Evaluating Feature Extraction in Ovarian Cancer Cell Line Co-Cultures Using Deep Neural Networks

Despite genome sequencing, molecular profiling and numerous new cancer drugs, most ovarian cancer (OvCa) patients are treated in the same way as 30 years ago. We do not understand cancer signaling dependences, cannot predict drug sensitivities nor create tailored drug combinations. Here, we apply ovarian cancer patient cell models and naturally occurring 3D spheroids (Oids) taken directly from patient ascites samples to determine the effects of >500 clinical and emerging oncology drugs and their combinations. We have established a translational flow of fresh (living) cancer samples from the clinic and a capability for high-throughput drug screening and multi-omics profiling. This research has already led to functional taxonomy of high-grade and low-grade OvCa and the discovery of critical pathways that are explored in more detail here. Our first aim is to identify synergies of Wee, SMAC and MEK signaling pathway inhibitors with other cancer drugs in the functional taxonomic subgroups of Ovca. Second, we will create a next-generation cancer drug testing technology using 3D-oids from ascites, where it will be possible to identify the impact of drugs on cancer-host/stroma interactions and therapeutic mechanisms involved. Finally, we will explore if further technology development could lead to the creation of a new diagnostic platform to better support the selection of drugs and drug combinations for each individual cancer patient.