Iacobas Dumitru Andrei
Personalized Genomics Laboratory, CRI Center for Computational Systems Biology, Roy G Perry College of Engineering, Prairie View A&M University, Prairie View, TX 77446, United States.
World J Clin Oncol. 2020 Sep 24;11(9):679-704. doi: 10.5306/wjco.v11.i9.679.
Every day, investigators find a new link between a form of cancer and a particular alteration in the sequence or/and expression level of a key gene, awarding this gene the title of "biomarker". The clinician may choose from numerous available panels to assess the type of cancer based on the mutation or expression regulation ("transcriptomic signature") of "driver" genes. However, cancer is not a "one-gene show" and, together with the alleged biomarker, hundreds other genes are found as mutated or/and regulated in cancer samples. Regardless of the platform, a well-designed transcriptomic study produces three independent features for each gene: Average expression level, expression variability and coordination with expression of each other gene. While the average expression level is used in all studies to identify what genes were up-/down-regulated or turn on/off, the other two features are unfairly ignored. We use all three features to quantify the transcriptomic change during the progression of the disease and recovery in response to a treatment. Data from our published microarray experiments on cancer nodules and surrounding normal tissue from surgically removed tumors prove that the transcriptomic topologies are not only different in histopathologically distinct regions of a tumor but also dynamic and unique for each human being. We show also that the most influential genes in cancer nodules [the Gene Master Regulators (GMRs)] are significantly less influential in the normal tissue. As such, "smart" manipulation of the cancer GMRs expression may selectively kill cancer cells with little consequences on the normal ones. Therefore, we strongly recommend a really personalized approach of cancer medicine and present the experimental procedure and the mathematical algorithm to identify the most legitimate targets (GMRs) for gene therapy.
每天，研究人员都会发现一种癌症形式与关键基因序列或/和表达水平的特定改变之间的新联系，从而赋予该基因“生物标志物”的称号。临床医生可以从众多可用的检测组中进行选择，以根据“驱动”基因的突变或表达调控（“转录组特征”）来评估癌症类型。然而，癌症并非“单基因表现”，除了所谓的生物标志物外，在癌症样本中还发现数百个其他基因发生了发生突变或/和受到调控。无论使用何种平台，精心设计的转录组研究都会为每个基因产生三个独立的特征：平均表达水平、表达变异性以及与其他每个基因表达的协调性。虽然所有研究都使用平均表达水平来确定哪些基因上调/下调或开启/关闭，但其他两个特征却被不公平地忽视了。我们利用这三个特征来量化疾病进展过程中的转录组变化以及对治疗的反应恢复情况。我们发表的关于手术切除肿瘤的癌结节及其周围正常组织的微阵列实验数据证明，转录组拓扑结构不仅在肿瘤的组织病理学不同区域有所不同，而且对每个人来说都是动态且独特的。我们还表明，癌结节中最具影响力的基因[基因主调控因子（GMRs）]在正常组织中的影响力明显较小。因此，对癌症GMRs表达进行“智能”操控可能会选择性地杀死癌细胞，而对正常细胞影响甚微。所以，我们强烈推荐一种真正个性化的癌症治疗方法，并介绍识别基因治疗最合理靶点（GMRs）的实验程序和数学算法。
