Neoantigen peptide vaccines

NeoVax, NEO-PV-01, personalised neoantigen long-peptide vaccine, patient-specific synthetic long peptide vaccine; related mRNA context: autogene cevumeran (BNT122/RO7198457), individualised neoantigen therapy (mRNA-4157/V940)

Oncology & imaging

Awaiting medical reviewEN ES DE FR PT ZH+ Follow for updates

Note on this profile. These are investigational, individually manufactured immunotherapies given only inside clinical trials under oncology supervision. Nothing here describes self-preparation.

Summary

Class: Personalised mixtures of synthetic long peptides (typically 15–30 amino acids, up to 20 peptides per patient) encoding mutations found by sequencing that patient's own tumour, given with a strong adjuvant — usually poly-ICLC (a TLR3 agonist)
Mechanism: Vaccinates each patient against the specific mutations in their own tumour, so the targets are sequences that exist nowhere in normal tissue and against which the immune system has not been tolerised
Routes studied: Subcutaneous and intradermal injection, split across several limb sites, on a priming-then-booster schedule over several months; poly-ICLC given with each dose
Half-life: Not the operative measure. Long peptides persist at the injection site for hours to days; the relevant durations are the T-cell response and memory, which have been detected four or more years after vaccination
Storage: Individually manufactured, frozen or refrigerated per protocol; prepared by specialist pharmacy. Manufacturing turnaround from biopsy to first dose has historically been 10–16 weeks, which is itself a clinical constraint
Regulatory status: Not approved anywhere. See region table

Doses are as reported in studies or by the community — never recommendations.

1What it is

This is the current frontier of therapeutic cancer vaccination, and the honest framing is that it is a strong idea with thin randomised evidence. Every tumour carries somatic mutations; a fraction of those produce peptides that the patient's HLA molecules can present and that T cells can recognise. These are neoantigens — genuinely foreign sequences, absent from normal tissue, against which no thymic tolerance exists. Where profiles 231 through 236 targeted shared self-antigens that the immune system has been trained to ignore, this approach manufactures a different vaccine for every patient.

The practical pipeline is: sequence the tumour and matched normal DNA, call the mutations, predict which resulting peptides will bind the patient's HLA alleles, rank them, synthesise the best 10–20 as long peptides, and vaccinate with a strong adjuvant. Long peptides are used deliberately — they must be taken up and processed by professional antigen-presenting cells rather than binding HLA directly on any passing cell, which is the mechanism by which short class I peptides (profile 231) can induce tolerance instead of immunity.

The same neoantigen concept is also being delivered as mRNA rather than peptide — autogene cevumeran and mRNA-4157 are the leading examples — and it is important not to conflate the two evidence streams. The peptide route has the longer immunological follow-up; the mRNA route has produced the first randomised efficacy signals.

2How it's thought to work

Long synthetic peptides injected with poly-ICLC are taken up by dendritic cells, processed, and presented on both class I and class II HLA.

Both CD8+ cytotoxic and CD4+ helper T-cell responses are generated; in practice the CD4+ response has usually been the larger one, which was not what the field initially expected.

Because the sequences are tumour-exclusive, the response is not limited by self-tolerance and on-target autoimmunity is mechanistically unlikely.

Over time the response broadens to mutations that were not in the vaccine — "epitope spreading" — which has been documented years after vaccination and suggests the vaccine initiates a self-sustaining process rather than only supplying the targets it contains.

The approach is best suited to minimal residual disease after surgery, and to tumours with enough mutations to find good neoantigens; the long manufacturing time makes it poorly suited to rapidly progressing advanced disease.

Peptides made to match the mutations in one patient's own tumour. Because the sequences exist nowhere else in the body, there is no self-tolerance to overcome.Schematic — not to scale

3What the research shows

Melanoma, adjuvant, peptide vaccine

The original NeoVax phase 1 study vaccinated six patients with resected stage III/IV melanoma with up to 20 personalised long peptides plus poly-ICLC. All developed neoantigen-specific T-cell responses; four were recurrence-free at follow-up and two who later progressed responded to subsequent checkpoint blockade (Ott 2017, Nature). A four-year follow-up found neoantigen-specific memory T cells still circulating, with epitope spreading to mutations not included in the vaccine, and all eight patients in the extended cohort alive (Hu 2021, Nature Medicine). Longer follow-up continues to report persistence. This is a six-to-eight patient uncontrolled study with exceptional immunological depth — grade B, and the small numbers must be stated every time it is cited. [1][2]

