标记物网络识别方法的建立及在正常核型急性髓系白血病预后评估中的应用

Tumor molecular biomarkers are characteristic molecular signals along with tumor progression. Gene expression levels are important biomarkers for prognostic evaluation of malignant tumors. Only single or several genes are independently tested in nowadays clinical applications, which could difficultly reveal the heterogeneity of malignant tumors. Cytogenetically normal acute myeloid leukemia (CN-AML) has no cytogenetic abnormalities, but has strong heterogeneity. Known single-gene biomarkers can’t effectively guide the choice of treatment intensity and personalized therapy. This project tries to present some new explorations based on the fusion of gene expression and protein interacting data. Firstly, theories and algorithms are developed for constructing the disease-associated network and identifying biomarker network via the combination of expression patterns and network topological features. Secondly, via extracting quantitative parameters for a biomarker network, the practicability of using these biomarker networks to CN-AML risk stratification, prognostic evaluation is explored by studying the relevance between these quantitative parameters and clinical parameters. Focusing on CN-AML can reduce the possible perturbations of different molecular mechanisms for different tumor types, and increase the sensibility and specificity of biomarker signals. Results of this study can present novel potential prognostic biomarkers, and help to promote the understanding of leukemia pathogensis. The algorithms and methodology developed are universal and could be extended to prognostic evaluation of other tumors.

肿瘤分子标记物指的是伴随肿瘤的发生发展，机体所表现出的特征性分子信号。基因表达水平是评估恶性肿瘤预后的重要标记物。目前临床主要用单个或少数几个基因的独立表达评价预后，很难全面反映肿瘤的异质性。正常核型急性髓系白血病（CN-AML）不包含细胞遗传学变异，但却具有极强的异质性，已有的单基因标记物无法有效指导治疗强度及个体化治疗方案的选择。本项目拟融合全基因组表达水平和蛋白作用数据，建立构建疾病关联网络的理论及分析处理算法，综合表达模式与网络拓扑信息识别特异的标记物网络，并进一步提取可反映标记物网络表达特征的定量指标，探讨其应用于危险度分层、预后评估的可行性。聚焦CN-AML保证了研究对象的均一性，可提高标记物网络的一致性和特异性。研究成果能够为CN-AML诊疗提供新的备选标记物，有助于进一步探讨CN-AML发生发展的分子机理。所设计的方法体系具有通用性，可推广应用至其它恶性肿瘤的预后评估。

肿瘤分子标记物对于疾病精准诊断、治疗方案优选具有重要应用价值，高通量测序技术广泛应用产生的高密度多组学数据为设计标记物识别算法奠定了良好的数据基础。针对急性髓系白血病（Acute Myeloid Leukemia，AML）现有肿瘤标记物无法覆盖全部病人、难以有效指导治疗方案抉择等问题，本项目围绕多组学数据整合分析、标记物识别及AML精准诊疗应用等开展了大量工作，整理形成了AML专病多组学数据库、设计标记物识别算法并编程实现方法框架、识别出多种新型分子标记物、探索AML发病及进展机制、探索AML治疗方案等，发表SCI检索论文21篇，申请国家发明专利5项，获批解放军总医院科技进步一等奖1项（相当于军队科技进步二等奖）。研究成果对深入理解AML发病及进展机理、探索形成新的精准诊疗技术方案具有重要意义。