Advanced mRNA vaccines targeting new cancer markers: Immune system design strategies for precise cancer treatment

TSAs represent a biologically privileged class of neoantigens that arise exclusively in malignant cells and are absent from normal tissues. Their tumor-restricted expression enables recognition as non-self, permitting high-avidity CTL responses without the central tolerance constraints that frequently limit TAA-directed immunity [39]. These antigens originate from diverse genomic alterations, including SNVs, indels, gene fusions, intron retention, and aberrant splicing, which generate novel peptides presented on HLA moleculesand targeted by T cells [6]. Immunopeptidomic analyses have further expanded this landscape by identifying cryptic peptides derived from non-canonical translation events, alternative reading frames, and unconventional transcripts [32, 40–42]. Such peptides are efficiently presented on tumor HLA molecules and recognized by tumor-infiltrating lymphocytes [43], and accumulating evidence suggests that cryptic antigens may be as abundant as mutation-derived TSAs in certain malignancies [44, 45]. This expanded repertoire is particularly relevant in tumors with low mutational burden, where classical mutation-derived targets may be limited.

The identification of TSAs has been accelerated by advances in NGS and computational pipelines enabling matched tumor–normal analyses. Predictive tools such as dbPepNeo [46], NetMHCpan [47], and MHCflurry [48] prioritize candidate epitopesbased on peptide–HLA binding affinity, and recently developed frameworks optimized for direct tumor–normal sequence comparison improve precision in identifying tumor-exclusive neoepitopes [49]. However, in silico prediction remains imperfect, frequently generating false positives that require validation using patient-derived T cells [50]. This validation step is not merely technical but clinically essential, as off-target recognition of self-antigens has been associated with severe immune-related toxicity in prior clinical experiences [51]. Accurate antigen prioritization therefore remains central to both efficacy and safety.

Clinical translation of TSA-based mRNA vaccines has demonstrated that these biological principles can be operationalized. Autogene cevumeran (BNT122) evaluated in resected pancreatic ductal adenocarcinoma induced de novo neoantigen-specific T-cell responses in approximately half of treated patients, and responders exhibited prolonged recurrence-free survival relative to non-responders [32, 33]. These findings suggest that TSA vaccination can introduce immunogenic epitopes into otherwise poorly infiltrated tumors and may sensitize immunologically “cold” malignancies to subsequent immune modulation (Table 1; Fig. 2).

Despite this promise, both logistical and biological challenges persist. Personalized workflows—from tumor sampling through sequencing, antigen prioritization, mRNA synthesis, LNP encapsulation, and GMP release—require several weeks, limiting applicability in rapidly progressing disease [52]. In addition, immune pressure may select for antigen loss, particularly when vaccines target subclonal mutations. Multi-epitope designs incorporating both clonal and subclonal targets have therefore been proposed to enhance response durability and mitigate immune escape [53]. The vaccinable landscape continues to expand through identification of fusion-driven TSAs such as EWS-FLI1 and BCR-ABL, as well as non-canonical sources including upstream open reading frames, endogenous retroelements, intronic regions, and lncRNA transcripts [6, 19, 54, 55] (Fig. 3). Oncogenic pathway alterations, including PIK3CA mutations in TNBC, may further reshape tumor antigenicity and generate shared tumor-restricted epitopes suitable for semi-universal incorporation [56–59], paralleling observations in BRAF-driven melanoma [33, 57].

Looking forward, modular vaccine architectures integrating patient-specific and shared TSAs aim to balance personalization with manufacturability. Combination strategies with checkpoint inhibitors, type I interferon agonists, and agents targeting immunosuppressive pathways such as A2AR and TGF-β signaling are increasingly pursued to enhance efficacy and prevent immune escape [60]. Real-time ctDNA monitoring offers an additional layer of adaptability, enabling dynamic tracking of clonal evolution and potential refinement of vaccine composition during therapy [61]. As antigen prioritization, manufacturing workflows, and combination paradigms continue to mature, TSA-based mRNA vaccines are likely to find their most immediate impact in adjuvant and minimal residual disease settings, where tumor burden is limited and immune pressure can be strategically leveraged [62].

