Developing neoantigen cancer vaccines: where are we now? Authors: Nune Markosyan, Robert H Vonderheide Published: 2026 ABSTRACT AbstractThe success of immunotherapies such as immune checkpoint blockade and engineered T-cells has demonstrated the immune system’s anticancer potential. Although prophylactic vaccines against oncogenic viruses such as human papillomavirus prevent cancer by blocking infection, therapeutic vaccines targeting tumour-associated or tumour-specific antigens (TAAs/TSAs) have not reliably produced meaningful tumour regressions. Advances in DNA sequencing and computational pipelines for predicting individual, MHC-restricted neoantigens now enable personalized cancer vaccines (PCVs). By targeting neoantigens arising from somatic mutations, PCVs escape the central tolerance that limits anti-TAA vaccines. Multiple clinical trials have shown promise, but PCV design, delivery, and optimal disease settings require further refinement.Subject terms: Tumour vaccines, Immunosurveillance Introduction Whether or not the immune system recognises and resists invasive tumours has been uncertain, even doubted, for decades. The discovery of immune checkpoint molecules, such as CTLA-4, PD-1 and PD-L1, and the clinical benefit of immune checkpoint inhibitors (ICIs) in advanced cancer overturned this view. Antibody ICIs reactivate pre-existing anti-tumour T cells, driving regression of even large tumours. Tumour mutational burden (TMB) and T-cell infiltration proved predictive of ICI response, illuminating the underlying T-cell specificity1–4. In this review, we will focus on one notion that has emerged over the last 15 years—T-cells reinvigorated by ICI therapy target tumour neoantigens arising from new mutations or aberrant expression of non-mutated genes, and that novel therapeutic vaccines can be designed specifically against them. Tumour antigens are classified as (i) tumour-associated antigens (TAA) differentially expressed in the tumours compared to the normal tissues (Table 1) and (ii) tumour-specific antigens (TSA) uniquely expressed in cancer cells (Table 2). High expression of TAAs, with minimal presence in normal tissues and the prospect of generating a de novo or reviving suppressed T-cell response, has been the basis for anti-cancer vaccines. As TAAs are shared across cancer types (TERT, EGFR and CTA) or within the tumour type (HER2, PAP), these antigens have been considered for the development of broadly applicable 'off-the-shelf' vaccines. Many targets and numerous clinical trials later, TAA-based therapeutic vaccines have become less common due to modest clinical responses in established diseases5–10. Nearly all TAA vaccines were safe, and many elicited T-cell responses, but the objective clinical responses were rare in patients with advanced disease. Twenty years of attempts to develop shared anti-cancer vaccines have revealed that central tolerance, tumour-cell-intrinsic resistance, immune exclusion, direct immunosuppression by established tumours, and prior immune-toxic chemotherapies can compromise the effect of the vaccines11,12. Table 1TAA – proteins and glycoproteins present in normal tissues but overexpressed in cancersPAP (prostatic acid phosphatase)Enzyme present in normal prostate, but levels proportionally increase with prostate cancer progression93Muc-1Highly glycosylated protein with low luminal or apical expression in healthy epithelial cells. The hypoglycosylated form of MUC-1 with no polarity is overexpressed in adenocarcinomatous cells94.CTA (cancer/testis antigens)The expression of these proteins (NY-ESO-1, MAGEs, CAGE, BAGE, GAGE, CEA and AFP) is restricted to fetal tissues and germ cells in adult and fetal testis, fetal ovary and trophoblasts, but genes encoding for CTA are aberrantly activated in various types of cancers95. NY-ESO-1 and MAGE are overexpressed in melanoma, bladder, NSCLC and other cancers. CEA (carcinoembryonic antigen) is overexpressed in gastrointestinal cancers, and AFP (alpha-fetoprotein) levels are elevated in early-stage hepatocellular carcinoma42,96GP100Melanocyte lineage glycoprotein used for the diagnosis and prognosis of melanoma. High expression in melanoma elicits HLA-A2.1 restricted cytotoxic T-cell response97.EGFR (epidermal growth factor receptor 1) and HER2 (human epidermal growth factor receptor 2)Receptor tyrosine kinases involved in normal cell differentiation, growth, survival and proliferation. These genes are amplified, mutated or overexpressed in breast cancer, NSCLC, glioblastoma and other cancers98,99.TERT (telomerase reverse transcriptase)Expressed at very low levels in healthy stem cells, bone marrow and testis, but upregulated in 85% of cancers, making it a universal tumour-associated antigen100,101. TAA – proteins and glycoproteins present in normal tissues but overexpressed in cancers Table 2TSA – proteins uniquely expressed by tumour cells and not found in normal tissuessTSA (shared tumour-specific antigen)Neoantigens arising from tumour driver mutations. Most of these mutations are exclusive to a particular patient’s tumour, but others are in recurrent 'hot spots' in oncogenes such as Kras mutations in pancreatic, lung and colorectal cancers, BRAF in melanoma, IDH1 in gliomas and AML, PIK3CA in breast and BCR-ABL fusion in chronic myeloid leukemia18. Human papilloma virus (HPV) DNA and E6 and E7 epitopes can be considered TSAs as they are present in various cancers, reaching 91% in cervical and anal cancers102. Like TAAs, sTSAs can be targets for 'of-the-shelf' vaccines.iTSA (individual tumour-specific antigen)Neoantigens arising from somatic mutations that occur in non-reproductive or germline cells and are acquired during a person’s lifetime. These neoantigens are present in individual patient tumours, are absent in the healthy tissues of the same patient and are not shared as they arise from genomic, transcriptomic and proteomic variants in individual tumours18,103. iTSA are targets for personalised anti-tumour vaccines. TSA – proteins uniquely expressed by tumour cells and not found in normal tissues Unlike TAAs, TSAs arise from cancer-cell-specific mutations or viral infections, are absent in normal tissues, and are potentially more suitable therapeutic targets. Human papillomavirus (HPV) vaccines that prevent 97% of infections caused by nine HPV strains were repurposed to target HPV-16 E6 and/or E7 antigens present in cervical tumours13. Despite encouraging results in preclinical models of large HPV-positive tumours, the vaccine produced poor clinical responses in patients14,15. Major histocompatibility complex class I (MHC-I) downregulation, defective antigen processing in cancer cells, and the immunosuppressive tumour microenvironment (TME) mask these viral antigens, rendering the tumours resistant to HPV antigen vaccines. However, the vaccine still holds promise for treating precancerous lesions, especially when combined with other immunotherapies16. Recently, vaccine development has shifted toward a more individualised approach, targeting individual tumour-specific neoantigens selected in the context of patients’ MHC types (Table 3). This approach, combined with the emergence