Abstract

Melanoma, owing to its high tumour mutational burden (TMB) and inherent immunogenicity, has emerged as a prime target for neoantigen-based customised cancer vaccines. Such vaccines may synergise with immune checkpoint inhibitors (ICIs) by harnessing patient-specific mutations to trigger targeted T-cell responses. This review systematically summarises and evaluates the clinical evidence and molecular mechanisms underlying customised neoantigen vaccines in melanoma, based on key clinical trial data. A central finding is that vaccine platform choice strongly influences, rather than rigidly determines, the dominant immunological pathway. Messenger RNA (mRNA) platforms generally favour endogenous antigen expression and MHC class I presentation, often eliciting robust CD8+ cytotoxic T-cell responses. By contrast, synthetic long peptide (SLP) platforms are typically processed as exogenous antigens and frequently engage MHC class II presentation, thereby promoting substantial CD4+ T-helper responses. However, this distinction is not absolute, because exogenous peptides can also be cross-presented on MHC class I by professional antigen-presenting cells, enabling CD8+ T-cell priming under appropriate conditions. Clinical data reflects this, with the mRNA vaccine mRNA-4157 (KEYNOTE-942) demonstrating a significant recurrence-free survival (RFS) benefit in the adjuvant setting. This efficacy, however, is contingent on the “hot” tumour microenvironment (TME) of melanoma; “cold” tumours like glioblastoma (GBM) and ovarian cancer (OvCa) present TME-specific barriers (e.g., the Blood-Brain Barrier, immune exclusion) that demand distinct, combination-based vaccine strategies. This review deconstructs this heterogeneity and defines the primary bottlenecks to broad clinical adoption: (1) the need to bridge the “validation gap” by correlating AI prediction accuracy with clinical outcomes; (2) the formidable economic and logistical barriers, including a clinically vulnerable 8–16 week manufacturing wait that poses psychological and clinical risks to patients; and (3) navigating adaptive regulatory pathways for “n-of-1” therapeutics. The field awaits the pivotal Phase III clinical trial of V940-001 (NCT05933577), whose timeline has been extended to 2029. This reflects the logistical and biological complexities inherent in developing personalised vaccines, highlighting challenges in both manufacturing and subject recruitment. These remain key obstacles impeding the widespread clinical application of such vaccines.

1 Introduction

Melanoma, as an aggressive malignant tumour of melanocytes, exhibits one of the highest somatic tumour mutational burdens (TMB) among human cancers, primarily attributable to ultraviolet radiation-induced DNA damage. This high TMB results in a disproportionately high proportion of neoantigens—mutated peptides absent from the normal human proteome and possessing strict tumour specificity. The “foreign” nature of these neoantigens renders them ideal targets for T-cell-mediated immune recognition (1, 2).

The intrinsic immunogenicity of melanoma has been validated by the clinical success of immune checkpoint inhibitors (ICIs). Since 2011, immune checkpoint monoclonal antibodies targeting CTLA-4 (ipilimumab) and PD-1 (pembrolizumab, nivolumab) have revolutionised the prognosis for metastatic melanoma, enabling some to achieve durable long-term survival. However, a significant proportion of patients exhibit primary resistance (non-response) or develop acquired resistance following an initial response, mechanisms often involving immune evasion such as loss of T-cell infiltration or downregulation of antigen presentation pathways (3–6).

Personalised neoantigen vaccines, as tailored therapeutic strategies, hold promise in overcoming these limitations. By identifying patient-specific mutation profiles through next-generation sequencing and screening immunogenic epitopes via bioinformatics, vaccines can be engineered to precisely expand and diversify the patient’s endogenous anti-tumour T-cell repertoire. Unlike vaccines targeting tumour-associated antigens (TAAs)—which are typically overexpressed self-antigens within tumours—neoantigen-specific T cells are not constrained by central tolerance. Theoretically, they can induce high-affinity responses with minimal risk of off-target autoimmune reactions. Preclinical models and early trials demonstrate that such vaccines are safe and reliable, capable of inducing potent CD4+ and CD8+ T-cell responses. They may synergise with immune checkpoint inhibitors by activating novel T-cell clones that are “unleashed” following checkpoint blockade (7, 8).

