The 5-Year Personalized mRNA Cancer Vaccine Data Just Read Out — Here's What It Means

In early 2026, Moderna and Merck released five-year follow-up data from the KEYNOTE-942 trial of intismeran autogene — a personalized mRNA cancer vaccine — combined with the checkpoint inhibitor pembrolizumab in high-risk melanoma patients after surgery. The combination cut the risk of recurrence or death by roughly 49% compared with pembrolizumab alone. Five years is the standard cancer-trial horizon at which most people stop calling something promising and start calling it durable.

The vaccine itself is unusual, even by mRNA standards. Each dose is custom-manufactured per patient from a tumour biopsy. The result is a personal therapy that took, less than a decade ago, more time to design than the patient had to live. The numbers below are about that approach finally working at scale.

How a personalized cancer vaccine is built

Cancer cells accumulate mutations that change the proteins on their surface. Most of those mutations are silent or shared with healthy tissue, but some produce neoantigens — protein fragments that look foreign to the immune system. The question is how to point the immune system at those specific neoantigens. The answer encoded in intismeran autogene is: sequence the patient''s tumour, identify the top neoantigens computationally, and load them into an mRNA construct that the patient''s own cells translate after injection.

• Step 1. Surgical tumour sample is collected, with a matched blood sample for germline reference.
• Step 2. Whole-exome sequencing identifies tumour-specific mutations.
• Step 3. An algorithm predicts which mutant peptides are most likely to be displayed on MHC-I and trigger T-cell responses.
• Step 4. The top candidates — up to 34 in this vaccine — are encoded into a single mRNA strand wrapped in a lipid nanoparticle.
• Step 5. The patient receives the vaccine alongside pembrolizumab, which removes the PD-1 brake that tumours otherwise use to silence newly-activated T cells.

What the five-year data actually shows

KEYNOTE-942 was a randomised Phase 2b trial in stage III-IV melanoma patients who had completed surgical resection — meaning the visible tumour was removed but the recurrence risk was high. Participants received either pembrolizumab alone or pembrolizumab plus intismeran autogene. The primary endpoint was recurrence-free survival.

At five years of follow-up, the combination arm showed a 49% relative reduction in the risk of recurrence or death compared with pembrolizumab alone. Distant metastasis-free survival followed the same pattern. The benefit was largest in the patients with the highest tumour-mutation burden — exactly the subgroup the neoantigen approach should work best for.

Where the platform goes next

Melanoma is, in many ways, the easy mode for cancer immunotherapy. It carries a high mutational burden, generates abundant neoantigens, and historically responds well to checkpoint inhibition. The platform''s value depends on whether it can move into cancers where checkpoint monotherapy works less well.

• Non-small cell lung cancer. A Phase 3 program is enrolling, focused on the adjuvant setting after resection or chemoradiation.
• Renal cell carcinoma. Phase 2 trials are studying the combination with checkpoint inhibitors in localised and advanced disease.
• Bladder cancer. Early-phase work is targeting muscle-invasive disease, where neoadjuvant immunotherapy is increasingly standard.
• Pancreatic cancer. Distinct from the melanoma platform, but proof-of-concept Phase 1 data from a different group has shown durable T-cell responses in a small cohort.

The manufacturing question

Personalised vaccines have an obvious scale problem: each dose is bespoke. The current turnaround from biopsy to first dose is measured in weeks. That is fast by historical standards, but it is also too slow for cancers with rapid trajectories. Reducing manufacturing time, hardening the supply chain, and shifting from research-grade to commercial-grade reliability are the next operational hurdles.

There is also a quieter consideration: the cost of goods. Building a vaccine per patient is more expensive than a small-molecule pill, but if the platform works across multiple cancer types it amortises across a much larger population. Whether the resulting price stays within reach of public health systems is one of the defining policy questions of the next five years.

What this means for bioscience students

Personalised cancer vaccines sit at the intersection of three skill sets that were considered separate fields when most current undergraduates started university: sequencing and bioinformatics, immunology, and pharmaceutical manufacturing. A working understanding of HLA-peptide presentation, mutational signature analysis, and mRNA-lipid nanoparticle formulation are now all part of the same conversation.

If you are choosing a research direction and want a field that is unlikely to be quiet for the next decade, neoantigen-directed therapy is one of the strongest candidates. The five-year data is the moment the field moved out of the speculative column. The question now is which tumour types follow.