TOPIC 438 - PREDICTION OF CANCER DRUG RESISTANCE TO AID IN CLINICAL DECISION MAKING

根据美国疾病控制与预防中心的数据，癌症是仅次于心脏病的第二大死因，每年约有60万人死亡。大约90%的癌症死亡归因于耐药性，使其成为一个主要的健康问题。癌症的内在和获得性耐药都归因于参与生长或凋亡的基因中存在遗传变异。然而，在患者肿瘤中发现的许多变异意义未知。提出的研究开发了一种计算方法，该方法利用机器学习应用于作为药物靶点的野生型和变异蛋白的分子动力学模拟，以预测耐药性及其严重程度。这些定量信息将被纳入描述癌症生长和细胞凋亡的蛋白质网络模型，以预测脱靶变异如何通过途径相互作用引起耐药性。该提案汇集了分子模拟，机器学习，途径建模，软件设计和开发以及系统生物学方面的专家合作团队，完成了这一范式转换工作。在这项工作的最后，一个原型将被开发出来，它可以帮助肿瘤学家和他们的团队了解并向患者传递有关患者肿瘤中可能存在的耐药性的信息，并做出临床治疗决策。

Cancer is the second leading cause of death behind heart disease with ~600,000 deaths annually according to the CDC. Approximately 90% of cancer deaths are attributed to drug resistance making it a major health problem. Both intrinsic and acquired drug resistance in cancers have been attributed to the presence of genetic variant in the genes involved in growth or apoptosis. However, many of the variants found in patients’ tumor are of unknown significance. The proposed research develops a computational method that leverages machine learning applied to molecular dynamics simulations of wild-type and variant proteins that are drug targets to predict drug resistance and its severity. This quantitative information will be incorporated into protein network models describing cancer growth and apoptosis to predict how off-target variants can cause drug resistance through pathway interactions. This proposal brings together a collaborating team of experts in molecular simulation, machine learning, pathway modeling, software design and development and systems biology accomplishing this paradigm shifting work. At the conclusion of the proposed work, a prototype will be developed that can help oncologists and their team to understand and deliver information to patients about possible drug resistance in the patient’s tumor and to make clinical treatment decisions.