Using Intrathecal Dendritic Cells for Leptomeningeal Disease From Breast Cancer

At the 2026 SNO ASCO CNS Metastases Conference, Peter Forsyth, MD, discussed a single-arm, dose-escalation, phase 1 trial (NCT05809752) evaluating the intrathecal delivery of conventional type 1 dendritic cells (cDC1s) in patients with HER2-positive or triple-negative breast cancer and leptomeningeal disease. As of data reported in his abstract from March 26, 2026, among 14 patients, grade 3 headache occurred in 57%, and dose-limiting toxicities (DLTs) occurred in 14% of patients treated at the highest dose level (5 × 10⁷ cells). This prompted de-escalation to 1 × 10⁷ cells, at which no DLTs had been observed to date. Cytokine and transcriptomic analyses in a subset of patients showed increased Th1-associated cytokines without a corresponding rise in IL-4, and single-cell RNA sequencing demonstrated expansion of CD4-positive, CD8-positive, γδ T cells, and B cells.

Forsyth discussed the immunologic rationale for pursuing a dendritic cell–based approach in leptomeningeal disease, how the trial was designed, and why HER2-positive and triple-negative breast cancer populations were selected. He also addressed the practical burden of weekly treatment on patients, outlined the phase 2 trials now open building on this platform, and closed by discussing other trials in the leptomeningeal disease space that he is following closely.

Forsyth is chairman of the Neuro-Oncology Program at Moffitt Cancer Center and a professor of Oncology at the University of South Florida.

CancerNetwork: Leptomeningeal disease has traditionally been treated with intrathecal chemotherapy. What was the rationale for pursuing an immune-based approach here?

Forsyth: It's a really bad disease, and it's been recently discovered that there are a lot of immune cells and inflammatory cells in the cerebral spinal fluid (CSF) space, which is surprising because I always thought that the CSF was just a bunch of tumor cells floating around in this watery stuff. When we did single-cell analysis, as did others, there were a whole bunch of immune cells. To answer your question, we thought, with all those immune cells, we should do something to modify the immune environment, and, to keep it simple, enhance the good responses and try to get rid of the bad responses in the CSF.

And it is—just to go off for a second—super interesting, because one of my colleagues found that in that environment, the immune cells are kind of trying to kill the cancer cells. They know they're there, and they mount a very weak innate response, which isn't sufficient, but it helps a little. What we're trying to do, and what we hoped—and I think it's true—is that we can augment this with this immunotherapy by enlisting an adaptive immune response, or an antibody or tumor-cell-specific response. That's our hope. We have to do all the lab work to figure that out, but we have some pretty good data.

How was the phase 1 trial designed, and why specifically did you go with patients with HER2-positive and triple-negative breast cancer?

First of all, one of the main topics is that we're using the patient's own immune system to turn it against the tumor cells, which uses dendritic cells, which I call the “Paul Revere” of the immune system. That recognizes common peptides, presents them, and makes them kill them. We chose HER2-positive patients because HER2 is obviously there, and HER3 is commonly there in triple-negative tumors too. We thought that was most likely to respond if we're using peptides to HER2 and HER3.

It's a pretty standard dose escalation trial. First of all, we take patient cells from their arm through apheresis, and then grow them up over 2 or 3 weeks, and train them—like, “Hey, these are your immune cells, now let's make them mad and angry about these HER2 and HER3 positive peptide things”—and then we put them back in an Ommaya, which is a brain port that goes into the spinal fluid. It is a bit of a wild thing to do, to put all these immune things in the CSF directly. It started off at very low doses, and then escalated to much higher doses. Of course, safety was number one, so we went pretty low to begin with—a million cells. Then it's a standard phase 1 dose expansion if we get a dose-limiting toxicity.

How safe was the treatment?

We found it was safe, but there were, at the highest doses—which was 50 million cells—patients who had a prolonged headache and didn't feel well for like a week, or in some cases 2 weeks. Of course, what we're doing is inducing a meningitis, but not a bacterial or viral meningitis. We're putting all these immune cells in there, so of course every patient gets headaches. It doesn't feel well for at least a few hours and probably the rest of the day, and then the headaches go and they feel a bit off. That's universal, everybody in the trial gets those, and they're not insignificant. There were a couple patients who developed symptoms that lasted a long time. We pegged the dose at the second-highest dose, which is 10 million cells. The setup was that it's given once a week in that Ommaya reservoir for 12 weeks. It's a pretty big investment in time.

You dropped the dose from 50 million cells to 10 million cells. Did the treatment maintain a meaningful immune response?

We believe so. We just completed the phase 1 portion, and now we're conducting single-cell cytokine analysis and profiling the immune cells in the CSF. We've done this in the first 6 patients, and we're now getting data back on the full cohort, which is 13 [evaluable patients] out of 15, since 2 had CSF samples that weren't usable. That changes things somewhat. But in the first 6, it was genuinely exciting: one patient, who has now survived past 3 years, had HER2-positive breast cancer and developed an antibody-dependent cellular cytotoxicity signature. We were able to run an assay using her CSF, and before treatment there was no ADCC signal; after treatment, it rose using a fluorescent readout system. So in at least a few cases, it looks like we're generating an adaptive immune response.

There's still a great deal of science left to confirm this, but something is clearly happening. We believe that, at least in part, and there's already a phase 2 study open testing this, the treatment prepares the CSF to mount these responses, making it more receptive to antibody therapy. So it's not a standalone treatment in and of itself.

