Precision medicine for glioblastoma patients: a patient-based pre-clinical platform for immuno-oncology

Glioblastoma (GBM) remains one of the most aggressive cancers, with patients facing a grim prognosis and minimal therapeutic progress over the past two decades that inevitably always leads to recurrence. While immune-based strategies have shown promise in other cancers, they have largely failed in GBM clinical trials due to the tumor's profoundly immunosuppressive and “cold” nature. Moreover, the inadequacy of current preclinical models significantly hinders the development of novel immunotherapies for GBM. Immunocompetent mouse models fail to replicate the unique brain-specific tumor microenvironment (TME), often producing misleading preclinical results. Meanwhile, classical patient-derived models lack essential immune components, making them unsuitable for studying immunotherapeutic responses. This thesis aims to address these limitations by examining immune-tumor interactions within traditional GBM patient-derived orthotopic xenografts (PDOXs) and advancing these models through the development of humanized mice, creating a more representative platform for immunotherapy research. In immunocompromised GBM PDOXs, tumor-associated macrophages (TAMs), predominantly of microglial origin, exhibited GBM-specific adaptations, including antigen presentation and dendritic-like features. These signatures were amplified under therapeutic pressure from temozolomide, the standard-of-care chemotherapeutic for GBM. To further assess the translational potential of PDOXs, a clinically relevant surgical resection protocol was developed, successfully modeling tumor recurrence alongside TME adaptation. Building on these findings, GBM PDOXs were generated in humanized mice to introduce an adaptive human immune system. Human CD34+ hematopoietic stem cell-engrafted (HU-CD34+) and human peripheral blood mononuclear cell-engrafted (HU-PBMC) mice demonstrated to be a successful host for GBM PDOXs. Human immune cells identified within GBM tumors were predominantly organized in clusters, featuring myeloid CD68+ CD11c+ populations and lymphoid CD4+ profiles. Among CD4+ T cells, we observed distinct phenotypes associated to exhausted and memory-like characteristics, as well as regulatory T cells, reflecting features commonly observed in human GBM. The murine-derived innate immunity was dominated by immunosuppressive TAMs, as in previous classical non-humanized GBM PDOXs. These humanized GBM PDOXs offer a clinically relevant platform for preclinical evaluation of immunotherapeutics by integrating a reconstituted human adaptive immune system while preserving patient tumor characteristics. Overall, this work highlights how GBM PDOX models recapitulate the immunosuppressive milieu observed in patients by facilitating a functional crosstalk between human GBM cells, human immune cells, and the murine brain TME. By addressing the translational gap between preclinical models and patient outcomes, these models provide a robust foundation for advancing precision immuno-oncology strategies, ultimately hoping to improve the success rates of GBM clinical trials.