Neoantigen vaccines train a patient’s immune system to recognize and attack proteins found only on cancer cells, and early clinical trials suggest they can meaningfully delay cancer recurrence in cancers as different as melanoma and pancreatic adenocarcinoma. Unlike conventional vaccines that target infectious agents, these are therapeutic: given after a cancer diagnosis, they teach T cells to hunt down tumor-specific markers that healthy tissue never displays. The concept has moved rapidly from laboratory curiosity to Phase 2 and Phase 3 trials, with results that have generated genuine excitement among oncologists, though the science is still working through significant manufacturing, cost, and biological hurdles.
What Makes Neoantigens Different from Other Cancer Targets
Cancer cells accumulate mutations as they grow, and some of those mutations change the proteins displayed on the cell surface. These mutated proteins, called neoantigens, look foreign to the immune system because they are foreign: they arise from the tumor’s own genetic errors and don’t appear on normal cells. That distinction matters enormously. Older immunotherapy approaches often targeted tumor-associated antigens, proteins that cancer cells overexpress but that healthy cells also produce at lower levels. Attacking those targets risks collateral damage to normal tissue. Neoantigens, by contrast, are genuinely unique to the tumor, which means the immune response they provoke is highly specific and carries a lower risk of autoimmunity.
The immune system already recognizes neoantigens as “non-self” and can mount T-cell responses against them, but tumors develop ways to suppress or evade those responses. A neoantigen vaccine works by amplifying the signal: it presents carefully selected neoantigens to the immune system in a context that triggers a strong, targeted attack. The goal is to wake up or expand T cells that can find and kill cancer cells displaying those specific markers.
Finding the Right Neoantigens
Not every mutation produces a useful vaccine target. The standard pipeline starts by sequencing both the patient’s tumor and a sample of their normal tissue (usually blood), then comparing the two to identify somatic mutations unique to the cancer. From that list of mutations, computational tools predict which mutated peptides will bind tightly to the patient’s own immune molecules, known as MHC proteins, because only peptides that bind well will be visible to T cells. The pipeline also evaluates whether the T-cell receptor is likely to recognize and respond to each candidate peptide.
Artificial intelligence is increasingly central to this process. Machine learning models now help rate the binding affinity between mutated peptides and MHC molecules and predict which candidates will actually provoke a T-cell response, a property called immunogenicity that has historically been difficult to forecast accurately.
A newer and complementary approach skips prediction entirely and looks directly at what the tumor is actually displaying. Mass spectrometry-based immunopeptidomics physically captures and identifies the peptides sitting on MHC molecules on tumor cells. One pipeline called NeoDisc used this technique to successfully identify confirmed immunogenic neoantigens in cervical cancer samples, demonstrating that direct detection can pinpoint targets that computational prediction alone might miss.
Researchers are also expanding the search beyond simple point mutations. A large-scale analysis of over 2,500 tumors found that structural variations in the genome, such as deletions, duplications, and rearrangements, account for a substantial pool of neoantigens that conventional pipelines often overlook. Rearrangements within single genes, which typical gene-fusion analyses tend to ignore, accounted for over 40% of the neoantigens identified from structural variants.
The Different Vaccine Formats
Once the right neoantigens are selected, they need to be delivered to the immune system in a way that provokes a strong response. Several delivery platforms are in active development, each with trade-offs in speed, potency, and manufacturing complexity.
- mRNA vaccines: These encode the selected neoantigens as messenger RNA, packaged in lipid nanoparticles that protect the fragile molecule and help it enter cells. Once inside, the patient’s own cells manufacture the neoantigen proteins, which then get presented to the immune system. The mRNA platform benefited enormously from the COVID-19 vaccine infrastructure and is currently the furthest along in clinical trials for neoantigen cancer vaccines.
- Synthetic long peptides: These deliver the neoantigen directly as a protein fragment. Extending short peptides into longer ones helps overcome immune tolerance and can activate both the killer T cells that destroy tumors and the helper T cells that coordinate the broader immune response.
