mRNA cancer vaccines: teaching the immune system to recognise a tumour

On 19 August 2026 something happened that oncology had been waiting decades for: Moderna and Merck announced that their individualized mRNA vaccine — intismeran autogene (previously known as mRNA-4157) — given together with pembrolizumab met the endpoints of a phase 3 clinical trial in patients after surgical removal of high-risk melanoma. It is the first positive phase 3 result in history for an mRNA-based cancer therapy. In this article we explain — step by step, without jargon — how such a vaccine actually works.

A vaccine that treats rather than prevents

The word "vaccine" misleads. The vaccines we know — against measles, flu or COVID-19 — are given to healthy people so they don't fall ill. A therapeutic cancer vaccine works the other way round: it is given to someone already ill, and its job is to help the body fight the existing disease and prevent it from coming back.

A: a preventive vaccine — a shield raised before disease appears. B: a therapeutic vaccine — a precision weapon aimed at an existing tumour.

Only the underlying mechanism is shared: in both cases we teach the immune system to recognise the enemy by its distinguishing marks.

First, know your enemy: biopsy and sequencing

An individualized mRNA vaccine starts not in a factory but with a specific patient's tumour. A fragment of the removed cancer (plus a sample of healthy tissue for comparison) goes to a laboratory, where the DNA of the tumour cells is read out — "sequenced".

1: the tumour sample from surgery. 2: reading the DNA of the cancer cells. 3: comparison with healthy DNA — the letters that differ in the tumour, i.e. mutations, are highlighted in red.

Cancer arises through mutations — typos in DNA. By comparing the tumour sequence with healthy tissue, a computer finds those typos. An average melanoma carries hundreds or even thousands of them; most are meaningless, but some change the proteins the cell produces.

Neoantigens: the tumour's fingerprints

A mutated protein can be chopped into fragments and displayed on the cell surface like a flag. Such fragments — neoantigens — appear exclusively on tumour cells, because only they carry the given mutation. For the immune system this is a potentially perfect target: an "I am foreign" sign that no healthy cell has.

The golden flags on the tumour-cell surface are neoantigens. Algorithms sift through hundreds of candidates and pick up to 34 that will stimulate the immune response most strongly.

The problem is that a tumour is usually good at hiding those flags or dampening the reaction to them. This is where vaccine design comes in: out of all detected mutations, algorithms (including machine learning) select up to 34 neoantigens — ones the patient's cells will genuinely display and lymphocytes will most likely recognise. The list is different for every patient; that is why the therapy is called individualized.

A recipe instead of a protein: mRNA in a lipid capsule

Rather than manufacturing those 34 protein fragments in a lab, they are encoded in a single mRNA molecule — a molecular "recipe" that human cells know how to read and execute. It is the same platform the world came to know through the COVID-19 vaccines, except that here the recipe is bespoke, written for one patient.

The mRNA strand carrying the neoantigen recipe is enclosed in a lipid nanoparticle (LNP) — a microscopic ball of fat that protects the recipe and delivers it into cells.

mRNA on its own is fragile, so it is packed into a lipid nanoparticle (LNP) — a fatty envelope that protects the contents and helps them slip inside cells after an intramuscular injection.

Training the army: dendritic cells and T lymphocytes

After the injection, the recipe reaches — among others — dendritic cells, the immune system's professional "instructors". A dendritic cell reads the mRNA, produces the neoantigen fragments and mounts them on its surface in special holders (MHC molecules), showing them to T lymphocytes like a wanted poster.

A dendritic cell (purple) reads the mRNA and then presents the neoantigen (gold) to a T lymphocyte (green). From this moment on, the lymphocyte "knows" its target.

A T cell that recognises the presented neoantigen becomes activated and starts multiplying. Single cells grow into a trained army whose only task is to find and destroy cells bearing the same flags — the tumour's cells.

The attack — and releasing the brake

The trained T cells patrol the body and destroy any cells with the neoantigens they meet — including microscopic deposits invisible on any scan. Those deposits are what drives relapse after surgery; the vaccine is meant to catch them before they regrow.

The T-cell army attacks tumour cells carrying the recognised neoantigens. In parallel, the anti-PD-1 drug (pembrolizumab) "opens the padlock" — it blocks the PD-1 brake with which the tumour silences the immune response.

Tumours, however, have one more defensive trick: they press a molecular PD-1 brake on lymphocytes, putting the attackers to sleep. That is why the vaccine is given together with pembrolizumab — an anti-PD-1 antibody that blocks this brake. The duo works like navigation plus a released handbrake: the vaccine points at the target, the checkpoint inhibitor lets the response run at full power.

What exactly did phase 3 show?

The INTerpath-001 trial enrolled 1,137 patients after complete resection of stage IIB–IV melanoma — a group at high risk of relapse. They were randomised 2:1 to receive intismeran with pembrolizumab, or pembrolizumab alone (the current standard of adjuvant care).

Schematic view: 1,137 patients randomised 2:1; the combination (gold line) extended recurrence-free survival (RFS) versus pembrolizumab alone (blue line). The curves are illustrative — full numerical data will be presented at a scientific congress. Timeline: trial phases, with the positive phase 3 (✓) in August 2026.

The companies announced that the trial met its primary endpoint — prolonging recurrence-free survival (RFS) — and the key secondary endpoint of distant metastasis-free survival (DMFS), with an improvement described as "statistically significant and clinically meaningful". The detailed numbers will be presented at an upcoming major oncology congress; regulatory filings come next. The earlier phase 2 (KEYNOTE-942) had shown roughly a 44% reduction in the risk of recurrence or death and about a 65% reduction in the risk of distant metastasis versus pembrolizumab alone.

Honest caveats

The excitement deserves a frame of facts:

• This is not "a cure for cancer." The result concerns one indication: adjuvant treatment after surgery for high-risk melanoma. Trials in lung, kidney and other cancers are still ongoing.
• The therapy is not yet approved. Regulators (FDA, EMA) must review the full data before it reaches practice.
• Manufacturing is demanding: each vaccine is produced separately, for a single patient, on a timescale of weeks — a logistical and cost challenge that will decide real-world availability.
• Not every patient responds — work is ongoing to predict who benefits most.

Caveats aside, the direction is genuinely groundbreaking: for the first time, a therapy tailored to the genome of a specific tumour has confirmed its efficacy in a phase 3 trial. Personalised immunotherapy has stopped being a promise — it has become data.

This article is educational and does not constitute medical advice. Always make treatment decisions with your oncology team.

Sources

1. Merck & Moderna — phase 3 INTerpath-001 results press release (19 Aug 2026): merck.com
2. STAT News: mRNA cancer vaccine succeeded in late-stage melanoma trial (19 Aug 2026): statnews.com
3. Chemical & Engineering News: In a first, an mRNA cancer vaccine succeeds in late-stage clinical trial (Aug 2026): cen.acs.org
4. National Cancer Institute — Immunotherapy to Treat Cancer: cancer.gov
5. Weber J.S. et al., Individualised neoantigen therapy mRNA-4157 (V940) plus pembrolizumab versus pembrolizumab monotherapy in resected melanoma (KEYNOTE-942), The Lancet 2024: thelancet.com