微RNA风险评分模型预测肝细胞癌预后的价值分析

Abstract

目的 下载癌症基因组图谱(TCGA)数据库中肝细胞癌(HCC)miRNA转录组数据进行数据挖掘，筛选与HCC预后相关的miRNA，构建miRNA风险评分模型，并评估其对HCC预后的预测价值。 方法 从TCGA数据库下载HCC样本miRNA表达量及临床数据，应用R语言筛选出HCC与癌旁组织的差异表达miRNA，将其与临床数据整合后，随机分为训练集和测试集，对训练集依次进行单因素Cox及LASSO-Cox回归分析，筛选与HCC预后相关的miRNA并构建风险评分模型，并使用Kaplan-Meier分析评估模型的稳健性及是否可以预测同一临床分期的患者预后，最后通过受试者工作特征曲线(ROC)计算曲线下面积(AUC)，在训练集、测试集及二者的合集中比较该模型和传统TNM分期的预测准确性。 结果 共筛选出300个差异基因，LASSO-Cox回归分析显示hsa-miR-139-5p、hsa-miR-1180-3p、hsa-miR-1269b、hsa-miR-3680-3p、hsa-miR-509-3-5p、hsa-miR-31-5p与HCC预后相关，根据构建的miRNA风险评分模型，计算每个样本的风险得分，并根据中位风险得分值将样本划分为高风险组和低风险组。Kaplan-Meier曲线显示，训练集与测试集的高风险组患者的生存率显著低于低风险组患者(P＜0.05)。ROC曲线结果显示，训练集、测试集及合集样本中，miRNA模型及TNM分期的AUC分别为0.817、0.667，0.808、0.665与0.814、0.663。独立预后分析结果显示，该miRNA评分模型可作为HCC的独立预后因子(P＜0.05)。 结论 hsa-miR-139-5p、hsa-miR-1180-3p、hsa-miR-1269b、hsa-miR-3680-3p、hsa-miR-509-3-5p、hsa-miR-31-5p与HCC预后相关，在训练集、测试集及合集样本中miRNA风险评分模型的预测准确性均优于TNM分期。分层分析表明，该模型还可预测同一TNM分期患者的预后，在临床工作中具有一定参考价值，可作为独立预测HCC患者预后的模型。

Alternate abstract:

Objective To screen out the microRNAs (miRNAs) associated with the prognosis of hepatocellular carcinoma (HCC) through data mining of miRNA transcriptome data of HCC downloaded from The Cancer Genome Atlas (TCGA) database, to establish a miRNA risk score model, and to investigate its value in predicting the prognosis of HCC. Methods The miRNA expression data and clinical data of HCC samples were downloaded from TCGA database and R language was used to screen out differentially expressed miRNAs between HCC tissue and adjacent tissue, which were randomly divided into training set and testing set after being integrated into clinical data. Univariate Cox regression analysis and least absolute shrinkage and selection operator (LASSO) Cox regression analysis were performed for the training set to screen out the miRNAs associated with the prognosis of HCC, and then a miRNA risk score model was established. The Kaplan-Meier method was used to evaluate the robustness of the model and whether it could predict the prognosis of patients in the same clinical stage. Finally, the receiver operating characteristic (ROC) curve was plotted and the area under the ROC curve (AUC) was calculated to compare the predictive accuracy of the model versus TNM staging in the training set, the testing set, and the entire set. Results A total of 300 differentially expressed miRNAs were screened out and the LASSO Cox regression analysis revealed that hsa-miR-139-5p, hsa-miR-1180-3p, hsa-miR-1269b, hsa-miR-3680-3p, hsa-miR-509-3-5p, and hsa-miR-31-5p were associated with the prognosis of HCC. The risk score was calculated for each sample according to the established miRNA risk score model, and the samples were divided into high-risk group and low-risk group according to the median risk score. The Kaplan-Meier curve showed that in both training and testing sets, the high-risk group had a significantly lower survival rate than the low-risk group (P < 0.05). The ROC curve was used to evaluate the prediction efficiency of this model, and the results showed that in the training set, the testing set, and the entire set, the miRNA model had an AUC of 0.817, 0.808, and 0.814, respectively, while TNM staging had an AUC of 0.667, 0.665, and 0.663, respectively. The results of independent prognostic analysis also showed that this miRNA score model could be used as an independent prognostic factor for HCC (P < 0.05). Conclusion Hsa-miR-139-5p, hsa-miR-1180-3p, hsa-miR-1269b, hsa-miR-3680-3p, hsa-miR-509-3-5p, and hsa-miR-31-5p are associated with the prognosis of HCC, and the miRNA risk score model has a better prediction accuracy than TNM staging in the training set, the testing set, and the entire set. The stratified analysis also shows that the model can predict the prognosis of patients within the same TNM stage, and therefore, it has a certain reference value in clinical practice and can be used as an independent model for predicting the prognosis of HCC patients.