Individualized mRNA cancer vaccines make strides

Biopharmas and academic investigators are vaccinating cancer patients with individualized mRNA vaccines encoding novel antigens derived from mutations arising in their own tumors. Studies in pancreatic cancer, in kidney cancer and across a range of solid tumors that are traditionally recalcitrant to treatment have already shown positive results, promising to push the limits beyond what is possible with checkpoint inhibitors and tackle those tumors that escape immune system. But tumors have an enormous molecular diversity, and selecting the new antigens (neoantigens) for immune recognition is a hugely complex task. For the most advanced candidates, Moderna’s and Merck’s mRNA-4157, in phase 3 studies, the timing of a regulatory filing remains uncertain. A first read-out, in melanoma, is expected next year.

Neoantigens are new proteins on cancer cells that are immunogenic. They arise from mutations within the tumor genome and make attractive targets because they not only generate anti-tumor immune responses but are also specific to a patient’s individual tumor fingerprint. Not all tumor-associated antigens (TAAs) can stimulate the immune response. Most common TAAs will have been encountered by the immune system’s T cells in the thymus during development and be recognized as ‘self’. If this is the case, those autoreactive T cells that recognize them are deleted before they can attack the body’s tissues. In contrast, neoantigens — particularly those that are immunologically distant from endogenous proteins — are likely to appear as ‘foreign’ to the immune system, an advantage for vaccine targeting cancer. “It gets at one of the fundamental weaknesses of the field,” says Patrick Ott of Dana-Farber Cancer Institute. Most conventional cancer vaccines are not truly tumor-specific as their targets are not truly foreign. “The ‘neons’ don’t have that issue.”

Identifying which neoantigens stimulate a robust T cell response is not easy. The field has yet to define a complete set of ‘rules’ for selecting neoantigens, and different groups working on the same datasets have produced highly divergent results. A key issue is whether neoantigens bind to major histocompatibility complex (MHC) proteins, to ensure immune activation. Predicting whether a neoantigen binds to MHC class I to activate CD8 tumor-destroying cells is largely a solved problem, but it has proven trickier to anticipate whether a neoantigen will bind to MHC class II molecules to stimulate CD4 helper T cells. Myriad other parameters are also associated with neoantigen processing, presentation and T cell recognition, and today vaccine developers are employing artificial intelligence to optimize them. Many of the fine details are still under wraps, however. “We don’t know exactly what other companies are doing,” says Alessandro Riva, chairman and CEO of Strasbourg, France-based Transgene, which has entered an alliance on neoantigen identification with Tokyo-based electronics and AI firm NEC.

To make an mRNA vaccine, the first step is to take tumor samples from an individual, typically from a biopsy or during surgical removal of the primary tumor. DNA from these tumors is sequenced to identify TAAs and tumor neoantigens. Researchers typically select 20–40 neoantigens, which they then design into an mRNA sequence, allowing many to be targeted at once. The mRNA molecules encoding the neoantigens are formulated in lipid nanoparticles. Dendritic cells take up and present the neoantigens, priming T cells. The turnaround time from tumor removal to vaccine administration is currently about one to two months for mRNA vaccines.

It’s been a painstaking effort, however, to establish whether neoantigen vaccines work in the clinic. Beatriz Carreno, then of Washington University School of Medicine (now at the University of Pennsylvania), and colleagues demonstrated the safety and feasibility of neoantigen-directed vaccines, in a clinical trial of a dendritic cell vaccine a decade ago. They tested it in three patients with melanoma, reporting that it increased the breadth and diversity of the T cell response against each patient’s neoantigens.

Soon after, BioNTech, Neon Therapeutics (which BioNTech acquired in 2020) and Moderna began their clinical programs. The field’s progress has not been all positive, however. There has been a string of notable casualties along the way, including Aduro Biotech, Advaxis Immunotherapies, Genocea Biosciences and Pact Pharma. Gritstone Bio, another of the early pioneers, entered bankruptcy proceedings in October 2024, after interim data from a phase 2 trial in patients with metastatic colorectal cancer of its Granite immunotherapy regimen failed to convince investors that it had a viable path to an approval. Despite this setback, Andrew Allen, the company’s founder and former CEO, remains optimistic that the ongoing trial will ultimately yield a positive overall survival (OS) signal, particularly in patients who had a low disease burden to begin with. “Those OS data are going to be critical, and they should become available in the second half of this year,” he says. Whatever the outcome, it will be too late for Gritstone, most of whose assets were acquired earlier this year by Seattle Project Corp.