- DNA vaccines: These use plasmids or viral vectors to deliver genes encoding tumor neoantigens. One advantage is the capacity to deliver multiple neoantigen targets simultaneously, along with built-in molecular adjuvants that boost the immune response.
- Dendritic cell vaccines: Dendritic cells are the immune system’s most powerful antigen-presenting cells: their job is to capture, process, and display foreign proteins to T cells. In this approach, a patient’s own dendritic cells are harvested, loaded with neoantigens in the lab, and then reinfused. Because dendritic cells are the natural initiators of immune responses, this platform offers a direct route to activating tumor-specific T cells.
Each format is being tested in early-phase trials, often in combination with other immunotherapies. The mRNA and synthetic peptide approaches currently have the most published clinical data.
What the Clinical Trials Show So Far
The most closely watched trial involves an individualized mRNA vaccine called mRNA-4157 (also known as V940), developed by Moderna and Merck, tested in patients with resected high-risk melanoma. In the Phase 2b KEYNOTE-942 trial, patients who received the personalized vaccine plus pembrolizumab (a checkpoint inhibitor) had a roughly 44% lower risk of recurrence or death compared to those receiving pembrolizumab alone. At 18 months, about 79% of patients in the combination arm remained recurrence-free versus 62% in the pembrolizumab-only group.
Longer follow-up data strengthened these results. At a three-year update, the combination maintained a 49% reduction in the risk of recurrence or death, with a two-and-a-half-year recurrence-free survival rate of about 75% versus roughly 56% for pembrolizumab alone. The vaccine also showed a meaningful benefit in reducing distant metastasis, the spread of cancer to other organs, which is the outcome that most directly relates to long-term survival.
Pancreatic cancer, widely considered one of the hardest cancers to treat with immunotherapy, has also shown promising early signals. In a trial of a personalized mRNA neoantigen vaccine called autogene cevumeran, given alongside the checkpoint inhibitor atezolizumab and chemotherapy after surgery, half the patients developed strong T-cell responses against multiple vaccine neoantigens. Those vaccine-expanded T cells made up as much as 10% of all circulating T cells, a striking level. Patients whose immune systems responded to the vaccine (the “responders”) had not yet reached median recurrence-free survival at 18 months of follow-up, while non-responders had a median of about 13 months.
Separately, two Phase 1 trials tested synthetic long peptide and DNA personalized vaccines in pancreatic cancer patients after surgery and chemotherapy. The vaccines were well tolerated, with no serious adverse events, and induced measurable neoantigen-specific T-cell responses. Vaccinated patients showed a trend toward longer median overall survival compared to a matched comparison group: about 4.4 years versus 3.5 years, though the difference did not reach statistical significance in such a small study.
Why Pairing with Checkpoint Inhibitors Matters
Tumors don’t just sit passively while the immune system gears up. They deploy molecular brakes, checkpoint signals like PD-1 and CTLA-4, that tell attacking T cells to stand down. A neoantigen vaccine can generate an army of tumor-specific T cells, but if the tumor suppresses those T cells once they arrive, the vaccine’s impact is blunted. Checkpoint inhibitors release those brakes, and combining them with neoantigen vaccines produces what appears to be a synergistic effect: the vaccine provides the soldiers, and the checkpoint inhibitor clears the obstacles.
This synergy has been observed across multiple settings. In preclinical models, neoantigen vaccination combined with anti-PD-1 antibodies prevented the late-phase tumor relapses that occurred in roughly half of mice receiving the vaccine alone. The combination also accelerated tumor elimination and broadened the immune response, with T cells expanding to attack additional tumor targets beyond what the vaccine originally contained, a phenomenon called epitope spreading.
Similar synergy has been demonstrated with anti-CTLA-4 therapy. In one preclinical study, a neoantigen vaccine combined with anti-CTLA-4 produced significantly greater tumor growth reduction than either treatment alone, with the combination also generating stronger T-cell responses in the spleen.
