Modeling of Drug Delivery in Brain Tumors and Validation with In-vivo Imaging

Patients diagnosed with glioblastoma have a grim prognosis with a median overall survival of less than 15 months. Vasculature within these tumors is typically abnormal, with increased tortuosity, dilation and disorganization. These tumors typically exhibit a disrupted blood brain barrier. Although it has been hypothesized that the “normalization” of the vasculature resulting from anti-angiogenic therapies could improve drug delivery, there is also evidence that suggests that the restoration of blood brain barrier integrity might limit the delivery of therapeutic agents. We have developed models of blood flow, permeability and diffusion through the tumor microenvironment to investigate the effect of these competing factors on drug delivery. Preliminary results from the modeling indicate that delivery of the agent to the tumor bed is affected by flow, permeability and diffusion in the interstitial space. Reductions in the permeability of the vasculature, as would be expected from restoration of the blood brain barrier in response to anti-angiogenic therapy, appeared to reduce the delivery of the drug to the tumor while increased flow could increase drug delivery.
Simultaneous MR/PET imaging allows us to explore the validity of these models through the use of specialized MR sequences that provide in-vivo measures of permeability, perfusion and diffusion while PET studies using radiolabeled drugs provide information about the spatial distribution of the drug into the delivery of therapeutic agents during concomitant therapy with anti-angiogenic agents. We performed MR-PET scans in patients receiving bevacizumab and radiolabeled [11C] temozolomide, pre and post bevacizumab therapy. In addition to standard anatomical MRI, we also performed perfusion/permeability imaging. Parametric maps were generated for cerebral blood flow, volume and permeability as well as SUV for uptake of the radiolabeled temozolomide and quantified within regions of interest identified by the clinician. Results from our preliminary imaging data were concordant with the modeling.