A "vaccine" — but not in the usual sense
Preventive vaccines prepare you before an infection. This one is built after a person already has cancer, and after doctors have analysed their tumour. The word "vaccine" is doing unusual work here: it is a treatment, not a precaution.
What it is
- A treatment for someone already diagnosed
- Designed separately for each patient
- Made from harmless, short-lived mRNA instructions
- Aimed at cancer cells left behind after surgery
- Studied together with Keytruda, never alone
What it isn't
- Not a vaccine that prevents cancer
- Not an injection containing cancer cells
- Not something that changes your DNA
- Not a replacement for surgery
- Not approved, and not proven to extend life
Cancer cells are your own cells, with typos
Every cell in your body works from the same instruction manual. A cancer cell is not an invader from outside — it is one of your cells whose copy of the manual has picked up typos. Those typos are why it grows when it shouldn't.
This is exactly why cancer is hard for your immune system. Your immune system is very good at spotting things that are foreign. A cancer cell is 99.9% you. There is almost nothing to spot.
Almost. Because when a cell builds a protein from a page with a typo in it, the protein comes out slightly misshapen — and a misshapen protein is the one thing in a cancer cell that isn't you.
Those misshapen proteins get held up outside the cell
Cells constantly hold up samples of what they're building on their outer surface — a running display of "here's what I'm making right now." It's a security system. Patrolling immune cells walk past, glance at the display, and move on if everything looks normal.
A tumour cell displays the misshapen proteins too. So the evidence is already on show. The problem is that the immune system has never been taught that those particular shapes mean danger, so it walks past.
The vaccine's entire job is to teach it what to look for.
So they read your tumour, and write a drug from it
After surgery removes the tumour, a piece of it goes to a lab along with a blood sample. Comparing the two shows which typos belong to the cancer and which were always in your normal DNA.
Software then ranks the typos: which ones actually get built into proteins, which will end up on display, and which shapes your particular immune system is best equipped to recognise. Up to 34 winners are written into a single custom instruction — the drug.
Surgery supplies tumour tissue; blood provides a healthy-DNA comparison.
Sequencing identifies mutations present in the cancer but not in healthy cells.
Software ranks which changed protein fragments the immune system may actually see.
Up to 34 targets are encoded into one patient-specific mRNA recipe.
Nine doses, three weeks apart, teach immune cells the target pattern.
The injection is a recipe, not an ingredient
Here is the part people usually get wrong. The vial does not contain the misshapen proteins. It contains mRNA — a short-lived recipe card, wrapped in a tiny bubble of fat so it survives the trip into your cells.
Your own muscle and immune cells read the recipe and build the shapes themselves, then hold them up on their surface exactly as the security system was designed to do. The immune system sees them in the place it trusts, in a body that is otherwise clearly under attack, and draws the obvious conclusion: these shapes are the enemy.
The recipe cards are then destroyed, as all mRNA is, within days. They never enter or alter your DNA. What lasts is not the drug — it's the lesson.
This is the same trick as the COVID vaccines, which is why the technology moved so fast. The difference is that a COVID vaccine's recipe is the same for eight billion people. This one's is written for you.
A target list is useless if the guards are asleep
The vaccine is never given alone. It's paired with an established immunotherapy called pembrolizumab, better known as Keytruda. They solve two completely different problems, and you need both.
Teaches immune cells which specific misshapen shapes, out of everything in the body, belong to this patient's cancer. It does nothing about the tumour's defences.
Tumours press a chemical "stand down" button on approaching immune cells. Keytruda blocks that button. It never says what to attack — it just stops the attacker being switched off.
Six weeks, and why that quietly shapes everything
Reading the tumour, choosing the targets, building the vial and testing it takes roughly six weeks. Nobody wants a cancer patient waiting six weeks doing nothing — so Keytruda starts almost immediately after surgery, and the vaccine joins in once it's ready.
That single practical fact explains several things at once. It's why the two drugs are always studied together. It's why this approach targets patients whose visible tumour has already been removed, rather than fast-moving advanced disease. And it's the hardest thing to scale: six weeks is achievable for a few hundred trial patients, and an enormous logistics problem for tens of thousands at once, each with their own vial that fits nobody else.
What actually happened to the patients
The clearest picture so far comes from a mid-stage trial of 157 people whose high-risk melanoma had been surgically removed. Two thirds got the combination; one third got Keytruda alone. Five years later, here is how many in each group were still free of any return of their cancer.
In 1,137 people with surgically removed high-risk melanoma, recurrence-free and distant-metastasis-free survival both improved.
Still missing: hazard ratios, event counts, survival curves, subgroups and complete safety results.
It is not yet proven that people live longer because of the treatment.
Melanoma was chosen because it's the easy case
Skin cancer caused by sun damage accumulates an unusually large number of typos — which means an unusually long list of targets to choose from. Melanoma also already responds to immunotherapy, so the guards are known to work once woken. And surgery yields plenty of tissue to read.
Most importantly, the drug is given after the visible tumour is gone. The immune system is being asked to find a scattering of stray cells, not to demolish a large, well-defended tumour. That is a far easier assignment.
Does the method travel?
Possibly — but "the same platform" does not mean "the same result." Every cancer type and treatment setting needs its own trial, and none of the others has reported yet.
Strongest evidence. Positive Phase 3 topline result, detailed data pending.
Phase 3Several trials running, including Phase 3 studies after surgery.
Phase 3Being tested after surgery and in high-risk early disease.
Phase 2–3Being tested after tumour removal.
Phase 2Harder: large tumours suppress immunity and contain diverse cell populations.
ExperimentalNo evidence for a universal product. Some tumours have too few visible targets, or escape recognition entirely.
Not establishedFour ways it can still fail
Enthusiasm for this result has been enormous, and some of it runs ahead of what has actually been shown. These are the real weak points.
Predicting which typos the immune system will actually recognise is still imperfect. Some chosen targets simply won't work.
Not every cancer cell carries every typo. Cells without the targeted ones can survive and regrow — or stop displaying them at all.
Some tumours build an environment that exhausts immune cells before they can act, no matter how good the target list is.
Too little tissue, a failed batch, a shipping error. A drug manufactured per person has failure modes no ordinary pill has.
And then it has to scale
Thousands of unique, individually traceable manufacturing jobs running at once.
Sequencing, design, manufacture and QA must all finish before the disease returns.
Target prediction has to work across the full diversity of human HLA immune types.
Sequencing, custom production, a year of Keytruda and clinical monitoring — as one bill.
- Does it delay cancer coming back?
- Yes — shown in two trials now
- Does it stop cancer spreading to other organs?
- Yes — same two trials
- Does it help people live longer?
- Not yet proven
- Does it work on its own, without Keytruda?
- Unknown — first trials just begun
- Does it work in other cancers?
- Unknown — each must be proven separately
- Can you get it?
- No — not approved anywhere
- What will it cost?
- Not announced
The simplest accurate explanation
After surgery removes visible melanoma, a made-to-order mRNA treatment shows the immune system what that patient's cancer looks like. Keytruda keeps the trained immune cells switched on. The goal is to destroy the microscopic cells left behind before they can restart the disease.
A medicine designed from one person's tumour beat an already-effective standard treatment in a large, properly controlled trial. That has never happened before, and it validates a way of making drugs, not just one drug.
It is also not a cure, not available, not proven to extend life, and not yet shown to work anywhere except the one form of skin cancer chosen precisely because it was the most winnable fight.
Both halves of that are true at the same time.