Computational modelling of perivascular-niche dynamics for the optimization of treatment schedules for glioblastoma

Glioma stem-like cells dynamically transition between a chemoradiation-resistant state (preferentially in the perivascular niche) and a chemoradiation-sensitive state (away from blood vessels). However, physical barriers in the tumour microenvironment restrict the delivery of chemotherapy to tumour compartments distant from blood vessels. Here we show that a computational stochastic model of the spatiotemporal dynamics of the perivascular niche that incorporates glioma stem-like cells, differentiated tumour cells, endothelial cells and stromal cells as well as relevant subcellular, cellular and tissue-level phenomena can be used to optimize the treatment schedules of chemoradiation and of standard radiation fractionation with the administration of temozolomide. In mice with platelet-derived-growth-factor-driven glioblastoma, the model-optimized treatment schedule increased the survival of the animals. For standard radiation fractionation in patients, the model predicts that chemotherapy may be optimally administered about an hour before the application of radiation. Massively parallel simulations of the spatiotemporal dynamics of the tumour microenvironment might predict tumour responses to a broader range of treatments and be used to optimize treatment regimens.