Phase 1 B7-H3 CAR T-cell trial in recurrent glioblastoma reports disease control signals and serious adverse events
Background
A phase 1 trial evaluated the initial therapeutic potential and safety of intracranial B7-H3-targeted chimeric antigen receptor T-cell (CAR T-cell) therapy in patients with recurrent glioblastoma. No dose-limiting toxicities were observed, but three grade 3 treatment-related adverse events classified as serious occurred. Among 14 patients with measurable lesions, 8 had stable disease or a better response.
Recurrent glioblastoma is a malignant brain tumor with limited treatment options. The investigators explored the safety and efficacy of TX103, an autologous CAR T-cell therapy made from patients’ own cells. Eligible patients were 18–75 years old and had recurrent glioblastoma meeting a B7-H3 positivity threshold of at least 30%. The study used an open-label 3+3 dose-escalation design. Primary endpoints included safety, maximum tolerated dose, and recommended phase 2 dose; survival, pharmacokinetics, and immune responses were also assessed.
Key findings
Fifteen patients received a total of 72 intracranial administrations, and 13 received repeated administrations. The cell doses assessed per administration were 20 million, 60 million, and 150 million. The supplied abstract did not specify the precise administration site, such as the ventricles, or the device used.
No dose-limiting toxicity was observed, and the maximum tolerated dose was not identified. This does not mean that there were no serious adverse effects. Treatment-related adverse events included low-grade cytokine release syndrome in 86.7% of patients; sinus tachycardia, vomiting, and hypertension in 53.3% each; and increased intracranial pressure in 46.7%. The grade 3 events classified as serious were increased intracranial pressure, seizure, and decreased consciousness; 2 of the 3 events occurred at the highest dose level.
Eight of the 14 patients with measurable lesions achieved disease control. Here, disease control means stable disease or a better response, so it does not mean that all 8 patients’ tumors shrank. One of these patients had a complete response, which was ongoing at the most recent follow-up. Measured from the first administration, the 12-month overall survival rate was 66.7%, and median overall survival was 19.1 months. The lower bound of the 95% confidence interval for median overall survival was 8.93 months; the upper bound had not yet been estimated.
CAR gene copy numbers and cytokine release increased markedly in cerebrospinal fluid, while peripheral activity was reported to be minimal. No cumulative toxicity was observed after repeated administrations. However, these findings alone do not establish that intracranial administration is less toxic than intravenous administration or more effective at killing cancer cells.
Implications and outlook
These results support further clinical evaluation of TX103 in B7-H3-positive recurrent glioblastoma. The investigators proposed the intermediate dose of 60 million cells per administration as the recommended phase 2 dose. Although the maximum tolerated dose was not identified, the study proposed a dose for evaluation in a follow-up trial.
The findings remain limited to an early-phase clinical study. This was an open-label phase 1 trial involving 15 treated patients, and the supplied abstract reported no comparison with a control group. The reported survival therefore cannot readily be interpreted as an improvement over existing treatments, nor can safety be considered established. The abstract also does not provide enough information to assess the precise duration of the complete response, the characteristics of responders, or long-term safety. Follow-up trials should assess the reproducibility and durability of responses alongside the pattern of serious adverse events.
Source: Europe PMC bibliographic record and abstract for the Nature Medicine paper.
Nature Medicine, Published online: 06 August 2026; doi:10.1038/s41591-026-04557-6In this phase 1 dose-escalation trial, intracranial delivery of B7-H3-targeting chimeric antigen receptor T cell therapy in patients with glioblastoma showed no dose-limiting toxicity signals and encouraging signals of clinical benefit.
This study provides a rationale and a dose for follow-up trials of repeated intracranial CAR T-cell administration in patients with B7-H3-positive recurrent glioblastoma. It may inform trial designs that assess disease control and survival at 60 million cells per administration while closely monitoring serious adverse events such as increased intracranial pressure, seizures, and decreased consciousness. This is a recommended dose for further clinical evaluation, not an established dose for routine care. Although experience with repeated administration has accumulated, the feasibility of outpatient treatment, reductions in manufacturing costs or hospitalization, and improvements in quality of life were not verified in the supplied abstract. Larger patient groups are needed to assess the durability of responses and the risks before clinical applicability can be judged.