摘要目的:鉴定肾透明细胞癌(ccRCC)甲基化生物标志物，并构建ccRCC甲基化预后分子标签。方法:提取TCGA数据库中ccRCC患者的临床数据以及ccRCC甲基化数据，筛选出ccRCC癌组织和正常肾组织中甲基化差异基因，然后将差异基因经过LASSO-Cox回归分析筛选ccRCC预后相关的基因并构建ccRCC预后分子标签。结果:首先筛选出485例ccRCC癌组织和正常肾组织中甲基化差异基因798种，差异基因结合ccRCC生存数据经过单因素Cox分析得到492种与ccRCC预后相关的甲基化差异基因( P<0.05)；再将325例肿瘤组织样品数据按30%和70%随机分为两组，70%组为训练组用于筛选预后分子标签，30%组为测试组用于测试筛选出的分子标签预测预后效果。采用LASSO-Cox回归分析从中筛选出5种与ccRCC预后相关的甲基化差异基因，分别为CMTM3、HAND2、HIST1H3E、MIR4425、SHOX2。利用5种基因联合多因素COX回归系数构建分子标签：-5.36342273779505×DML CMTM3+3.260844271×DML HAND2+4.995985347×DML HIST1H3E-1.84339230384795×DML MIR4425+3.323857473×DML SHOX2代入相关数据后，计算分子标签分值。训练组、测试组、全部数据组的分子标签值与ccRCC患者生存期显著相关，分子标签值越高患者预后越差。 结论:通过对TCGA数据库的挖掘，发现影响ccRCC预后的有5种基因，且联合5种基因构建的分子标签与ccRCC患者的预后有显著性关联，构建的ccRCC预后模型有望应用于临床。

abstractsObjective:To identify methylation biomarkers for clear cell renal cell carcinoma and construct molecular label for prognosis of clear cell renal cell carcinoma methylation.Methods:The clinical data and methylation data of renal clear cell carcinoma were extracted from TCGA database. Firstly, the differential genes of methylation in renal clear cell carcinoma were screened out, and then the genes related to the prognosis of renal clear cell carcinoma were screened by LASSO-Cox regression, and the prognosis molecular label of renal clear cell carcinoma were constructed. Finally, the molecular label combined with clinical data was used to construct the prognosis nomogram of renal clear cell carcinoma.Results:Firstly, 798 kinds of the differential genes of methylation were screened from 485 patients with renal clear cell carcinoma. The difference genes combined with survival data of renal clear cell carcinoma was analyzed by single factor Cox analysis to obtain 492 genes related to prognosis of renal clear cell carcinoma ( P<0.05). Then, 325 tumor tissue samples were randomly divided into two groups according to 30% and 70%，and 70% group was used to screen the molecular labels of prognosis, 30% group was used to test the molecular label of prognosis. Five genes related to the prognosis of renal clear cell carcinoma were screened by LASSO-Cox regression analysis, which were CMTM3，HAND2，HIST1H3E，MIR4425，SHOX2. A molecular label for the prognosis of renal clear cell carcinoma was constructed as follow: -5.36342273779505×DML CMTM3+3.260844271×DML HAND2+4.995985347×DML HIST1H3E-1.84339230384795×DML MIR4425+3.323857473×DML SHOX2 after substituting relevant data, the molecular label score was calculated. The molecular label value was significantly correlated with the survival of ccRCC patients in the training group, test group and all data group, and the higher the molecular label value, the worse the prognosis of patients. Conclusions:We found 5 genes that affect the prognosis of renal clear cell carcinoma and constructed molecular label to predict the prognosis of patients with renal clear cell carcinoma. There was a significant correlation between the molecular label and the prognosis of patients with renal clear cell carcinoma.The prognosis model of renal clear cell carcinoma may be used in clinic in the future.