胶质母细胞瘤 CAR-T 治疗中的内源性免疫招募：细胞因子、髓系与趋化因子环路

Glioblastoma (GBM) has remained relatively unresponsive to immunotherapy, with scattered durable responses reported in early CAR T-cell studies, but without clear benefit at the population level. The major challenge for GBM has been its heterogeneous nature with a significantly immunosuppressive microenvironment that is predominantly composed of myeloid cells, inhibiting T-cell infiltration, function, and providing a rapid pathway for adaptive resistance. The focus of this review is to reposition GBM CAR T-cell therapy as a systems-level issue, turning localized CAR T-cell cytotoxicity into sustained control of the disease by engaging endogenous antitumor immunity via cytokine myeloid chemokine networks.

We integrated both mechanism- and translation-oriented evidence for how inflammatory mediators derived from CAR T cells (Type I IFNs, IFN- , TNF) may license microglia/tumor-associated macrophages for antigen presentation and chemokine secretion, thus recruiting host effector cells and promoting antigen epitope spreading. To place this work within the context of current engineering trends, the current paper undertook a structured meta-synthesis on registry trials for interventional CAR T therapy for GBM using ClinicalTrials.gov. Using a structured advanced search strategy, we searched 91 registry records, found 44 trials for interventional CAR T therapy, and evaluated 23 active trials commenced after January 2020. Trials were classified based on target antigen choice, multi-antigen OR-gated approaches, conditional AND-gated synNotch logic, as well as safety and controllability measures (inducible off-switches).

The effectiveness of CAR T cells for GBM is not likely to be actualized by targeting alone and needs to incorporate both killing and productive self-reinforcing endogenous immunity via myeloid licensing and chemokine amplification. Current trials are increasingly integrating this paradigm with a focus on more comprehensive antigens, gated CARs, immune-conjugate payloads, and safety designs amenable to the CNS without major toxicity such as ICANS. Future translation will require a focus on implementing endogenous immune activation as a quantified endpoint (including cytokine and chemokine analysis within CSF) and a simultaneous focus on immune set points that maintain cross-priming and memory without unmasking neuroinflammatory toxicity.