A new vaccine for pancreatic cancer has produced one of the clearest immune responses ever seen in this disease — and, at the same time, a result that should stop anyone from calling it a breakthrough too soon. In a randomised phase 2 trial, nearly every patient who could be assessed developed T-cell responses against mutated KRAS, the gene fault that drives most pancreatic cancers. Yet the trial did not meet its main goal: across all patients, it did not significantly delay the return of cancer.
That tension is why the study, published in Nature Medicine on 23 October 2026 and presented at ESMO Congress 2026, matters. It shows the immune system can be taught to recognise pancreatic cancer. Whether that lesson means longer, cancer-free life is still unanswered, and a larger phase 3 trial is planned to test it.
Why pancreatic cancer is so hard to treat
Pancreatic cancer is feared for a simple reason: even when surgery appears successful, the disease often comes back. Microscopic cancer cells can remain in the body after an operation and chemotherapy, invisible on scans, and later grow into a recurrence. Researchers have therefore searched for a treatment that can hunt down those remaining cells, rather than only shrinking a visible tumour.
Immunotherapy has transformed some cancers, but historically worked poorly in pancreatic cancer: its tumours tend to shield cancer cells from immune attack. As Mayo Clinic lead author Dr Ben George put it, pancreatic cancer has been very difficult to treat with immunotherapy.
KRAS: the fault found in about 90% of pancreatic cancers
KRAS is a gene that normally helps control how cells grow and divide. When it mutates, that control breaks down and cells receive a constant signal to grow and survive. Mutations in KRAS are found in about 90% of pancreatic ductal adenocarcinomas, the most common type of pancreatic cancer.
That prevalence makes KRAS an attractive target: a treatment teaching the immune system to recognise it could apply to most patients with this cancer type. As our report on what the average human genome reveals about genetic risk notes, one genetic change rarely tells the whole story.
How the AMPLIFY-7P vaccine works
The investigational vaccine, called camupepimut (also known as ELI-002 7P), targets seven of the most common KRAS mutations. It is designed to travel to the lymph nodes, where immune responses often begin. There, it helps T cells — immune cells that can recognise and attack abnormal cells — learn to identify cancer cells carrying mutated KRAS.
Crucially, this is an "off-the-shelf" vaccine, not one made for each patient. A personalised vaccine can take time and resources to produce; a ready-made product, if it works, could be given more quickly and widely.
That contrast matters because cancer vaccines come in very different forms. Our coverage of the Moderna-Merck mRNA cancer vaccine for melanoma describes a personalised approach, tailored to each patient's tumour and given alongside immunotherapy. AMPLIFY-7P tests the opposite bet: a ready-made vaccine aimed at a shared target.
Who took part, and what was measured
The trial enrolled 144 patients with stage 1 to 3 KRAS-mutated pancreatic cancer. All had undergone surgery and chemotherapy and had no cancer visible on imaging when they entered the study. Researchers randomly assigned 96 patients to receive the vaccine and 48 to observation alone.
The main measure was disease-free survival: how long patients lived without their cancer returning. Overall survival results are not yet available. Importantly, the vaccine was tested as an additional step after standard treatment, not a replacement for it.
The headline result: the main goal was not met
The trial did not meet its pre-specified primary endpoint. Median disease-free survival was 23.8 months for patients who received the vaccine and 21.2 months for those in the observation group. The hazard ratio was 0.85, with a p-value of 0.5329 — meaning the difference was not statistically significant and could plausibly reflect chance in a trial of this size.
A hazard ratio below 1 points in a favourable direction, but without statistical significance, researchers cannot conclude from this trial alone that the vaccine delays recurrence. Calling it proven to prevent recurrence would go beyond the evidence.
There were earlier signals worth noting. At six months, disease-free survival was 75.7% in the vaccine group compared with 61.7% in the observation group, a 14-point difference. An exploratory early-period analysis gave a hazard ratio of 0.55, though its p-value of 0.0557 also fell just short of conventional significance. These are clues, not conclusions — findings that justify a bigger trial, not a change in care today.
The immune response: strong, broad and lasting
Where the trial was most convincing was in its laboratory results. Of the 90 vaccinated patients who could be evaluated for immune response, 89 — or 98.9% — developed increased T-cell responses against mutated KRAS. More than eight in ten evaluable patients developed a KRAS-specific T-cell response more than nine times greater than before vaccination.
