TMOD-35. Introducing MADR: A next-generation EGFR-driven glioblastoma model that recapitulates human immunotherapy resistance

Early pre-clinical studies suggested that immune checkpoint blockade (ICB) would be a viable treatment for glioblastoma (GB). However, ICB and other immunotherapies have largely failed to provide meaningful survival benefits in patients. This stark disconnect underscores a desperate need for improved models that better recapitulate human disease. Here, we characterize and compare a novel syngeneic and orthotopic pre-clinical GB model, MADR-mEGFRvIII, with the existing CT-2A, GL261, and SB28 models. We demonstrate that this new model closely resembles the histopathologic, molecular, immunologic, and functional features of primary human GB. Unlike existing models, MADR-mEGFRvIII was developed through targeted deletions in Pten and Cdkn2a and ectopic expression of a murine form of the oncogenic EgfrvIII. We demonstrate that mice bearing MADR-mEGFRvIII tumors fail to respond to dual anti-PD-1 and anti-CTLA-4 ICB, with treated mice showing a similar 30-day median overall survival time as the IgG control group. Using scRNA-sequencing and flow cytometry, we compare the composition of the tumor immune microenvironment (TIME) across all models. We reveal that MADR-mEGFRvIII tumors have significantly fewer infiltrating lymphocytes. Notably, MADR-mEGFRvIII tumors were highly enriched in microglia, as opposed to bone marrow-derived macrophages seen in other models, accounting for over 70% of all immune cells. Leveraging the hgp100-antigen specific Pmel-1 T cell model system, we show that MADR-mEGFRvIII cells more readily avoid mounting T cell pressure in vitro despite high levels of secreted IFN-γ, resulting in less cytotoxicity compared to other models. Therefore, we interrogated tumor-intrinsic responses to IFN-γ treatment, which identified a comparatively stronger transcriptional response in MADR-mEGFRvIII cells. Overall, the MADR-mEGFRvIII model shares many characteristics of primary human GB, suggesting this model is highly suitable for pre-clinical GB research. We hope to leverage this model to gain meaningful insight into the immunological determinants that govern immunotherapy resistance and help guide new therapeutic strategies to treat GB.