While mid-stage randomized trials such as KEYNOTE-942 study of mRNA-4157 and the peptide-based trial NCT01970358, have shown encouraging signals of safety and efficacy, the protracted timeline of the pivotal V940–001 trial (now extended to 2029) underscores that translation to routine clinical practice remains a decade away, contingent on resolving manufacturing, economic, and regulatory bottlenecks. Significant variability in outcomes across different vaccine platforms (e.g., mRNA, peptide, dendritic cell) and methodological challenges in data synthesis have obscured a clear understanding of the true clinical potential. This review aims to critically synthesize the current clinical, mechanistic, economic, and regulatory landscape for personalised neoantigen vaccines in melanoma. It will deconstruct the observed heterogeneity by first establishing the core immunological principles differentiating the major vaccine platforms before performing a critical, corrected analysis of the clinical trial data (9–13).

2 The neoantigen immunotherapy pipeline: from silico to clinic2.1 Genomic identification and in silico prediction

The generation of a personalised neoantigen vaccine is a complex, multi-step bioinformatic and manufacturing process (Figure 1). The workflow begins with the collection of tumour tissue and a matched normal sample (e.g., peripheral blood). Tumour and normal DNA are subjected to whole-exome sequencing (WES) or whole-genome sequencing (WGS) to identify somatic (tumour-specific) mutations. Simultaneously, tumour RNA sequencing (RNA-seq) is performed to confirm that these mutations are expressed and to quantify their transcript abundance (8, 13).

Neoantigen identification to immune activation process. This schematic illustrates the workflow: neoantigen identification (NGS, variant calling, HLA typing, epitope prediction) → vaccine preparation (platform-specific) → immune activation (TMB influence, MHC binding, T-cell activation pathways). Arrows depict sequential steps with key tools and challenges noted.

Following variant calling, the patient’s Human Leukocyte Antigen (HLA) haplotype is determined. Bioinformatic algorithms (e.g., NetMHCpan) are then used to predict which of the thousands of mutated peptides (neoepitopes) will bind with high affinity to the patient’s specific MHC Class I or Class II molecules. Rule-based pipelines that prioritize predicted HLA binding often exhibit low positive predictive value for true T-cell immunogenic neoantigens; benchmarking studies show that many top-ranked candidates fail downstream validation. They frequently fail to account for subsequent critical steps in antigen processing and presentation, such as proteasomal cleavage, peptide transport via TAP (Transporter associated with Antigen Processing), and the stability of the final peptide-MHC complex, leading to a high false-positive rate (12, 14).

2.2 The AI-driven prioritization challenge: accuracy vs. reproducibility

To overcome the limitations of simple binding-affinity algorithms, a new generation of Artificial Intelligence (AI) and machine learning models has been developed. These tools, such as imNEO, DeepNeoAG, and ImmuneMirror (Table 1), integrate multi-omic data, including mass spectrometry (MS)-verified immunopeptidomes, gene expression levels, and features of T-cell recognition. These models aim to provide a more composite prioritization score rather than a simple binding score (12, 15).

ModelInput featuresAUCExperimental validationLimitationsDeepNeoAGPeptide sequences from melanoma antigens (no MHC allele info)~ 0.905-fold cross-validation on CEDAR dataset; in vitro binding assaysReproducibility issues in diverse HLA types; limited to melanoma sequencesImmuneMirrorMHC binding affinity, stability rank, agretopicity, multi-omics (e.g., MS data)0.87Validated with hotspot mutations in ESCC/CRC/HCC; binding affinity assays with HLA-A02Training data biases; poor transferability to non-hotspot mutationsimNEOEpitope properties, antigen processing/presentation, T-cell interaction, tumour microenvironment, mutant-wildtype differential>0.85In vivo tumour growth inhibition models; antibody secretion tests; confirmed immunogenicity in multiple cancer datasetsOverfitting to specific cancer types; lacks independent clinical outcome validation

Comparison of AI-based neoantigen prediction models.

While these models report superior performance, with Areas Under the Curve (AUCs) often exceeding 0.85 in in vitro validation datasets (Table 1), a critical “validation gap” persists. The high AUCs reported by these models primarily reflect their accuracy at predicting in vitro peptide-MHC binding, an essential but insufficient proxy for in vivo T-cell activation and, more importantly, clinical efficacy. The translational link between a high prioritization score and a patient’s recurrence-free survival (RFS) has not yet been prospectively established (12).