The response was also broad. The vaccine activated both CD4 and CD8 T cells and produced responses against several KRAS mutations. In a small group of patients, researchers observed antigen spreading: after vaccination, the immune system began recognising other cancer-related targets not included in the vaccine itself. That could matter, because cancers can evolve and lose a single target; a wider response may be harder to escape.
Durability was also encouraging: among patients evaluable at least 46 weeks after starting, 95.5% still had a detectable KRAS-specific T-cell response. Stronger responses were associated with longer disease-free survival in exploratory analyses, about a 53% reduced risk after adjustment. But that comparison was within vaccinated patients, not by random assignment, so it cannot prove the response itself caused longer survival.
Who might benefit most? Clues from blood tests and past treatment
Two clues may help shape the phase 3 trial. The first involves circulating tumour DNA, or ctDNA — small pieces of DNA released by cancer cells into the bloodstream. Detecting ctDNA after treatment can signal that cancer remains in the body even when scans look clear.
In a pre-specified analysis, patients who had no detectable ctDNA at the start of the study showed a signal of longer disease-free survival with the vaccine. The hazard ratio in the ctDNA-negative subgroup, comprising 93 patients, was 0.51, with a p-value of 0.0280. That is more promising, but subgroup results in a phase 2 trial still need confirmation in a larger study designed to test them.
The second clue involves previous radiation therapy. Patients who had received radiation had fewer lymphocytes — a type of white blood cell central to immunity — and developed weaker immune responses after vaccination. Among patients who had not received radiation, researchers saw a signal of longer disease-free survival with the vaccine. That pattern makes biological sense: a vaccine depending on lymphocytes may work less well if those cells have been depleted.
There was also a baseline imbalance that complicates interpretation. Detectable ctDNA, a strong predictor of recurrence, was more common in the vaccine arm (20.8%) than in the observation arm (12.5%). When researchers adjusted for prognostic factors, the hazard ratio improved to 0.47, with a p-value of 0.0086. That adjustment is informative, but it is not a substitute for a randomised phase 3 trial with balanced groups.
Our article on how a DNA test can change health habits explains what such tests can — and cannot — tell people, a distinction that matters for ctDNA too.
Safety and the limits of a phase 2 trial
Safety results were reassuring in this small study, but they should not be overstated. Grade 3 or higher adverse events were reported in 10.4% of vaccinated patients compared with 14.6% of those under observation. A trial of 144 patients cannot establish that a new treatment is as safe as observation, and rarer side effects may only appear in larger studies. Longer follow-up will also be needed to judge overall survival, which remains immature.
Other limits matter too: the trial was open-label, the main endpoint was not met, and several interesting findings came from exploratory or subgroup analyses, which carry a higher risk of chance findings. Teaching the immune system to recognise KRAS in nearly every evaluable patient is still a real step forward — but it is not yet a proven clinical benefit.
What happens next: a phase 3 trial, and combination ideas
A phase 3 trial is planned to test whether the vaccine can delay or prevent pancreatic cancer from returning. The phase 2 results give its designers several choices to refine: which patients to enrol, whether to use ctDNA to select or stratify patients, how to handle prior radiation, and how to measure immune response alongside survival.
Separately, V Foundation-funded research at the University of Miami will study trial tumour samples and test camupepimut with a RAS inhibitor, daraxonrasib, with or without nivolumab. The question, in researcher Dr Peter Hosein's words, is why T cells taught to recognise KRAS could not always finish the job. Whether a combination can release that brake is a question for trials, not an assumption from early data.
What this means for patients and families right now
First, this vaccine is investigational: not approved and not available as standard treatment. Patients interested in trials should discuss options with their own oncology team. This article is general information only, not medical advice.
Second, the standard of care has not changed. Surgery and chemotherapy remain the backbone of treatment for resectable pancreatic cancer, and follow-up after treatment remains essential. A promising phase 2 result does not replace those steps, and it does not mean patients who did not receive the vaccine missed a proven benefit.
Third, the right way to read future headlines is to ask three questions: did the trial meet its main endpoint, was the benefit seen in all patients or only a subgroup, and has overall survival been shown? For AMPLIFY-7P, the answers today are no, mainly subgroups and exploratory signals, and not yet. A phase 3 trial that answers yes to the first question would change the story fundamentally.
The bottom line
AMPLIFY-7P did not prove that a KRAS vaccine delays pancreatic cancer recurrence. It did prove something almost as important for science: the immune system can be taught, reliably and durably, to recognise the mutation that drives most pancreatic cancers. The next trial must now show whether that education saves time, and lives. Until then, the honest summary is hopeful but unproven — and that honesty is what makes the science worth following.