Modeling of the spatio-temporal evolution of tumor vasculature to improve predictions of breast cancer response to neoadjuvant chemotherapy regimens

One of the great challenges for treating cancer is the inability to design optimal therapeutic regimens for individual patients. Without a reasonable mathematical framework, selecting treatment regimens for the individual patient is fundamentally limited to trial and error. We have previously established a mechanically coupled, reaction-diffusion model at the tissue scale for predicting breast tumor response to therapy. The patient-specific, 3D model is initialized with tumor cell number estimated from quantitative, diffusion-weighted magnetic resonance imaging (DW-MRI) data. Additionally, the model includes a tumor cell reduction term due to drug delivery as estimated from dynamic contrast-enhanced (DCE-) MRI data (per individual clinical patient treatment schedules). We have expanded this model to differentiate between the effects of different chemotherapies to generate personalized and, potentially, optimized regimens for individual patients. This original model’s predictions have been found to be highly correlated to actual tumor response, but one limitation is that it does not account for the spatio-temporal changes of the tumor vasculature. Therefore, we now seek to extend this work by explicitly including the dynamics of an evolving vasculature to better simulate delivery of chemotherapies and account for the effect of these drugs on the vasculature itself. Importantly, by adding a second governing equation to the mathematical model representing the vasculature, we are able to reduce the parameter space of the model by coupling proliferation to the vasculature component—instead of defining proliferation as a local parameter in space. For an initial cohort of nine breast cancer patients, we evaluate the performance of the extended model by comparing its predictive ability to that of the original model (without vasculature). We report preliminary findings that the extended model’s results have lower median errors for its predictions. Future work will focus on expanding the model to account for targeted therapies and the simulation of alternative treatment regimens. We propose that an integrated mathematical-experimental approach leveraging patient-specific imaging data can provide optimal strategies for delivering therapy for breast cancer.