基于TCGA数据库构建肺腺癌预后相关的微小RNAs风险模型

目的寻找肺腺癌(lung adenocarcinoma，LUAD)特异性的预后相关微小RNAs(microRNAs, miRNAs)，为LUAD患者预后预测及个性化治疗方案制定提供依据。 方法下载TCGA数据库中522例LUAD患者组织标本的miRNA-Seq数据和临床病理及生存时间数据，用R语言对LUAD与癌旁组织中差异miRNAs进行分析。采用LASSO & COX回归模型在训练集(245例LUAD)中进行LUAD预后相关miRNAs筛选，并构建基于7个miRNAs表达谱的线性风险模型。根据风险值的高低，以中位风险值为界将患者分为高、低风险组，并分别在测试集(245例LUAD)和总体标本(490例LUAD)中对风险模型预测患者预后的有效性进行验证。采用COX回归分析miRNAs风险模型是否是独立的预后因子。 结果LUAD组织与癌旁组织中共有72个差异表达的miRNAs(上调45个、下调27个)。从训练集中确定miR-101-3p、miR-148a-3p、miR-192-5p、miR-193b-3p、miR-505-3p、miR-584-5p和miR-99a-5p 7个与总生存期相关的miRNAs构建预后风险模型。在训练集、测试集及总体标本中，高风险组患者与低风险组患者相比，总体生存时间均显著降低(P均<0.05)。经多因素COX回归分析，风险模型在训练集、测试集及总体样本中均是一个独立的预后因子(训练集HR=1.97，P=0.02；测试集HR=1.927，P=0.009；总体HR=1.909，P=0.001)。 结论研究确定了7个与LUAD患者预后相关的miRNAs，基于7个miRNAs构建的风险模型是1个独立的预后因子。

Construction of risk model associated with prognosis of lung adenocarcinoma based on a set of microRNAs by analyzing TCGA database

ObjectiveTo explore the specific prognosis related microRNAs (miRNAs) of lung adenocarcinoma (LUAD), and provide basis for prognosis prediction and individualized treatment. MethodsThe miRNA-Seq data and clinical information of LUAD patients were downloaded from the TCGA database, and the differentially expressed miRNAs between LUAD and adjacent normal tissues were identified by R Language. The LASSO & COX regression was used to develop a miRNA-based model for predicting patients? survival in the training set (n=245) and to carry out LUAD prognostic related miRNAs screening, and to construct a linear risk model based on 7 miRNAs expression profiles. According to the risk values, the patients were divided into high and low risk groups with the median risk value as the boundary, and the effectiveness of the prognosis was verified by the risk model in the test set (n=245) and the total specimens (n=490) respectively. COX regression analysis was used to determine whether the miRNAs risk model was an independent prognostic factor. ResultsSeventy-two differentially expressed miRNAs were identified between LUAD and adjacent normal tissues. forty-five miRNAs were up-regulated and 27 were down-regulated in LUAD tissues. Seven survival-related miRNAs (miR-101-3p, miR-148a-3p, miR-192-5p, miR-193b-3p, miR-505-3p, miR-584-5p, and miR-99a-5p) were identified in the training set and a prognostic model based on the expression of the 7 miRNAs was developed and its coefficient was evaluated. It showed that the overall survival time of the high-risk group was significantly lower than that of the low risk group in the training set, test set and whole cohort (P<0.05). Multivariate cox regression analysis indicated that the risk model was an independent prognostic factor in the training set, test set and whole cohort (training set HR=1.97, P=0.02; test set HR=1.927, P=0.009; overall HR=1.909, P=0.001). ConclusionSeven miRNAs are identified to be significantly associated with the prognosis of LUAD patients and the risk model based on the 7 miRNAs could be an independent prognostic factor.