Predictive computer simulations of tumor drug response demonstrate that 3-D hypoxic gradients significantly increase drug resistance

Abstract

2071 Background: We created a three-dimensional physiologically based computer (in-silico) model of cancer based on a description of biological events at the cellular scale with input variables determined from patient specific information, such as in-vitro drug response experiments and in-vivo tumor imaging, with the long term goal of individualized treatment selection. The central hypothesis is that such a model that incorporates basic tumor growth kinetics information is capable of representing and predicting tumor response to chemotherapy. Methods: We measured in-vitro tumor growth and drug response for Doxorubicin sensitive and resistant MCF-7 breast cancer cells through trypan blue exclusion counts, tridiated thymidine incorporation, and the XTT assay. We used these results of parameter-based statistics to define input variables to our in-silico model of cancer, and ran computer simulations to measure the drug response predicted by the model. Results: The computer model could accurately predict the in-vitro response of drug sensitive and resistant MCF-7 breast cancer cells. The model also predicted that gradients of oxygen and nutrient in a tumor microenvironment, whether naturally occurring or induced by treatment, and which in previous work we found could increase the invasive capability of tumor cells and destabilize tumor morphology, could also contribute to acquired drug resistance by increasing the population of quiescent cells. Conclusions: We demonstrated that a rigorously, experimentally calibrated computer model of cancer is accurately predictive of in-vitro tumor response to chemotherapeutic drugs, and established that this model offers a means to quantitatively study tumor drug response. We did this through a grounds-up physical representation of tumor biology, not by fitting to experimental data. This validation begins the path to computational modeling and more efficient prediction of in-vivo tumor response to chemotherapy. No significant financial relationships to disclose.