Combining with malignant tumor staging,which should be emphasized in predicting prognosis for LUSC patients futurely.