IO102-IO103
IO102-IO103, two peptide immune modulatory cancer vaccine targeting IDO1 and PD-L1 expressing cells
Written by Aaron CuhaReviewed Sep 2026
Also known as: IO102, IO103, IDO/PD-L1 peptide vaccine
A subcutaneous vaccine of two peptides from IDO1 and PD-L1, given with pembrolizumab for advanced melanoma. In a 407 person phase 3, median progression free survival was 19.4 versus 11.0 months, hazard ratio 0.77 (95% CI 0.58 to 1.00), p = 0.0558 against a threshold of 0.045: the primary endpoint was missed.
Overview
Most cancer vaccines try to raise T cells against the tumour. IO102-IO103 does something different: its two peptides come from indoleamine 2,3-dioxygenase 1 (IDO1) and programmed death ligand 1 (PD-L1), proteins expressed by the immunosuppressive cells that protect a tumour, so the T cells it raises attack the tumour's defences rather than the tumour itself. It is given subcutaneously at 85 mcg of each peptide alongside a PD-1 blocking antibody. IO Biotech developed it; the founding science and the first trial came from the melanoma immunotherapy group at Copenhagen University Hospital, whose lead investigator founded the company.
The human evidence runs from a striking small trial to a near miss in a large one. MM1636, a single arm phase 1/2 in 48 people with metastatic melanoma given the vaccine with nivolumab, reported at five years of follow up a median progression free survival of 25.5 months, a figure that drew attention but has no randomised comparator behind it. The phase 3 that followed, IOB-013 / KEYNOTE-D18, randomised 407 people with untreated advanced melanoma to the vaccine plus pembrolizumab or pembrolizumab alone. Median progression free survival by blinded central review was 19.4 months against 11.0, hazard ratio 0.77 with a 95 percent confidence interval of 0.58 to 1.00 and p = 0.0558. The prespecified threshold for significance was p of 0.045 or less, and the authors state that it was missed. Longer progression free survival in the vaccine arm was seen across most subgroups, most markedly in PD-L1 negative tumours (16.6 versus 3.0 months, HR 0.54), but every subgroup analysis is exploratory. Grade 3 or worse treatment related adverse events were 14.5 versus 15.6 percent and serious adverse events 32.0 versus 32.3 percent; 56 percent of vaccinated patients had injection site reactions, mostly mild. Separately, an academic phase 2 in head and neck cancer at Saint-Luc in Brussels was terminated with the reason unmet primary endpoint after 17 participants, and a 15 person window of opportunity study in the same disease reported a peptide specific T cell response in one patient.
IO102-IO103 is not approved anywhere. What happens next depends on whether regulators or a further trial will act on a hazard ratio whose confidence interval touches 1.00.
Mechanism of action
In people, vaccination with IO102 and IO103 raises circulating T cells that recognise IDO1 and PD-L1 derived peptides; this has been measured by ELISpot in the phase 1/2 and window of opportunity studies, and the 2025 Nature Communications paper reports serum increases in CCL3, CCL4 and TNF-alpha in responders that were not seen in a matched anti-PD-1 monotherapy cohort. The proposed downstream effect, killing or reprogramming of IDO1 and PD-L1 expressing myeloid cells and tumour cells in the tumour microenvironment, is supported by functional assays on patient samples and by spatial transcriptomics of resected head and neck tumours showing more inflammation related gene expression in vaccinated patients. The rationale for pairing it with PD-1 blockade is that the antibody releases the brake on existing T cells while the vaccine removes the cells applying it. That is a mechanism in the sense that each step has been observed in patients; whether the steps add up to longer survival is what the phase 3 was designed to test.
Human evidence
Substantial for an investigational vaccine: a 407 person randomised phase 3 with a published result, a 48 person single arm phase 1/2 with five year follow up, and two small head and neck studies. The phase 3 missed its primary endpoint by a margin of 0.011 on the p value, with a hazard ratio whose upper confidence bound is exactly 1.00.