Glioblastoma, peptide vaccine

A phase 1/1b study of personalised neoantigen peptide vaccination in newly diagnosed glioblastoma showed that patients not on dexamethasone generated circulating neoantigen-specific T cells, and that those T cells could be found infiltrating the tumour on repeat sampling (Keskin 2019, Nature). Importantly, patients on dexamethasone largely did not respond — a concrete demonstration of the steroid problem that limits all glioblastoma immunotherapy. No survival claim. [3]

Combination with checkpoint blockade

NEO-PV-01 plus nivolumab was studied in advanced melanoma, NSCLC and bladder cancer, reporting feasibility, immunogenicity and epitope spreading without a randomised comparison. Other small studies have combined personalised peptide vaccines with anti-PD-1 in several tumour types. Consistently immunogenic, consistently uncontrolled.

Graded A/B

Randomised evidence, from the mRNA route

The randomised data in this field so far comes from mRNA neoantigen therapies, not peptides. A randomised phase 2b of individualised mRNA neoantigen therapy plus pembrolizumab versus pembrolizumab alone in resected high-risk melanoma reported improved recurrence-free survival; a phase 3 programme is running. In pancreatic cancer, a phase 1 study of autogene cevumeran after surgery, with atezolizumab and mFOLFIRINOX, induced neoantigen-specific T cells in about half of the 16 patients and reported longer recurrence-free survival in those who responded, with follow-up reports out to several years describing persistent vaccine-induced T-cell clones; a randomised phase 2 is ongoing and has not reported (Rojas 2023, Nature; later follow-up reports 2025–2026 †). None of this is peptide-vaccine evidence, and it is listed here because it is the context in which the peptide work is now read. [5]

Not graded

What has not been shown — stated plainly

There is no completed of a personalised neoantigen peptide vaccine demonstrating improved survival or recurrence-free survival in any cancer. Every clinical claim in this profile rests on single-arm studies of between six and about sixty patients, or on a different delivery technology.

4Reported side effects and interactions

Literature-reported

the safety profile across studies has been notably mild for an oncology intervention — injection-site reactions (common, sometimes with persistent nodules), flu-like symptoms, fatigue, low-grade fever, myalgia, transient rash. Poly-ICLC accounts for most of the systemic reactogenicity. No vaccine-attributed grade 3–4 immune toxicity has been consistently reported, and no autoimmune disease caused by neoantigen vaccination has been described. When given with checkpoint inhibitors, immune-related adverse events occur at the rate expected from the checkpoint inhibitor.

Community-reported

Community use: essentially none documented. These products cannot exist outside a sequencing-and-manufacturing pipeline tied to an individual patient's tumour.

Theoretical. a predicted neoantigen could cross-react with a similar self-peptide and cause autoimmunity; this has been a theoretical concern throughout and has not materialised. Epitope spreading means the eventual response is wider than what was deliberately administered, which is the intended benefit and also the least controllable aspect.

Interactions. corticosteroids, especially dexamethasone, substantially blunt the response and this has been directly observed in the glioblastoma data — the most clinically actionable interaction in this profile. Cytotoxic chemotherapy and lymphodepletion reduce the available T-cell pool. Checkpoint inhibitors are studied as deliberate partners.

Markers researchers commonly monitor

From trial protocols:

Marker	Why it's watched
Tumour mutational burden and whole-exome/RNA sequencing output	determines whether a usable vaccine can be designed at all; low-mutation tumours yield few candidate neoantigens.
HLA genotype (class I and II)	required for epitope prediction, not as a restriction on eligibility.
Neoantigen-specific CD4+ and CD8+ T-cell frequency over time (ELISpot, tetramer, single-cell TCR sequencing)	the pharmacodynamic readout, and the main scientific output of this field to date.
Epitope spreading (responses to non-vaccine mutations)	tracked as evidence that the vaccine initiated a broader process.
T-cell receptor clonality in blood and in post-vaccination tumour tissue	used to show vaccine-induced clones actually reach the tumour; the strongest mechanistic evidence available.
Circulating tumour DNA	the practical minimal-residual-disease endpoint in adjuvant trials, and probably the surrogate that will make randomised trials feasible.
Corticosteroid exposure, particularly dexamethasone	tracked as a covariate because it demonstrably suppresses the response.
Recurrence-free and overall survival with protocol imaging	the endpoints that have not yet been met in a randomised peptide-vaccine trial.
CBC, CMP, thyroid function and the standard immune-related adverse event panel	when a checkpoint inhibitor is part of the regimen.

