基于miRNA和临床特征建立可手术肺鳞癌预后的预测模型

Establishment and validation of a prognostic model for resectable lung squamous cell carcinoma patients based on miRNA and clinical features

目的探讨miRNA和临床特征联合用于预测可手术肺鳞癌患者预后的性能。方法从TCGA数据库下载了310例Ⅰ~Ⅲ期可手术肺鳞癌患者的miRNA变异数据建立了训练集。回顾性收集解放军总医院第八医学中心近5年134例Ⅰ~Ⅲ期肺鳞癌患者组织样品,进行miRNA测序,建立了验证集。用R软件和PRISM5.0进行分析作图。结果与正常组织比较,Ⅰ~Ⅲ期可手术肺鳞癌患者显示大量高表达或低表达miRNA,其中显著高表达miRNA数量(45个)远多于显著低表达miRNA数量(6个)。富集分析揭示了多个肿瘤相关的功能或通路,主要包括PI3K-Akt、MAPK、Ras通路,RHO GTPase循环及神经系统功能等。单因素Cox回归分析发现肿瘤分期和肿瘤部位是显著影响预后的因素。多因素分析发现肿瘤部位为影响预后的独立因素。单因素Cox回归分析发现7个miRNA是影响预后的因素。多因素Cox回归分析发现miR-374a、miR-378c、miR-4638、miR-4746、miR-656、miR-99b和肿瘤部位是影响预后的独立因素。这些因素被用于建立个体预后预测的Cox模型。随后的内、外部验证显示,模型1~5年预测性能良好,内部验证最大AUC来源于第3年预测,外部验证最大AUC来自于第5年预测,在内、外部验证中AUC分别达到0.77和0.75。结论基于miRNA和临床病理因素成功建立了肺鳞癌患者个体预后预测模型,将有助于患者预后预测和治疗策略选择。

Objective To investigate the prognostic performance of miRNA combined with clinical features in patients with resectable lung squamous cell carcinoma(LUSC).Methods The miRNA sequencing data of 310 patients with stageⅠ-Ⅲresectable LUSC were downloaded from the TCGA database to establish a training cohort.Tissue samples from 134 patients with stageⅠ-ⅢLUSC in our hospital in the past 5 years were retrospectively collected and miRNA sequencing was performed to establish a validation cohort.All analysis and figure plot were performed by R software or Graphpad PRISM 5.0.Results Compared with normal tissues,patients with stageⅠtoⅢoperable lung squamous cell carcinoma showed a large number of miRNAs with high or low expression,in which the number of significantly high expression miRNAs(45)was much more than that of significantly low expression miRNAs(6).Enrichment analysis revealed a number of tumor-related functions or pathways,including PI3K-Akt,MAPK,Ras pathway,RHO GTPase cycle,and nervous system functions.Univariate Cox regression analysis showed that tumor stage and location were significant prognostic factors.Multi-factor analysis showed that site was an independent factor affecting prognosis.Univariate Cox regression analysis of mirnas found that 7 mirnas were prognostic factors.Multivariate Cox regression analysis showed that miR-374a,miR-378c,miR-4638,miR-4746,miR-656,miR-99b and location were independent factors affecting prognosis.These factors were used to establish a Cox regression model for individual prognostic prediction.Subsequent internal and external validation showed that the prediction performance of the model from 1 to 5 years was good.The maximum AUC of the internal validation came from the prediction of the third year,and the maximum AUC of the external validation came from the prediction of the fifth year,and the AUC of the internal and external validation reached 0.77 and 0.75 respectively.Conclusion The individual prognosis prediction model for LUSC was successfully established based on miRNA and clinicopathological factors,which may be helpful for prognosis prediction and treatment strategy selection.