of new tools, initiated the boom of personalised cancer vaccines (PCV). Next-generation sequencing (NGS) has been the catalyst for PCV development. Since its emergence in the mid-2000s, this platform of massively parallel DNA and RNA deep sequencing has evolved to better recognise the complexities of genomes. The unrivaled capacity and cost-effectiveness of NGS have made it an invaluable research and clinical tool, permitting the fast detection of tumour-specific mutations in individual patients17,18. Tumour and healthy tissue whole-genome, exome, or transcriptome sequencing combined with variant calling tools, HLA typing and MHC binding prediction pipelines identify non-synonymous somatic mutations and their product peptides with potential immunogenicity18–20. Advances in predicting immunogenic mutant peptides, neoantigens, have propelled preclinical studies and clinical trials that test various cell-, nucleic acid- and peptide-based vaccine platforms (Table 4), routes of administration, and vaccine-drug combinations. The goal remains to identify optimal strategies for neoantigen repertoire and vaccine platform selection, as well as the therapeutic window, to maximise the benefits from this promising approach. Major cancer-specific efforts in PCV development are discussed below. Table 3Neoantigens: sources and selection for PCVSomatic genome mutationsInclude single nucleotide variants (SNV), insertions and deletions (INDEL) and gene fusions, which together with viral sequences and endogenous retroviral (ERV) genes can either generate mutated versions of existing peptides or create new open reading frames (ORF). Classically defined neoantigens have mutations at the T-cell receptor binding interface, whereas alternatively defined neoantigens arise from mutations at the MHC groove-anchoring amino acids, resulting in improved MHC binding and higher immunogenicity compared to their wild-type counterparts104,105. In addition, mutations in non-coding regions of the genome generate cancer-specific and immunogenic non-canonical MHC-I-binding peptides, referred to as cryptic neoantigens74.RNA variantsResult from alternative splice sites, exon skipping/inclusion, intron retention and A-to-I editing that can lead to the production of novel mature mRNAs and peptides.Translational modificationsAlternative start sites, defective translation, non-coding open reading frames (nuORFs) and coding long non-coding RNA (lncRNA), together with post-translational hyper- and hypo-glycosylation, can generate neoantigens18,106. Various bioinformatics pipelines perform neoantigen identification (variant calling) by comparing an individual’s tumour whole-genome or whole-exome sequences (WGS and WES, respectively) to the healthy tissue genome19. The expression of novel sequences is confirmed by RNA sequencing and proteomics. While mass spectrometry can confirm neoantigens predicted from genome and transcriptome sequencing, it can also identify novel peptides generated through aberrant translation18.Neoantigen selection for PCV typically requires patient HLA typing, MHC binding prediction and epitope prioritisation based on MHC binding affinity19,106,107. Further prioritisation of neoantigens based on dissimilarity to self and homology to infectious disease-derived peptides, protein abundance, proteasomal processing potential and T-cell recognition are steps that can allow for selecting the most immunogenic neoantigens20,49,61. Neoantigens: sources and selection for PCV Table 4Vaccine formats and adjuvantsCell vaccinesAntigen-presenting cells (APC), mostly dendritic cells (DC), have been used as cell vaccines. Autologous blood monocyte-derived DC, in vitro matured and activated, can be pulsed with neoantigens or loaded with whole tumour lysates and infused back into the bloodstream40,108. Otherwise, peptide-loaded DCs or immortalised B cells can be co-cultured with T-cells isolated from the tumours (TIL) to activate and expand them for downstream adoptive transfer28,41,83. Using autologous, peptide-loaded APCs for neoantigen delivery is the most physiological way, but the required ex vivo manipulations are labour and time-demanding, especially when combined with TIL extraction and ex vivo activation.Nucleic acid platformsThe advantage of DNA and mRNA platforms is the potential to encode the neoantigens together with DAMPS (damage-associated molecular patterns), cytokines and co-stimulatory molecules32,50,67,70,109. DNA molecules have longer expression than the mRNA, but they require electroporation for delivery and cannot be targeted to APCs. The inherent instability of mRNA molecules and delivery to APCs was addressed by including them in liposome formulations that are preferentially internalised by DCs110. In addition, the lipids in nanoparticles and mRNA molecules themselves have immunostimulatory effects that need to be dampened or tweaked for optimal results110,111.Proteins and peptidesRecombinant proteins and synthetic long peptides (SLP) are so far the most common neoantigen delivery vectors34,36,38,63. These modular vectors that deliver neoantigens can include domains that target cross-presenting APC. SLP of up to 30 amino acids can contain MHC-I and II epitopes. However, these vectors have low solubility and lack intrinsic immunogenicity.Microbial vectorsDue to their tropism for hypoxic tumour regions and cancer cells and high immunogenicity, bacteria and viruses expressing neoantigens, cytokines and immune checkpoint inhibitors (ICI) can be leveraged as potent cancer vaccines. The highly complex manufacturing, viability, genetic variability between batches, and additional time that increases the already lengthy PCV-generating processes are the difficulties that limit the use of microbial vectors for neoantigen delivery112.Vaccine adjuvantsStimulate innate immune response, help to activate adaptive immunity, and improve vaccine delivery. Various adjuvants used in clinical trials: TLR3/MDA5 agonist poly-ICLC to trigger innate immunity and activate APC37, recombinant GM-CSF Sargramostim to support DC63, agonists to CD40 for DC licensing113, IL-12 to activate T-cells41, liposome-based cationic inducer of DC uptake and cross-presentation CAF®09b39, tetanus toxoid peptide to boost helper T-cell activity72, Montanide, a water-in-oil emulsion to create a depot and slowly release the antigens37. In addition, antigen-agnostic in situ strategies to modify TME through intra-tumoural administration of adjuvants, oncolytic viruses, autologous and allogeneic DCs are used in patients with accessible diseases103. Vaccine formats and adjuvants Melanoma The data from 4742 tumour-normal tissue pairs revealed extraordinary variation in mutation frequency and spectrum among and within human cancer types21,22. Melanoma, lung, and urothelial cancers demonstrated high tumour TMB, making them a common focus for PCV development. Mutanomes—unique sets of tumour-specific neoantigens—were identified in three independent murine models of melanoma, colon and breast cancers23. Mutations in the B16F10 melanoma cell line were validated and prioritised based on their MHC-I and MHC-II binding predictions. Fifty 27-mer