基于DNA甲基化的分子亚型预测头颈鳞状细胞癌患者预后分析

Study of molecular subtypes based on DNA methylation on prediction of the prognosis of patients with head and neck squamous cell carcinoma

目的探讨基于DNA甲基化状态的亚型在177例头颈鳞状细胞癌(head and neck squamous cell carcinoma,HNSCC)样本中的预后价值。方法从TCGA数据库下载HNSCC的甲基化芯片数据、临床数据及转录组数据,通过Cox比例风险回归模型筛选出与预后显著相关的甲基化位点,然后利用基因组注释将这些甲基化位点注释为其相对应的基因,并进行基因本体(gene ontology,GO)功能注释和京都基因与基因组百科全书(kyoto encyclopedia of genes and genomes,KEGG)通路分析,同时基于一致性聚类,用上述甲基化位点将HNSCC样本分成若干亚组,并计算得出每个亚组相比其他亚组的差异甲基化位点,最后选择一个最合适的亚组来构建Cox比例风险预测模型。结果 共筛选出613个预后显著相关的甲基化位点,GO分析显著基因主要富集于钾离子跨膜转运活性和减数分裂细胞周期的正调控等生物过程,KEGG通路分析显著富集在细胞因子受体相互作用和Rap1信号通路等信号通路,613个甲基化位点通过一致性聚类确定了7个亚组,最后基于亚组3的6个特异甲基化位点构建了一个最优的Cox比例风险预测模型。结论基于DNA甲基化的分子分型构建的由6个特异甲基化位点组成的模型可以预测HNSCC患者的生存结局,并有助于推进HNSCC患者的精准医疗。

OBJECTIVE To investigate the prognostic value of subtypes based on DNA methylation status in patients with head and neck squamous cell carcinoma(HNSCC).METHODS The methylation chip data,clinical data and RNA-sequencing data were downloaded from the TCGA database.According to T,N,and M category,age,stage,grade,prognosis and methylation level,Cox proportional risk regression model was used to screen methylation sites significantly correlated with prognosis.Then genome annotation was performed on these methylation sites to identify the corresponding genes and GO functional annotation and KEGG pathway analysis were performed.At the same time,HNSCC samples were divided into several subgroups by these methylation sites based on consensus clustering,and the difference of methylation sites between each subgroup and other subgroups was calculated.Finally,the most suitable subgroup was selected to construct a Cox Proportional Hazard Model.RESULTS A total of 613 methylation sites that significantly influenced survival were identified.The genes were significantly enriched in GO terms of potassium ion transmembrane transporter activity and positive regulation of meiotic cell cycle.The pathways were enriched in Cytokine-cytokine receptor interaction and Rap1 signaling pathway.Seven subgroups were identified based on consensus clustering using 613 methylation sites.Finally,an optimal Cox Proportional Hazard Model was constructed based on 6 methylation sites in subgroup 3.CONCLUSION Based on the molecular subtypes of DNA methylation,a model consisting of 6 specific methylation sites may help to predict survival outcomes of patients with HNSCC and help promote the accurate medicine of HNSCC patients.