Peptide News Digest
Evidence Brief 11 min read

Galinpepimut-S Phase 3 REGAL Results: An Off-the-Shelf WT1 Peptide Vaccine Extended Survival in AML

SELLAS says its four-peptide vaccine cut the risk of death 41% in an open-label trial of 126 people with acute myeloid leukemia in second or later remission. This guide covers how peptide cancer vaccines work, what the topline release leaves out, the field's record of Phase 3 failures, and the peptide vaccines still in late-stage testing.

Key points

  • In REGAL (126 adults with AML in second or later remission who could not have a donor stem cell transplant), median overall survival was 24.0 months with galinpepimut-S maintenance versus 14.7 months with investigator's choice of best available treatment, according to SELLAS's November 17, 2026 release.
  • SELLAS reported a hazard ratio of 0.59 (described as a 41% reduction in the risk of death) with a two-sided p-value of 0.035, but no confidence interval.
  • The trial was open-label. SELLAS cut the deaths required for the final analysis from 105 to 80 in 2022 after blinded pooled survival ran about twice as long as planned, and the control arm's 14.7 months exceeded the 8 months the revised plan assumed.
  • GPS combines two natural and two modified (heteroclitic) WT1 peptides. WT1 is found in more than 95% of AML cases, according to MSK, which invented the vaccine and has a financial interest in SELLAS.
  • Several large peptide vaccine trials have failed, including tecemotide in lung cancer, rindopepimut in glioblastoma, nelipepimut-S in breast cancer, IMA901 in kidney cancer, and the WT1 vaccine DSP-7888 in glioblastoma.
  • SELLAS plans to meet with the FDA and targets a BLA submission in the first half of 2027. GPS is not approved anywhere.

REGAL's topline result: 24.0 months versus 14.7

On November 17, 2026, SELLAS Life Sciences reported that its Phase 3 REGAL trial of galinpepimut-S (GPS) met its primary endpoint, overall survival. REGAL enrolled 126 adults with acute myeloid leukemia (AML) whose cancer had come back and then gone into a second or later complete remission after salvage treatment. All of them were unable to have, or not eligible for, an allogeneic stem cell transplant, the donor transplant that gives people with relapsed AML their best chance of a cure.

Patients were randomized 1:1. One group received GPS alone as maintenance therapy, a vaccine made of four peptides taken from a protein called WT1. The other group received whatever their doctor judged the best available treatment: observation, a hypomethylating agent such as azacitidine, venetoclax, and/or low-dose cytarabine.

According to the company's release, filed with the SEC, median overall survival was 24.0 months on GPS and 14.7 months in the control group. The hazard ratio was 0.59, which SELLAS describes as a 41% reduction in the risk of death, with a two-sided p-value of 0.035. Side effects were mostly grade 1 or 2 (mild to moderate), and the most common were reactions at the injection site. SELLAS plans to meet with the FDA about what it calls an "accelerated filing strategy" and is preparing a Biologics License Application (BLA), the filing needed to approve a biologic drug, with submission targeted for the first half of 2027.

GPS was discovered in David Scheinberg's lab at Memorial Sloan Kettering Cancer Center (MSK), which licenses it to SELLAS. MSK and Scheinberg have financial interests in SELLAS, and both MSK and the company disclose this. "These data bring us closer to having the first off-the-shelf vaccine for AML patients," Scheinberg said in MSK's announcement. News coverage is collected at #galinpepimut-s.

How a vaccine can treat cancer that is already there

A flu shot is preventive: it trains the immune system before the virus arrives. A therapeutic cancer vaccine goes to someone who already has cancer, and its job is to train T cells, the immune system's hunters, to find and kill cancer cells. REGAL gave GPS during remission, when leftover leukemia cells may be too few to detect but can still cause a relapse.

T cells do not see whole proteins. Every cell chops some of its proteins into short fragments called peptides and displays them on its surface in holders called MHC molecules (in humans, HLA). Killer T cells, also called CD8+ T cells, check fragments held by class I molecules. Helper T cells (CD4+) respond to fragments on class II molecules and help sustain the attack. A peptide vaccine injects synthetic copies of chosen fragments under the skin, usually mixed with an adjuvant, an ingredient that draws immune attention. In REGAL, GPS was given with Montanide, an oil-based adjuvant, and GM-CSF, an immune-stimulating protein, according to the trial's registry entry.

WT1 (Wilms tumor 1) is in the leukemia cells of more than 95% of people with AML, according to MSK, and is also found in ovarian cancer, mesothelioma, and other cancers. A 2025 paper on the REGAL design, co-written by a SELLAS employee, describes WT1 as overexpressed in leukemic blast cells and nearly absent in normal blood-forming cells. MSK's 2018 Phase 2 report cites a National Cancer Institute project that ranked WT1 first among cancer immunotherapy targets.

The catch is that WT1 is one of the body's own proteins, and the immune system is trained not to attack "self." MSK researchers worked around this by changing one or two amino acids in some WT1 fragments so they would sit more firmly in HLA molecules and provoke a stronger response. T cells raised against these altered peptides also recognized natural WT1. GPS combines two of these modified ("heteroclitic") peptides with two natural WT1 sequences, and the longer peptides were chosen to engage both CD4+ and CD8+ T cells.