TAAs represent a distinct immunologic category characterized by aberrant expression, overexpression, mutation, or altered post-translational modification in malignant cells, with limited expression in normal adult tissues. Unlike TSAs, which are strictly tumor-restricted, TAAs are self-derived proteins and therefore exist within the spectrum of peripheral tolerance. Nevertheless, their recurrent expression across tumor types enables broader clinical applicability and more standardized manufacturing strategies. Frequently studied TAAs include CEA, HER2, WT1, and MUC1, all of which are widely expressed in solid tumors and hematologic malignancies [63]. Tumor-specific glycosylation changes, such as the hypoglycosylated form of MUC1 observed in adenocarcinomas, expose normally concealed peptide cores and create immunologically targetable epitopes [2, 64, 65]. Similarly, WT1 is overexpressed in glioblastoma and leukemia while remaining minimally expressed in most healthy tissues, supporting its development as a vaccine target [66, 67].

Clinical translation of TAA-based mRNA vaccination has demonstrated that shared antigen platforms can generate measurable immune responses. BNT111, encoding multiple melanoma-associated TAAs, showed early clinical activity in advanced melanoma, including in patients previously resistant to standard therapies [9]. Preclinical studies combining tumor-derived vesicle platforms with immune checkpoint blockade have further illustrated the capacity of TAA-directed vaccination to enhance cytokine production and T-cell activation in murine models [37]. These data underscore the potential of shared antigen constructs to amplify antitumor immunity when integrated into combination regimens.

Hybrid strategies further extend this concept. The mRNA-4157 platform integrates patient-specific neoantigens with shared TAAs, thereby balancing personalization with manufacturability. In the KEYNOTE-942 trial, combination therapy with pembrolizumab significantly improved recurrence-free survival and reduced distant metastasis risk in patients with high-risk resected melanoma, with a favorable safety profile [68]. These findings suggest that TAA-containing constructs may enhance immune breadth and durability when administered alongside checkpoint inhibition.

Despite these advances, TAA-directed vaccines face inherent biological challenges. Because TAAs are derived from self-proteins, immunodominance hierarchies and tolerance mechanisms can constrain response magnitude. Antigen loss, reduced epitope presentation, and defects in antigen-processing pathways further contribute to immune escape [69, 70]. Consequently, multi-epitope formulations and rational epitope prioritization are increasingly employed to broaden immune targeting and reduce vulnerability to single-antigen loss. Combination strategies with checkpoint blockade have shown preclinical efficacy, including restoration of CD8⁺ T-cell function in WT1-targeted models of leukemia [67]. In parallel, optimized LNP platforms, including PEGylated and pH-responsive formulations, enhance dendritic cell transfection and antigen expression, improving immunogenicity in vivo [71].

Table 1 summarizes ongoing clinical trials evaluating TAA-based mRNA vaccines across multiple malignancies, detailing antigen targets, construct design, and clinical stage. Collectively, TAA strategies occupy an intermediate position between individualized TSA platforms and fully shared neoantigen constructs, prioritizing scalability and antigen recurrence while requiring careful integration with immune-modulating therapies to overcome tolerance-related constraints.

Shared neoantigens arise from recurrent somatic driver mutations that are absent from the normal genome yet conserved across subsets of patients. Unlike TAAs, they remain tumor-specific and avoid central tolerance, while differing from fully individualized TSAs in their cross-patient recurrence. This dual property—specificity combined with scalability—positions shared neoantigens as strong candidates for off-the-shelf mRNA vaccine development [72].

Many shared neoantigens originate from oncogenic hotspot mutations. KRAS^G12C/D variants, frequently observed in pancreatic, colorectal, and non-small cell lung cancers, generate immunogenic epitopes presented by defined HLA alleles such as HLA-A*11:01 and HLA-C*08:02 [73, 74]. Similarly, recurrent TERT promoter mutations produce neoepitopes detected across glioblastoma, melanoma, and bladder cancer, supporting broader population-level targeting [75, 76]. Because these alterations often occur early in tumorigenesis and are clonally maintained, they may reduce intratumoral heterogeneity and mitigate the risk of antigen loss that complicates subclonal TSA strategies [77].

