Work

165 Engineering novel synNotch CAR T cells to precisely and effectively target medulloblastoma

Year: 2026

Type: conference-abstract

Abstract: Background Chimeric antigen receptor (CAR) T cells have shown promise for treating pediatric brain tumors; however, antigen heterogeneity, on-target, off-tumor toxicity, and an immunosuppressive microenvironment limit their efficacy. An emerging strategy to improve CAR T-cell therapy is the use of synthetic Notch (synNotch) receptors, which recognize a tumor- or tissue-specific antigen and locally induce CAR expression to kill tumor cells. Such systems have previously been shown to improve CAR T cell efficacy and mitigate toxicities in preclinical models of adult glioblastoma and pediatric diffuse midline glioma. Here, we investigated the anti-tumor efficacy of synNotch CAR-T cells in patient-derived models of Group 3 medulloblastoma, the deadliest medulloblastoma subtype.Methods We utilized a synNotch receptor specific to brevican, a proteoglycan expressed primarily in the central nervous system, for brain-specific priming. For tumor-specific priming, we developed a synNotch receptor that recognizes B7H3, an immune checkpoint protein commonly overexpressed in medulloblastoma. synNotch-CAR T cells were evaluated in vitro by co-culturing with mCherry+ D283 tumor cells and measuring fluorescent signal over time using IncuCyte live imaging. in vivo efficacy was assessed in a xenograft mouse model using luciferase-expressing D283 tumor cells. Following tumor engraftment, 1 × 106 synNotch-CAR T cells were infused intracerebroventricularly (ICV), and tumor growth was measured weekly using bioluminescent imaging.Results Despite effective cell-killing in vitro, BCAN→GD2 (α-BCAN synNotch-α-GD2) CAR T cells failed to clear tumors in vivo and induced off-tumor neurotoxicity following ICV delivery. In contrast, tumor-specific priming via B7H3 improved the in vivo efficacy of the α-GD2 CAR without neurotoxicity following local delivery. B7H3→GD2 CAR T cells homed more efficiently to the tumors than BCAN→GD2 CAR T cells and effectively controlled tumor growth, and improved survival in tumor-bearing mice.Conclusions Although α-BCAN synNotch-CAR T cells have shown significant preclinical efficacy in aggressive adult and pediatric glioma models, our findings indicate that the unique characteristics of medulloblastoma may require a more tumor-specific priming approach. Consistent with our previous findings, the toxicity observed from the BCAN→GD2 circuit suggests that broadly expressed priming antigens can activate CAR expression beyond the tumor, potentially leading to toxicity when the CAR target is also expressed at low levels on normal cells. Future studies are aimed at characterizing this neurotoxicity and investigating why BCAN→GD2 CAR T cells fail to effectively target medulloblastoma.Ethics Approval The present study followed institutional guidelines for humane animal treatment and complied with relevant legislation from the UCSF Institutional Animal Care & Use Committee.

Source: Journal for ImmunoTherapy of Cancer

DOI: 10.1136/jitc-2026-sitc2026.0165