Moderna/Merck are expected to present their data at this 2026 European Society for Medical Oncology (ESMO) Congress demonstrating that the combination of a personal neoantigen cancer vaccine, prepared as a messenger RNA (mRNA) in lipid nanoparticle (LNP) format, and standard-of-care anti-programmed cell death protein 1 (PD-1) antibody (pembrolizumab) is more effective than anti-PD-1 alone in the adjuvant setting of stage IIB-IV melanoma. This is a big win for the up-again, down-again field of cancer vaccines, a field in which past phase III trials have only led to disappointment. The promise of neoantigens as the go-to target has gripped attention for the past 15 years, and a failed trial now would be devastating.
Despite this vaccine being first out of the gate and a validation for the field, there are questions. How robust are the data? Will this particular format have staying power, or will the pioneers be replaced by fast followers with possibly better designs already in the works?
And what will we understand about the underlying immunobiology driving the effectiveness of this vaccine? We will address each of these topics below. But first things first, what is unique about this vaccine?
Ever since the first observations that T-cell responses in patients could be linked to mutations, initially by Wölfel et al.1 with mutated CDK4 and later, convincingly, by the same group in a patient with metastatic melanoma who was experiencing a favorable outcome,2 so-called neoantigens have been highly sought-after vaccine targets. Arising as somatic missense mutations in cancer cells, neoantigens were inherently exquisitely tumor-specific—found only in the cancer cells—and thus presumed to be safe targets and were thought to possess superior immunogenicity due to escape from central tolerance.3 Both features have held true as the genomic revolution made their identification possible, and multiple players, in both academia and the biopharmaceutical industry, pushed the boundaries of drug development and manufacturing to a new form of bespoke medicine.4, 5, 6 These efforts shifted the paradigm from ‘finding’ the right medicine for each person to ‘making’ the right medicine for each person. In parallel, the revolution in cancer therapy brought about by immune checkpoint blockade (ICB), led by anti-CTLA-4 and anti-PD-1 antibodies, has made the modulation of the immune system a mainstream player in cancer treatment and research.
These efforts have buoyed up excitement about neoantigens as vaccine targets, as evidence linking ICB efficacy to neoantigens in humans was found,7, 8 and their new availability provided fresh opportunities for synergy, which may be very important in the trial at hand.
Intismeran autogene is an mRNA-based vaccine manufactured by Moderna encoding up to 34 patient-specific neoepitopes derived from the sequence of each patient’s tumor cells and identified and selected based on proprietary algorithms, including features such as, from what can be gleaned from the literature,9 allele frequency of the mutation, expression levels, amino acid conservation (i.e. potential similarity of mutated to nonmutated sequence), and predicted human leukocyte antigen binding (primarily class I but also some class II). The epitope sequences are encoded in a plasmid vector arranged as ‘beads on a string’, synthesized as mRNA by in vitro transcription (IVT) of template DNA incorporating N1-methyl-pseudouridine in place of uridine, which is 5′ capped and polyadenylated (during or after IVT). The mRNA is formulated as an LNP with a proprietary lipid composition utilizing SM-102 as the ionizable lipid.
Patients in the vaccine arm of the phase III INTerpath-001 trial (NCT 05933577) received nine doses of the vaccine on an every-3-week schedule in alternating limbs. Each dose contained 1 mg of formulated RNA.10 Patients in the vaccine arm also received pembrolizumab (Keytruda; Merck’s approved version of anti-PD-1 therapy, the standard of care for fully resected melanoma) during vaccine manufacture, during vaccination, and afterward for up to 1 year in total. The control arm received pembrolizumab alone.
The fundamental question was thus: would addition of a neoantigen vaccine to standard-of-care anti-PD-1 therapy lead to statistically meaningful improvement in a meaningful clinical outcome?
Based on recent press releases from Moderna/Merck and the gestalt of the biotech news media, the answer to that question is yes. As per a prespecified interim analysis, Moderna/Merck have reported that compared with the pembrolizumab-alone control arm, the combination of vaccine and pembrolizumab led to statistically meaningful improvements in recurrence-free survival, which was a primary endpoint of the study, and in distant metastasis-free survival, a key secondary endpoint. Overall survival outcomes await further maturation of the data.
At this conference, we will finally have the opportunity to see the underlying data for ourselves. There are several questions to pay close attention to. The phase II study by Weber et al.11 (NCT03897881) enrolled patients with completely resected melanoma at high risk of recurrence.
The study was open to patients with stages IIIB-IV disease, with the majority of enrolled patients being IIIC, followed by IV and then IIID (no IIIB). These stages were well-balanced between the two arms. INTerpath-001 was open to patients with a wider range of stages (IIB-IV).