- Phase 3 IOB-013 / KEYNOTE-D18 (2026): 407 patients, first line advanced melanoma, vaccine plus pembrolizumab versus pembrolizumab. Median PFS 19.4 versus 11.0 months, HR 0.77 (0.58 to 1.00), p = 0.0558; threshold 0.045 missed. Exploratory subgroups favoured the vaccine, most strongly in PD-L1 negative tumours (HR 0.54). Toxicity not increased beyond injection site reactions in 56 percent. Sponsor funded.
- MM1636 phase 1/2 (2025 five year report): 48 patients, single arm, vaccine plus nivolumab, median PFS 25.5 months. No comparator. Academic trial with sponsor involvement in authorship.
- HN1901 window of opportunity: 15 head and neck cancer patients, T cell response in 1, tumour shrinkage in 2 of 5 IO103 patients, safe over three weeks.
- NCT04445064: academic phase 2 in head and neck cancer, 17 enrolled, terminated for unmet primary endpoint.
- Overall survival from the phase 3 has not been reported in the publication.
What this does not tell you: The phase 3 result is compatible with a real but modest benefit and with no benefit; that is what a confidence interval running from 0.58 to 1.00 means, and the trial's own significance rule says it did not demonstrate one. The subgroup findings, including the PD-L1 negative result, are exploratory and would each need their own trial. The trial was open label, which can affect the timing of progression assessments even with blinded central review, and it measured progression, not survival. The 25.5 month figure from MM1636 comes from 48 people with no control arm and cannot be compared with the phase 3 arms. Nothing here applies to any cancer other than melanoma, and the head and neck data are a null phase 2 and a 15 person descriptive study.
Reading the research record
IO102-IO103 is the rare investigational peptide with a completed, published, adequately powered randomised trial, and the trial produced the hardest kind of result to act on. A hazard ratio of 0.77 would be clinically meaningful in first line melanoma if real; a p value of 0.0558 against a threshold of 0.045 means the trial did not establish that it is. The authors report the miss plainly in the abstract, which is to their credit, and then describe the combination as showing potential benefit, which is the sponsor's interest speaking; IO Biotech funded the trial and the author list declares relationships across the industry. The company's founder and employees also appear on the MM1636 paper whose 25.5 month median set expectations for the phase 3.
What the record does not contain is an independent replication or overall survival. Regulators sometimes accept a near miss with supportive secondary data, and sometimes ask for another trial; either path is defensible on these numbers, and this page does not pick one. The PD-L1 negative subgroup, where checkpoint inhibitors alone do worst and where the vaccine's exploratory hazard ratio was 0.54, is the obvious population for a confirmatory study. The separate academic phase 2 in head and neck cancer, terminated for an unmet primary endpoint, is a reminder that the mechanism does not transfer automatically between tumour types. There is no consumer or grey market angle: this is a hospital administered oncology product that has never existed outside trials.
The evidence, charted
Fig. 1 · evidence composition
6of 6 citations (100%) are in people
Every citation cited here is Human work; no other study type is cited on this page. This count is of our own citation list and understates any literature larger than the sources we cite.
Fig. 2 · evidence over time
Evidence spans 5 distinct years, 2018 to 2026, counted from the citation list on this page.
Fig. 3 · legal status at a glance
US
Not approved
UK
Approved
AU
Approved
CA
Not approved
Approved in 2 of 4, prescription route in 0, not approved in 2. A jurisdiction's classification is a regulatory fact, not a verdict on the science; see Legal status below for the full text and any notes.
