TL;DR: Key Takeaways
- A therapeutic, not preventive, vaccine — given after melanoma is diagnosed or removed
- Built from neoantigens: protein fragments unique to that person's tumour
- Algorithms rank which neoantigens to include; manufacture is per patient
- Usually studied in combination with a checkpoint inhibitor, not alone
- All programmes remain investigational — none is commercially approved
The idea in one paragraph
Every melanoma carries mutations that healthy cells do not. Some of those mutations produce altered protein fragments — neoantigens — that the immune system could in principle recognise as foreign. A personalised melanoma vaccine takes a set of those fragments from one patient's own tumour and presents them back to that patient's immune system, with the aim of teaching it to find and attack cells carrying those markers.
How one is made, step by step
- A sample of the patient's tumour and their healthy cells are sequenced
- Mutations specific to the tumour, and actually expressed by it, are identified
- Predictive algorithms rank the resulting neoantigens by how likely they are to be presented by that person's immune system
- A limited set of selected neoantigens is encoded into a vaccine manufactured for that individual
- The vaccine is given, usually alongside another therapy such as a checkpoint inhibitor
Because every step after sequencing is specific to one person, there is no shelf stock. That is the defining difference from a conventional vaccine, and the reason production time and cost are central practical questions for the field.
mRNA, peptide or DNA?
Programmes differ mainly in how the selected targets are delivered. mRNA products encode the neoantigens and let the body's own cells produce them; peptide products deliver the fragments directly; DNA products encode a larger panel. The programme comparison table on our AI and melanoma hub sets out the main candidates, their trial phases and their published evidence side by side.
Read the full evidence overview
Programme-by-programme comparison of personalised neoantigen vaccines, with evidence strength and limitations stated for each.
What the evidence supports today
- Computationally selected targets can repeatedly generate patient-specific T-cell responses
- A personalised mRNA vaccine plus a checkpoint inhibitor produced an encouraging randomised Phase 2b signal in melanoma
- AI-assisted target selection is already operating inside human clinical trials
What is not established
Read trial claims carefully
- That the AI component alone caused any clinical benefit — a positive trial validates the whole chain: sequencing, algorithm, construct, formulation, manufacturing and companion therapy.
- That these vaccines work for every patient.
- That they cure melanoma.
- That response rates from small, single-arm trials can be compared directly between programmes.
What this means for you
Nothing on this page changes what to do about a spot on your skin. Treatment research sits at the far end of the melanoma journey; the part you control is noticing change early and getting it examined. If a lesion is new, changing, bleeding or simply bothering you, book a clinician. Photo monitoring — how to track a lesion over time — gives that clinician a comparable history instead of a memory.
Frequently Asked Questions
It is an investigational therapy made for one individual. Their tumour and healthy cells are sequenced, tumour-specific mutations are identified, and a limited set of the resulting protein fragments — neoantigens — is encoded into a vaccine intended to help their immune system recognise cancer cells carrying those targets.
No. These are therapeutic vaccines given to people who already have or have had melanoma, usually after surgery and usually alongside another treatment such as a checkpoint inhibitor. They are not preventive vaccines like those used for infectious disease.
No programme discussed here is commercially approved. Each remains investigational and available only inside clinical trials or expanded-access settings decided by an oncology team.
Manufacturing is per patient and begins after sequencing, so it takes weeks rather than being taken off a shelf. That production time is one of the practical limits on how widely this approach can be used today.
Algorithms rank which mutation-derived peptides are most likely to be presented by that patient's immune system, so a manageable number of targets can be selected for manufacture. The AI selects candidates; it does not generate the immune response or treat the cancer.
No. Photo-based AI screening sits at the awareness and monitoring end of the journey — it helps people notice a changing lesion earlier and take it to a clinician. It is a separate technology from genomic vaccine design and it is not a diagnosis.