Balancing of the disease stages between arms will be critical, as will the balance between the less and more challenging stages. At first pass, one might suspect that inclusion of lower-stage patients might have lowered the bar for a positive result, an important result to understand for real-world applications, and so, stratified analysis will be necessary, particularly to replicate the effect seen by Weber et al.11 in higher-risk patients.
The impact of baseline tumor mutational burden (TMB) was addressed in an earlier report from KEYNOTE-942,12 and although the benefit of combination therapy compared with monotherapy was similar between patients with high and nonhigh TMB, those with high TMB had better performance than those with nonhigh TMB in the pembrolizumab-alone arm. KEYNOTE-942 had some imbalance in the percentage of patients with high TMB between the combination and immunotherapy arms, slightly confounding the analysis. Hopefully, whether or not TMB is a useful biomarker will be further clarified in the analysis of the phase III study.
Similarly, although patient number was limited, the phase II study suggested that activity was independent of programmed death-ligand 1 status, another important point to validate.
The coronavirus 2019 (COVID-19) pandemic established mRNA vaccines as a (relatively) rapidly manufactured and adaptable delivery platform, in part based on the technical advances established during the development of personal neoantigen mRNA vaccines by Moderna and BioNTech, which preceded the pandemic. We hope, and mostly expect, that the INTerpath-001 data will establish the clinical utility of intismeran autogene as a cancer vaccine treatment. But is it the best vaccine platform?
And will effectiveness translate to other more challenging indications? These are unanswerable questions, for now. Well-controlled, head-to-head studies comparing platforms in the same model are lacking, and how well models reflect human biology and real-world human settings further confounds conclusions.
Further, comparative studies may never be carried out except in academia and, without direct industry cooperation and drug supply, might be done with potentially flawed versions of the approved drugs or those in development. With individualized neoantigen vaccines, the comparisons become even more complicated because the epitope set differs per patient, and immunological readouts vary by clinical trial.
All of this results in the default of handwaving, which we will do here, articulating reasons why certain platforms may be better than others. In one example, BioNTech was initially ahead of Moderna in the biopharma race to be first to the clinic.5 However, this was accomplished with naked mRNA cumbersomely injected directly into lymph nodes as the delivery format, and they subsequently rapidly switched to a more tractable format by delivering mRNA formulated as lipoplex nanoparticles. This formulation is similar in part to LNPs but contains a controlled amount of a positively charged ionizable lipid, which efficiently incorporates the RNA and allows intravenous delivery and, most importantly, favors delivery to macrophage and dendritic cells in the spleen, the largest lymphatic organ.13 Using this updated formulation, impressive clinical and immunological results have been observed in the adjuvant setting in pancreatic cancer,14 but even immunological responses are difficult to compare between studies due to methodological differences.
In another example, DNA platforms have the obvious advantage of simpler manufacturing—no need to produce DNA and RNA, especially as RNA manufacturing costs can be prohibitive. Delivery is more challenging, however, entailing potentially painful electroporation after intramuscular injection15 or the use of a device that delivers the vaccine to the muscle with high-pressure fluid flow.16 Thus far, neither delivery approach has shown immunological effects that are clearly superior to either the Moderna or BioNTech mRNA vaccine. Finally, synthetic long peptides (SLPs) were for years the dominant vaccine format but are still plagued by issues of cost, consistent synthesis and formulation across different peptides, and limitations in the number of deliverable epitopes.
In the adjuvant setting for melanoma and renal cell carcinoma, however, the clinical results with SLPs have been exciting but limited in number.4, 17, 18 Using a tactic different from ‘synthetic immunity’, in an effort to recapitulate nature, viral delivery formats have the potential to capitalize on the highly evolved immunological response pathways of viral response.16 Currently under evaluation is an off-the-shelf viral delivery platform utilizing 209 predicted neoepitopes [relevant to patients with Lynch syndrome (LS)] that are encoded in both an adenoviral vector and a modified vaccinia virus Ankara vector (MVA; a derivative of the vaccinia virus strain that contributed to the worldwide elimination of smallpox) in a heterologous prime-boost approach. Robust T-cell responses were observed,19 and we are awaiting release at this same ESMO conference of more clinical results of 1-year follow-ups in LS carriers. A phase III trial aimed at intercepting cancer in LS carriers is being planned, which, with Food and Drug Administration Fast Track Approval status, hopes to launch NOUS-209 as the first approved cancer-intercepting immunotherapy.
In a similar vein, an MVA-based personal neoantigen vaccine has eliminated recurrence compared with the control arm in a small number (n = 32) of patients with head and neck cancer.20
With all that being said, what can we glean regarding the hows and the whys the Moderna/Merck vaccine may work? While all eyes will be focused on Kaplan–Meier curves and P values, hazard ratios and patient statistics, there is still far more that remains unknown regarding the underlying biology of such vaccines. This trial provides fertile ground on which to explore that science and to link detailed immunology with clinical benefit.