Preclinical data support the immunogenicity of this approach. mRNA constructs encoding KRAS^G12D have elicited robust CD8⁺ T-cell responses and tumor regression in murine models, prompting early-phase clinical evaluation of KRAS-targeted vaccines in metastatic pancreatic cancer [78]. These findings illustrate the translational logic of shared neoantigen vaccination: by focusing on conserved driver mutations, vaccine design can shift from bespoke manufacturing to population-stratified deployment.

However, identification and prioritization of shared neoantigens remain technically demanding. Predictive algorithms such as NetMHCpan and MHCflurry frequently overestimate peptide–HLA binding affinity, necessitating ex vivo validation to confirm true immunogenicity [46–48, 50]. Tumors may also evade immune pressure through antigen editing or pathway adaptation, reinforcing the importance of multi-epitope constructs targeting several recurrent mutations simultaneously. Emerging strategies therefore emphasize modular vaccine libraries incorporating common driver mutations across tumor types, although most remain in preclinical or early clinical stages [33, 79].

Beyond point mutations, fusion-derived neoantigens such as BCR-ABL and recurrent splicing-derived epitopes represent additional shared targets with strict tumor specificity [42, 80]. Preclinical evidence supports their immunogenicity and potential integration into semi-universal vaccine platforms [4]. Collectively, shared neoantigen vaccination represents a strategic midpoint between individualized and broadly shared antigen approaches, prioritizing clonal stability and manufacturability while preserving tumor specificity.

Hybrid mRNA vaccine platforms seek to integrate individualized TSAs with recurrent TAAs or shared neoantigens within a single construct, thereby combining precision with scalability. Rather than positioning antigen categories as mutually exclusive, this strategy recognizes that tumor immunity may benefit from simultaneous targeting of clonal, subclonal, and shared determinants. By broadening antigenic coverage, hybrid constructs aim to enhance immune breadth, reduce the likelihood of antigen escape, and accommodate inter-patient heterogeneity.

Clinical translation of this concept has demonstrated feasibility. Platforms encoding patient-specific mutations alongside shared antigens have elicited polyfunctional CD8⁺ and CD4⁺ T-cell responses in melanoma, particularly when administered with anti–PD–1 therapy, and have been associated with improved recurrence-free outcomes in high-risk settings [68, 81]. These findings support the premise that combining antigen classes can strengthen immune activation while maintaining manufacturability and compatibility with checkpoint inhibition.

Hybrid approaches are also motivated by biological considerations. Tumors evolve under immune pressure, and targeting a single antigen class may allow selective outgrowth of antigen-negative clones. Incorporating mutation-derived epitopes together with recurrent or pathway-associated antigens increases epitope diversity and promotes recruitment of a broader T-cell repertoire. Preclinical evidence indicates that multi-antigen mRNA constructs can achieve more durable tumor control than single-target vaccines, consistent with reduced vulnerability to immune editing [82].

Beyond classical mutation- and overexpression-derived targets, hybrid platforms can incorporate unconventional tumor-specific epitopes identified through immunopeptidomics and transcriptomic profiling [40–42, 83]. Fusion-derived neoantigens and other non-canonical peptides expand the available antigen pool and may be encoded within modular mRNA architectures without requiring structural redesign of the delivery platform [6]. The flexibility of mRNA constructs permits rapid adaptation of epitope composition while preserving standardized manufacturing workflows.

A further advantage of hybrid vaccination is dynamic adaptability. Serial tumor sequencing or ctDNA monitoring enables real-time assessment of clonal evolution, allowing refinement of antigen selection during therapy [84, 85]. This adaptive framework aligns with the iterative nature of cancer progression and supports the concept of vaccines that evolve in parallel with tumor biology.

Collectively, hybrid neoantigen vaccination represents a convergence of individualized and shared strategies. By integrating multiple antigen sources within a modular mRNA backbone, these platforms seek to balance personalization, immune breadth, and clinical practicality. As sequencing pipelines, antigen prioritization algorithms, and manufacturing timelines continue to improve, hybrid designs may offer a pragmatic pathway toward scalable precision immunotherapy.