Free for everyone. Safety information is never behind a paywall.

5Formulation and administration context

Each vaccine is manufactured for one patient: tumour and normal tissue are sequenced, candidate neoantigens predicted and ranked, peptides synthesised to clinical grade, quality-tested and released — historically a 10–16 week process, and a central reason this approach is studied after surgery rather than in progressive disease. Doses are given subcutaneously or intradermally by trained staff, with the peptide pool split into several injection groups across different limbs so that multiple draining lymph-node basins are engaged, each with poly-ICLC. Priming doses are front-loaded over the first weeks with boosters months later. Steroid use is restricted by protocol wherever possible. There is no self-prepared, compounded or off-the-shelf version of a personalised vaccine, by definition, and no dose calculation appropriate to publish.

6Commonly studied alongside

Poly-ICLC (the standard adjuvant in this class); pembrolizumab, nivolumab and atezolizumab; mFOLFIRINOX and other adjuvant chemotherapy backbones. Conceptually related: KRAS-targeted peptide vaccines (profile 237) — the shared-neoantigen version of the same idea; IMA901 and GAPVAC (profile 236) — the warehouse-personalised predecessor; gp100 peptide vaccine (profile 231) — the short-peptide approach this class was designed to replace.

Informational cross-links. Not stack recommendations.

7Regulatory status by region

Region	Status (checked —)
United States	Not FDA-approved; investigational. Breakthrough therapy designation has been reported for an mRNA individualised neoantigen therapy in melanoma, which is a different product class from the peptide vaccines profiled here †
European Union	Not authorised
United Kingdom	Not licensed. A national programme to enrol patients into personalised cancer vaccine trials has been established †
Australia	Not approved
South Africa	Not registered with SAHPRA
Sport	Not listed by WADA; no performance application

8Open questions

Will a randomised trial of a personalised neoantigen peptide vaccine ever be run, or has the field moved decisively to mRNA delivery?
How accurate is neoantigen prediction? A large fraction of predicted epitopes generate no detectable response, and the field still cannot reliably pick the ones that matter.
Can manufacturing time be reduced enough to treat anything other than minimal residual disease?
Why is the response so CD4-dominated, and does that matter for killing tumour cells?

9References

1Ott PA et al. (2017). An immunogenic personal neoantigen vaccine for patients with melanoma. Nature. PubMed · DOI
2Hu Z et al. (2021). Personal neoantigen vaccines induce persistent memory T cell responses and epitope spreading in patients with melanoma. Nature Medicine. PubMed · DOI
3Keskin DB et al. (2019). Neoantigen vaccine generates intratumoral T cell responses in phase Ib glioblastoma trial. Nature. PubMed · DOI
4Sahin U et al. (2017). Personalized RNA mutanome vaccines mobilize poly-specific therapeutic immunity against cancer. Nature. PubMed · DOI
5Rojas LA et al. (2023). Personalized RNA neoantigen vaccines stimulate T cells in pancreatic cancer. Nature. PubMed · DOI
6Ott PA et al. (2020). A phase Ib trial of personalized neoantigen therapy plus anti-PD-1 in patients with advanced melanoma, non-small cell lung cancer, or bladder cancer. Cell. PubMed · DOI
7Weber JS et al. (2024). Individualised neoantigen therapy mRNA-4157 (V940) plus pembrolizumab versus pembrolizumab monotherapy in resected melanoma (KEYNOTE-942): a randomised, phase 2b study. Lancet. PubMed · DOI

† Being confirmed by our reviewers against current sources.

10Community

forum threads →

community protocols include this →

Threads and protocols are visible to members. Joining is free.

Self-reported side effects Phase 2

Not enough data yet. 0 of the 50 members needed have logged this compound for 4+ weeks. We don't show anything below 50, so no one can be identified.

11Page history

31 Oct 2026Profile published — awaiting medical review

Licensed CC BY-NC-SA 4.0. Found an error? Report it · Editorial standards — corrections are logged here.