The cancer peptide vaccine has spent the last thirty years as one of oncology's most stubborn unfulfilled promises. The basic idea—train a patient's own T cells to recognize and kill tumor cells using vaccine-delivered tumor antigens—is elegant, mechanistically obvious, and grounded in the same immunology that produces lifelong protection from measles or yellow fever. For decades it almost worked. Late-stage trials of shared tumor antigen vaccines like gp100, NY-ESO-1, MAGE-A3, telomerase, and survivin produced tantalizing immunologic responses, occasional dramatic clinical responses, and a long string of failed Phase 3 readouts. By the mid-2010s, the cancer vaccine field was widely considered a graveyard of hopeful immunology.
Then two things changed. First, immune checkpoint inhibitors like pembrolizumab and nivolumab solved the dominant problem (T cell exhaustion at the tumor) that had been silently sabotaging earlier vaccine attempts. Second, mRNA delivery technology matured during the COVID-19 pandemic to the point where personalized, patient-specific vaccines could be manufactured in weeks rather than years. The combination has produced what looks, in the December 2023 KEYNOTE-942 readout and its follow-up trials, like the first credible large-scale clinical signal for the entire cancer vaccine concept. mRNA-4157, Moderna and Merck's personalized neoantigen vaccine, is the lead asset, and the Phase 3 INTerpath-001 trial is the field's most-watched ongoing study.
What Is a Cancer Peptide Vaccine?
A cancer peptide vaccine is a therapeutic vaccine designed to train the patient's adaptive immune system to recognize and kill cancer cells by presenting tumor-associated peptide antigens to T cells in a context that drives durable cytotoxic T lymphocyte (CTL) responses.
The biological logic is simple. Every cell in the body continuously degrades a fraction of its intracellular proteins into short peptides (typically 8 to 11 amino acids), loads them onto major histocompatibility complex class I (MHC-I, also called HLA-I in humans) molecules, and displays them on the cell surface. CD8+ T cells continuously scan those displayed peptides looking for non-self sequences. If a tumor cell displays a peptide that the immune system recognizes as foreign—because the underlying gene has been mutated—the T cell can kill it.
That mutation-derived peptide is called a neoantigen. Every individual tumor has its own unique neoantigen catalog, derived from its specific somatic mutations. Identify those neoantigens, manufacture a vaccine that delivers them, and you can in principle teach the immune system to seek out and destroy that specific tumor. The personalization is the entire point.
Earlier cancer vaccine generations did not work this way. They targeted shared tumor-associated antigens—proteins that are overexpressed in tumors but also present in normal tissue (gp100 in melanoma, NY-ESO-1 in several cancers, telomerase, survivin, MUC1). The advantage was that one vaccine formulation could serve every patient with a given cancer type. The disadvantage was that the immune system had been tolerized to those self-derived sequences during normal development, so the resulting T cell responses were weak and prone to peripheral tolerance.
Mechanism: From Tumor Mutation to Cytotoxic T Cell
A modern personalized cancer peptide vaccine pipeline runs as follows:
- Tumor and normal sequencing. A biopsy of the tumor is whole-exome sequenced and matched against germline DNA from the patient's blood. The somatic mutation catalog is extracted.
- HLA typing. The patient's MHC class I and class II alleles are sequenced.
- Neoantigen prediction. Computational tools (NetMHCpan, MHCflurry, and increasingly AlphaFold-based binding predictors) score every mutation-derived peptide for predicted MHC binding, T cell receptor recognition probability, and processing efficiency.
- Selection. The top-ranked 20 to 34 candidate neoantigens are selected. Modern vaccines typically include both MHC-I and MHC-II epitopes to engage CD8+ and CD4+ T cell responses simultaneously.
- Manufacture. The selected neoantigens are encoded into a single mRNA construct (in Moderna's mRNA-4157) or assembled as a peptide cocktail (in older platforms). The mRNA is encapsulated in a lipid nanoparticle for delivery.