A case report of a pediatric patient with metastatic liver cancer (hepatocellular carcinoma) illustrated this principle in a clinical setting. After disease recurrence, the patient received a neoantigen peptide vaccine combined with checkpoint inhibitor therapy. Immune monitoring showed robust vaccine-induced T-cell responses that were further enhanced by the checkpoint blockade, resulting in tumor regression and long-term remission.
Off-the-Shelf Versus Fully Personalized
The vaccines described above are fully personalized: each patient gets a vaccine designed around their individual tumor’s mutations. That’s powerful but slow and expensive. An alternative approach targets shared neoantigens, mutations that occur frequently across many patients’ tumors. If common cancer-driving mutations like those in the KRAS gene produce predictable neoantigens, a standardized vaccine could be manufactured in advance and given to any patient whose tumor carries one of those mutations.
A Phase 1 trial tested exactly this concept using a vaccine encoding 20 shared neoantigens derived from common oncogenic driver mutations, delivered via a chimpanzee adenovirus prime and a self-amplifying mRNA boost, alongside checkpoint inhibitors. The vaccine was well tolerated, but results revealed an unexpected complication: the immune response was disproportionately directed against TP53-derived neoantigens rather than the KRAS neoantigens that the patients’ tumors actually expressed. This hierarchy of immunodominance, where the immune system preferentially targets certain neoantigens over others regardless of which ones the tumor displays, could undermine the therapeutic benefit of multi-target shared vaccines. The researchers subsequently developed an optimized version that exclusively targets KRAS-derived neoantigens.
The immunodominance finding highlights a fundamental tension in shared vaccine design. Cramming more targets into a single vaccine seems intuitively better, but the immune system doesn’t respond equally to all of them, and a dominant response against an irrelevant target can crowd out responses against the one that matters. Personalized vaccines sidestep this problem because each target is selected specifically for the individual patient’s tumor, but they carry their own logistical burdens.
Durable Immune Memory and Epitope Spreading
One of the most encouraging findings from neoantigen vaccine trials is that the immune responses they generate don’t fade quickly. A follow-up study of melanoma patients who received personal neoantigen peptide vaccines found that vaccine-specific memory T cells persisted for years after vaccination. These weren’t just lingering cells with no function; the T-cell clones remained active and broadened their targeting over time, a phenomenon where the immune system learns to attack additional tumor antigens beyond those in the original vaccine.
Preclinical work in liver cancer models provides mechanistic support for why combining vaccines with checkpoint inhibitors might produce particularly durable responses. When a neoantigen vaccine was paired with anti-PD-1 therapy in orthotopic liver cancer models, 80% of mice achieved durable tumor regression and long-term immune memory. Analysis of the tumor environment showed a marked increase in a specific population of immune cells called tissue-resident memory T cells, which patrol tissues locally and can mount rapid responses if cancer reappears. The abundance of these cells was directly associated with how well the treatment worked.
Who Stands to Benefit Most
Not every cancer produces enough neoantigens to make vaccination feasible. The number of neoantigens a tumor can generate is closely tied to its tumor mutational burden, essentially how many mutations it carries. Cancers with high mutational burdens, like melanoma, lung cancer in smokers, and certain colorectal cancers, tend to produce more potential vaccine targets. Cancers with low mutational burdens, like many pediatric tumors and some pancreatic cancers, present a harder challenge because there are fewer neoantigens to choose from.
Research has confirmed that even among tumors that are microsatellite stable (a group generally considered less responsive to immunotherapy), those with high mutational burden carry a substantially greater number of predicted neoantigens, including a higher proportion of high-affinity targets likely to provoke an immune response. This suggests that mutational burden could serve as a practical biomarker for selecting patients most likely to benefit from neoantigen vaccination, although the pancreatic cancer trials show that even some lower-mutation tumors can generate meaningful responses when the right targets are identified.
Manufacturing Timelines and Cost
The fully personalized approach requires a biopsy, genomic sequencing, computational neoantigen prediction, vaccine manufacturing, and quality testing, all before the first dose can be administered. This timeline, often measured in weeks to months, creates a real clinical problem: aggressive cancers don’t wait. During the manufacturing window, tumors can evolve, shed the very mutations being targeted, or further suppress the immune environment. These delays are a key reason why clinical benefit from personalized neoantigen vaccines has been inconsistent across trials.
