Thomas J. Lai

Glioblastoma (GBM) is the most common and deadly primary brain tumor in adults, characterized by diffuse infiltrative lesions, distinct oncogenic alterations, and a suppressive tumor immune microenvironment (TIME). These traits, however, are not fully recapitulated in current syngeneic murine glioma models, thereby limiting the pre-clinical development of novel therapies. Here we developed murine GBM models with patient-specific mutations that shared striking resemblance with human disease. Mosaic analysis with dual-recombinase mediated cassette exchange (MADR) and CRISPR-Cas9 were leveraged to generate autochthonous tumor models with single-copy somatic transgenesis of murine epidermal growth factor receptor variant III (mEGFRvIII) and loss-of-function mutations commonly observed in GBM. Characterization of a model lacking Pten and Cdkn2a (mEGFRvIII-PC) via histopathologic analyses, functional assays, and single-cell RNA sequencing showed these tumors developed as diffuse, infiltrative lesions with ill-defined margins and a necrotic core. The TIME was primarily comprised of microglia with minimal lymphocyte infiltration, recapitulating patient tumors. Furthermore, these tumors displayed intrinsic resistance to immune checkpoint blockade, potentially due to poor myeloid cell antigen presentation and T cell priming. Additionally, mEGFRvIII-PC cells were the most resistant to antigen-specific T cell-induced cytotoxicity and simultaneously promoted T cell exhaustion, while murine interferon-γ (mIFN-γ) stimulation induced a greater response in vitro with enhanced upregulation of PD-L1 relative to MHC-I. In sum, these findings demonstrate the improved fidelity of the mEGFRvIII-PC model to patient tumors and support its use in developing and refining novel therapeutic paradigms for GBM.

We conducted a randomized surgical window-of-opportunity trial ( NCT04606316 ) in recurrent, resectable glioblastoma. Between 2021 and 2024, 71 patients were screened, and 63 were randomized (intention-to-treat [ITT] population), and 58 received study treatment. Patients received pre-surgical immune checkpoint blockade (ICB) with dual anti-PD1 nivolumab + anti-CTLA4 ipilimumab (Arm 1), nivolumab alone (Arm 2), or placebo (Arm 3). Following surgery, Arms 1 and 3 received dual ICB, while Arm 2 continued nivolumab until progression or unacceptable toxicity. The primary endpoint, tumor-infiltrating lymphocyte (TIL) density, was met for Arm 1, as neoadjuvant dual ICB significantly increased TIL density compared with untreated control (Arm 3). As a secondary endpoint, median overall survival in the ITT population was 402 days (95% CI, 265–571) among patients who received dual ICB (Arms 1 and 3) and 273 days (95% CI, 166–506) for those assigned to nivolumab alone (Arm 2). No unanticipated toxicities were observed. Exploratory analyses showed that dual ICB elicited robust intratumoral and systemic immune activation, including increased interferon-related gene expression in blood. Higher TIL density and early systemic interferon-signature induction were associated with improved survival, whereas tumor mutational burden was not. Our results demonstrate pharmacodynamic activity of dual ICB in glioblastoma, with survival outcomes comparing favorably to similar studies.

Isoplexis fluorescence values per cytokine per cell

HERV raw counts and FPKM values