At least 6 or 7 of the 15 patients didn't do well, and that's heartbreaking. For full disclosure, one woman came from Minnesota, and her disease progressed out of control; she was only about 35 years old. By no means did everyone benefit, but some patients benefited substantially. Another patient, from out of state, was 64 with triple-negative disease. Patients with a triple-negative diagnosis typically survive about 2 months, and she passed away about 3 weeks ago, roughly a year after starting treatment. We're working hard to determine who is going to benefit and who isn't, which will probably take about a year to sort out. We literally started looking at some of this data yesterday.

Looking at the cytokine and transcriptomic analyses, what did that show specifically with this trial, and how confident are you that the immune shift is translating into tumor control?

On the cytokine analysis, overall, we're seeing a CD4, dendritic cell, B cell response, consistent with what we observed in animal models. Based on antibody blocking experiments, it appears to be CD4-dependent rather than CD8-dependent, and B cells play an important role. We were surprised by this finding in both mice and patients; B cells went from 1% of cells before treatment to 25% after, which is a substantial increase. We're now trying to determine whether the B cells are participating directly in the adaptive immune response, serving an antigen-presenting function, or both. The signal we're seeing in the single-cell transcriptomic analysis, along with the cytokine data, is consistent with a Th1 response, though these remain correlations rather than proof. We're doing further lab work to confirm it.

I don't believe the treatment is harmful to anyone. The patient I mentioned earlier who had such a difficult outcome––her family, donated her body to science afterward. That allowed us to examine her brain, since I was concerned there might have been some unusual immune-related complication, but there wasn't; it was simply tumor progression. So we're working to determine who will benefit and who won't, and I don't have early answers yet since we only started reviewing the data yesterday. We'll be looking at B cell activation and CD4 activity, as well as the myeloid cells and tumor-suppressive cell populations, which have received less attention so far. It's genuinely exciting work. I want to figure this out, because I don't want to put patients through this treatment if it isn't going to help them. For those it will help, they need to get started as soon as possible.

Practically and logistically, what does the delivery look like? How did it impact patient burden and quality of life during treatment?

It's once a week, so it's a considerable burden. Patients come in, receive treatment through the Ommaya, get connected to their port, and we keep them for 2 to 4 hours afterward in case of complications. They typically have a bad headache and feel unwell for a day or a day and a half. I can't speak to the broader quality-of-life impact beyond that because we didn't formally measure it. There's also a sociological factor we don't fully understand: patients with leptomeningeal disease from breast cancer are, almost universally, notably younger, roughly 14 to 17 years younger on average, than patients with HER2-positive metastatic breast cancer without leptomeningeal involvement. That biology needs further investigation. But the practical reality is that these patients are young, they've just started their lives, and they're not inclined to complain much because they genuinely want to continue treatment. There aren't many trials open in this space, probably only 2 or 3 others in the country, so patients don't have many alternatives.

I'll add one more point: the phase 2 study is already open. For patients with HER2-positive disease, we're adding intrathecal trastuzumab [Herceptin], since trastuzumab works. Animal data suggest it should make responses to therapeutic antibodies more effective, so we're testing that in patients. For triple-negative and HER2-negative patients, including those who are hormone receptor–positive, we're adding intrathecal nivolumab [Opdivo], a checkpoint inhibitor, to further enhance the immune response in the CSF, which again has worked in mice, though curing mice is considerably easier than curing patients. We've been fortunate, because research funding is becoming increasingly scarce, and we received a Department of Defense grant that funded the first clinical trial, followed by a second Department of Defense grant. That allowed us to close one study and open the next within about 6 weeks to 2 months, which is unusually fast. Typically you wait for results, then publication, then write a grant, and 2 to 3 years pass before the next step begins. We have a third trial planned to open next February or March, using a further variation on this approach.

If this continues to show promise, what would a phase 2 or phase 3 trial look like?

For HER2-positive patients, it's 12 weeks of dendritic cell treatment, followed by intrathecal trastuzumab starting after the first 6 weeks, which is effective against HER2 breast cancer. That combination continues for a full year, after which we'll evaluate outcomes. For HER2-negative patients, whether triple-negative or hormone receptor–positive, it's 6 weeks of dendritic cells followed by intrathecal nivolumab, also continuing for a year. It's a promising and interesting approach, but we still need to determine the underlying mechanism and confirm whether, and how well, it works.

What other trials in the CNS metastases space are you interested in or watching that you think might report good data in the near future?

There aren't many trials in leptomeningeal disease, so my apologies to colleagues I might omit, but I believe there are about 4 total, including ours. One is at Memorial Sloan Kettering, led by [Jessica Wilcox, MD], which I believe has about 2 enrollment slots remaining; they published a paper showing that iron levels were markedly low in the CSF and that an iron inhibitor could be used therapeutically, with promising early data, though we're still awaiting final results. [Jonathan Yang, MD, PhD], a radiation oncologist at NYU, has been using proton craniospinal radiation and now has trials combining that approach with DNA repair inhibitors, such as ATM inhibitors. The third is [Andrew Brenner, MD, PhD], at UT San Antonio, who developed the approach of delivering radioactive nanoparticles directly into the spinal fluid, so the radiation stays localized to the CSF rather than affecting the rest of the body. There may be 1 or 2 others, but there aren't many trials in this space nationally.

References

Forsyth P, Law V, Grogan P, et al. Intrathecal dendritic cell therapy in a first-in-human trial for breast cancer leptomeningeal disease induces adaptive immunity and transcriptomic immune awakening in the cerebral spinal fluid. Neurooncol Adv. 2026;8(Suppl 6):vdag161.013. doi:10.1093/noajnl/vdag161.013