CELL INTELLIGENCE · 肿瘤细胞治疗研究

Adoptive cellular therapies may expand treatment options for pediatric brain tumors by focusing activity on tumor antigens and limiting off-tumor effects. We systematically reviewed preclinical and clinical evidence for CAR T cells, TCR-engineered T cells, and NK or T-cell platforms directed against HER2, B7-H3 (CD276), EGFR806-reactive EGFR, GD2, IL13R 2, and EphA2 or EphA3, with attention to delivery route, safety, persistence, and combination strategies. Following PRISMA, we searched PubMed, Embase, and Scopus from inception through September 17, 2025, restricted to English. The search yielded 324 records; 103 duplicates were removed; 221 titles and abstracts were screened; 180 full texts were reviewed; and 34 studies were extracted by two independent reviewers. We captured design, tumor and molecular features, product engineering, route and schedule, lymphodepletion, toxicities including cytokine release syndrome, immune effector cell associated neurotoxicity, and tumor inflammation associated neurotoxicity, radiographic or clinical response, survival, and correlatives such as persistence or trafficking in blood, cerebrospinal fluid, or tumor tissue, cytokines, and antigen dynamics. In vivo studies showed reproducible antitumor activity for HER2 in medulloblastoma, GD2 in diffuse midline glioma, and multi-antigen constructs incorporating IL13R 2 and EphA2 in medulloblastoma and ependymoma, with significant survival advantages compared with controls. T cells targeting the EphA axis selectively killed medulloblastoma with neural sparing; GD2 CAR NK-92 inhibited diffuse intrinsic pontine glioma growth. In early clinical programs, route shaped safety and pharmacodynamics. For GD2, low-dose intravenous induction followed by repeated intraventricular dosing produced objective radiographic regressions and manageable tumor inflammation associated neurotoxicity, while dose-limiting cytokine release syndrome was confined to higher intravenous doses. Intraventricular B7-H3 CAR T cells, given without lymphodepletion, enabled multi-cycle dosing with mainly grade 1 to 2 events and cerebrospinal fluid localized persistence. Weekly intracranial EGFR806 CAR T cells were feasible and well tolerated, with stable disease as the best response in a small cohort. Across trials, persistence and immune activation were most evident in cerebrospinal fluid, supporting cerebrospinal fluid centered pharmacodynamic monitoring. Mechanism-based combinations, including IGF-axis inhibition in diffuse midline glioma and epigenetic priming of GD2 with an integrated safety switch in medulloblastoma, enhanced activity. The evidence supports pediatric-centric antigen selection and a CNS-first, locoregional dosing approach to increase on-tumor exposure and reduce systemic toxicity. Priorities include multi-antigen strategies to prevent escape, incorporation of safety switches, earlier deployment when tumor burden is low, and prospective cerebrospinal fluid pharmacodynamics in multisite phase II studies.