neoantigens were administered to mice as peptide or mRNA vaccines. Both vaccine formats elicited immune responses against 30% of neoantigens, mostly CD4+ T-cell-mediated. A single MHC-II epitope profoundly retarded the growth of B16F10 tumours in syngeneic mice. The study demonstrated that a single strong MHC binder with a high expression level can overcome the deeply immunosuppressive TME of B16F10 tumours. Substantial evidence supports the ability of CD8+ tumour-infiltrating lymphocytes (TIL) to recognise tumour neoantigens24–27. Early-stage clinical trials with cell-based PCVs indicated an equally important role for CD4+ TIL in neoantigen recognition. In a cholangiocarcinoma patient, adoptively transferred neoantigen-reactive CD4+ TIL induced tumour and metastasis regressions27. In a group of five stage IV melanoma patients, CD4+ TIL obtained from patients’ resected melanoma increased IFNγ production in response to stimulation with tumour-specific neoantigen (NCT001495066)28. Autologous neoantigen-specific CD4+ TILs increased IFNγ production when incubated with tumour cells in vitro, indicating MHC-II expression on melanoma cells. Earlier studies have shown constitutive or IFNγ-induced expression of MHC-II in melanoma cell lines29. Together, these studies provided a rationale to include MHC-II binders in PCV design. The mRNA PCV approach tested in mice23 was advanced to a clinical trial in which 13 patients with stage III and IV melanoma received 20 naked, non-modified mRNAs encoding 27-mer neoantigens predicted to bind MHC-I and II, via lymph node injections (NCT02035956)30. All patients responded to at least 3 neoantigens; 60% of the predicted neoantigens elicited T-cell responses, and 57% of which were recognised by CD4+ T-cells. The vaccine reduced the recurrence of metastatic events and sustained progression-free survival (PFS). Neoantigen-responsive T cells were PD-1-positive, and one patient with no clinical response to the vaccine responded to anti-PD-1 with 80% tumour reduction. An important observation was that one neoepitope can be presented on different HLA restriction elements and recognised by CD4+ and CD8+ T-cells with different TCR clonotypes. The study also provided a strong rationale for combining PCVs with ICI. A combination of an mRNA vaccine and the anti-PD-1 pembrolizumab was tested in a phase 1 trial (NCT03313778), in which 11 stage III/IV melanoma and four non-small cell lung cancer (NSCLC) patients received mRNA-4157, a modified mRNA encoding up to 34 neoantigens encapsulated in lipid nanoparticles (LNP)31. Ten patients with pre-treatment samples had low or no preexisting T-cell response to neoantigens, which did not change after run-in pembrolizumab administration. Post-mRNA, 3/3 vaccine-only patients showed responses to 22.7% of neoantigens, whereas 7/7 combination-treated patients responded to 29.5% of neoantigens. The vaccine elicited consistent de novo CD8+ and CD4+ T-cell responses in 60% and 40% of patients, respectively, and strengthened the pre-existing ones. Thirty-six months post-enrollment, recurrence was reported only in 1/15 subjects. The same modified mRNA-4157 vaccine, renamed V940, was evaluated in a two-arm, phase 2b randomised trial (NCT03897881), where 107 patients with high-risk, resectable, stage III/IV melanoma were treated with vaccine+pembrolizumab and 50 patients with pembrolizumab only32. At a median follow-up of 24 months, there was a 44% reduction in disease recurrence in the vaccine+pembrolizumab group compared with pembrolizumab alone. At the 3-year median follow-up, the risk reduction in recurrence and/or death in the combination group reached 49% and stayed stable at the 5-year follow-up33. The clinical benefit in the combination group was independent of TMB and PD-L1 expression, implying that the vaccine-broadened immune response can tip the balance when overlaid on strengthened pre-existing immunity. An ongoing phase 3 clinical trial (NCT05933577) aims to confirm the clinical results and provide data on rates, depth and durability of immune responses. Another vaccine format, NeoVax, based on synthetic long peptides (SLP), was assessed in a phase 1 trial in stage III/IV melanoma patients (NCT01970358)34. Up to 20 SLPs with predicted MHC-I neoepitopes were administered with poly-ICLC. The vaccine induced polyfunctional T-cell responses against 73/97 unique neoantigens, with 60% mediated by CD4+ T-cells, despite the use of predicted MHC-I binders. Twenty-five months post-vaccination, 4/12 patients had no recurrence; two with recurrent disease achieved a complete response after anti-PD-1. Four years post-vaccination, six patients remained disease free35. Blood analyses revealed T-cells with a memory phenotype and epitope spreading. It is worth emphasising that 15–30-amino-acid (aa)-long MHC-I-binders elicited MHC-II binder-specific responses. Building upon NeoVax, a phase 1b trial was conducted in patients with non-resectable melanoma, NSCLC and bladder cancer (NCT02897765)36. Fifty-one patients received 14–35 aa-long SLPs with high MHC-I presentation scores based on binding affinity, allele-specific expression and proteasomal cleavage potential (Neo-PV-01). The vaccine was combined with a run-in anti-PD-1 that continued for up to 2 years. An average of 42% and 24% vaccine peptides elicited CD4+ and CD8+ responses, respectively; cytolytic CD4+ and CD8+ T-cells were detected in serial tumour biopsies. PFS and overall survival (OS) rates compared favorably with historical data for anti-PD-1 monotherapy; longer PFS was correlated with epitope spreading to 33 non-immunising epitopes. The high TMB, a certain level of endogenous immunity, and the presence of heterogeneous cancer clones in these advanced tumours possibly enabled epitope spreading. The overall clinical response rate in melanoma patients was 59%—evidence that clinical response to PCV can be achieved in advanced disease. Further building on the SLP vaccine platform, nine patients with stage III/IV melanoma received PCV consisting of MHC-I-binding 16–26-aa-long peptides formulated with poly-ICLC and Montanide (NeoVaxMI) and co-administered with intramuscular ipilimumab and systemic nivolumab given before, during, and after vaccination (NCT03929029)37. All nine patients exhibited ex vivo CD4+ and CD8+ T-cell responses, with a predominance of CD4+ responses. As in NCT03313778, the vaccine was more successful at expanding existing clonotypes and generating hundreds of new ones than run-in nivolumab. At a median follow-up of 26.4 months, 6/9 patients had no evidence of active disease. Another peptide vaccine, EVX-01, was tested in five patients with metastatic melanoma treated with anti-PD-1 (NCT03715985)38. The 15–27 aa-long peptides were given with a CD8+ T-cell inducer, liposome-based cationic adjuvant CAF®09b. T-cell reactivity was detected to 27/31 peptides with more robust CD4+ T-cell responses. Clinical response included one complete and two partial responses. In the phase 1b dose escalation trial with EVX-01, in which the patients received anti-PD-1 pre- and post-vaccine, two complete responses correlated with the higher neoantigen dose and selection