基于高通量数据的间变脑胶质瘤分子分型与生物标记物挖掘

发布时间：2018-08-26 14:39

【摘要】：第一部分:基于基因表达谱的间变脑胶质瘤分子分型背景:既往在研究全级别脑胶质瘤(WHO II-IV级)或原发性胶质母细胞瘤(WHO IV级)分子分型方面,全基因组表达谱分析显示出了良好的稳定性与有效性,可用于揭示脑胶质瘤不同的细胞起源、预后差异及治疗敏感性差异。但是在间变性脑胶质瘤(WHO III级)分子分型方面,该方法尚未得到有效运用。方法:为了获得间变性脑胶质瘤的客观、可行的分子分型方案,我们在公共数据库中下载了三组相互独立的间变脑胶质瘤基因表达谱。通过联合应用单因素cox回归、线性结合危险评分及受试者工作特征曲线(ROC曲线),我们构建了具有最佳预后区分效果的基因标签。为了进一步解释不同分型预后差异的分子机制,我们继续应用了基因本体分析(Gene Ontology,GO),基因组变异分析(GSVA)及基因组富集分析(GSEA),用以揭示该分型相关的生物学功能。结果:本研究共收集到各平台的间变脑胶质瘤基因表达谱数据444例(CGGA数据库34例,GSE16011数据库80例,REMBRANDT数据库67例,TCGA数据库263例),通过上述分型标志物筛选方案,我们最终获得了基于三个基因的间变性脑胶质瘤分子分型方案。该方案可将间变性脑胶质瘤分为高危组和低危组,两组患者预后具有明显差异。从预后及分子遗传学特征上看,低危组患者与WHO II级患者类似,高危组患者与WHO IV级患者类似。高危组患者,其肿瘤更具侵袭性,遗传背景更复杂。结论:基于三个基因的分型方案可很好的区分患者临床预后。该高、低危患者与WHO IV级、II级患者具有一定一致性,提示需要给予不同的治疗方案。第二部分:磷酸化组蛋白H3在脑胶质瘤中的生物学意义背景:世界卫生组织(WHO)依照形态学特征,将脑胶质瘤分为不同的病理级别。但是,即使处于同一级别的肿瘤,其恶性度仍有较大差异。这提示我们要寻找更客观有效的指标,准确的判断肿瘤的恶性程度及生物学特征。磷酸化组蛋白H3(p HH3)是在多种体部肿瘤中广泛报道的生物标志物,在肿瘤细胞分裂期具有特征性表达。方法:为研究p HH3在脑胶质瘤中的预后价值及生物学功能,我们首先通过免疫组化方式,研究了61例间变脑胶质瘤中p HH3的表达趋势及预后意义。继而我们通过325例中国人群脑胶质瘤m RNA测序数据、169例美国TCGA中胶质母细胞瘤样本m RNA测序数据、305例中国人群脑胶质瘤m RNA芯片数据及603例美国TCGA中胶质母细胞瘤m RNA芯片数据,研究了p HH3及其相关基因在脑胶质瘤中的预后价值及生物学意义。最后,我们结合多因素cox回归及基因组富集分析(GSEA)等方法,进一步确证了上述研究结果。结果:p HH3蛋白表达水平、m RNA表达水平及p HH3相关基因在多个数据集中都被证实可预测患者预后。其表达水平较高者,预后较差。另外,在GSEA结果中,我们发现p HH3不仅与细胞周期相关,也与肿瘤的上皮间质转化密切相关。结论:p HH3及其相关基因可将脑胶质瘤分为两种不同的亚型。两种亚型的肿瘤具有截然不同的临床预后与生物学特征。第三部分:肿瘤细胞纯度在脑胶质瘤中的生物学意义背景:脑胶质瘤组织中不仅包含肿瘤细胞,还包含间质细胞、免疫细胞等非肿瘤细胞。这些非肿瘤细胞稀释了肿瘤细胞的纯度,对于构成肿瘤生存的微环境具有重要作用。目前,肿瘤细胞纯度这一指标在脑胶质瘤中的生物学意义尚未阐明。方法:我们收集了5个临床研究队列、2249例样本中的脑胶质瘤细胞纯度。依照5个研究队列中研究方法的不同,将RNA测序来源的数据(CGGA、TCGA RNA测序数据)作为发现组,m RNA芯片来源的数据(CGGA、TCGA、REMBRANDT、GSE16011 m RNA芯片数据)作为验证组。另将CGGA来源的肿瘤样本通过HE及免疫组化方法验证。结果:我们发现肿瘤细胞纯度与脑胶质瘤主要的临床及分子病理事件均显著相关。低纯度的样本多为高级别肿瘤、生存期较短。将肿瘤纯度纳入到预后预测模型中,可显著提高预测的准确度。在纳入肿瘤纯度这一指标后,目前已知的多个预后因素其效力明显降低,进一步验证了肿瘤纯度的重要生物学意义。另外,通过联合分析样本相关基因组学数据,我们发现了肿瘤细胞纯度相关的特征性基因组学变化,并发现低纯度的脑胶质瘤,更多的体现出免疫相关表型的的增强,特别是巨噬细胞、小胶质细胞和中性粒细胞相关表达谱的表达增强。联合预后分析,我们发现巨噬细胞及中性粒细胞表达谱的高低与预后相关。结论:肿瘤细胞纯度及肿瘤微环境中的非肿瘤细胞的比例与临床、基因组学特征等具有显著相关性,这对于脑胶质瘤精确分型与临床诊疗具有重要意义。
[Abstract]:Part I: Molecular typing of anaplastic gliomas based on gene expression profiles: Previous genome-wide expression profiles have shown good stability and effectiveness in the study of molecular typing of full-grade gliomas (WHO II-IV) or primary glioblastomas (WHO IV), and can be used to reveal different cell origins of gliomas. Methods: To obtain an objective and feasible molecular typing scheme for anaplastic gliomas, we downloaded three sets of mutually independent metaplastic gliomas gene tables from a public database. By combining univariate Cox regression, linear combination of risk score and ROC curve, we constructed gene tags with the best prognostic discrimination effect. To further explain the molecular mechanism of prognostic differences among different subtypes, we continued to apply gene ontology analysis (GO), genome. Variation analysis (GSVA) and genome enrichment analysis (GSEA) were used to reveal the biological functions associated with the classification. Results: A total of 444 cases of anaplastic glioma gene expression profiles were collected from different platforms (34 cases in CGGA database, 80 cases in GSE16011 database, 67 cases in REMBRANDT database and 263 cases in TCGA database). Finally, we obtained a three-gene-based molecular typing scheme for anaplastic gliomas. The scheme divides anaplastic gliomas into high-risk group and low-risk group, and the prognosis of the two groups is significantly different. The high-risk group had more aggressive