Abstract
BACKGROUND: Recurrent glioblastoma (GB) still carries a dismal prognosis with few established treatment options. In the multicenter CAR2BRAIN phase I first-in-human clinical trial, we investigated repetitive local adoptive transfer of clonal chimeric antigen receptor (CAR)-NK cells (NK-92/5.28.z) targeting HER2 alone and in combination with the anti-PD-1 checkpoint inhibitor ezabenlimab in patients with recurrent HER2-positive GB. METHODS: After establishing the dose of 1×10(8) irradiated CAR-NK cells as safe for intracerebral injection in the dose-escalation cohort in 9 patients, 6 patients were treated with repetitive doses of CAR-NK cells in the expansion cohort. In the subsequent CAR2BRAIN-Check cohort, 12 patients received a combination therapy with the anti-PD-1 checkpoint inhibitor ezabenlimab. CAR-NK cells were injected into the margins of the surgical cavity during relapse surgery, and repeatedly via an implanted reservoir into the resection cavity. Where feasible, we adhered to a biopsy-treat-resect-treat strategy, initiating study treatment before planned tumor resection. An in-depth analysis of tissue, cerebrospinal fluid (CSF) and blood samples before and after immunotherapy was performed to assess treatment-induced modulation of the tumor immune microenvironment. RESULTS: Repetitive intracranial injection of CAR-NK cells was feasible and safe, and none of the patients developed a cytokine release syndrome or immune effector cell-associated neurotoxicity syndrome. Combination immunotherapy induced a local immune response with elevated pro-inflammatory cytokines and cell counts measured in CSF sampled from the resection cavity. We observed an increase of CD4(+) and CD8(+) T cells, and a decrease of regulatory CD4(+)FoxP3(+) T cells. Median progression-free survival of the patients treated in the dose-escalation cohort was 7 weeks, compared to 10.5 and 14.5 weeks, respectively, for the patients of the expansion cohort and the CAR2BRAIN-Check chort. Median overall survival of the patients in the dose escalation cohort and the expansion cohort was 31 and 44.5 weeks, and has not yet been reached in the CAR2BRAIN-Check cohort. CONCLUSIONS: Immunotherapy with repetitive intracranial injection of HER2-targeted CAR-NK cells is feasible and safe both as monotherapy and in combination with the systemic checkpoint inhibitor ezabenlimab, and favorably modulates the intratumoral immune microenvironment. Given the highly promising results obtained so far, further trials are warranted to confirm the potential clinical benefit of this strategy.