Developing Safe and Effective GD2-CAR T Cell Therapy for Diffuse Midline Gliomas

PROJECT SUMMARY Brain tumors are the leading cause of cancer related death in children; among these, diffuse intrinsic pontine glioma (DIPG) and other histone-3 K27M (H3K27M) mutated diffuse midline gliomas (DMGs) are the most aggressive and are universally fatal with current standard therapies. Despite several decades of investigational trials testing dozens of therapeutic approaches, median overall survival for DIPG is 11 months. Chimeric antigen receptor (CAR)-expressing T-cells have mediated impressive clinical activity in B-cell malignancies, and recent preclinical and early clinical results suggest benefit in CNS malignancies. We discovered homogenous, high overexpression of the GD2 ganglioside on H3K27M DMGs and demonstrated impressive antitumor effects in xenograft models of H3K27M-mutant DIPG following treatment with GD2-CAR T cells (GD2-CART, Mount, Nat Med 2018). Significant clinical experience with GD2 targeting CAR T cells, available mostly from studies in neuroblastoma, demonstrate safety and some early signals of antitumor activity. Safe and effective translation of these findings to children with DMGs would transform the landscape for this universally lethal pediatric brain tumor. This bench-to-bedside-to-bench project will conduct three aims in parallel leveraging a recently launched single institution Phase I trial of GD2.BB.z.iCasp9-CAR T cells administered intravenously following a lymphodepleting preparative regimen in children and young adults with H3K27M DMGs. The first aim focuses on safety, integrating insights gleaned in our preclinical models into trial design to diminish the risk of tumor inflammation associated neurotoxicity (TIAN), to establish best practices and to develop improved grading and treatment algorithms for this novel toxicity. The second aim focuses on efficacy, assessing clinical activity of GD2-CART in DMG and identifying biomarkers and clinical features associated with response. We further address the limitations of standard radiographic imaging in these infiltrative tumors using a novel machine learning aided MRI radiomics approach to quantify textural changes within the tumor and assess whether such changes correlate with clinical outcome, and we assess whether GD2-CART induced changes in CSF cell free DNA can provide a rapid quantitative assessment of antitumor response. Our third aim is a discovery aim, focused on improving understanding of the biology associated with myeloid cell activation following GD2-CART therapy for DMGs, which we observe in preclinical models and we observed in the first patient treated. Here we undertake comprehensive single cell profiling of CSF myeloid cells emerging post-GD2-CART in patients enrolled on the study and in preclinical models, and bedside-to-bench translation using murine models to test the hypotheses that GD2-CART induced CNS myeloid cell expansion/activation limit the efficacy of GD2-CART, are modulated by corticosteroid therapy and that this obstacle can be overcome by engineering CD47 overexpression in the GD2-CART.