tumors and more complex genetic background. Conclusion: The three-gene typing scheme can well distinguish the clinical prognosis of patients. BACKGROUND OF BIOLOGICAL SIGNIFICANCE IN GLIOMA: The World Health Organization (WHO) classifies gliomas into different pathological grades according to their morphological characteristics. However, even in the same grade of tumors, their malignancy still varies considerably. This suggests that we should look for more objective and effective indicators to accurately judge the malignancy and biology of tumors. Phosphorylated histone H3 (p HH3) is a widely reported biomarker in various somatic tumors and has a characteristic expression in the stage of tumor cell division. Methods: To study the prognostic value and biological function of P HH3 in gliomas, we first studied the expression of P HH3 in 61 cases of anaplastic gliomas by immunohistochemistry. Then we studied P HH3 and its related genes in glia by using 325 Chinese glioma m RNA sequencing data, 169 US TCGA glioblastoma m RNA sequencing data, 305 Chinese glioma m RNA microarray data and 603 US TCGA glioblastoma m RNA microarray data. The prognostic value and biological significance of tumors were further confirmed by multivariate Cox regression and genome enrichment analysis (GSEA). Results: The expression of P HH3 protein, m RNA and P HH3-related genes could predict the prognosis of patients in multiple datasets. In addition, in the GSEA results, we found that P HH3 is not only associated with cell cycle, but also with epithelial-mesenchymal transformation of tumors. Conclusion: PHH3 and its related genes can be divided into two different subtypes of gliomas. The two subtypes of tumors have distinct clinical prognosis and biological characteristics. The biological significance of tumor cell purity in glioma: glioma tissue contains not only tumor cells, but also stromal cells, immune cells and other non-tumor cells. These non-tumor cells dilute the purity of tumor cells and play an important role in the microenvironment of tumor survival. Methods: We collected the purity of glioma cells from 5 clinical cohorts and 2 249 samples. According to the different research methods in the five cohorts, the data from RNA sequencing sources (CGGA, TCGA RNA sequencing data) were used as discovery group and the data from m RNA microarray sources (CGGA, TCGA, REMB). RANDT, GSE16011 m RNA chip data) were used as the validation group. In addition, CGGA-derived tumor samples were validated by HE and immunohistochemical methods. Results: We found that tumor cell purity was significantly correlated with major clinical and molecular pathological events of gliomas. After incorporating the tumor purity index, the effectiveness of many known prognostic factors has been significantly reduced, further validating the important biological significance of tumor purity. In addition, through joint analysis of sample-related genomic data, we found the purity phase of tumor cells. Associated with prognostic analysis, we found that the expression profiles of macrophages, microglia and neutrophils were correlated with the prognosis. Conclusion: The purity of tumor cells and the proportion of non-tumor cells in tumor microenvironment are significantly correlated with clinical and genomic characteristics, which is of great significance for accurate classification and clinical diagnosis of glioma.
【学位授予单位】：首都医科大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：R739.41

【相似文献】