- Administration. The vaccine is injected (typically intramuscularly), the encoded peptides are translated and processed by antigen-presenting cells, the patient's HLA molecules display the resulting epitopes, and naive T cells specific for those neoantigens are primed and expanded.
- Combination with checkpoint blockade. Almost all current cancer peptide vaccine programs are combined with anti-PD-1 antibodies (pembrolizumab, nivolumab) to release T cell exhaustion at the tumor and let the vaccine-primed T cells do their job.
The full cycle from tumor biopsy to first vaccine dose takes 6 to 9 weeks in the current Moderna manufacturing process and is the most logistically complex personalized medicine workflow in mainstream oncology drug development.
Clinical Evidence: KEYNOTE-942 and the mRNA-4157 Story
The pivotal data point for the modern cancer peptide vaccine field came from KEYNOTE-942, a Phase 2b randomized trial of mRNA-4157 (also called V940) plus pembrolizumab versus pembrolizumab alone in patients with resected high-risk stage III/IV melanoma. The trial enrolled 157 patients, 107 in the combination arm and 50 in pembrolizumab monotherapy.
The headline result, presented at AACR in April 2023 and published in The Lancet in early 2024, was a 44 percent reduction in the risk of recurrence or death for the mRNA-4157 plus pembrolizumab combination versus pembrolizumab alone (hazard ratio 0.561, p = 0.0266) at a median follow-up of about 23 months. Distant metastasis-free survival also favored the combination with a hazard ratio of 0.347. The combination was generally well tolerated, with the additional adverse events attributable to the vaccine being mostly low-grade injection-site reactions, fatigue, and chills.
This was the first randomized clinical trial to show a clinically meaningful improvement from a personalized neoantigen vaccine in any cancer indication. It triggered FDA Breakthrough Therapy Designation and EMA PRIME designation. The Phase 3 confirmatory trial, INTerpath-001 (V940-001), is enrolling roughly 1,089 patients with resected high-risk melanoma, with similar trials now expanding into adjuvant non-small cell lung cancer (INTerpath-009), renal cell carcinoma, urothelial carcinoma, and cutaneous squamous cell carcinoma.
In parallel, BioNTech's BNT122 (autogene cevumeran) is in Phase 2 trials for adjuvant pancreatic cancer, melanoma, and colorectal cancer. A small Phase 1 pancreatic cancer trial published in Nature in 2023 by Vinod Balachandran's group at Memorial Sloan Kettering reported that 8 of 16 patients who received BNT122 plus chemotherapy plus atezolizumab developed strong T cell responses, and those responders had significantly longer recurrence-free survival than non-responders. Pancreatic cancer is one of the most treatment-resistant solid tumors, which makes any vaccine signal in this disease particularly notable. Gritstone bio's GRANITE platform, Transgene's TG4050 individualized vaccine in head and neck cancer, and Nouscom's viral-vector personalized vaccines are all in active clinical development.
For older shared-antigen platforms, the picture is more sobering. The melanoma vaccine gp100 failed multiple Phase 3 trials. MAGE-A3 failed Phase 3 in non-small cell lung cancer (MAGRIT) and melanoma (DERMA). Telomerase vaccines like GV1001 produced disappointing late-stage results. The neoantigen revolution exists in part because the shared-antigen approach exhausted its credit.
Approved Uses
As of early 2026, no personalized neoantigen cancer peptide vaccine has received full FDA approval. The category that comes closest to historical approval is the autologous cellular vaccine sipuleucel-T (Provenge), FDA-approved in 2010 for metastatic castration-resistant prostate cancer—technically a dendritic cell vaccine rather than a peptide vaccine, but conceptually adjacent. Talimogene laherparepvec (T-VEC, Imlygic), an oncolytic herpes virus engineered to express GM-CSF, was approved for melanoma in 2015 and is sometimes grouped with cancer immunotherapies of the vaccine family.
mRNA-4157 in combination with pembrolizumab is the leading candidate for full FDA approval if the INTerpath-001 Phase 3 readout (expected 2027–2028) confirms the Phase 2b benefit. BNT122 is following a similar path on the BioNTech side.
