Engineering Novel 3D Tumor-Stroma Models to Bridge the Gap Between Preclinical Drug Development and Human Clinical Outcomes

While much progress has been made in the war on cancer, shortcomings in the drug development process have kept anti-cancer clinical trial success rates low. One of the many factors implicated in this is the lack of pathophysiologically relevant and predictive preclinical models. Specifically, traditional preclinical tumor models do not capture tumor microenvironment complexity and heterogeneity, while advanced three-dimensional (3D) models suffer from poor reproducibility, lack of relevant and standardized extracellular matrix (ECM), and inability to interface with automated, high-throughput systems. Because of this, it has been suggested that developing novel phenotypic tumor models which balance the need for complexity and relevance with the ability to scale-up and translate, may help reduce the high attrition rates of clinical trials. Toward this end, this work describes the development and validation of a novel preclinical tumor model striving to achieve this balance. Model development was specifically focused on metastasis, as it remains the main cause of cancer deaths and has few good preclinical models. Since one major shortcoming of 3D in-vitro models is a lack of standardized, relevant ECM, initial work focused on defining the role of ECM composition and biophysical properties in guiding invasive phenotypes. Using a customizable and standardized oligomeric type I collagen, we demonstrated that 3D collagen fibril architecture and model geometry were key determinants of phenotypic trends and important design considerations for future model development. This work was followed by the design and validation of a custom fabrication platform to enable the rapid, reproducible embedment of tunable tumor-tissue spheres within a customizable 3D ECM. It was validated that this model can distinguish various metastatic phenotypes, is compatible with low-passage, patient-derived cells, and is able to interface with automated imaging systems. Overall, this work represents the first steps of design, verification, and validation of a novel 3D metastasis model which can serve as a relevant and predictive tool for high-throughput, high-content preclinical drug development