Terminology used in this review: Binding refers to predicting peptide–MHC affinity/stability in vitro. Presentation refers to predicting whether a peptide is generated and displayed on the cell surface (processing, transport, and ligand elution), which is closer to in vivo biology but still not an immune response. Immunogenicity refers to demonstrable T-cell recognition/activation (functional assays and/or clinical immunomonitoring), and therefore should not be inferred from binding AUC alone.

An additional application of AI-guided prioritization that warrants explicit consideration is the selection of short peptides, or minimal epitopes, designed primarily for HLA class I presentation. In contrast to synthetic long peptides, which often require endosomal uptake and may preferentially expand CD4+ T-helper responses, short-peptide strategies can be used to enrich for candidate neoepitopes with high predicted HLA-I binding, favourable processing features, and greater likelihood of eliciting cytotoxic CD8+ T-cell responses. In this context, AI models are not merely ranking peptide–MHC affinity, but are increasingly being used to refine epitope length, anchor-residue suitability, presentation probability, and, in some cases, T-cell recognition features. These short-peptide approaches therefore represent a mechanistically distinct design strategy within the broader peptide-vaccine landscape, and their inclusion helps explain why peptide platforms should not be treated as uniformly CD4+-dominant (12, 14, 15). Nevertheless, short-peptide approaches are not universally superior, because their performance remains constrained by HLA restriction, peptide stability, and the risk of incomplete helper T-cell support.

This validation gap is exacerbated by what has been termed the “neoantigen algorithm reproducibility crisis”. Many AI models exhibit performance bias due to their training data. For example, models trained on datasets enriched for specific HLA alleles or on melanoma-specific sequences (which are abundant) may not generalize well to other cancer types or patients with different HLA haplotypes. This lack of standardization and poor transferability remains a major scientific and regulatory hurdle (12).

Although Table 1 summarizes predictive performance, underlying architectural diversity contributes significantly to the observed spread in AUCs. DeepNeoAG utilizes a recurrent convolutional neural network trained on CEDAR peptide datasets, focusing on motif recognition independent of HLA allele context. ImmuneMirror instead applies ensemble gradient-boosting with integrated binding-stability and ligand-elution data, achieving higher biological interpretability. imNEO extends this framework by introducing TCR–epitope co-features, improving recall but at risk of melanoma-biased overfitting (12).

2.3 Addressing the reproducibility crisis in neoantigen AI models

While the field has made significant progress, two structural weaknesses underlie the reproducibility crisis in neoantigen prediction.

First, most machine learning pipelines are trained on heavily biased datasets—particularly overrepresented HLA-A02:01* alleles and melanoma-derived immunopeptidomes. As a result, model accuracy often collapses when applied to rarer HLA haplotypes or non-melanoma tumours. Second, benchmark datasets such as IEDB or CEDAR lack unified standards for peptide length, affinity thresholds, or negative-sample definition, inflating in vitro AUCs without reflecting true immunogenicity (12).

Emerging solutions emphasize cross-cohort benchmarking and synthetic augmentation of rare HLA alleles to enhance generalizability. In parallel, regulators have begun framing AI-based pipelines as “software as a medical device (SaMD)” components under Chemistry, Manufacturing and Controls (CMC) standards. This shift allows algorithmic validation to become a formal part of regulatory review, potentially transforming the current ad hoc research tools into auditable, clinical-grade systems (16–18).

3 Core immunological mechanisms: differentiating platform efficacy

The clinical heterogeneity observed in neoantigen vaccine trials is not random; it is, in large part, shaped by the distinct antigen-processing and presentation pathways preferentially engaged by different platforms. The choice of an mRNA- or peptide-based vaccine can bias the immune response toward particular T-cell compartments, but does not rigidly confine it to either CD8+ cytotoxic or CD4+ helper immunity (7, 8, 13).

3.1 The endogenous pathway: mRNA vaccines and CD8+ T-cell priming

Messenger RNA (mRNA) vaccines, typically encapsulated in lipid nanoparticles (LNPs), are delivered directly into the cytoplasm of cells, primarily antigen-presenting cells (APCs) such as dendritic cells. Once inside, the mRNA is translated by the host cell’s own ribosomes, producing the neoantigen protein endogenously (i.e., inside the cell) (8, 13).