Off-the-shelf mRNA vaccines represent the most scalable extension of neoantigen immunotherapy, prioritizing standardized manufacturing and rapid deployment over individualized tailoring [86]. While shared neoantigen strategies focus on recurrent driver mutations within defined patient subsets, off-the-shelf platforms are designed for immediate clinical availability, often stratified by common oncogenic alterations and HLA distribution patterns across populations.

Recurrent mutations such as KRAS codon 12 variants and TERT promoter alterations are particularly attractive in this context due to their prevalence across colorectal, pancreatic, lung, and glioblastoma tumors [76]. Unlike individualized constructs requiring sequential sequencing and custom synthesis, these vaccines can be produced in bulk, stored, and administered without delay [20]. The feasibility of industrial-scale deployment is supported by infrastructure established for pandemic mRNA vaccine production, including advances in formulation, cold-chain optimization, and regulatory standardization. Efforts to improve storage stability—such as spray-dried and thermostable formulations capable of room-temperature preservation and mucosal delivery—further enhance accessibility, particularly in resource-limited settings [87].

Early clinical investigations provide proof of immunogenicity. Off-the-shelf constructs such as GRT-C901/GRT-R902 have demonstrated acceptable safety profiles and measurable immune activation, including induction of CD8⁺ T-cell responses and reductions in ctDNA in subsets of patients with metastatic solid tumors [36, 37]. Although durable tumor regressions remain limited in early studies, these data suggest that standardized neoantigen platforms can generate biologically meaningful immune responses outside a personalized framework.

However, this approach faces distinct biological constraints. HLA polymorphism can restrict epitope presentation, limiting universal applicability, and tumor heterogeneity may lead to variable antigen expression across patients. These factors can attenuate T-cell priming and reduce therapeutic consistency. To address these limitations, current development efforts emphasize multi-epitope payloads incorporating frequently occurring driver mutations, stratified by population-level HLA frequencies to maximize coverage [88]. Combination with checkpoint blockade and other immune-modulating strategies may further enhance response durability.

In aggregate, off-the-shelf neoantigen vaccination represents a population-oriented strategy that leverages standardized production and rapid availability [89]. While immunogenetic variability remains a challenge, continued refinement of antigen selection, formulation stability, and combination paradigms may enable broader and more equitable access to mRNA-based cancer immunotherapy [90].

The clinical utility of different neoantigen vaccine formats may depend on disease context and tumor burden. Personalized neoantigen vaccines are most suitable for the adjuvant or minimal residual disease setting, where tumor burden is low and sufficient time exists for individualized antigen discovery and vaccine manufacturing. In contrast, shared or off-the-shelf vaccines may be more appropriate in advanced or metastatic disease, where rapid treatment initiation is required and recurrent driver mutations can be targeted across patients. Hybrid vaccine approaches, combining TSAs with shared antigens, may provide a balanced strategy that maintains specificity while enabling faster deployment [91]. Such context-dependent selection of vaccine strategies may improve clinical translation by aligning antigen design with tumor biology, treatment timing, and immune responsiveness. Biomarkers such as TMB, HLA genotype, and baseline immune infiltration may further guide the selection of the most appropriate vaccine strategy for individual patients [92].

Although multiple neoantigen vaccine platforms have been explored, including peptide-based vaccines, DNA vaccines, and viral vector approaches, mRNA vaccines offer several advantages that make them particularly attractive for personalized cancer immunotherapy. Unlike peptide vaccines, which are restricted by HLA binding and often require complex adjuvant formulations, mRNA vaccines enable endogenous antigen production within host cells and support presentation through both MHC class I and class II pathways. Compared with DNA vaccines or viral vectors, mRNA platforms avoid risks of genomic integration and allow rapid, cell-free manufacturing through in vitro transcription, enabling faster adaptation to patient-specific mutational profiles. These features have positioned mRNA vaccines as one of the most flexible and scalable platforms for neoantigen-based cancer immunotherapy [93, 94].