score39. Unlike mRNA and SLP vaccines, cell-based neoantigen vaccines have used peptide-pulsed DCs or adoptive transfer of autologous activated TILs28,40,41. A small clinical trial was conducted in three stage III melanoma patients treated with anti-CTLA-4 (NCT006833670)40. Autologous DCs were pulsed with seven tumour-specific peptides verified to bind HLA-A02 with high affinity (IC50 < 500 nm). Neoantigen-specific CD8+ T-cells were identified two weeks post-vaccination and peaked 8–9 weeks after the initial vaccine dose. Memory T-cells were detected up to 4 months after the final vaccine dose. Overall, PCV trials in melanoma patients have demonstrated that an immune response can be safely induced in high-TMB tumours. The studies are small, with mostly short follow-up periods, precluding definitive conclusions about clinical responses. However, the clinical response persistence for 5 years after the V940+pembrolizumab vaccine33 indicates an overall right direction for the SLP vaccines. So far, stage III/IV melanoma has been the most common disease setting and mRNA and SLP platforms the most common delivery methods in PCV trials. A direct comparison between these two vaccine formats or PCV versus PCV + ICI assessments are complicated because of (i) the small number of patients, (ii) mostly one-arm trial design, (iii) different formulations, vaccine adjuvants, delivery routes, and vaccine/drug combinations and (iv) different readouts of immune responses. One prominent feature of all the vaccines listed in Table 5 is the presence of immune responses, CD4+ and/or CD8+-mediated, in all tested patients, with a caveat that responses were assessed ex vivo, often after stimulation, using various methods. The percent of immunogenic neoantigens in melanoma patients varies from 29.5% (mRNA-4157 + pembrolizumab)42 to 62.9% (SLP, NeoVaxIM + ipilimumab + nivolumab)37. This difference is hard to attribute to the mRNA vaccine format, because 60% of neoantigens in another mRNA vaccine, IVAC MUTANOME, were immunogenic30. Important distinctions between the two mRNA vaccines may explain similar high rates of immune responses with significantly different proportions of immunogenic neoantigens: IVAC MUTANOME was administered as naked, non-modified mRNA via intranodal injections, whereas intramuscularly administered mRNA-4157, a modified mRNA, was delivered in LNP particles that protect the mRNA from degradation and may act as a vaccine adjuvant43. The quality of the immunogenic neoantigens rather than their quantity is more important, as mRNA-4157 induced T-cell responses were detected in all tested patients31. Table 5mRNA and SLP trials in melanomaVaccineplatformTumour typeand diseasestageStudyClinicaltrialNumberofvaccinatedpatientsPCVtypeNumberof neo-antigensCombi-nationtreat-mentsCD8+and CD4+responserateCD4+neo-antigensC8+neo-antigensTotalimmuno-genic neo-antigensClinicalresponsemRNAMelanomastage III/IVSahin et al.30NCT02035956phase 113Naked,non-modifiedmRNA20100%57%17%60%Reduction inrecurrentmetastatic eventsand sustained PFS. Anti-PD-1 response invaccine non-responder.mRNAMelanoma stage III/IV NSCLCGainor et al.31NCT03313778Keynote 603phase 1 PCV with orw/o pembroArm 1- NSCLC4 patients,PCV Arm 2-Melanoma,11 patients,PCV+ pembroModifiedmRNA-4157-LNPs34PembroArm 1-100%tested(3/3)Arm 2-100%tested(7/7)24%60%Arm 1-22.7%Arm 2-29.5%Evaluated inNCT03897881mRNAMelanoma stage III/IVWeber et al.32NCT03897881 Keynote 942 phase 2 PCV with orw/o pembro157ModifiedmRNA-4157-LNPs(V940)34PembroN/AN/AN/AN/A49% decrease inrecurrencein combo groupvs pembrolizumabaloneSLPMelanoma stage III/IV previously untreatedOtt et al.34NCT01970358 phase 1 PCV + poly-ICLC6SLPNeoVax20100%60%16%N/A4 patientsrecurrence-free at 25 monthsfollow-up. Tworecurred patients responded toanti-PD-1SLPMelanoma, NSCLC,bladdercancer(advancedormetastatic)Ott et al.36NCT02897765 phase 1 PCV + poly-ICLC26 melanoma, 14 NSCLC, 11 bladdercancerSLP(Neo-PV-01)20Nivo100%42%24%52%Overall responserate in melanoma patients 59%.PFS and 1-yearOS rates comparefavorably withhistorical data foranti-PD-1monotherapy.SLPMelanoma stage III/IVBlass et al.37NCT03929029 phase 1 PCV + poly-ICLC + Montanide9SLP(NeoVaxMI)20Nivorun-in,localipi andsystemicnivo100%N/A(ex vivo)mostlyCD4+responseN/A(ex vivo)CD8+responsein 66.7%ofpatients62.9%After medianfollow-up of 26.4months –1 completeresponse,6 patients with noevidence ofdisease, and 3patients withprogressivediseasePembrolizumab (pembro), nivolumab (nivo), ipilimumab (ipi). mRNA and SLP trials in melanoma Melanoma stage III/IV NCT02035956 phase 1 Naked, non- modified mRNA Reduction in recurrent metastatic events and sustained PFS. Anti- PD-1 response in vaccine non- responder. NCT03313778 Keynote 603 phase 1 PCV with or w/o pembro Arm 1- NSCLC 4 patients, PCV Arm 2- Melanoma, 11 patients, PCV+ pembro Modified mRNA- 4157 -LNPs Arm 1- 100% tested (3/3) Arm 2- 100% tested (7/7) Arm 1- 22.7% Arm 2- 29.5% Evaluated in NCT03897881 NCT03897881 Keynote 942 phase 2 PCV with or w/o pembro Modified mRNA- 4157 -LNPs (V940) 49% decrease in recurrence in combo group vs pembrolizumab alone NCT01970358 phase 1 PCV + poly- ICLC SLP NeoVax 4 patients recurrence- free at 25 months follow-up. Two recurred patients responded to anti-PD-1 Melanoma, NSCLC, bladder cancer (advanced or metastatic) NCT02897765 phase 1 PCV + poly- ICLC 26 melanoma, 14 NSCLC, 11 bladder cancer SLP (Neo- PV-01) Overall response rate in melanoma patients 59%. PFS and 1-year OS rates compare favorably with historical data for anti-PD-1 monotherapy. NCT03929029 phase 1 PCV + poly- ICLC + Montanide SLP (NeoVax MI) Nivo run-in, local ipi and systemic nivo N/A (ex vivo) mostly CD4+ response N/A (ex vivo) CD8+ response in 66.7% of patients After median follow-up of 26.4 months – 1 complete response, 6 patients with no evidence of disease, and 3 patients with progressive disease Pembrolizumab (pembro), nivolumab (nivo), ipilimumab (ipi). Another prominent feature of both mRNA and peptide vaccines is the predominance of responsive CD4+ over CD8+ T-cells. Higher proportion of neoantigens in SLP vaccines NeoVax34,36 and Neo-PV-0137, and in the IVAC MUTANOME mRNA vaccine30 induced CD4+ responses. While the IVAC MUTANOME design included predicted MHC-I and MHC-II neoantigens, the other vaccines delivered only MHC-I binders. In almost all trials, in addition to the neoantigens that elicited CD4+ and/or CD8+ responses, there were CD4+ only responses, indicating that, together with the shared neoepitopes, the vaccines contained MHC-II-only binders in an unplanned fashion. The data so far demonstrate that the immune response to the vaccine is necessary for clinical response, but does not guarantee it, likely because of insufficient strength and persistence of the response and limited T-cell clonality. Combining ICI with the vaccine is expected to broaden and strengthen the immune response by expanding pre-existing clones and improving T-cell persistence and function. In the NeoVaxMI trial, run-in nivolumab expanded pre-existing T-cell clones and induced previously undetected ones37. In the advanced high TMB melanoma, 5/19 patients showed a major pathological response after