This intracellular origin is immunologically critical. Endogenously synthesized proteins are processed by the proteasome into short peptides. These peptides are then transported by TAP into the endoplasmic reticulum, where they are loaded onto MHC Class I molecules. The peptide-MHC-I complex is then trafficked to the cell surface.

Presentation on MHC class I provides a major route for CD8+ cytotoxic T-lymphocyte (CTL) priming. This helps explain why mRNA vaccine trials often show strong CD8+-skewed responses, although accompanying CD4+ responses can also contribute meaningfully to anti-tumour immunity. This direct priming of CTLs—the immune system’s primary tumour-killing cells—provides a strong mechanistic rationale for the synergy observed between mRNA vaccines and anti-PD-1 ICIs. The vaccine primes an army of new tumour-specific killers, and the ICI releases the PD-1 “brake,” allowing them to execute their function (8, 9, 19).

3.2 The exogenous pathway: peptide vaccines, helper T-cell priming, and cross-presentation

In contrast, synthetic long peptide (SLP) vaccines, which are co-administered with an adjuvant (e.g., Poly-ICLC, Montanide) to stimulate APCs, are exogenous antigens. They are taken up from the extracellular space by professional APCs via endocytosis or phagocytosis (7, 13). It is also important to distinguish SLP vaccines from short-peptide or minimal-epitope formulations, which are often intentionally designed for MHC class I loading and CD8+ T-cell activation and therefore should not be mechanistically collapsed into the same category.

These exogenous peptides traffic through the endolysosomal pathway, where they are processed and loaded onto MHC Class II molecules. The peptide-MHC-II complex is then presented on the APC surface.

Presentation on MHC class II is a major mechanism through which SLP vaccines activate CD4+ T-helper cells, and this likely contributes to the high-frequency CD4+ responses reported in several peptide-vaccine trials. However, this pathway should not be interpreted as exclusive. After uptake by professional APCs, exogenous peptide antigens can also enter the MHC class I pathway through cross-presentation, thereby generating CD8+ T-cell responses under favourable biological and adjuvant conditions. Accordingly, peptide-based vaccines should be viewed as platforms that often favour CD4+ helper immunity but remain capable of inducing mixed CD4+/CD8+ responses, with the balance depending on peptide design, APC subset engagement, adjuvant choice, and antigen-processing efficiency. This mechanistic difference may underlie the more variable efficacy signals observed with peptide-based platforms (7, 13, 20).

Although the majority of current pipelines emphasize MHC-I–restricted CD8+ T-cell epitopes, the contribution of CD4+ T-cell responses via MHC-II presentation remains underexplored.

Prediction algorithms such as NetMHCIIpan 4.1 now allow high-throughput identification of HLA class II–restricted epitopes, though their accuracy remains lower than class I counterparts (14).

Incorporating MHC-II predictions may help explain why several peptide vaccine trials demonstrated robust CD4+ responses without corresponding clinical benefit, suggesting a need for balanced epitope selection (21, 22).

3.3 Vaccine platforms: a mechanistic and logistical comparison

The choice of platform involves a trade-off between the desired immune response, manufacturing speed, cost, and logistical stability. mRNA vaccines, for example, offer rapid manufacturing but require a stringent cold chain, whereas peptides are more stable but have a longer synthesis time. These differences are summarized in Table 2 (13, 16).

PlatformKey immune pathwayAntigen processingDominant/typical T-cell responsePreparation timeAdvantagesLimitationsmRNAPredominantly MHC Class I, with secondary MHC II engagementEndogenous (cytosolic)Often CD8+-skewed, with supportive CD4+ responses4–6 weeksStrong cellular immunogenicity; rapid, scalable manufacturingHigh cost; requires cold chainPeptide/SLPPredominantly MHC Class II, but may access MHC I via cross-presentationExogenous (endolysosomal; cross-presentation possible)Frequently CD4+-dominant, but mixed CD4+/CD8+ responses are possible6–10 weeksStable; simple production; flexible epitope designCD8+ priming may be variable and depends on cross-presentation efficiencyDendritic Cell (DC)MHC Class I & IIEx vivo loadingDual (CD4+/CD8+)8–12 weeksPrecise antigen loading; potent dual activationDifficult to scale; high cost; complex ex vivo manufacturing

Mechanistic and logistical comparison of vaccine platforms.