So far, the field’s most convincing evidence of efficacy has come from a phase 2b trial of mRNA-4157 in 157 patients with high-risk melanoma who had already undergone surgery. After three years, a combination of the vaccine plus Keytruda reduced the risk of recurrence or death by 49% compared with Keytruda alone. The 30-month recurrence-free survival rates were 74.8% for those on the combination therapy and 55.6% for those on Keytruda only. It’s the first randomized trial of a neoantigen-targeted vaccine to demonstrate clinical benefit. “That’s why this study is so exciting to us, but the study is small,” says Ott. Moreover, the heterogeneity of the participants and difficulties in maintaining randomization (the study was conducted during the COVID-19 pandemic) also introduce potential bias. “We know that just a small number of patients can take the results in one direction,” he says. The phase 3 data are keenly awaited, therefore.

The complexity of individualized vaccines — and of the immune system — can make it difficult for clinical investigators to make firm predictions about their likely efficacy in a given cancer indication or at a given stage of treatment. But it appears that neoantigen-targeted vaccines are best deployed soon after a tumor is surgically removed. Ott has recently co-authored a report on BioNTech and Genentech’s ongoing phase 1 trial of autogene cevumeran in over 200 patients with advanced solid tumors. Although the study detected neoantigen-specific immune responses in 71% of participants, only three had any kind of clinical response. “It’s probably not the right population,” says Ott, referring to the potential drawback of including patients with advanced disease, who have already received several lines of treatment. In early-stage disease, immune fitness is less compromised, and, what’s more, early-stage tumors have less clonal heterogeneity and may be more easily eradicated by a relatively narrow immune response than heavily mutated late-stage tumors. BioNTech and Genentech are now testing the therapy in patients with tumors at earlier stage across several indications (Table 1).

CureVac is another pioneer of mRNA-based therapeutics, although its neoantigen-based cancer vaccine efforts are still preclinical. The company has developed a distinctive approach. Whereas most neoantigen discovery efforts involve exome sequencing only, which largely limits the search to the protein-coding regions of patients’ genes, CureVac conducts whole-genome sequencing of the tumor to uncover additional antigen classes that can be found in other regions of the genome. It is standard practice to perform RNA transcriptomic analyses, to confirm that putative neoantigens are actually expressed as proteins. Here again, CureVac differentiates itself from the other players, who generally perform short-read RNA sequencing involving several hundred nucleotides. “We do long RNA reads, because in these long RNAs you often find those novel classes of antigens that are translated into proteins expressed or overexpressed in tumors,” says CureVac’s CSO, Myriam Mendila. The company has disclosed one such class, which it calls neo-open reading frame peptides. These are typically derived from structural alterations in the genome, small genomic insertions or deletions, mutations that alter mRNA splicing, or disruptions to stop codons. Its analysis of 61 tumors across six different cancer types revealed a tumor ‘framome’ — the collection of all potential neo-open reading frame peptides in a tumor’s genome — of up to 2,000 amino acids.

Although mRNA-based individualized vaccines are most advanced, other vaccine platforms for delivering neoantigens are also in clinical development. Transgene uses modified vaccinia Ankara (MVA) virus to express up to 30 neoantigens in its individualized vaccine TG4050. In an ongoing phase 1/2 trial in patients with newly diagnosed, operable stage III or IV head and neck cancer, all 16 evaluable patients who received TG4050 immediately after surgery, chemotherapy and radiotherapy remained relapse-free after a median 24.1 months of follow-up, according to the company’s last update, in November 2024. In contrast, 3 of 16 patients in a control arm, who received standard of care only, had relapses within the same time frame.

