Liu Danqi, Zhou Boting, Liu Rangru
Department of Pharmacy, Xiangya Hospital, Central South University, Changsha, Hunan, People's Republic of China.
Institute for Rational and Safe Medication Practices, National Clinical Research Center for Geriatric Disorders, Xiangya Hospital, Central South University, Changsha, Hunan, People's Republic of China.
PeerJ. 2020 Jul 21;8:e9422. doi: 10.7717/peerj.9422. eCollection 2020.
Bladder cancer (BC) is the ninth most common malignancy worldwide. Bladder urothelial carcinoma (BLCA) constitutes more than 90% of bladder cancer (BC). The five-year survival rate is 5-70%, and patients with BLCA have a poor clinical outcome. The identification of novel clinical molecular markers in BLCA is still urgent to allow for predicting clinical outcomes. This study aimed to identify a novel signature integrating the three-dimension transcriptome of protein coding genes, long non-coding RNAs, microRNAs that is related to the overall survival of patients with BLCA, contributing to earlier prediction and effective treatment selection, as well as to the verification of the established model in the subtypes identified. Gene expression profiling and the clinical information of 400 patients diagnosed with BLCA were retrieved from The Cancer Genome Atlas (TCGA) database. A univariate Cox regression analysis, robust likelihood-based survival modelling analysis and random forests for survival regression and classification algorithms were used to identify the critical biomarkers. A multivariate Cox regression analysis was utilized to construct a risk score formula with a maximum area under the curve (AUC = 0.7669 in the training set). The significant signature could classify patients into high-risk and low-risk groups with significant differences in overall survival time. Similar results were confirmed in the test set (AUC = 0.645) and in the entire set (AUC = 0.710). The multivariate Cox regression analysis indicated that the five-RNA signature was an independent predictive factor for patients with BLCA. Non-negative matrix factorization and a similarity network fusion algorithm were applied for identifying three molecular subtypes. The signature could separate patients in every subtype into high- and low- groups with a distinct difference. Gene set variation analysis of protein-coding genes associated with the five prognostic RNAs demonstrated that the co-expressed protein-coding genes were involved in the pathways and biological process of tumourigenesis. The five-RNA signature could serve as to some degree a reliable independent signature for predicting outcome in patients with BLCA.
膀胱癌（BC）是全球第九大常见恶性肿瘤。膀胱尿路上皮癌（BLCA）占膀胱癌（BC）的90%以上。其五年生存率为5% - 70%，BLCA患者的临床预后较差。在BLCA中鉴定新的临床分子标志物对于预测临床结果仍然至关重要。本研究旨在鉴定一种整合蛋白质编码基因、长链非编码RNA、微小RNA三维转录组的新特征，该特征与BLCA患者的总生存期相关，有助于早期预测和有效治疗选择，并在已鉴定的亚型中验证所建立的模型。从癌症基因组图谱（TCGA）数据库中检索了400例诊断为BLCA患者的基因表达谱和临床信息。使用单变量Cox回归分析、基于稳健似然的生存建模分析以及用于生存回归和分类算法的随机森林来鉴定关键生物标志物。利用多变量Cox回归分析构建风险评分公式（训练集中曲线下面积最大值为AUC = 0.7669）。该显著特征可将患者分为高风险和低风险组，总生存时间存在显著差异。在测试集（AUC = 0.645）和整个数据集（AUC = 0.710）中证实了类似结果。多变量Cox回归分析表明，五RNA特征是BLCA患者的独立预测因子。应用非负矩阵分解和相似网络融合算法鉴定三种分子亚型。该特征可将每个亚型的患者分为高风险和低风险组，差异明显。对与五个预后RNA相关的蛋白质编码基因进行基因集变异分析表明，共表达的蛋白质编码基因参与肿瘤发生的途径和生物学过程。五RNA特征在一定程度上可作为预测BLCA患者预后的可靠独立特征。
