A novel GD2-directed CAR T-cell therapy demonstrated preliminary antitumor activity, including neurological benefit, in patients with H3K27M-mutant diffuse midline gliomas, including diffuse intrinsic pontine glioma and spinal diffuse midline glioma, according to Michelle Monje, MD, PhD.1
A phase 1 study (NCT04196413) evaluated the safety and efficacy of the GD2-directed CAR T-cell therapy in patients with a diagnosis of H3K27M-mutant diffuse midline glioma who were between 2 and 30 years of age.1,2 Patients on the study were required to have received prior therapy at least 4 weeks after completion of standard upfront radiation therapy and at least 3 weeks post chemotherapy or 5 half-lives, whichever was shorter.2 Of note, the median overall survival (OS) for patients in arm A treated with the intravenous (IV) GD2 CAR T-cell therapy was 20.6 months from diagnosis; of these patients, 2 with diffuse intrinsic pontine glioma remained alive and were followed to data cutoff.1 Patients with diffuse intrinsic pontine glioma had a median OS of 17.6 months, and those with spinal diffuse midline glioma had a median OS of 31.96 months.
“[We have] seen the promise of this therapy and have seen some patients get a lot better for a substantial amount of time, and even 1 patient having a durable complete response [CR],” Monje said in an interview with OncLive® for Childhood Cancer Awareness Month, observed annually in September. “We know that this is a therapy with activity that needs to be optimized so that every patient has this kind of response.”
During the interview, Monje discussed the background and rationale for evaluating the novel GD2-directed CAR T-cell therapy, key efficacy results from the phase 1 trial, and notable safety findings from the study.
Monje is the Milan Gambhir Professor of Pediatric Neuro-Oncology and a professor by courtesy of neurosurgery, pediatrics, pathology, and psychiatry and behavioral sciences at Stanford University, as well as a member of the Institute for Stem Cell Biology and Regenerative Medicine, the Maternal & Child Health Research Institute, the Wu Tsai Neurosciences Institute, and the Stanford Cancer Institute in Palo Alto, California.
OncLive: What was the background and rationale of the phase 1 study?
Monje: Diffuse midline gliomas are a terrible primary brain cancer, and these typically occur in children. They [also] can and do occur in adults, particularly young adults; we now recognize that we see these diseases across the lifespan. These are cancers that occur in the midline of the nervous system, hence the name, and include diffuse intrinsic pontine glioma, as well as thalamic and spinal cord versions. They’re high-grade, aggressive cancers. [Patients with] diffuse intrinsic pontine glioma have a median OS of only approximately 11 months, and [patients with] the non-pontine versions in the thalamus and spinal cord have a median OS of only approximately 13 months, with a less than 1% 5-year survival [rate].
Over the years, my laboratory has studied these cancers and discovered that their tumors integrate into the nervous system. This is part of why they’re so diffusely infiltrative. They form synapses with neurons, and they spread out widely and integrate into the neural circuits that they’re invading.
How do you go in and pick out each individual cancer cell? Surgery is not an option. Radiation is only temporary, and no chemotherapeutic agents have been helpful to date.
On that backdrop, in approximately 2015, as CAR T-cell therapy was [demonstrated to be] promising for hematological malignancies, in my lab, we started thinking about ways that we might leverage cancer cellular therapy to address diffuse intrinsic pontine glioma and other diffuse midline gliomas. We started by doing a cell surface screen, looking for antigens present on the surface of patient-derived tumor cultures from patients with diffuse intrinsic pontine glioma and other diffuse midline gliomas. There was a high and uniform expression of a disialoganglioside glycolipid called GD2. This was exciting, because it was the perfect endotherapeutic target in that it was highly expressed on every single malignant cell. The reason it’s so highly and invariably expressed is that the expression of the synthesis enzymes that make GD2 are driven by the pathognomonic mutation that causes this kind of cancer, which is a mutation in genes encoding histone H3 that cause prominent epigenetic dysregulation, and consequently, upregulation of GD2. Every cell type that has the mutation has high GD2 levels.
What were the key efficacy findings from the phase 1 trial?
This was, early on, a hopeful clinical experience. [Among] the first few patients we treated, several had promising therapeutic responses. What was incredibly encouraging was that we didn’t just see tumor shrinkage; we saw children and young adults get dramatically better, neurologically.
After an initial period of inflammation, in which you expect patients to be sick and need support throughout, we saw patients regain major neurological functions. We saw a patient with a spinal cord tumor that had caused dense paraparesis and incontinence of urine and stool regain the ability to move her legs, the ability to stand with support, and the ability to have full continence. We saw a young man who was so ataxic when he came to the hospital that he needed to be in a wheelchair for long distances. Within 2 weeks, [he was] going on 2-mile hikes with his family.
We saw dramatic improvements in neurological function, return of hearing, return of movement, return of taste sensation, and return of touch sensation. Then, patient 10 on the trial had a dramatic CR. We saw the tumor disappear completely, and he is still tumor free. It’s [been] 4 years since we began treating him, which is amazing for this disease.
