Personalized mRNA Cancer Vaccines: 5-Year Data Rewrites the Rules of Oncology

49%. That is the reduction in the risk of recurrence or death for patients with high-risk melanoma who received a personalized mRNA cancer vaccine alongside Merck's Keytruda, compared to the PD-1 inhibitor alone. The data, presented at ASCO 2026 in Chicago and published simultaneously in the Journal of Clinical Oncology, is the longest follow-up for an mRNA-based cancer immunotherapy — and the effect did not fade between year three and year five.

The KEYNOTE-942 Phase 2b trial enrolled 157 patients with resected high-risk Stage III/IV melanoma. After a median follow-up of 60.3 months, the combination therapy also reduced the risk of distant metastasis or death by 59% and improved overall survival by 53% in an exploratory analysis. These figures are not incremental. They are the first demonstration that an mRNA cancer vaccine can produce durable, years-long benefit in a randomized controlled setting.

Jefferies analysts called the results "encouraging" in a June 8 note, adding that the data "helps instill confidence in pivotal Phase III data" expected later this year. William Blair noted that the stability of the treatment effect from year three to year five "suggests the personalized vaccine successfully reprograms the adaptive immune system for long-term surveillance."

How Personalized mRNA Neoantigen Vaccines Work

Intismeran autogene, formerly mRNA-4157 or V940, belongs to a class of therapies called individualized neoantigen vaccines. The process begins with a tumor biopsy. Researchers perform whole-exome and RNA sequencing on the resected tissue, comparing the tumor's genetic code against the patient's healthy cells to identify somatic mutations unique to the cancer. These mutations produce neoantigens — abnormal proteins displayed on the surface of cancer cells that the immune system can learn to recognize as foreign.

A bioinformatics pipeline prioritizes the most immunogenic neoantigens based on predicted MHC binding affinity, tumor mutational burden, and expression levels. Up to 34 neoantigens are encoded into a single mRNA construct, which is manufactured within four to eight weeks.

The lipid nanoparticle (LNP) delivery system — the same technology validated across billions of COVID-19 vaccine doses — carries the mRNA to antigen-presenting cells (APCs) in lymph nodes. These APCs produce the encoded neoantigens and present them to CD8+ cytotoxic T cells and CD4+ helper T cells, training the immune system to seek out and destroy cells displaying those same neoantigens anywhere in the body.

KEYNOTE-942: The 5-Year Data Breakthrough

The treatment effect did not fade. At the three-year analysis, the combination showed a 49% reduction in recurrence or death. At five years, the exact same 49% figure held. In oncology, where treatment effects commonly wane over time as tumors evolve resistance, a flat survival curve from year three to year five is unusual.

While incremental, we view today's 5-year data update on intismeran as positive maintenance of an impressive effect over Keytruda monotherapy, the first therapy, to our knowledge, to show added benefit in this post-surgical setting.— William Blair analysts, June 2026

The impact on distant metastasis was even more pronounced: a 59% reduction in risk of metastatic spread or death. For melanoma patients, distant metastasis is the event that transforms a manageable disease into a life-threatening one. Cutting that risk by nearly 60% and maintaining the effect for five years shifts the risk-reward calculus for adjuvant treatment.

The AI Neoantigen Prediction Pipeline

What makes this class of therapy scalable is not the mRNA manufacturing alone — it is the computational biology layer upstream of the production line. Its neoantigen prediction engine uses machine learning models trained on mass spectrometry data, MHC binding assays, and clinical response data from prior vaccine trials to select which mutations to include in each patient's construct.

Yale researchers published Immunostruct, a deep learning model that incorporates peptide three-dimensional structure into neoantigen prediction, in Nature Machine Intelligence in February 2026. The model outperformed sequence-only approaches in identifying peptides likely to trigger CD8+ T cell responses. Washington University in St. Louis described in Nature in April 2026 how mRNA cancer vaccines engage an unconventional dendritic cell pathway, providing mechanistic insight into why vaccines like mRNA-4157 generate durable T cell memory.

The convergence of AI-driven neoantigen design and mRNA delivery platforms gives the KEYNOTE-942 results weight beyond a single trial. Better prediction means fewer neoantigens wasted on non-immunogenic targets. Faster manufacturing means shorter wait times after surgery. Durable T cell responses mean the therapy may work across tumor types beyond melanoma.

Beyond Melanoma: The Pipeline Expands

Moderna and Merck now have eight Phase 2 and Phase 3 trials underway for intismeran autogene. The pivotal Phase 3 INTerpath-001 trial in adjuvant melanoma has enrolled approximately 1,089 patients. Enrollment is also underway for Phase 3 trials in non-small cell lung cancer and Phase 2 studies in renal cell carcinoma, bladder cancer, and cutaneous squamous cell carcinoma.

BioNTech and Genentech are pursuing a parallel path with autogene cevumeran (BNT122). In pancreatic ductal adenocarcinoma, the Phase 1 data showed that 8 of 16 patients who generated T cell responses after vaccination had significantly longer recurrence-free survival — none of those eight saw their cancer return. A Phase 2 trial launched in 2026. Providence Therapeutics started the PaedNEO-VAX trial, the world's first pediatric mRNA cancer vaccine study, in March 2026 across seven Australian hospitals, targeting relapsed brain tumors in children.

What happens to the market a year from now?

Development scenarios