initial therapy with nivolumab36. It is important to mention that neoantigen-reactive T-cells have been shown to be PD-1+30. In at least two melanoma trials, in the absence of clinical response to the vaccine, after anti-PD-1 treatment, one patient responded with 80% shrinkage of the tumour30 and two patients broadened their T-cell repertoire, increased the immune response persistence, and achieved complete radiographic responses34. Whether subcutaneously co-administered ipilimumab contributed to the improved immune response in the NeoVaxMI trial is difficult to discern; the increased myeloid and T/NK-cell populations observed at the vaccine injection site might also be attributable to the vaccine adjuvant Montanide. In the renal clear cell carcinoma (RCC) trial, local ipilimumab did not improve the response to NeoVax + poly-ICLC44. So far, the data suggest that systemic ICI, given alone or in combination with PCV, can be particularly helpful in patients with high-TMB tumours and pre-existing T-cell responses, which may include tumour driver mutations, whether or not they are included in the vaccine. Currently, the effectiveness of pairing PCVs with ICIs is under evaluation in a phase 3 trial with V940 + pembrolizumab (NCT05933577). Pancreatic adenocarcinoma In the mutational spectrum, pancreatic adenocarcinoma (PDAC) has low TMB, with fewer than one somatic mutation per megabase of the genome, compared to more than 10 somatic mutations per megabase in melanoma22,45. About 93% of PDACs are driven by mutant KRAS (mKRAS), which creates an immunosuppressive niche and confers resistance to immunotherapies46. Recent research from BioNTech demonstrated that KRAS G12D and G12V-reactive T-cell are remarkably prevalent, appearing in all 20 of the healthy donors screened across diverse HLA backgrounds47. This universal presence supports the development of 'off-the-shelf' vaccines for mKRAS-driven cancers. ELI-002 2P vaccine, including lymph node-targeted 18-mer mKRAS amphiphile peptides and the CpG-7909 adjuvant, was administered to 25 treated PDAC patients with no or minimal residual disease48. All patients responded with mKRAS directed CD4+ and CD8+ T-cells. The vaccine led to epitope spreading; ex vivo T-cell responses against tumour neoantigens were detected post-ELI-002 2P. Following tumour biomarker increase, 3/11 patients got chemotherapy only, and another three received chemotherapy + ICI and remained free of radiographic progression. The study demonstrated that a strong anti-mKRAS immune response is achievable, durable, resistant to cytotoxic chemotherapy, and responsive to ICI. Per the company’s June 15, 2026, press release, the follow-up study with ELI-002 7P did not meet the pre-specified primary endpoints but allowed to refine the target population for a phase 3 study based on HR 0.65, p = 0.048, in the completely resected population. As the vaccine targets to expand pre-existing, exhausted T cell clones—no epitope spreading was reported in the update—the combination with ICI seems necessary to improve the clinical response. In low TMB tumours, it is important to identify high-quality neoantigens before committing to the lengthy and expensive therapy with PCV. Balachandran et al. identified neoantigen qualities that promote T-cell activation in long-term PDAC survivors49. Unlike the neoantigen quantity model, a fitness model based on differential neoantigen presentation and homology to infectious disease-derived peptides predicted longer survival. These high-quality neoantigenic clones were lost in metastatic progression, suggesting immunoediting49,50. In preclinical studies, the PancVax SLP vaccine, a mix of 20 aa-long neoantigens identified in the low/moderate mutational burden Panc02 mouse cell line51,52, was evaluated together with a STING activator in an implanted tumour model53. Of the 239 MHC-I-binding peptides, only 16% were immunogenic. CD8+ immune responses were detected with pulsed splenocytes but not with APCs alone, suggesting an induction of CD4+ T-cells. PancVax alone caused transient tumour regressions, but co-administration of anti-PD-1 and agonistic OX40 resulted in durable regressions and prolonged survival. MHC-I and II neoantigens containing PancVax2 demonstrated better results; MHC-II neoantigens induced CD4+ and CD8+ responses, and tumour suppression and survival benefits were similar to PancVax, without the concurrent administration of ICI54. These proof-of-concept studies showed that low TMB tumours harbour high-quality neoantigens and advanced the understanding of the role of CD4+ reactivity in responses to PCVs. In a phase 1 clinical trial, autogene cevumeran, a non-modified mRNA-lipoplex (LPX) vaccine against tumour neoantigens, was tested in an adjuvant setting in 16 patients with surgically resected PDAC (NCT05968326)50. mRNA-LPX formulations generally exhibit greater immunogenicity than mRNA-LNP by driving potent innate immune activation and IFN/T-cell responses, especially when administered intravenously43. Patients with minimal disease burden were sequentially administered anti-PD-L1, cevumeran with up to 20 MHC-I and MHC-II neoantigens, and modified FOLFIRINOX. The vaccine induced long-lived, polyfunctional, neoantigen-specific effector CD8+ T-cell responses in 8/16 patients, targeting approximately 11% of the tested neoantigens. After 18 months, responders with vaccine-expanded T-cells had a longer median recurrence-free survival (RFS) compared with nonresponders. After 3.2 years, 6/8 responders remained recurrence-free with 86% of vaccine-induced CD8+ clones per patient still observed, including the high-avidity clones with estimated lifespan of 7.7 years55. These were tissue-resident memory-like CTLs with preserved neoantigen-specific effector function. Two responders with recurrent disease were pruned of vaccine-targeted cancer clones, implying either immune editing or tumour adaptation through neoantigen masking. Some notable differences and similarities between ELI-002 2P and cevumeran vaccines warrant further investigation. No epitope spreading and predominantly CD8+ T-cell responses were reported in cevumeran-treated patients, and T-cell persistence and function were not dampened by cytotoxic chemotherapy in both studies. The success of cevumeran in PDAC opens the prospect of PCVs across cancers, regardless of TMB. However, both cevumeran and ELI-002 2P and 7P showed promise in patients with minimal disease burden. Only 10–15% of PDAC patients present with resectable disease56. The efficacy of these vaccines in patients with metastatic and higher disease burden will determine their real impact on this tumour type with limited treatment options. A randomised phase 2 trial (NCT05968326) in 260 PDAC patients is currently evaluating the efficacy of cevumeran+chemotherapy versus standard-of-care chemotherapy. Two more autogene cevumeran trials are underway: a phase 2 trial of the vaccine with and without pembrolizumab in patients with previously untreated advanced melanoma (NCT03815058) and cevumeran versus watchful waiting in patients with resected stage II/III ctDNA+ colorectal cancer (NCT04486378). Glioblastoma Like PDAC, glioblastomas (GBM) are very low (<1/megabase) TMB tumours22. IDH1 mutations, present in only 12% of primary GBM, but occur in 70% of astrocytomas and secondary GBM57, creating an opportunity for off-the-shelf vaccines. A phase 1 trial in 33 patients with grade III/IV astrocytomas treated with standard-of-care (SOC) chemo and radiation therapies before enrollment (NCT02454634), tested an IDH1(R132H) vaccine consisting of a 20-mer peptide administered with topical imiquimod58. The peptide harboured a promiscuous MHC-II epitope and induced an immune response in 93.3% of patients, independent of their HLA alleles, through clonal expansion of IDH1(R132)-specific Th cells. The two-year PFS rate of 0.82 was reported in patients who mounted an immune response to the vaccine. A different approach to overcoming the low TMB of GBM was tested in the GAPVAC-101 study (NCT02149225, 16 patients with newly diagnosed GBM, post resection and SOC); the off-the-shelf APVAC-1 vaccine (19 commonly overexpressed, non-mutated peptides) was combined with 1–2 neoantigen-containing APVAC-2, poly-ICLC, and GM-CSF, and given with the maintenance chemotherapy59. More than 90% of patients had CD8+ responses to APVAC-1 MHC-I peptides alone. The addition of APVAC-2 induced de novo CD4+ Th1-cell responses in 84.7% of patients; de novo CD8+ responses were detected in 16.7% of patients and always in combination with a CD4+. The surprisingly high response to overexpressed, non-mutated APVAC-1 peptides by CD8+ T-cells suggests the pre-existence of endogenous immunity invigorated by the vaccine, which could benefit from combination with ICI. The fact that de novo CD8+ responses were never isolated from CD4+ reactivity delineates the requirements and differences between expanding existing reactive T-cell clones and generating new ones. The study reported a median overall survival of 29 months in vaccinated patients, a significant increase compared with 12–15 months in patients treated with SOC60. SLP vaccine NeoVax, tested in melanoma, NSCLC and renal carcinoma34,36,37,44,61, was evaluated in 8 patients with newly diagnosed MGMT-unmethylated GBM (NCT02287428)62. In this phase 1/1b trial, the cohort 1 patients received radiation therapy with or without 1–8 mg dexamethasone post-resection and before vaccination with 12 24-aa-long MHC-I-binding peptides. Only patients who did not receive dexamethasone generated polyfunctional, neoantigen-specific T cells, with a predominance of CD4+. Although TCR analysis revealed that neoantigen-specific T-cells had migrated to the tumours, all patients experienced progressive disease. Four additional cohorts were added to this still ongoing study; each new cohort receives NeoVax and radiation therapy with different schedules of pembrolizumab or daily temozolomide. High-affinity MHC-I and II neoantigens with expected high expression in patients’ primary and recurrent GBM were administered in another single-arm trial63. Twenty peptides were given with recombinant GM-CSF (sargramostim) and topical imiquimod. Eighty-eight percent of patients (87/97) had post-vaccine T-cell responses against at least one neoantigen. At median follow-up of 21.3 months, 54% of patients were alive with a median OS of 31.9 months, which compared favorably with historical OS data. In a recently reported clinical trial in MGMT-unmethylated GBM, patients received DNA-based PCV with an IL-12 plasmid (NCT04015700)64. Multi-centre tumour sampling at the time of resection enabled the identification of up to 40 spatially representative neoantigens per patient. The vaccination induced strong T-cell responses significantly suppressed by dexamethasone and enhanced by PD-1 blockade, resulting in 24 months survival in 1/3 of patients. The modest clinical responses to PCVs in GBM may be related to low MHC-I and absent MHC-II expression, and profound T-cell exhaustion in these tumours65,66. The majority of GBM patients receive steroids to manage tumour-associated edema and symptoms. More selective and limited use of steroids is one of the obvious steps to improve the response to PCVs in GBM. Further trials with different PCV formats and combinations with ICI, targeted, chemo, and/or radiation therapies need to be evaluated in this notoriously treatment-resistant disease. Breast, hepatocellular, renal and hematological malignancies Tumours such as breast, hepatocellular, renal, colorectal and hematological cancers fall within the intermediate TMB group, which encompasses more than one and fewer than 10 somatic mutations per megabase. This still makes them very challenging to target with PCVs. A neoantigen DNA vaccine was evaluated in triple-negative breast cancer (TNBC) patients with persistent disease on surgical pathology (NCT02348320)67. Eleven MHC-I neoantigens were administered to 18 patients via electroporation after completion of adjuvant SOC. T-cell responses were detected in 14/18 patients. After 36 months, 87.5% of vaccinated patients were recurrence-free—a significant increase compared to the reported 50% complete response rate after neoadjuvant therapy67,68. Pending results of phase 1 trial (NCT03199040) will demonstrate DNA neoantigen vaccine effects with and without anti-PD-L1 durvalumab. As part of the TNBC-MERIT non-comparative umbrella study exploring various types of mRNA cancer vaccines in early-stage TNBC, NCT02316457 enrolled 14 T1N0-T2N1M0 patients, within one year of completing adjuvant chemo and radiation therapies. Non-modified mRNA-LPX vaccine delivered up to 20 neoantigens per patient. All patients mounted T-cell responses against neoantigens, of which 44.6% were recognised by CD4+ and 20% by CD8+ T-cells. The immune responses persisted up to 3.5 years. Three years later, 10 patients were recurrence-free; one of the three recurred patients had the weakest immune response but developed a response to anti-PD-1. The trial achieved its goal of demonstrating feasibility, safety, and the vaccine’s ability to induce a neoantigen-specific immune response. However, the clinical response did not improve compared to the 70–85% 5-year overall survival rate reported for stage IIB TNBC patients receiving SOC69. A DNA vaccine encoding up to 40 neoantigens and IL-12 was administered through electroporation to 36 previously treated patients with hepatocellular carcinoma (NCT04251117)70. The vaccine neoantigen repertoire was much broader than in other studies due to less stringent MHC affinity selection criteria and included individual and shared, driver and passenger mutations. Eighty-six percent of evaluated patients (19/22) responded to the vaccine with tumour-penetrating, vaccine-specific, proliferative and cytotoxic CD4+ and CD8+ T-cells; 30% of patients had an objective response, with 8.3% achieving a complete response, a significant improvement over the 12–18% historical objective response rate to anti-PD-1 monotherapy71. The different approach to selecting neoantigens in this trial paid off: an increased number of neoantigens correlated with a higher number of reactive epitopes, and the de novo response rate against medium- and low-affinity epitopes was similar to high-affinity epitopes. This observation