However, we saw a wide range of responses. As encouraging as those hopeful responses were, we also saw patients who didn’t respond, who had early and transient benefit and then lost it, or who never experienced benefit to begin with. There weren’t outliers. We evaluated the tumor volume and change as 1 measure of tumor response. There was a Gaussian distribution; there was a normal distribution of responses from across the patients who we treated in the first cohort of the trial.
What accounted for this difference in response? [These were] the same tumors [as each other]. The antigen expression [levels were] similar [between them]. Whenever we studied that on postmortem samples, it looked like the antigen was highly expressed in the tumors. [We treated patients with] the same therapy given the same way. What was the difference? We’re working hard by studying the correlative samples to try to identify factors that predict and mediate better response, so we can work toward a combination strategy to achieve that for all patients. That’s still ongoing.
What were notable safety findings?
The way we designed the trial [for this] first cohort of patients was to first give the therapy IV after lymphodepleting chemotherapy, and then if patients exhibited benefit, either radiologically or clinically, they were eligible for subsequent doses given intracerebroventricularly through an Ommaya catheter. We had anticipated that this [therapy, which causes] therapeutic inflammation in midline structures like the brain stem and the spinal cord, would require multidisciplinary, neurocritical care support. Therefore, we administered this therapy in an inpatient setting, often in the pediatric intensive care unit with neurology, neurosurgery, and oncology all involved.
What we saw was, unsurprisingly with the IV administration of the therapy, a lot of cytokine release syndrome [CRS]. Reliably, a few days after we gave the therapy, we would see patients spike a high fever. Sometimes there was low blood pressure associated with that, and in some cases, there was also pulmonary edema—high-grade CRS.
CRS was the dose-limiting toxicity in this first cohort of the trial. We only saw CRS after IV therapy. We never saw it after intracerebroventricular administration of the CAR T cells. Only with IV therapy did we see the CRS.
We anticipated seeing a high rate of immune effector cell–associated neurotoxicity syndrome [ICANS]. We thought, with this degree of inflammation in the nervous system, we were likely to see this acute neurotoxicity syndrome that’s characterized by confusion, delirium, aphasia, and ataxia. However, surprisingly, we saw little ICANS.
What we did see, though, was [another syndrome that was also] anticipated. We decided in the course of this trial that we needed to name this syndrome. We anticipated seeing neurological symptoms attributable to inflammation in the place where the tumors were located. For every neurologist, that’s intuitive: if you see a big inflammatory response in a focal region of the nervous system, you expect to see that part of the nervous system not function well for a period. That’s informed by our experience with inflammation in many different contexts.
We saw a transient worsening of pre-existing neurological symptoms and other neurological symptoms attributable to the place where the tumor was. [We also saw] that the inflammation was happening. We did not see evidence—either clinically or correlatively, including on post-mortem studies of some patients—of off-tumor inflammation. [Nevertheless,] we saw tumor inflammation–associated symptoms, which we decided to name tumor inflammation–associated neurotoxicity, so we could begin to grade and discuss it. An important aspect that came out of this trial was describing [a finding that] is intuitive for neurologists but important to quantify and describe: the syndrome of tumor inflammation–associated neurotoxicity.
Mechanistically, tumor inflammation–associated neurological symptoms can take on 1 of 2 major flavors. One, there can be swelling in that region of brain with shifts in tissue and mechanical issues, given the closed compartment of the skull and the spinal canal. Mechanical issues that include hydrocephalus can be caused by expansion of the region of tumor and impingement on the ventricular system, tissue shifts, and herniation. There’s also an electrical problem where parts of the nervous system don’t function properly. Inflammatory molecules can affect the function of neurons, and [this effect is] transient. It’s not a destruction, it’s a dysregulation.
However, in general, the nervous system doesn’t work well when there are marked levels of cytokines, chemokines, and immune cells around. Therefore, we saw a transient worsening in neurological functions, focally attributable to the location where the tumor was and where inflammation was ongoing, that typically lasted days to weeks and were reversible. [These] can be clinically important—particularly if the tumor is located in critical areas, like areas of the brain stem—for swallowing or the strength of the muscles of respiration. It’s important to keep an eye on [these functions]. It’s important to anticipate issues based on where the tumor is and be prepared to support the patient through that period of inflammation.
In almost every patient, we saw tumor inflammation–associated neurotoxicity. Sometimes it was low-grade weakness in 1 limb that was a little worse in the context of inflammation and then got better. Sometimes it was high grade, and we would see obstructive hydrocephalus that needed to be treated through the Ommaya catheters that we placed in every patient in anticipation of that potential application. [We also saw the potential for] herniation syndromes. This inflammation occurred in a bad location in the brain stem and spinal cord, and we needed to support these patients in an intensive way.
References
- Monje M, Mahdi J, Majzner R, et al. Intravenous and intracranial GD2-CAR T cells for H3K27M+ diffuse midline gliomas. Nature. 2025;637(8046):708-715. doi:10.1038/s41586-024-08171-9
- GD2 CAR T cells in diffuse intrinsic pontine gliomas (DIPG) & spinal diffuse midline glioma (DMG). ClinicalTrials.gov. Updated March 17, 2025. Accessed September 2, 2025. https://clinicaltrials.gov/study/NCT04196413