4 A critical review of clinical evidence in melanoma4.1 Methodological note on synthesis: a narrative review, not a meta-analysis

A quantitative meta-analysis of neoantigen vaccine trials is precluded by the profound heterogeneity across studies. Trials differ in vaccine platform (mRNA vs. peptide), adjuvant used (e.g., Poly-ICLC, Montanide, CDX-1140), patient population (adjuvant stage III or IV vs. metastatic), comparator (ICI monotherapy vs. single-arm), and primary endpoints (RFS vs. ORR) (9, 11, 21–23).

Presenting these data in a single “forest plot” or calculating a “weighted average” effect size, as has been attempted, is statistically invalid and highly misleading. Combining heterogeneous endpoints like Hazard Ratios (HRs), which are time-to-event measures, and Objective Response Rates (ORRs), which are dichotomous proportions, on a single visual axis is methodologically unsound. Furthermore, attempting to “approximate” an HR from a Kaplan-Meier proportion ignores censored data and violates the proportional hazards assumption, rendering the estimate uninterpretable. Any weighting by simple sample size (), rather than by the inverse variance of the effect estimate, is not a valid meta-analytic technique.

Therefore, this review will not conduct a meta-analysis. Instead, it adheres to the Synthesis Without Meta-analysis (SWiM) guidelines by presenting a critical narrative synthesis, with results from the six key trials summarized descriptively in Table 3. A risk-of-bias assessment (Table 4) highlights that all single-arm trials included suffer from a high risk of selection and reporting bias (24).

Trial IDPlatform/agent(s)nPopulationKey efficacy and immunogenicity outcomesStatus/limitationsKEYNOTE-942 (NCT03897881)mRNA (mRNA-4157) + Pembrolizumab157Adjuvant (Resected Stage IIIB-IV)Efficacy (3-yr): RFS HR 0.510 (95% CI 0.288–0.906); DMFS HR 0.384. Sustained CD4+/CD8+ responses.Positive. Phase 2b, Randomized.NCT01970358Peptide (NeoVax) + Poly-ICLC8 (LTFU)Adjuvant (Resected Stage IIIB-IV)Efficacy (~4-yr): 6 of 8 (75%) patients disease-free. Immunog: Persistent, diversified memory T-cell responses.Positive. Phase 1, Single-arm.NCT05309421Peptide (EVX-01) + Pembrolizumab16Metastatic (Unresectable)Efficacy (2-yr): 75% ORR (12/16) (95% CI 0.51–0.90). Durable: 92% (11/12) of responses sustained at 24 mos.Positive. Phase 2, Single-arm.NCT03929029Peptide (NeoVax) + Montanide + Ipi/Nivo11MetastaticEfficacy (Final): 36% ORR (4/11) (95% CI 0.15–0.65). Immunog: Responses in 8/11.Inconclusive. Phase 1b. Final results posted Oct 2024.NCT04072900Peptide + Toripalimab (Anti-PD-1)30MetastaticEfficacy: 10% ORR (3/30) (95% CI 0.03–0.26).Negative. Phase 1, Single-arm.NCT04364230Peptide (Shared Ag + neoAg-mBRAF) + Adjuvants22Adjuvant (Resected, Disease-Free)Immunog: 27% (6/22) ex vivo CD4+ response to 6MHP (shared Ag). No ex vivo response to neoantigen (1 patient IVS-positive).Negative (Immunog.). Phase 1/2. No efficacy endpoint; 73% (16/22) was adjuvant dose, not RFS.

Summary of key clinical trials of personalised neoantigen vaccines in melanoma.