The earlier human evidence was small. In a single-arm Phase 2 study at MSK, 22 adults with AML in first remission received GPS; 9 of 14 patients tested (64%) showed an immune response, and median disease-free survival from remission was 16.9 months. Without a control group, that study could not show that the vaccine extended anyone's life. REGAL was designed to answer that question.

Off the shelf or built for one patient

GPS is the same product for every patient. MSK notes that an off-the-shelf vaccine can be mass-produced, stored, and given right away. Its target, WT1, is shared by most AML cases, so no tumor sequencing is needed to make it.

Personalized vaccines take the opposite route. Merck and Moderna's intismeran autogene starts by sequencing a patient's tumor, then encodes up to 34 neoantigens (pieces of mutated proteins found only in that tumor) in an mRNA molecule made for that one person. On August 19, 2026, the companies said intismeran plus Keytruda (pembrolizumab) beat Keytruda alone on recurrence-free survival in the 1,137-patient Phase 3 INTerpath-001 trial in surgically removed stage IIB-IV melanoma. Full results have not been published.

Neoantigens look foreign to the immune system, which may make them easier targets, but each batch has to be built from one patient's tumor. WT1 vials can sit on a shelf, at the cost of asking the immune system to attack a self protein.

Many peptide vaccines have another limit: they only fit certain HLA types. Greenwich LifeSciences' FLAMINGO-01 trial at first randomized only patients with the HLA-A*02 type, and OSE Immunotherapeutics' Tedopi trial enrolls only HLA-A2-positive patients. A 2023 MSK paper describes GPS as "non-HLA restricted," meaning it is meant to work regardless of a patient's HLA type. More background is on the #cancer-vaccine and #neoantigen tag pages.

Why the topline numbers need a second look

REGAL was open-label: patients and their doctors knew who got the vaccine. Death is an objective outcome, so overall survival is harder to bias than endpoints that depend on someone's judgment. An open-label design can still influence how closely patients are followed, who stays in the study, and which later treatments they receive. The control arm was "investigator's choice," a mix of observation and several drugs, and the release does not say how many control patients got each one.

A hazard ratio (HR) compares how often deaths occur over time in one group versus the other. An HR of 0.59 means that at any given point during follow-up, patients on GPS were dying at about 59% of the rate of the control group. It does not mean 41% fewer patients died overall.

SELLAS reported the HR and the p-value but no confidence interval, the range that shows how precise the estimate is. Tech Times flagged the gap, and the SEC filing confirms it. Without it, readers can't tell how close the result came to showing no difference. A p-value of 0.035 in a trial of 126 patients and 80 deaths points to real uncertainty about the size of the benefit, and the full data at a medical meeting or in a journal should settle that.

The trial's statistical plan also changed. In November 2022, after a blinded review of pooled data showed median survival across both arms running about twice as long as the design assumed, SELLAS cut the deaths needed for the final analysis from 105 to 80, set an interim analysis at 60 deaths, and raised enrollment from 116 to between 125 and 140. CEO Angelos Stergiou said then that the cause of the longer survival was unknown and could include benefit from GPS or better care for control patients. The revised plan still assumed the control arm would live a median of 8 months, and in December 2025 REGAL steering committee member Yair Levy cited "around eight months" as the expected survival for transplant-ineligible patients on a hypomethylating agent and/or a BCL-2 inhibitor such as venetoclax. The final analysis was expected by the end of 2024; SELLAS announced the 80th death on October 28, 2026.

REGAL's control group reached 14.7 months, and the GPS group's median was 9.3 months longer still. The distance between the plan's assumptions and what happened is one more reason to see the full dataset before drawing firm conclusions.

Survival curves, median follow-up, and all secondary endpoints are still unpublished. Those include leukemia-free survival, T-cell responses, and measurable residual disease (leukemia found only with sensitive lab tests). SELLAS says they will come at a medical meeting and in a peer-reviewed paper.

Peptide vaccines have missed in Phase 3 before

REGAL's result stands out partly because so many large peptide vaccine trials failed. A few examples from the published record:

  • Tecemotide (START trial, lung cancer): In a double-blind trial of 1,513 people with stage III non-small-cell lung cancer, Merck KGaA's MUC1 lipopeptide produced median survival of 25.6 months versus 22.3 on placebo (HR 0.88, p=0.123), short of significance.
  • Rindopepimut (ACT IV trial, glioblastoma): This EGFRvIII peptide vaccine, given with temozolomide to 745 people with newly diagnosed glioblastoma, was stopped for futility. Median survival was 20.1 months versus 20.0 months in the main analysis group.
  • Nelipepimut-S (PRESENT trial, breast cancer): A HER2-derived peptide vaccine in 758 women with node-positive, HER2-low breast cancer was stopped early when an interim analysis found no difference in disease-free survival.
  • IMA901 (IMPRINT trial, kidney cancer): Adding this multi-peptide vaccine to sunitinib in an open-label trial of 339 patients did not improve survival.
  • DSP-7888 (glioblastoma): Sumitomo Dainippon Pharma Oncology's WT1 peptide vaccine, the same target as GPS, was stopped in December 2021 after an interim analysis in recurrent glioblastoma. Posted registry results show median survival of 10.2 months with the vaccine plus bevacizumab versus 9.4 months with bevacizumab alone.

