Engineering personalized micro-tumor ecosystems

摘要 癌症是美国的主要死亡原因之一，占每4例死亡中的近1例。 然而，尽管最近发展了亚型特异性个性化治疗， 尽管在分子和遗传分析领域，许多癌症治疗仍然具有低功效， 因为预测患者肿瘤对治疗剂的反应的能力有限。当前的主要原因 治疗往往不能转化为一个成功的临床结果是因为复杂的肿瘤 微环境和异质性限制了生物标志物指导策略的预测能力， 化疗因此，个性化三维（3D）肿瘤生态系统的成功工程 强烈希望能够在体外重现肿瘤微环境和异质性， 预测患者对抗癌药物的反应，从而进一步改善患者的预后。在此，我们建议 开发一个个性化的乳腺癌生态系统芯片平台，用于癌症的个性化筛查 通过使用患者来源的肿瘤外植体，定义的肿瘤等级匹配， 生物材料基质和自体患者血清以模拟患者特异性肿瘤标志。拟议 芯片上的癌症生态系统也将在生理流体动力学下受到严格调节。本课题 我假设1）使用肿瘤外植体将包括肿瘤的关键成分 患者的异质性，2）定义的肿瘤等级匹配基质，自体患者 血清，并且微流体生物反应器将防止肿瘤外植体的表型改变，以及3） 机器学习算法与癌症生态系统芯片平台的集成将提供更多 根据从以下方面收集的数据，准确、无偏地预测患者对化疗药物的反应： 工程肿瘤模型。我们的初步结果表明，肿瘤外植体培养和 肿瘤来源的基质成分预测了100%灵敏度的治疗反应。我们的初步 结果显示，在包括乳腺癌、结肠直肠癌和头颈癌在内的一系列癌症中， 和颈部鳞状细胞癌，因此，研究结果可以有广泛的应用，并可以成为一个 癌症管理的模式转变。

Abstract Cancer is one of the leading causes of death in the United States, accounting for near 1 in every 4 deaths. However, despite the recent development of subtype-specific personalized therapy based on achievements in the fields of molecular and genetic profiling, many cancer treatments still have low efficacy which mostly arise from the limited ability to predict the patient tumor responses to therapeutic agents. The major reason that current therapeutics often cannot translate into a successful clinical outcomes is because of the complex tumor microenvironment and heterogeneity that limit the predictive power of the biomarker-guided strategies for chemotherapy. Therefore, the successful engineering of personalized three-dimensional (3D) tumor ecosystem that can recapitulate the tumor microenvironment and heterogeneity in vitro is strongly desired to accurately predict patients’ responses to anti-cancer drugs and thus further improve patient outcome. Here we propose to develop a personalized breast-cancer-ecosystem-on-a-chip platform for personalized screening of cancer chemotherapeutics with high accuracy by utilizing patient-derived tumor explant, defined tumor grade-matched biomaterial matrices and autologous patient serum to mimic patient-specific tumor hallmarks. The proposed cancer-ecosystem-on-a-chip will also be tightly regulated under physiological fluid dynamics. In this project, we have hypothesized that 1) the use of tumor explant will embrace the critical components of the tumor heterogeneity of the patient, 2) the combination of defined tumor grade-matched matrix, autologous patient serum, and a microfluidic bioreactor will prevent the phenotype alteration of the tumor explant, and 3) the integration of a machine-learning algorithm with the cancer-ecosystem-on-a-chip platform will provide more accurate, unbiased prediction of the patient responses to chemotherapeutics based on the data gathered from the engineered tumor model. Our preliminary results show that the combination of tumor explant culture and tumor-derived matrix constituents had predicted therapeutic responses with 100% sensitivity. Our preliminary results show high specificity throughout a range of cancers including breast cancer, colorectal cancer, and head and neck squamous cell carcinoma, and thus the findings can have broad applications, and can emerge as a paradigm shift in the management of cancer.