DomainKEYNOTE-942NCT01970358NCT03929029NCT04364230NCT04072900NCT05309421Overall assessmentSelection BiasLow (Randomized)High (Single-arm)High (Single-arm)High (Single-arm)High (Single-arm)High (Single-arm)HighPerformance BiasModerate (Partial blinding)Moderate (Partial blinding)Moderate (Partial blinding)Moderate (Partial blinding)Moderate (Partial blinding)Moderate (Partial blinding)ModerateDetection BiasLow (Objective endpoints)Low (Objective endpoints)Low (Objective endpoints)Low (Objective endpoints)Low (Objective endpoints)Low (Objective endpoints)LowAttrition BiasLow (Low dropout)Low (Low dropout)Low (Low dropout)Low (Low dropout)Low (Low dropout)Low (Low dropout)LowReporting BiasModerate (Selective outcomes)High (Selective outcomes)High (Selective outcomes)High (Selective outcomes)High (Selective outcomes)Moderate (Selective outcomes)HighOther (Heterogeneity)High (Patient variability)High (Patient variability)High (Patient variability)High (Patient variability)High (Patient variability)High (Patient variability)High

Risk of bias assessment (RoB 2 for randomized trials; ROBINS-I domains for non-randomized single-arm studies).

Bias assessment followed qualitative domains adapted from the ROBINS-I framework, including selection bias, confounding, and measurement bias. Trials with non-randomized design were rated high in at least one domain but varied in magnitude. A semi-quantitative bias score was intentionally omitted in accordance with SWiM narrative-synthesis guidance; qualitative grading was used to reflect methodological robustness without introducing pseudo-numeric precision.

4.2 Summary of clinical trial evidence

The clinical evidence for neoantigen vaccines in melanoma is derived from a small number of key Phase I/II trials (Table 3). A critical re-examination of the data from these trials, including recent 2024–2025 updates, reveals a more complex and nuanced picture than previously reported (9, 11, 21–23, 25, 28).

4.3 Visualizing platform-specific clinical outcomes

To clarify the relationship between vaccine type, prediction pipeline, and outcome, Figure 2B stratifies the key melanoma trials by platform and AI model sophistication.

Durability of T-cell responses in selected neoantigen vaccine trials. X-axis: Time (Months); Y-axis: T-cell Response Rate (%). Data for KEYNOTE-942: ~80% at 6 months, 75% at 12, 70% at 24, ~65% at 36 (preliminary data from conference abstract; final results may update); NCT01970358: 100% at 6/12, 67% at 24; NCT03929029: 73% at 6/12; others limited. (B). Platform-specific clinical outcomes by AI sophistication. Each data point represents one clinical trial: the x-axis denotes vaccine platform (mRNA, peptide, or dendritic cell), the y-axis represents the principal endpoint (RFS or ORR), and bubble size corresponds to sample size. Colour intensity reflects the level of AI integration—ranging from rule-based affinity predictors (light) to deep-learning platforms (dark).

Although the clonal architecture of tumours has been recognized as a major determinant of vaccine efficacy, only a minority of ongoing trials explicitly incorporate clonal analysis in antigen selection. Although tumour clonality is increasingly recognized as important, explicit clonality-informed antigen selection is not consistently reported across trials, and thresholds/pipelines remain non-standardized.

In practice, clonality-informed antigen selection depends not only on mutation detection by NGS, but also on how variant allele fraction is interpreted after accounting for tumour purity, local copy-number status, sequencing depth, and sampling bias. As a result, there is no universally accepted NGS-based cutoff that cleanly separates clonal from subclonal neoantigens across studies. For this reason, clonality should be treated as a probabilistic prioritization feature rather than a binary rule and ideally interpreted together with RNA expression and antigen-presentation likelihood.

Incorporating clonal frequency estimates into antigen prioritization pipelines may improve the likelihood of targeting truncal neoantigens that persist across metastases. However, real-time implementation remains constrained by the computational cost and lack of standardized thresholds for clonality calls (26, 27).

Trials employing deep-learning–guided neoantigen selection (e.g., EVX-01, imNEO) cluster in the upper-right quadrant, reflecting higher ORR and durability, even with comparable baseline TMB and disease stage (Supplementary Table 1) (28).

4.4 Synthesis of clinical signals

The corrected trial data (Table 3) reveals distinct patterns.