emphasises the need to understand the characteristic features of reactive epitopes better. Perhaps evaluating TCR binding and generating scores based on combined MHC and TCR binding affinities will identify neoantigens that would otherwise be excluded from vaccine design. The NeoVax SLP vaccine with poly-ICLC was administered to nine patients with high-risk stage III/IV renal clear cell carcinoma (NCT02950766)44. Like the NeoVax effect in melanoma and GBM, the vaccine induced durable neoantigen-specific, predominantly CD4+ T-cell responses, including those targeting driver mutations. In the PGV001 study, an adjuvant 25-aa-long SLP vaccine targeting up to 10 neoantigens per patient was administered to patients with solid and hematologic tumours (NCT02721043)72. The pool of strong MHC-I binders was further refined based on the gene of origin (oncogene or tumour suppressor), differential from the wild type binding, and manufacturability. The SLPs were formulated with poly-ICLC and tetanus toxoid peptide. Forty-four percent of the 126 neoantigens induced polyfunctional CD4+ and CD8+ T-cell responses (68% and 38%, respectively) in all 13 subjects with a 16–18-fold increase in T-cell activation. Thirty-nine percent of responses were detectable 3–5 months after the last vaccine dose. Eleven of 12 subjects analyzed for neoantigen-specific IgG and IgA antibodies had detectable vaccine-specific IgG, and five peptides across four patients induced IgA responses. Although the study was not powered to assess clinical response, an exploratory analysis showed that patients with more polyclonal T-cell responses trended toward improved OS: 50% were alive at 5 years, and 3 had no evidence of disease recurrence and required no follow-up treatments. Overall, in these phase 1 clinical trials, PCVs induced significant immune responses in the intermediate TMB group. Long-term follow-ups and larger studies powered to assess clinical response are necessary to evaluate the concordance between immune and clinical responses. Future directions Identifying quality neoantigens remains a work in progress In all trials, vaccines safely induced immune responses against a wide range of neoantigens, nearly all of which were patient-specific. Selection of neoantigens is key to the success of PCVs. Strategies that increase the number of high-quality neoepitopes can potentially narrow the gap between delivered and immunoreactive neoantigens (Fig. 1). Various methods can be applied to increase the range and quality of neoantigens: (i) proteomics (immunopeptidome) and direct HLA-bound peptide target analysis (HLA ligandome) to identify neoantigens arising from differential protein processing not captured by genome sequencing73, (ii) multi-centred sampling to increase the number, spatial representation and clonality of neoantigens64, (iii) inclusion of medium and low MHC-binders70, (iv) sequencing of non-coding regions (dark DNA) enriched with immunoreactive cryptic neoantigens74,75, (v) incorporating foreignness, high mutant allele expression76,77, and TCR-epitope recognition and binding affinity78 to the neoantigen fitness score. The neoantigen prediction pipelines are the crux of PCV development and have improved significantly by incorporating the above-discussed features into algorithms79. But there is still ample room for improvement, as the proportion of immunogenic neoantigens in PCVs against melanoma remains 22–62% (Table 5). In a recent press release, Evaxion Biotech A/S reported an impressive 86% rate of immunogenic neoantigens in the EVX-01 vaccine developed using its AI-ImmunologyTM platform; the vaccine was tested in patients with advanced melanoma in a phase 2 trial (NCT05309421). Quality neoantigen selection. Workflow of the identification and selection of neoantigens for PCVs. Blue boxes indicate procedures and analyses to identify neoantigens and select high-quality neoepitopes for inclusion in PCV. CD4+ T-cell responses are common and critical With few exceptions30,38,63,67 PCVs were based on MHC-I-binding neoantigens. Nevertheless, almost all trials reported CD4+ T-cell responses, including cases where a CD4+ T-cell response was generated in the absence of CD8+ T-cell reactivity. In many cases, MHC-I-binders were extended at either terminus, potentially increasing the likelihood of incorporating an existing CD4 epitope. If this is the case, the relationship of the CD4 epitope to the predicted CD8 epitope is unclear, and the value added by certain prediction algorithms remains to be demonstrated. CD4+ T-cell responses are critical; helper function of CD4+ T-cells skews the APCs toward Th1 cytokines and upregulates co-stimulatory signals leading to durable and effective CD8+ responses12. In vaccinated patients, CD4+ T-cells may acquire effector and cytotoxic functions and directly kill melanoma cells30,61; the adoptive transfer of autologous tumour lysate-activated CD4+ T-cells was sufficient to shrink the melanoma tumours28. Adding MHC-II-only binders can enrich and enhance the neoantigen repertoire and strengthen the CD8+ T-cell reactivity. Reported high constitutive expression of MHC-II in melanoma and its further upregulation by IFNγ28,29 likely contributes to better immune responses compared to other tumour types. If true, tumour cell MHC-II expression can potentially predict a better response to PCVs. Concomitant treatments seem important, but still much to learn Treatments administered before PCV or concurrently can significantly alter the immune and clinical response. There is growing evidence that glucocorticoids can dampen ICI-induced immune activation80,81. In two GBM PCV clinical trials, immune responses were detected primarily in subjects who did not receive dexamethasone62,64. Chemotherapy, on the other hand, has a complex effect on the anti-tumour immune response; after initial upregulation of inflammatory signatures and tumour T-cell infiltration, residual tumours become immunosuppressed relative to baseline82. Modified FOLFIRINOX administered right after cevumeran priming doses may have contributed to the lower immune response in PDAC patients50, but the ex vivo immune response to mKRAS was not different before and after chemotherapy in patients who relapsed after the ELI-002 2P vaccine48. Hence, the timing of chemotherapy relative to PCV should be carefully considered. Combining PCV with ICI is a justified strategy: repeated or strong antigen stimulation pushes responding T-cells to upregulate exhaustion signature and become hypofunctional. Combination has been shown to be safe and feasible in multiple single-arm clinical trials37,38,50,61,72,83. Even though preclinical data had suggested that ICI exposure before vaccination induces dysfunctional T-cells84, a two-armed trial with anti-melanoma mRNA-4157-LNP and pembrolizumab given before and with the vaccine, showed clinical response improvement compared with pembrolizumab alone32. The rationale behind giving ICI before the vaccine is to induce and expand endogenous immunity. As expected, in the absence of pre-existing immunity, the run-in pembrolizumab did not affect the depth of immune response31, but in patients with reactive