The development of effective neoantigen mRNA vaccines begins with sequence engineering, which tailors the encoded transcript to the biological characteristics of each antigen class. TSAs, TAAs, hybrid constructs, and shared or off-the-shelf antigens differ substantially in expression level, immunogenic potential, and immune tolerance risk. These differences influence codon optimization strategies, nucleotide modification choices, and regulatory sequence design [95]. Accordingly, sequence engineering aims to maximize antigen expression and immunogenicity while maintaining transcript stability and avoiding excessive innate immune activation [96] (Fig. 4).

For TSAs, the primary objective is to preserve mutation-containing epitopes while ensuring efficient protein translation. Because many neoantigens arise from single nucleotide variants embedded within otherwise normal protein sequences, codon optimization must maintain the mutated residue and surrounding epitope context while adjusting other codons to enhance translational efficiency. Patient-specific optimization pipelines increasingly incorporate machine-learning–guided algorithms that preserve mutation-adjacent amino acid motifs while modifying synonymous codons to improve ribosome processivity and protein output [97, 98].

In contrast, TAA constructs often encode longer or full-length proteins, such as WT1 or MUC1, which may be susceptible to extensive RNA secondary structure formation. To mitigate this effect, sequence designs frequently incorporate moderate GC content (approximately 45–50%), which reduces stable RNA folding and improves ribosomal accessibility, thereby enhancing translation efficiency in antigen-presenting cells [99].

In addition to codon selection, nucleoside modification is a central component of sequence engineering. Modified nucleotides can improve transcript stability, enhance translational output, and modulate innate immune sensing. Among these, N1-methyl-pseudouridine (m¹Ψ) is widely used in experimental mRNA vaccine platforms because it reduces recognition by innate immune receptors such as Toll-like receptors while maintaining high translational efficiency. This modification has been incorporated in several neoantigen vaccine platforms, including shared or off-the-shelf constructs targeting driver mutations such as KRAS^G12D [100].

Different antigen classes may require distinct nucleoside modification strategies. In many TAA-based constructs, pseudouridine or 5-methoxyuridine is used to maintain efficient translation while preserving a moderate level of immune stimulation that supports antigen presentation without provoking excessive inflammatory signaling. Hybrid vaccine constructs, which combine TSAs and TAAs within a single transcript, sometimes employ tiered modification strategies in which m¹Ψ is preferentially applied to TSA regions to maximize translation efficiency, while alternative modifications such as pseudouridine are used in TAA segments to maintain balanced immune activation [101].

Overall, sequence engineering provides the first level of immunoengineering control in neoantigen mRNA vaccine design. By integrating codon optimization, nucleotide modification, and antigen-specific sequence constraints, these strategies ensure that encoded antigens are expressed efficiently while preserving the mutation-specific epitopes required for effective anti-tumor immune recognition.

The untranslated regions (UTRs) of mRNA transcripts regulate translation by controlling ribosome recruitment, transcript stability, and interactions with RNA-binding proteins and microRNAs. The 5′ UTR primarily influences translation initiation through its secondary structure and Kozak consensus sequence, whereas the 3′ UTR modulates transcript stability and translational efficiency via binding factors such as HuR and poly(A)-binding protein (PABP) (Fig. 4) [102, 103]. In mRNA vaccine platforms, rational UTR engineering enables precise control of antigen expression kinetics while minimizing excessive innate immune activation.

Several studies have shown that UTRs derived from dendritic cell–associated transcripts can enhance translation within antigen-presenting cells (APCs). For example, regulatory regions from genes such as CD40 or LAMP3 can recruit cell-specific RNA-binding proteins that promote efficient translation in dendritic cells, thereby improving antigen presentation without altering the delivery tropism of the LNP carrier [104, 105].

UTR design is often adapted to the immunological properties of the encoded antigen. TSA constructs typically encode low-abundance mutation-derived epitopes that require efficient translation to achieve sufficient antigen presentation [106]. To support rapid protein synthesis, UTRs derived from immune-response transcripts such as IFN-β or IL-12 are sometimes incorporated because these regulatory regions promote strong translation in activated dendritic cells—the principal APCs responsible for priming naïve CD8⁺ T cells [107]. These UTRs enhance antigen expression in immune-active tissues, particularly lymph nodes where dendritic cell density and antigen-processing capacity are highest [108]. In addition, synthetic 5′ UTRs optimized for rapid ribosome scanning are frequently combined with relatively short-lived 3′ UTR elements to achieve strong but transient antigen expression. This strategy promotes efficient cytotoxic T-cell priming while limiting prolonged cytokine production that could contribute to immune exhaustion [109].

