High grade serous ovarian cancer (HGSOC) is the deadliest gynecological malignancy. Despite successful first line therapies, most patients relapse and develop more chemoresistant disease. This recurrence and development of chemoresistance is attributed to a rare population of tumor cells, termed cancer stem cells (CSCs), which are more chemoresistant, have the capacity to self-renew, and can repopulate the entire tumor. Research has shown that CSCs are maintained by the non-cancer cells in the tumor microenvironment (TME), such as mesenchymal stem cells (MSCs), endothelial cells (ECs), and immune cells. Furthermore, the role of non-cancer cells in clinical outcomes and chemoresistance has been highlighted by recent evidence showing that classification of HGSOC molecular subtypes, which have variable clinical prognoses, are influenced by the presence of non-cancer cells in the tumor. However, it is currently unclear exactly how CSCs and the nuanced cell composition of the TME work together to promote chemoresistance. Current models used to study these phenomena either suffer from a lack of cellular complexity in the case of many in vitro models or impractical experimental constraints such as long latency periods and poor control over cell composition in patient-derived xenografts. To better understand the role of CSCs and the TME cells in chemoresistance, practical in vitro model systems that more closely represent in vivo processes and microenvironments are needed. We hypothesize that the development of these in vitro model systems will contribute novel insights into TME-mediated CSC regulation and the development of chemoresistance in HGSOC.

In aim 1 we examined the emergence of chemoresistance in the context of CSCs by developing a 3D in vitro serial passaging model system that allows for long term culture of patient-derived tumor cells with periodic evaluation of stemness and chemoresistance. Using this model system, we demonstrated increased proliferation, expression of CSC markers, tumorigenicity, and chemoresistance over the course of long-term passaging, reflective of emerging chemoresistance in vivo. Furthermore, this system enabled us to define a malignant gene expression signature that is associated with chemoresistance, tumorigenicity, and stemness and to evaluate patient-specific chemoresistance development following treatment. Finally, we demonstrated the translational value of this model system by showing that Metformin treatment can hinder CSC driven development of chemoresistance in a phase II clinical trial.

In aim 2 we developed a heterogeneous tumoroid culture system that enabled culture of patient-derived tumor cells with controlled ratios of MSCs, ECs, and immune cells to study TME-mediated maintenance of CSCs and chemoresistance. Using this model, we found that changes in CD133+/-ALDH+/- CSC phenotypes in response to TME cells varied depending on the patient sample. We also observed increased tumorigenicity and chemoresistance in tumoroids compared to spheroids cultured with patient-derived tumor cells alone. Furthermore, we found evidence of epithelial-to-mesenchymal transition (EMT) in tumoroids accompanied by altered CSC phenotypes and a malignant matrisome signature. All of this together supports idea that the non-cancer cells in the TME contribute to the development of advanced, chemoresistant disease and implicates EMT, changes in CSC phenotypes, and matrix remodeling as the primary culprits.

Finally, in aim 3, we utilized this tumoroid system to generate tumoroids with 23 different cell compositions to evaluate the role of TME cell composition in response to therapy. Drug assays with novel and traditional chemotherapies revealed that tumoroids with different compositions respond differently to therapy and that the number of monocytes included in the culture was associated with the greatest resistance to therapy. Furthermore, our random forest models trained on the drug responses of each cell composition were able to predict drug response with moderate success. With these models we identified that nuanced differences in cell composition can influence drug response and that the strongest predictor of response to therapy was the total quantity of non-cancer cells. Overall, this model demonstrates the potential of using the TME composition to predict patient drug response and direct clinical management.

In these aims we demonstrate the clear utility of complex and realistic, yet practical in vitro model systems in the study of chemoresistance and CSC maintenance in ovarian cancer. Specifically, we identified the link between CSCs and the development of chemoresistance in long term 3D in vitro serial passage culture. Furthermore, we showed that the non-cancer cells in the TME can confer chemoresistance and promote EMT associated with altered CSC phenotypes and matrix remodeling. Lastly, we demonstrated the potential of TME composition in predicting drug response. Overall, the model systems presented in this study provide platforms that can be used to better understand the role of CSCs and the TME in chemoresistance and poor clinical outcomes. This could ultimately lead to the development of novel therapies, enhanced clinical management, and improved clinical outcomes.

Date: Wednesday, July 21, 2021
Time: 9:00 AM
Zoom: https://umich.zoom.us/j/96111622879
Password: 326862
Chair: Dr. Geeta Mehta