Cost is equally daunting. Current mRNA cancer immunotherapies can exceed $100,000 per dose, and treatment typically requires multiple doses. At that price point, widespread use is unrealistic without significant cost reductions. Researchers and industry groups point to artificial intelligence, streamlined manufacturing processes, and economies of scale as potential pathways to affordability, but these remain aspirational rather than demonstrated at this stage. Shared neoantigen vaccines offer a potential cost advantage by allowing batch manufacturing, though the immunodominance challenges described earlier may limit their effectiveness.
How Tumors Fight Back
Even when a neoantigen vaccine successfully generates an immune response, tumors have evolved multiple strategies to resist destruction. One of the most fundamental is immunoediting: as the immune system kills cancer cells displaying certain neoantigens, it effectively selects for cancer cells that have lost or downregulated those same targets. Longitudinal multi-omics studies in glioma have documented reduced neoantigen expression over time, suggesting this isn’t a theoretical concern but an observed pattern.
The tumor microenvironment poses another obstacle. Many solid tumors are surrounded by immunosuppressive cells and molecular signals that dampen T-cell function right where it’s needed most. Regulatory T cells, a population that normally prevents autoimmunity, are co-opted by tumors to shut down anti-cancer immune responses. Research has shown that depleting these regulatory cells can improve the priming and expansion of anti-tumor immune cells triggered by vaccines, particularly when the neoantigens involved have weak immunogenicity on their own.
Loss of the MHC presentation machinery is yet another escape route. If a tumor cell stops displaying its MHC molecules, it becomes invisible to T cells regardless of how many neoantigens it harbors. This is why many researchers view neoantigen vaccines not as standalone therapies but as one component of a broader strategy that might include checkpoint inhibitors, drugs targeting the tumor microenvironment, and therapies designed to prevent antigen loss.
Safety Profile So Far
Across the trials published to date, neoantigen vaccines have been consistently well tolerated. The triple-negative breast cancer DNA vaccine trial reported that neoantigen vaccines can induce or enhance highly specific antitumor immune responses with minimal risk of autoimmunity, a finding that aligns with the theoretical advantage of targeting tumor-exclusive proteins.
The pancreatic cancer peptide and DNA vaccine trials reported no grade 3 or higher adverse events. The shared neoantigen vaccine trial reported side effects consistent with what you’d expect from viral vector-based vaccines and checkpoint inhibitors: acute inflammation, mostly mild to moderate. The mRNA melanoma vaccine’s side effects have been broadly consistent with those seen in the pembrolizumab control arm plus typical injection-site reactions.
This favorable safety profile makes sense biologically. Because neoantigens are truly tumor-specific, the vaccine-trained immune cells shouldn’t attack healthy tissue. That’s a meaningful advantage over some other forms of immunotherapy, particularly CAR-T cell therapy and certain checkpoint inhibitor regimens, which can cause severe autoimmune complications. Whether this safety advantage holds in larger Phase 3 trials with longer follow-up remains to be confirmed, but the early data is reassuring.
The Economics of Scaling Up
Beyond per-dose pricing, the logistics of personalized cancer vaccines create bottlenecks that conventional drug manufacturing doesn’t face. Each patient needs an individual biopsy processed, sequenced, analyzed, and used to produce a unique product. That workflow doesn’t benefit from the efficiencies of mass production. Hospitals and cancer centers would need new infrastructure for rapid tumor profiling and coordination with vaccine manufacturers, which adds institutional costs on top of the drug price itself.
A modeling study examining the potential public health impact of mRNA cancer immunotherapies in the United States projected that continued investment in the technology could enhance its economic viability over time, but noted that personalized mRNA immunotherapy faces logistical bottlenecks, with increased biopsy and processing requirements potentially lengthening turnaround times even as production scales up. The authors suggested that AI-driven improvements in neoantigen prediction and manufacturing efficiency are the most plausible routes to bringing costs down to a level where broad access becomes realistic.