Glioblastoma Organoids Track Bioactivity of Patient-Matched CAR-T Products Alongside a Clinical Trial

Matching the Tumor and the Clinical Cell Product

A University of Pennsylvania study paired glioblastoma organoids with autologous chimeric antigen receptor T-cell (CAR-T) products from the first six patients in a phase 1 trial for recurrent glioblastoma. The CAR-T cells targeted epidermal growth factor receptor (EGFR) and interleukin-13 receptor α2 (IL13Rα2). Organoid assays ran alongside clinical treatment, allowing responses to the same manufactured product to be examined in a matched experimental tissue model.

Freshly resected tumor fragments generated organoids within two to three weeks. At day 21, staining for the targeted antigens remained qualitatively comparable to parental tumor tissue. The core paired cohort comprised six treated patients; a seventh patient who did not receive trial treatment supplied material for additional control experiments.

The paired series combined the first three patients at a clinical dose of 10 million cells and the next three at 25 million cells. These clinical dose cohorts remain distinct from the fixed estimated culture ratio, which does not model each patient’s delivered tumor exposure.

This design connects a patient-derived model to a clinical product without assuming that an experimental culture reproduces every aspect of delivery in the patient. The culture used an estimated CAR-T-to-tumor-cell ratio of 1:10 because the corresponding ratio in vivo could not be measured accurately.

Electrical and Tissue Measurements Support Tumor Killing

Organoids adhered to coated impedance plates before CAR-T addition and were monitored for six days. The assay normalized electrical changes against untreated organoids and full-lysis controls. In the main figure, each patient had at least 12 treated organoids and at least six untreated controls. Cleaved-caspase-3 staining in four to six organoids per condition supplied a separate apoptosis-associated check.

All six paired samples showed significant cytolysis over time. In the off-trial control material, on-target CAR-T cells produced substantial killing, whereas unmodified T cells and CD19-directed CAR-T cells did not. Additional staining showed infiltration of CD3-positive cells and loss of Nestin-positive tumor cells, particularly near organoid edges.

At 24 hours, the magnitude of organoid cytolysis correlated with peak CAR-T engraftment measured in patient cerebrospinal fluid. That clinical comparator was CAR-copy abundance, not a direct measure of patient tumor killing or survival. Changes in magnetic-resonance contrast enhancement reported in the trial were suggestive but could not definitively establish tissue killing; the organoids supplied a separate direct experimental observation.

Antigen and Cytokine Assays Add Distinct Information

After six days, more than 20% of organoid-associated CD3-positive cells were Ki-67-positive across all six samples. Granzyme B and CD69 staining also supported activation and cytotoxic activity. These measurements complement the tumor-response assay rather than replace it.

Flow cytometry after treatment was possible for five patients, with insufficient organoids from the sixth. Double EGFR/IL13Rα2-positive populations decreased and double-negative populations increased. Four of five samples showed decreased EGFR-only and increased IL13Rα2-only fractions. The heterogeneous changes are consistent with differential antigen-directed effects, but the study did not obtain matching post-treatment patient tumor tissue to establish the same antigen shifts in vivo.

Daily measurements of IFNγ, TNFα and IL-2 in conditioned medium showed peaks around days 1 or 2 and broadly similar temporal patterns to patient cerebrospinal fluid. Comparisons through day 4 did not find statistically significant concentration differences. With six patients and different sampling compartments, absence of statistical significance is not evidence of equivalence; later comparisons also paired organoid day 6 with clinical day 7.

A Correlative Trial Companion with a Small Paired Cohort

The study demonstrates that organoids can be generated quickly enough to investigate a clinical cell product in parallel with patient treatment. It also exposes practical limits: scarce tissue prevented every assay from being completed in every patient, the estimated effector ratio could not mirror individual in vivo exposure, and the correlative experiments were neither randomized nor blinded.

The strongest clinical link is between early ex vivo cytolysis and subsequent cerebrospinal-fluid CAR-T abundance. The work remains exploratory and hypothesis-generating. It supports a trial-companion model for studying bioactivity, antigen selection and activation dynamics, while prospective treatment selection and prediction of durable patient benefit require further validation.