For mRNA Platform (KEYNOTE-942), the randomized Phase 2b KEYNOTE-942 trial provides the highest-quality evidence to date. The combination of the mRNA vaccine mRNA-4157 with pembrolizumab resulted in a clinically and statistically significant improvement in RFS. The 3-year update confirmed a durable benefit, with an RFS HR of 0.510 (49% risk reduction) and a distant metastasis-free survival (DMFS) HR of 0.384. In the randomized phase 2b KEYNOTE-942 trial, the observed 49% relative reduction in recurrence or death was demonstrated within a cohort of completely resected high-risk stage IIIB-IV melanoma. Because recurrence risk differs across TNM categories, direct comparison with external populations should be interpreted cautiously unless stage-matched subgroup data are available. This result strongly supports the hypothesis that vaccine-induced cellular immunity, including a substantial cytotoxic T-cell component, can add tangible benefit to checkpoint blockade (9, 19, 25).

For Peptide Platforms (A Heterogeneous Picture), they show significant variability.

4.4.1 Positive signals

NCT01970358, the first-in-human NeoVax peptide study, demonstrated remarkable long-term persistence of T-cell memory and durable disease control, with 75% of patients (6/8) remaining disease-free at a median follow-up of nearly 4 years. More recently, NCT05309421 (EVX-01) showed a high 75% ORR in metastatic patients, and critically, these responses were highly durable, with 92% (11/12) sustained at 2 years. This trial’s success was explicitly linked to its AI-driven prediction platform, which reportedly achieved 81% accuracy in predicting T-cell responses (7, 11, 20, 28).

4.4.2 Negative/inconclusive signals

The narrative is balanced by clear negative signals. NCT04072900, which combined a peptide vaccine with the anti-PD-1 inhibitor toripalimab in metastatic patients, reported a dismal 10% ORR (3/30) (21).

4.4.3 The NCT04364230 correction

A critical re-analysis of NCT04364230 is required. This trial was previously misinterpreted as showing a 73% (16/22) relapse-free rate. However, recent 2024 data (SITC poster 1466) clarifies this is factually incorrect. The 22 patients were enrolled disease-free. The “73%” figure (16/22) referred to the proportion of patients who received the maximum dose of the adjuvant, not an efficacy outcome. The actual primary endpoints were safety and immunogenicity, and the latter was poor: a CD4+ T-cell response to the shared antigen component was seen in only 27% (6/22) of patients ex vivo, and responses to the neoantigen component (neoAg-mBRAF) were essentially undetectable ex vivo (22).

This synthesis suggests that vaccine success is not a simple “mRNA vs. peptide” dichotomy. Rather, it is contingent on the quality of the antigen prediction (e.g., the 81% accuracy in the successful EVX-01 trial) and the biological context of the patient (Figure 5). As summarized in Supplementary Table 1 and visualized in Figure 2B, trials using deep-learning–guided neoantigen selection appear to cluster with higher response rates and durability, supporting the view that prediction quality may be a dominant driver of clinical benefit (26, 28).

4.5 Emerging biomarkers for neoantigen vaccine response

While neoantigen-based vaccines have demonstrated promising efficacy in select melanoma cohorts, predictive biomarkers that delineate responders from non-responders remain underdefined. A multidimensional biomarker strategy is therefore essential to guide patient selection (Table 5), monitor vaccine-induced immunity, and optimize clinical outcomes.

BiomarkerTypeEvidence of predictive valueKey limitationsTumour Mutational Burden (TMB)PredictiveCorrelates with neoantigen load and ICI responseNot all mutations generate immunogenic peptidesNeoantigen ClonalityPredictiveTruncal antigens linked to durable T-cell responsesRequires deep sequencing; difficult to quantify subclonal targetsIFN-γ Signature/MHC-I ExpressionPredictiveReflects immune readiness and antigen presentationInfluenced by inflammation and therapy-induced modulationctDNA ClearanceDynamicTracks vaccine-induced tumour regressionLow sensitivity in minimal disease statesTCR Repertoire DiversityDynamicIndicates clonal expansion and persistenceRequires paired pre/post samplesComposite AI ModelsIntegrativeMultimodal predictor of clinical benefitLack of standardization and external validation

Emerging biomarkers for neoantigen vaccine response.

a. Predictive biomarkers

Tumour mutational burden (TMB) has been consistently correlated with the abundance of neoantigens; however, its predictive precision is limited by intertumoural heterogeneity and non-immunogenic passenger mutations. Clonality of neoantigens—particularly those derived from truncal mutations—has shown stronger association with durable responses compared to subclonal targets. In parallel, baseline interferon-γ (