Multi-scale models of ovarian cancer

In ovarian cancer, disease and treatment can be examined across multiple spatial scales including molecules, cells, intra-tumor vasculature, and body-scale dynamics of circulating drugs. Survival of primary tumor cells and their development into disseminated tumors is related to adhesion between the cells, attachment, and invasion. Growth of new tumors depends on the delivery of nutrients, which depends on the tumor diameter and the tumors vasculature. Drug delivery also depends on tumor diameter and vasculature, and molecular- and gross-scale drug processes. A cellular Potts simulation integrated data at these multiple scales to model microscopic residual disease during relapse after a primary surgery. The model generated new hypotheses about tumor cell behavior, and the effectiveness of drug delivery to tumors disseminated in the peritoneal cavity. First, the model required high intra-tumor adhesion in ovarian tumors, the existence of an unknown factor that drew tumor cells to vessels, a threshold of vascular endothelial growth factor (VEGF) for initiation of endothelial sprouting, and constitutive expression of angiogenic chemical messengers by tumor cells prior to needing oxygen. Alteration of the model incorporated drug delivery by the two standard routes, intraperitoneal and intravenous, from tumor vasculature parameterized from real patient data. Delivery of both small- and large-molecular weight therapies was superior during intraperitoneal therapy. Finally, empirical and theoretical distributions of vessel radii were considered. Samples from tumors with each type of vascular morphology were run as though too distant from the peritoneal cavity to receive peritoneal delivery, with three results: first, intravenous delivery was superior to the secondary delivery into the circulatory system from a primary intraperitoneal delivery. Second, small molecules penetrated homogeneously across all cells, regardless of vascular volume or morphology, while antibodies penetrated heterogeneously, particularly in low-vessel-volume samples. Third, when each of the whole tumors was considered, this heterogeneity resulted in a large sub-population of cells that accumulated non-therapeutic levels of antibody, even during the best delivery scenario (IV). Fourth, delivery of antibodies was poorest in the empirical distribution. Finally, hypotheses were generated about the impact of heterogeneity of drug delivery, to be addressed as future questions