The most obvious advantage of viral vectors is their strong immunogenicity. However, the manufacturing process takes about 28 days longer than mRNA production because of the need to scale up from the initial seed stock to a clinical dose. The key, says Riva, is to have the vaccine ready after patients have completed first line therapy — typically surgery, chemotherapy and radiotherapy. “If you are able to produce the vaccine in three months or less, you can cover the largest part of the patient population in the early setting.” The company is further optimizing its process to bring the turnaround time closer to two months, he adds.

Nouscom, of Basel, Switzerland, has adopted a hybrid approach in its most advanced program, NOUS-209, for Lynch syndrome, a genetic condition caused by mutations in several genes involved in DNA mismatch repair. This leads to an accumulation of DNA replication errors, which puts those affected at a risk of developing colorectal and other types of cancer, including endometrial cancer in women. NOUS-209 is an off-the-shelf vaccine encoding 209 neoantigens — novel peptides arising from frameshift mutations — that are shared across the broad patient population. An individual patient typically has about 50 of them, says Nouscom CEO Marina Udier. The vaccine is administered in a prime–boost regimen, comprising a great ape adenovirus vector followed by an MVA vector. Because the neoantigenic payload is so large, the company generates four different versions of each of these vectors. All four can be combined in a single cocktail for ease of administration at both the prime and boost steps.

The goal of the Lynch therapy is not to treat an existing cancer but to trigger an immune response that will intercept cancerous cells expressing neoantigens before they develop into full-blown tumors. “There is nothing today to offer these people, other than frequent and invasive surveillance, which, of course, doesn’t prevent cancer or deal with the disease,” she says. Nouscom recently reported that NOUS-209 elicited a neoantigen-directed CD8 and CD4 T cell response in all 37 patients it evaluated in a phase 1/2 trial. It also showed that the induced CD8 T cells were capable of killing cancer cells ex vivo. It now aims to move NOUS-209 toward a phase 2/3 trial that could serve as the basis of a regulatory filing.

Ygion Biomedical, of Vienna, raised $17 million last year to take forward a targeted delivery technology called Cargonaut. The company has not fully disclosed its mechanism, but it comprises a carrier molecule onto which peptides encoding neoantigens are loaded. The construct also contains a targeting and activation moiety, designed to ensure that patients’ antigen-presenting cells initiate a potent immune response when exposed to their tumor neoantigens. “Our data show that we can trigger a rapid — very rapid — immune response through this targeted way,” says Ygion’s chief business officer, Sophie Zettl.

The company, which is still in preclinical development. is considering glioblastoma, an aggressive brain cancer that has a poor prognosis and limited treatment options, as its lead indication. A recently published observational study, conducted at the Centre for Human Genetics Tübingen, in Germany, showed that a peptide-based neoantigen-directed vaccine — which lacked Ygion’s immunostimulatory delivery technology — improved survival in those patients who mounted a strong immune response.

Meanwhile, Moderna is building a dedicated production facility for mRNA-4157 in Marlborough, Massachusetts, in advance of an anticipated US Food and Drug Administration (FDA) approval in the not-too-distant future. A complicating factor is that processes used to manufacture products at scale do not readily apply to personalized vaccines. How the FDA will handle Moderna’s application will be keenly watched, particularly as Vinay Prasad, a critic of therapeutic cancer vaccines, is now leading the agency’s Center for Biologics Evaluation and Research. Europe’s current approval pathway is not straightforward either. “If you look at manufacturing timelines, I would say almost a third to a half are related to regulatory requirements,” says Ygion’s co-founder and CEO Wolfgang Fischl. Sterility testing alone can take up to two weeks under the present regime. Draft guidance from the UK’s Medicines & Healthcare Products Regulatory Agency — which addresses mRNA-based individualized vaccines only — proposes to regulate the product design aspects of the therapies, including sample collection and storage, genome sequencing, bioinformatics analysis, and neoantigen identification and selection, under medical device regulations. But the actual therapies will be regulated as human medicines. The complexity of the regulations mirrors the complexity of the products. But convincing clinical data can accelerate the field’s evolution.