Abstract 2441: Modeling glioblastoma in the mouse: Insights from noninvasive imaging

Abstract

Abstract What can noninvasive imaging reveal about mouse models of brain cancer and how closely they resemble their human counterparts? Glioblastoma multiforme (GBM) is the most aggressive form of brain cancer. Genetically engineered mouse models provide an appealing experimental framework in which to address the critical need for improved diagnostic and therapeutic strategies for GBM in a preclinical setting. An important step in the validation of such models is to determine how well they recapitulate salient features of human disease, most commonly from a molecular or genetic perspective. While molecular characterization is certainly imperative, we contend that it is also essential to validate mouse models of GBM from an in vivo imaging perspective. In vivo imaging can reveal not only tumor location and size, but also important biological and physiological processes including vascularity, blood brain barrier (BBB) permeability and necrosis. In addition, given the important role of magnetic resonance imaging (MRI) in the management of patients with GBM, there is a wealth of human imaging data that can be readily compared with analogous mouse data. Thus, our objective was to develop a clinically relevant framework for performing (i) image-based characterization of mouse models of GBM and (ii) comparisons with human disease. We have utilized novel mouse models wherein genetic alterations in the pRb, PTEN and Ras signaling pathways yield GBM. Clinically motivated 3T MRI acquisitions were performed including pre-contrast T1 and T2 datasets, dynamic contrast enhanced MRI (DCEMRI) and a post-contrast T1 acquisition. Image analysis was also guided by clinical relevance. T1 and T2 images were qualitatively analyzed using the VASARI lexicon developed for human GBM in the TCGA project. DCEMRI data was mathematically analyzed using a two compartment model yielding Ktrans parametric maps that are proportional to the BBB permeability. Human GBM MRI data and VASARI features from the TCGA study were downloaded from caIntegrator and compared with murine features using the χ2 test. We found that MRI has high sensitivity for GBM (97%) and accurately reflects histologic extent (≥ =0.88) in mice. We observed marked inter and intra lesion heterogeneity in BBB permeability, with some tumors showing over 10 times higher Ktrans in the rim compared to the center. Interestingly, many VASARI features exhibited similar distributions in both mouse and human GBM, however one feature was rarely found in mice (p< 0.001): a thick enhancing margin which is a salient feature of human GBM and reflects a central region of necrosis and a proliferative, vascular rim. In summary, we have developed a new standardized framework for performing clinically relevant MRI comparisons of mouse and human GBM. Remaining challenges include further investigation on the molecular mechanisms for the observed heterogeneous patterns of BBB permeability and the lack of central necrosis in mice. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 2441. doi:1538-7445.AM2012-2441