1.Arion Cancer Center,Beijing,China;2.920th Hospital of Joint Logistics Support Force;3.The 3rd Xiang-Ya Hospital, Central South Universi;4.Medical School of Chinese PLA;5.Chinese PLA General Hospital
目的 构建细胞外基质（Extracellular matrix，ECM）相关基因预后模型，评价其预测胶质瘤患者预后的能力，探索基于该模型的胶质瘤免疫微环境特征。 方法 基于肿瘤基因组图谱（The cancer genome atlas，TCGA）和基因型组织表达（Genotype-tissue expression，GTEx）数据库中胶质瘤以及正常脑组织数据，筛选获得差异基因（Differential expressed genes，DEGs）。基于基因本体论（Gene ontology，GO）数据库获取ECM相关基因，基于单因素COX回归分析获取胶质瘤预后基因。将上述三部分取交集获得重叠候选基因，再经由Lasso分析获取最佳的4基因预后模型，并于TCGA以及中国脑胶质瘤图谱数据库（Chinese glioma genome atlas，CGGA）中胶质瘤队列中进行生存分析和Cox回归分析。基于4基因预后模型以及TCGA患者预后数据构建预后列线图，并在CGGA胶质瘤队列中进行验证。最后，基于富集分析、免疫检查点分析以及免疫浸润分析探索4基因预后模型相关的免疫微环境特征。 结果 22个重叠候选基因经由lasso分析后获得最佳4基因预后模型，该模型的风险评分能够较好的预测TCGA以及CGGA胶质瘤患者的预后，并且是患者预后不佳的独立危险因素。细胞系验证实验中提示U251细胞系（胶质瘤细胞）最佳4基因表达均高于HMC3细胞系（小胶质细胞），符合TCGA以及CGGA数据库分析结果。于4基因预后模型构建的预后列线图同样具有较好的预测患者预后的能力。高风险组患者肿瘤组织内具有较高水平的M2型巨噬细胞浸润且免疫检查点相关分子（PD-L1，B7-H3，CTLA4，PD1，TIM3以及LAG3）显著高于低风险组。 结论 ECM相关基因模型以及预后列线图均能够较好的预测胶质瘤患者的预后，高风险组患者具有抑制性免疫微环境特征，免疫检查点抑制剂可能是该类患者的潜在治疗方式。
Aim An extracellular matrix (ECM) - related gene prognostic model was constructed to evaluate its ability to predict the prognosis of glioma patients and explore the glioma immune microenvironment characteristics based on this model. Methods Based on glioma as well as normal brain tissue data from the tumor Genome Atlas (TCGA) and genotype tissue expression (GTEX) databases, differential expressed genes (DEGs) were filtered. ECM-related genes were obtained based on the Gene Ontology（GO） database, and glioma prognostic genes were obtained based on cox proportional-hazards model. Overlapping candidate genes were obtained by intersection of the above three parts, and then the best 4-gene prognostic model was obtained by lasso analysis, followed by survival analysis and Cox regression analysis in the glioma cohort in TCGA as well as the Chinese glioma Genome Atlas (CGGA) database. A prognostic nomogram was constructed based on the 4-gene prognostic model as well as TCGA patient prognostic data and validated in the CGGA glioma cohort. Finally, the immune microenvironment features associated with the 4-gene prognostic model were explored based on enrichment analysis, immune checkpoint analysis as well as immune infiltration analysis. Results The 22 overlapping candidate genes were subjected to lasso analysis to obtain the best 4-gene prognostic model, and the risk score of this model was able to better predict the prognosis of TCGA as well as CGGA glioma patients, and was an independent risk factor for poor patient outcome. Cell line validation experiments suggested that the U251 cell line (glioma cells) had a higher expression of all optimal 4 genes than the HMC3 cell line (microglia), which is consistent with the results of the TCGA as well as the CGGA database analysis. The prognostic Nomogram constructed based on the 4-gene prognostic model similarly had better ability to predict patient prognosis. Patients in the high-risk group had higher levels of M2 macrophage infiltration and significantly higher levels of immune checkpoint related molecules (PD-L1, B7-H3, CTLA4, PD1, tim3 as well as Lag3) in tumor tissues than those in the low-risk group. Conclusion Both the ECM related gene model as well as the prognostic Nomogram were able to better predict the prognosis of glioma patients, and patients in the high-risk group had a suppressive immune microenvironment characteristic, and immune checkpoint inhibitors may be a potential treatment modality for such patients.