The first critical information to be gathered will be how the patients react to the vaccine immunologically. To our knowledge, the only available information for intismeran autogene comes from 10 patients in the first-in-human phase I study9 (KEYNOTE-603, NCT 03313778) and longitudinal sequencing of the T-cell receptor beta subunit (TCRβ) in patients in the phase II trial.21 Weak to moderate ex vivo enzyme-linked immunospot reactivity was detected in most patients, responses were primarily not present or weak before immunization, only 20%-30% of epitopes were detectably immunogenic (after in vitro stimulation), and responses appeared durable (to at least 100 days after last vaccination). Compared with other vaccine formats and trials, this is not a standout.
TCRβ sequencing was more revealing in showing that combination therapy led to higher clonality and higher expansion of novel clonotypes, with lack of recurrence associated with a greater number of new clonotypes. No information on T-cell phenotype or quality is available for intismeran autogene, which is an important gap to fill. Analysis of pretreatment samples, and of recurrent tumors in treatment failures, can provide valuable insights and hypotheses.
Vaccine-specific T-cell infiltration and spatial localization are becoming valuable indicators of effective immune activity.22, 23 Another lesson from failures is investigating whether and how tumor evolution has occurred and how it may have impacted vaccine efficacy.24 Hopefully, these lines of investigation can be pursued. Sample collection is critical for such analysis, including blood, original tumor resections, and biopsies/resections of recurrent tumors, and it is unknown whether such samples have been widely collected in INTerpath-001.
A major, currently unanswerable clinical question, which will require a new trial, is whether the vaccine alone will produce the same benefit. The strong safety profile of the vaccine and the high cost of ICB make this a question that should be answered. Drilling down further, similar questions relate to whether ICB specifically before, during, or after vaccination (or at all timepoints) is contributing to the clinical impact of addition of the vaccine.
Additionally, intismeran autogene delivers neoantigen epitopes and has an adjuvant effect from the mRNA/LNP formulation, and recent studies suggest,25 or not,26 that receipt of a COVID-19 mRNA/LNP vaccine positively impacts the activity of ICB. Formally, INTerpath-001 would not be controlled for the potential contribution of this adjuvanticity on some or all of the activity of pembrolizumab in the combination arm unless an appropriate epitope-less mRNA/LNP placebo was used for the pembrolizumab monotherapy arm (unlikely because the trial began well before this potential confounding data was first presented in 2025).
Intismeran autogene, now the poster child for personalized neoantigen vaccines, is in phase III studies in three non-small-cell lung cancer adjuvant settings and in half a dozen phase II studies in other indications. Thus, there is a breadth of research activity in disease settings beyond melanoma, which has classically been the low-hanging fruit for immunotherapy, and these results will also be much anticipated and provide learning opportunities. There are many barriers to effective vaccine immunotherapy, such as tumor heterogeneity and tumor-intrinsic and extrinsic factors in the tumor immune microenvironment, which will be more or less dominant by disease type and stage, and even within disease type and stage.
Understanding which levers (e.g. disease stage; inclusion and timing of ICB, and which ICB; clinical trial endpoints) to utilize generically or by situation will be increasingly necessary.27 For example, despite apparently strong results in pancreatic cancer with the BioNTech personalized neoantigen cancer vaccine autogene cevumeran (Rojas et al.14), a trial with the same vaccine design in fully resected colorectal cancer patients at high risk of recurrence was recently halted28 because of early data suggesting lack of efficacy. Whether this was due to absence of ICB combination therapy or limited neoantigen numbers or quality in this disease setting, tumor evolution in recurrent tumors, or other factors awaits analysis by the developer. Additionally, in the setting of unresectable or metastatic human papillomavirus (HPV)-positive head and neck cancer, a blinded, phase III randomized trial with an HPV E6 and E7 (which are targets highly analogous to cancer neoantigens) oncogene vaccine in combination with pembrolizumab versus pembrolizumab alone showed no benefit in the first primary endpoint of progression-free survival.29 Whether this is indicative of the challenge of vaccines in general in the active disease setting, or of a deficiency of the vaccine design, is yet to be understood.
The trial will continue until the overall survival endpoint can be evaluated. A positive outcome might suggest that vaccine immunotherapy may require endpoints that require longer timeframes to fully develop.
We all should (hopefully) celebrate for now what is expected to be a huge step forward for the field. At the same time, we cannot take our eyes off the work that will be needed to reveal and demonstrate durable success and even stronger outcomes.