Abstract Chimeric antigen receptor (CAR) T cell therapies have transformed outcomes for several hematological malignancies, and this has led to recent efforts developing CAR T cell therapies for children with brain tumors. Early-phase clinical trials in pediatric brain tumors have demonstrated feasibility and safety, and indicators of response have been observed. However, few objective responses have occurred, and we hypothesize that poor inherent T cell fitness may contribute to minimal response in these patients. Herein, we investigated the functionality of B7-H3 CAR T cells generated from 9 pediatric brain tumor patients on our ongoing Loc3CAR clinical trial (NCT05835687). We evaluated brain tumor patient CAR T cells in vitro and in vivo and compared them to CAR T cells generated from 20 healthy donors. Despite high in vitro expansion, persistence, and killing that was indistinguishable from healthy donor CAR T cells, brain tumor patient CAR T cells were ineffective at controlling tumors in vivo. Phenotypic evaluation revealed that brain tumor patient CAR T cells had a low proportion of CD4+ T cells, secreted low overall levels of cytokines upon antigen stimulation, and exhibited high exhaustion after interacting with tumor cells. Interestingly, these phenotypes were shared with healthy donor CAR T cells that were ineffective in DIPG and medulloblastoma brain tumor models. In concordance with this, single cell RNA sequencing identified that brain tumor patient CAR T cells were more transcriptionally similar to ineffective healthy donor CAR T cells than effective healthy donor CAR T cells. These results demonstrate that CAR T cells derived from pediatric brain tumor patients have inherently limited fitness, leading to poor tumor control in vivo and possibly contributing to the lack of clinical responses. Future work will focus on identifying targetable mechanisms to improve the fitness of brain tumor patient CAR T cells to enhance therapeutic efficacy.
Meehl et al. (2026) studied this question.