T-cells at baseline, run-in nivolumab expanded existing clones and induced de novo responses37. More studies comparing PCV and PCV+anti-PD1/PD-L1 are needed to understand the additive effects of combination therapy. The scenario where PCVs sensitise tumours to ICI becomes more obvious: melanoma and TNBC patients who failed to respond to PCVs or recurred post-vaccination responded to anti-PD-1 with tumour shrinkage or complete response30,34,85. Preclinical data have shown strong evidence of improved T-cell responses and suppression of tumour growth when a single dose of anti-CTLA4 was administered with a DNA neoantigen vaccine to tumour-bearing mice86. In non-human primates, simian immunodeficiency virus model antigens inserted into chimpanzee adenovirus and a self-amplifying mRNA vaccine were administered with and without subcutaneous ipilimumab; the addition of anti-CTLA4 generated a more robust and persistent T-cell response87. In-depth immune response to neoantigens delivered via the same vaccine platform with nivolumab and ipilimumab is being assessed in a first-in-human phase 1/2 trial (NCT03639714). A clinical trial in nine patients has tested the anti-CTLA4 + PCV combination; based on the magnitude of the peripheral immune response, ipilimumab given adjacent to subcutaneous or intradermal SLP vaccine failed to improve the vaccine-only effect44. More multi-arm clinical trials with similar vaccine and disease settings are needed to hone down PCV + ICI regimens. Clinical setting matters The timing and deployment of PCVs are crucial for a successful clinical response. PCV + ICI given in adjuvant settings are expected to elicit better immune responses based on the absence of immunosuppressive TME, ICI-mediated invigoration of exhausted T-cells, and vaccine-induced de novo immune responses. Tumour burden has been shown to impact ICI and conventional therapy responses and outcomes; patients with more pronounced surrogates of higher tumour burden and systemic inflammation are more likely to experience disease progression88. Baseline tumour burden is strongly associated with response to immunotherapy, PFS and OS in melanoma and NSCLC patients89. Patients with lower minimal residual disease burden assessed by ctDNA at the start of adjuvant mRNA-4157 and pembrolizumab had more favourable RFS and distant metastasis-free survival compared to the ICI-only group32. At the same time, based on preclinical data, there is growing interest in PCV use in the neoadjuvant setting in the context of a bigger neoantigen supply and an immune system not affected by prior therapies90. While fewer tumour recurrences seem achievable with PCVs, preclinical and clinical trials are evaluating their efficacy in advanced and metastatic diseases (Fig. 2). Treatments by disease stage. Disease progression stages (top row boxes), current treatment strategies (second row boxes) and proposed interception, prevention and treatment windows for PCVs (third row boxes). Consensus remains elusive on optimal delivery platforms The comparison of PCV delivery platforms is complicated due to numerous variables across preclinical and clinical trials; head-to-head trials keeping the tumour type, disease setting and antigen repertoire constant, with standardised immune and clinical readouts and benchmarking, will be needed to draw definitive conclusions. mRNA vaccines, with their short manufacturing time and ability to activate the innate immune system, are an attractive option. In addition, induction of unconventional pathways by mRNA-LNP vaccines, such as cDC cross-dressing to prime CD8+ T-cells and broaden the immune response beyond vaccine reactive T-cells, can have a significant impact on the clinical response91. Considering new findings, targeting mRNA expression to myocytes instead of professional antigen-presenting cells may enhance mRNA-LNP immunogenicity92. Which mRNA vaccine formulation holds greater potential - modified versus unmodified nucleosides, and LNP encapsulation versus LPX - remains to be determined. For viral disease prevention, modified mRNA-LNPs are preferable due to their milder effects on the innate immune system. In contrast, non-modified mRNA-LPX vaccines may be more effective in immunosuppressed cancer patients. Pending results from INTerpath 001-009 trials with modified mRNA-LNP vaccine and pembrolizumab, and NCT03815058, NCT05968326, and NCT04486378 trials with non-modified uridine-rich mRNA-LPX trials with and without ICI will shed more light on the future of mRNA vaccines. Given the importance of CD4+ T-cell responses for positive clinical outcomes, SLP-based vaccines also hold promise for the future of anti-tumour PCVs; peptide length and extracellular delivery to antigen-presenting cells strongly favour MHC II presentation, whereas mRNA vaccines can be more efficient in delivering MHC I-binders. So far, mRNA and SLP platforms have been tested separately; it would be interesting to see preclinical data on a combined SLP/mRNA platform. Administration of DNA vaccines via electroporation may limit their broad use, and more data are needed to evaluate their effectiveness relative to SLP and mRNA vaccines. Similarly, limited data on cellular vaccines and their complex manufacturing processes complicate predictions about their further use in patients. PCVs hold great promise, but their design tools remain a work in progress; pipelines need better predictive accuracy and selection criteria. PCV manufacturing is complex, costly, and time-consuming, allowing tumours to evolve and undergo immunoediting that affects the clinical response to vaccines. If the effectiveness of vaccines proves to be better in the neoadjuvant setting, it is hard to imagine their use outside of cancers with an inherited genetic basis of common mutations. Increasing the immunogenicity of off-the-shelf vaccines against TAAs and TSAs is a viable alternative for standalone use or as a run-in to PCVs. Given the remarkable adaptability of tumours and their wide arsenal of mechanisms to evade immune surveillance, a hybrid regimen combining off-the-shelf vaccines targeting shared driver and passenger mutations, ICI and PCV could be sufficiently potent for use throughout the various stages of the disease. Author contributions R.H.V.: conceptualisation. N.M.: literature review, writing original draft. R.H.V.: review and editing. All authors have read and approved the final manuscript. Peer review Peer review information Nature Communications thanks Vinod Balachandran and the other anonymous reviewer for their contribution to the peer review of this work. Competing interests Vonderheide has received consulting fees from Astellas, EMD Serono, Grey Wolf Therapeutics and Crossbow Therapeutics, research funding from Revolution Medicines, is an inventor on patents relating to cancer cellular immunotherapy, cancer vaccines, cancer interception and KRAS immune epitopes and receives royalties from Children’s Hospital Boston for a licensed research-only monoclonal antibody. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Nune Markosyan, Email: nune@pennmedicine.upenn.edu. Robert H. Vonderheide, Email: rhv@upenn.edu References