Recent attempts have come close or fallen short. IO Biotech's IO102-IO103, built from IDO and PD-L1 peptides, was added to Keytruda in an open-label Phase 3 trial of 407 people with advanced melanoma. Median progression-free survival was 19.4 months versus 11.0 with Keytruda alone, but the p-value of 0.056 missed the 0.045 threshold, and in September 2025 the FDA advised the company not to file. In June 2026, Elicio's ELI-002 7P, a KRAS peptide vaccine, missed its disease-free survival goal in a 144-patient randomized Phase 2 trial after pancreatic cancer surgery.

None of this predicts what the FDA will do with REGAL. The record does show how often peptide vaccines that looked promising in smaller studies came up empty in large randomized trials, and REGAL, at 126 patients, is far smaller than START or ACT IV.

Where other peptide vaccines stand

Greenwich LifeSciences' GLSI-100, for HER2-positive breast cancer, pairs GP2, a 9-amino-acid HER2 peptide, with GM-CSF. Its Phase 3 FLAMINGO-01 trial is double-blind and placebo-controlled, enrolling people who still had cancer at surgery (or had a high-risk complete response) after trastuzumab-based treatment, and it measures invasive breast cancer-free survival. In July 2026, the company said its data monitoring board recommended in May that the trial continue unchanged, that an amended protocol opens randomization to all HLA types, and that interim analyses are planned at 28 and 56 events. ClinicalTrials.gov lists a primary completion date in 2031.

OSE Immunotherapeutics' Tedopi, for lung cancer, combines nine HLA-A2-restricted epitopes (peptide fragments that T cells recognize) drawn from p53, HER2, CEA, MAGE-2, and MAGE-3, plus one epitope designed to bind many HLA-DR types. The Phase 3 ARTEMIA trial compares it with docetaxel chemotherapy in 363 HLA-A2-positive patients with metastatic non-small-cell lung cancer that became resistant to checkpoint inhibitor drugs after first responding. On September 28, 2026, OSE said enrollment should finish by the end of December and that a futility analysis has moved to early 2027 because fewer deaths have occurred than expected.

Enterome's EO4010, for colorectal cancer, uses peptides from gut bacteria that mimic HLA-A2-restricted fragments of five tumor proteins, plus a helper peptide. Its Phase 1/2 AUDREY trial in previously treated metastatic colorectal cancer was terminated with 25 patients enrolled; the registry says the sponsor stopped recruitment as part of its overall development plans and that safety was not the reason.

Peptides also reach cancer through other routes, including peptide-drug conjugates and radioligand therapies such as Lutathera and Pluvicto. The #peptide-vaccine tag page collects new readouts as they arrive.

What the FDA could do next

GPS is not approved anywhere. SELLAS has not explained what it means by an "accelerated filing strategy." In FDA terms, accelerated approval lets a drug for a serious condition be approved based on a surrogate endpoint, such as a lab marker, before benefit on survival is shown. REGAL measured survival directly, so that pathway may not be the relevant one. Priority review is a separate designation that sets an FDA goal of acting on an application within 6 months instead of the standard 10.

Either way, FDA reviewers will see the patient-level data the public has not. Durability is also unknown: a median of 24 months says nothing yet about how many patients are alive at three or five years, or whether vaccine-induced T-cell responses last.

Looking for a peptide vaccine trial

GPS is unapproved, and ClinicalTrials.gov lists REGAL (NCT04229979) as active but no longer recruiting. To look for other vaccine trials, search ClinicalTrials.gov by condition, such as "acute myeloid leukemia," and add terms like "peptide vaccine," "WT1," or "cancer vaccine." Filter to recruiting studies and read the eligibility criteria closely, since peptide vaccine trials often require a specific remission status, prior treatments, a tumor that expresses the target protein, or a particular HLA type, which requires a lab test.

Each study has an NCT number to bring to your oncologist, who can judge whether a trial fits and what it would mean for your standard care. The National Cancer Institute's Cancer Information Service (1-800-4-CANCER, or LiveHelp chat on cancer.gov) answers questions about clinical trials, and NCI's website has its own trial search. Questions worth asking: Does my cancer express the vaccine's target? Is my HLA type required? Will I give up a standard treatment to join? Who pays for tests and travel?

Updates and corrections

No corrections yet.

Limitations

  • All REGAL efficacy numbers come from a company topline release. No peer-reviewed paper, survival curves, confidence interval, median follow-up, or secondary endpoint results were available when we wrote this.
  • We do not know how many control patients received each treatment option, which affects how the comparison should be read.
  • SELLAS's release says 126 patients were randomized; the ClinicalTrials.gov record lists 127 enrolled. We use the company's figure.
  • Pipeline details for GLSI-100, Tedopi, and EO4010 come mainly from company releases and trial registries, not published Phase 3 results.

Citations

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.