Dogma in clinical neuro-oncology holds that Gadolinium (Gd) contrast on magnetic resonance imaging (MRI) in tumor regions confirms that the blood-brain barrier (BBB) is locally compromised, and thus sufficient levels of drug are being distributed within these tumor regions. However, drug distribution data indicate the importance of the local microenvironmental heterogeneity and other physical factors that lead to differential distribution of therapeutic agents relative to Gd contrast. Non-invasively acquired imaging features can provide a snapshot of tumor microenvironment and ultimately a better understanding of drug distribution. The goal of this project is to develop and validate a “minimal” model that will capture intra- and inter-tumor heterogeneity to predict clinically relevant levels of drug distribution using routine imaging. In this project, we will use a combination of patient data, GBM patient-derived xenografts (PDXs), matrix-assisted laser desorption/ionization mass spectroscopy imaging (MALDI-MSI), and stimulated raman spectroscopy (SRS) to quantify the differences in drug distribution within and across tumors and, in doing so, develop a computational framework for predicting the efficacy of BBB-penetrant and BBB-impenetrant drugs for the treatment of GBMs.
Our hypothesis is that mathematical models based on multiparametric high content imaging techniques will predict spatially distinct drug distribution patterns in invasive primary and metastatic brain tumor models for both small molecule and macromolecular therapeutics, and therefore be pivotal to predicting the in vivo efficacy of targeted therapies. In the first phase of the project we will build a computational framework that quantitatively connects imaging features with differences in drug distribution within and across tumors. This part of the project involves experiments to quantify differences in drug distribution across tumors in a series of PDXs with MALDI mass spectrometry imaging (MSI), physical tissue features with stimulated Raman scattering (SRS), development/calibration of imaging-driven models for drug distribution incorporating BBB permeability, and extending our results to patients through a Phase 0 trial. In the second phase of the project we will construct a computational framework that quantitatively connects differences in drug distribution with imageable response within and across tumors. The second phase will include the investigatation of treatment response using BLI imaging, development/calibration of models of treatment response connecting drug distribution and tumor kinetics, and extension of our results to patients by determining sub-cohorts of patients most likely to respond to therapies. Overall, this project will provide a quantitative connection between imaging features and drug distribution at levels sufficient to predict heterogeneous treatment response across patients. The ultimate vision is to provide clinicians an accessible decision-making tool to help choose relevant targeted therapies that will be tailored for an individual GBM patient.