Personalized Brain-Cancer Vaccine Shows Promise at ASCO 2026

For three decades, the playbook for treating glioblastoma has barely changed, and neither has the grim arithmetic that follows a diagnosis. So when a Dana-Farber Cancer Institute researcher stepped to the podium at the American Society of Clinical Oncology annual meeting in late May, the message he carried was unusual for a disease that has humbled so many promising therapies. A personalized vaccine, custom-built from the unique genetic fingerprint of each patient's tumor and combined with a widely used immunotherapy drug, kept some glioblastoma patients alive and stable longer than doctors would normally expect.

The findings, drawn from an early-phase trial of a vaccine known as NeoVax given alongside the checkpoint inhibitor pembrolizumab, are preliminary. The patient numbers are small, and the survival comparison relies on matched historical patients rather than a randomized control group. But in a cancer where caution is usually code for disappointment, the durability of the responses gave the audience reason to take notice.

A cancer that resists almost everything

Glioblastoma is the most common and most aggressive malignant brain tumor in adults, and it remains one of the hardest cancers in all of medicine to treat. With the current standard of care, surgery followed by radiation and the chemotherapy drug temozolomide, median survival sits at roughly 15 months. Fewer than 5 percent of patients survive five years. High-profile patients including Senators John McCain and Ted Kennedy, and Beau Biden, brought the disease into public view, but did little to change its trajectory.

Several features make glioblastoma so lethal. The tumor grows with tentacle-like projections that weave into healthy brain tissue, making complete surgical removal essentially impossible and recurrence nearly universal. The blood-brain barrier, the brain's protective filter, blocks many cancer drugs from reaching the tumor in meaningful concentrations. And glioblastoma is what oncologists call an immunologically cold tumor, meaning it does a remarkable job of hiding from and suppressing the body's immune defenses. Each tumor is also genetically chaotic, with different populations of cells carrying different mutations, so a treatment that knocks out one part of the tumor often leaves another to grow back stronger.

How a personalized vaccine takes aim

NeoVax attacks the problem from a different angle than chemotherapy or radiation. Rather than poisoning fast-dividing cells, it tries to teach the patient's own immune system to recognize the cancer as foreign. The target is a class of molecules called neoantigens, abnormal proteins that appear only on tumor cells as a result of the mutations driving the cancer. Because they do not exist on healthy cells, neoantigens are, in theory, ideal flags for an immune attack that spares the rest of the body.

Building the vaccine begins with the patient's own tumor. After surgery, scientists sequence the DNA of the cancer and compare it against the patient's normal tissue to map mutations unique to the tumor. Analysis then predicts which mutated fragments are most likely to be displayed on the cell surface and recognized by the immune system, and the strongest candidates, roughly 7 to 20 peptides per patient, are manufactured into a personalized vaccine. Once injected, the vaccine prompts immune cells to present these neoantigens to T cells, training the immune system to hunt down anything carrying the same markers.

The combination with pembrolizumab is the strategic twist. The vaccine acts as an accelerator, priming an army of tumor-specific T cells. Pembrolizumab, sold as Keytruda, acts on the brakes, blocking the PD-1 pathway that tumors exploit to switch off immune attacks. The idea is that the vaccine summons the right soldiers while the checkpoint inhibitor keeps them from being shut down once they reach the brain.

What the trial actually showed

The approach was pioneered in the laboratory of Catherine Wu at Dana-Farber, in partnership with the Broad Institute and Brigham and Women's Hospital. Foundational work published in the journal Nature in 2019 established that the vaccine could generate T cells that crossed the blood-brain barrier and infiltrated the tumor, a milestone for a cancer long considered beyond the reach of immune therapy. "This is the first time it has been shown that a vaccine can generate immune cells against the tumor that can traffic from the bloodstream into a glioblastoma tumor," David Reardon, the trial's senior investigator and director of the Center for Neuro-Oncology at Dana-Farber, said of that earlier finding.

The data presented at ASCO 2026 build on that proof of concept by following patients longer and by reporting on the vaccine paired with pembrolizumab. Among patients with newly diagnosed MGMT-methylated glioblastoma, a subgroup with more favorable biology, median overall survival reached 36.9 months, compared with 25.3 months in propensity-matched standard-of-care controls at the same institution. In patients with MGMT-unmethylated disease, a harder-to-treat group, median survival was 19.0 months versus 16.7 months in matched controls. Vaccine-stimulated anti-tumor activity remained evident in some patients more than a year after treatment, and the strength of neoantigen-specific T-cell responses correlated with longer survival.

One detail of trial design deserves emphasis. The study avoided high doses of dexamethasone, a steroid commonly prescribed to reduce brain swelling but which also blunts the immune system. That choice appeared to give the vaccine room to work, underscoring how delicately the therapy depends on an unimpeded immune system and how everyday clinical decisions can either help or hinder it. The data also hinted that starting pembrolizumab before vaccine priming may yield longer survival, a question being explored in additional cohorts.

Why the optimism comes with caveats

The reasons for restraint are real, and Reardon did not hide them, cautioning that the results are encouraging but must be interpreted carefully because the study was not a direct, randomized comparison. This is a Phase 1 trial, designed first to test safety and feasibility, not to prove that patients live longer. Comparing results to matched historical controls, rather than to a randomized group treated at the same time, can flatter a new therapy, in part because trial participants are often healthier than the broader patient population. The patient numbers are small, and glioblastoma has a long history of early treatments that dazzled in initial studies only to fail when tested rigorously in larger trials.

There is also the sheer complexity of the manufacturing process. Sequencing a tumor, predicting neoantigens and producing a bespoke vaccine takes weeks to months, a difficult timeline for a cancer that can grow quickly in that span. Scaling a personalized product to thousands of patients, at a price health systems can absorb, remains an open question.

What comes next

The path forward runs through larger, controlled trials that can establish whether the survival signal holds up when the comparison is fair and the patient group is bigger. Researchers will be watching not only how long patients live, but which patients benefit, and whether the immune responses seen in the lab translate reliably into shrinking or stabilized tumors. Refining the timing of surgery, vaccine delivery, pembrolizumab and steroid use, and shortening the manufacturing window, will all factor into whether the approach can move from a research milestone toward routine care.

For now, the ASCO 2026 results are best understood as encouraging early evidence rather than a breakthrough at the bedside. In a disease defined by decades of dashed hopes, even cautious progress carries weight. The question that will define the next several years is whether teaching the immune system to read a tumor's genetic signature can finally bend glioblastoma's stubborn survival curve, or whether it joins the long list of ideas that worked better in theory than in the clinic.