Key studies & citations
- Human2026
IO102-IO103 immune-modulatory cancer vaccine and pembrolizumab in melanoma
Open label phase 3, 407 patients with untreated advanced melanoma randomised 1:1 to IO102-IO103 (85 mcg each) plus pembrolizumab or pembrolizumab alone for up to 2 years. Median progression free survival 19.4 versus 11.0 months, HR 0.77 (95% CI 0.58 to 1.00), p = 0.0558; the prespecified threshold of p at most 0.045 was missed. Grade 3 or worse treatment related adverse events 14.5 versus 15.6 percent; injection site reactions in 56 percent. Sponsored by IO Biotech; the authors declare extensive industry relationships.
Annals of Oncology - Human2022
IO102-IO103 in Combination With Pembrolizumab Versus Pembrolizumab Alone in Advanced Melanoma, IOB-013 / KN-D18 (NCT05155254)
Registry record for the phase 3: 407 enrolled, primary completion 30 May 2025, status active, not recruiting, as of September 2026. Sponsor funded.
ClinicalTrials.gov, IO Biotech - Human2025
Five-year clinical outcome and immune biomarkers of durable response from the MM1636 trial on IDO/PD-L1 vaccination and PD-1 blockade in first line metastatic melanoma
Single arm phase 1/2 (NCT03047928), 48 patients with metastatic melanoma given the vaccine with nivolumab. Median progression free survival 25.5 months at five year follow up. Rises in serum CCL3, CCL4 and TNF-alpha tracked long progression free survival and were not seen in a matched anti-PD-1 monotherapy cohort. No randomised comparator. Investigator initiated at Copenhagen University Hospital; authors include IO Biotech's founder and employees.
Nature Communications - Human2025
Translational investigations in the HN1901 phase II window-of-opportunity study of IDO1 (IO102) and PD-L1 (IO103) peptide vaccines in squamous cell carcinoma of the head and neck
15 patients: 10 randomised to IO102 or no preoperative treatment, 5 more given IO103, for three weeks before curative surgery. A peptide specific T cell response was detected in one IO103 vaccinated patient; two IO103 patients had tumour shrinkage. No grade 3 or 4 events attributed to the vaccine. Very small, largely descriptive.
European Archives of Oto-Rhino-Laryngology - Human2020
Activity and Safety of Peptide-based Immunotherapy in Patients With Squamous Cell Carcinoma of the Head and Neck (NCT04445064)
Academic phase 2 in head and neck cancer, 17 enrolled, terminated. Reason recorded by the sponsor: unmet primary endpoint.
ClinicalTrials.gov, Cliniques universitaires Saint-Luc - Human2018
Combination Therapy With Nivolumab and PD-L1/IDO Peptide Vaccine to Patients With Metastatic Melanoma, MM1636 (NCT03047928)
Phase 1/2, 48 enrolled, completed December 2022, results posted February 2025. Investigator sponsored (Inge Marie Svane).
ClinicalTrials.gov, Copenhagen University Hospital
Frequently asked questions
Did the IO102-IO103 phase 3 succeed?
No, by its own rule. Median progression free survival was 19.4 months with the vaccine plus pembrolizumab versus 11.0 months with pembrolizumab alone in 407 patients, hazard ratio 0.77 with a 95 percent confidence interval of 0.58 to 1.00 and p = 0.0558. The trial required p of 0.045 or less, and the authors state the threshold was missed.
Was the vaccine toxic?
Not systemically, on the published numbers: grade 3 or worse treatment related adverse events were 14.5 percent with the vaccine versus 15.6 percent without, and serious adverse events 32.0 versus 32.3 percent. Injection site reactions occurred in 56 percent of vaccinated patients, mostly grade 1 or 2.
What about the PD-L1 negative result?
In patients whose tumours were PD-L1 negative, median progression free survival was 16.6 months with the vaccine versus 3.0 months without, hazard ratio 0.54 (95% CI 0.35 to 0.85). It is an exploratory subgroup analysis from a trial that missed its primary endpoint, and it would need a trial of its own to be established.
Is IO102-IO103 approved or available?
No. It is investigational, has only ever been given in hospital trials with a PD-1 antibody, and has no existence outside them.