Safety and Side Effects
The safety profile of personalized peptide vaccines has been favorable in clinical trials so far. The most common adverse events are local injection-site reactions, low-grade fever, fatigue, chills, and myalgias—essentially the same reactogenicity profile that mRNA COVID-19 vaccines produced, scaled to the larger doses used in oncology. Serious autoimmune reactions appear to be uncommon, although the most aggressive risk to monitor is off-target T cell reactivity against normal tissues that share epitope similarity with the targeted neoantigens. Modern computational pipelines explicitly screen against this risk.
When combined with checkpoint inhibitors, the safety profile is dominated by the immune-related adverse events of the checkpoint blockade itself: thyroiditis, hypophysitis, colitis, hepatitis, pneumonitis, dermatitis. The vaccines do not appear to substantially increase the rate of these effects.
Connection to Gene Editing and Modern Peptide Therapy
Personalized cancer peptide vaccines sit at the precise intersection of three technologies that this site covers in depth: peptide chemistry, mRNA delivery, and AI-driven biology. They are peptide therapies in the sense that their active mechanism is the presentation of short peptide neoantigens on HLA molecules to T cells. They are mRNA therapies in the sense that the most advanced platforms (mRNA-4157, BNT122) use lipid nanoparticle-encapsulated mRNA to deliver the peptide-encoding sequences. They are AI therapies in the sense that they cannot exist without high-accuracy computational neoantigen prediction.
The gene-editing connection is less direct but increasingly relevant. CRISPR screens have been used to identify tumor-essential neoantigens that resist immune escape. AAV and lentiviral approaches are being explored for in vivo delivery of TCR transgenes that recognize personalized neoantigens. Several CRISPR-edited cell therapies in development use the same neoantigen prediction pipelines that power mRNA-4157 to select TCR specificities. The convergence of peptide presentation, mRNA delivery, and gene-editing modalities is one of the central themes of the next decade of oncology drug development. See our broader peptide-CRISPR convergence map for the full landscape.
FAQ
What is mRNA-4157?
mRNA-4157 (also called V940) is Moderna and Merck's personalized neoantigen cancer vaccine. It is an mRNA construct encoding up to 34 patient-specific tumor neoantigens, delivered in a lipid nanoparticle and combined with pembrolizumab.
What did KEYNOTE-942 show?
The Phase 2b KEYNOTE-942 trial in resected high-risk melanoma showed a 44 percent reduction in the risk of recurrence or death for mRNA-4157 plus pembrolizumab versus pembrolizumab alone, the first randomized evidence of clinical benefit from any personalized neoantigen vaccine.
How is a personalized cancer vaccine manufactured?
The patient's tumor and normal DNA are sequenced, the somatic mutations are identified, neoantigen candidates are computationally predicted from those mutations and the patient's HLA type, the top candidates are selected, encoded into a single mRNA molecule, and formulated into a lipid nanoparticle. The full process currently takes 6 to 9 weeks.
Why did older cancer vaccines fail?
Older vaccines targeted shared tumor antigens (gp100, MAGE-A3, telomerase) that are also present in normal tissue, so the immune system was tolerized against them. They were also developed before checkpoint inhibitors were available to release T cell exhaustion at the tumor.
Are cancer peptide vaccines FDA-approved?
Not yet for personalized neoantigen vaccines. Sipuleucel-T (Provenge) is an FDA-approved dendritic cell vaccine for prostate cancer, and T-VEC (Imlygic) is an FDA-approved oncolytic vaccine for melanoma. mRNA-4157 is in Phase 3 trials and could become the first approved personalized neoantigen vaccine.
How does AI improve neoantigen vaccine design?
Computational tools like NetMHCpan, MHCflurry, and AlphaFold-derived models predict which mutated peptides will bind a patient's specific HLA molecules, which will be efficiently processed and presented, and which will trigger durable T cell responses. Better prediction means fewer wasted slots in the vaccine and stronger clinical responses.