In contrast, TAA constructs often aim to sustain antigen expression to support efficient MHC class II presentation and CD4⁺ T-cell activation, which are essential for overcoming immune tolerance and reinforcing adaptive immunity. To achieve this effect, stabilizing UTRs derived from transcripts such as β-globin or GM-CSF are frequently incorporated to extend mRNA half-life and maintain translation within professional APCs including dendritic cells and macrophages. Longer-lived 3′ UTR elements prolong antigen persistence and enhance T-cell help and antibody responses [110]. For example, incorporation of β-globin UTRs has been reported to increase transcript stability by more than 2.5-fold in dendritic cells compared with certain viral-derived regulatory regions.

Thus, the distinction between TSA and TAA UTR designs primarily reflects differences in desired immune kinetics rather than differences in APC targeting. TSA constructs typically favor short-term, high-intensity antigen expression to rapidly activate CD8⁺ cytotoxic responses, whereas TAA constructs emphasize sustained translation to reinforce CD4⁺ helper T-cell activity and immune support [54].

Hybrid vaccine platforms introduce additional design complexity because they must balance these distinct antigen expression profiles within a single construct. In some strategies, differential UTR engineering is used to modulate antigen kinetics—for example, pairing shorter-lived regulatory elements such as truncated α-globin 3′ UTRs with TSA sequences while stabilizing TAA regions with longer β-globin UTR variants [111]. This asynchronous expression strategy can mimic physiological immune activation patterns, supporting coordinated CD8⁺ and CD4⁺ T-cell responses while reducing the risk of clonal anergy.

For shared or off-the-shelf neoantigen vaccine platforms, which rely on fixed antigen sequences intended for broad patient populations, UTR design must support consistent translation across diverse cell types. In these systems, consensus-engineered regulatory regions derived from housekeeping genes such as α-tubulin or eIF4G are often employed because they produce balanced translational activity in multiple cell types [112]. Combining optimized 5′ UTR Kozak sequences with GC-balanced 3′ UTR elements further stabilizes translation efficiency and reduces variability across patients.

Overall, UTR engineering represents a central component of mRNA vaccine optimization. By modulating translation kinetics and transcript stability, UTR design enables antigen-specific tuning of immune activation: TSA constructs prioritize rapid and transient expression to stimulate cytotoxic responses, whereas TAA constructs emphasize sustained antigen availability to reinforce helper T-cell activity. Hybrid and shared vaccine strategies integrate these approaches to achieve physiologically balanced antigen presentation and broad immunogenic coverage.Advances in RNA structure modeling and codon optimization further refine these design strategies. The folding pattern of an RNA transcript can influence ribosome accessibility and transcript stability, thereby shaping translational efficiency and antigen production [97]. As computational tools for RNA structure prediction improve, these approaches are increasingly integrated into the rational design of next-generation mRNA vaccine constructs.

Structural design refers to the internal organization of the mRNA sequence—including coding region arrangement, epitope order, linker composition, and ORF structure—rather than the three-dimensional conformation of the RNA molecule. These architectural features determine how encoded antigens are processed intracellularly and ultimately presented through MHC pathways [95]. Consequently, structural layout must be tailored to the immunological objective of the vaccine platform.

For TSA vaccines, tandem minigene constructs are commonly used to encode multiple patient-specific mutations within a single transcript. These minigenes are typically separated by short synthetic linkers to maintain balanced proteasomal processing and prevent epitope dominance that could restrict T-cell repertoire diversity [113]. Proper epitope spacing helps ensure that multiple neoepitopes are processed efficiently and presented across diverse human leukocyte antigen (HLA) alleles.

To enhance antigen expression levels, some platforms employ self-amplifying RNA (saRNA) constructs incorporating alphavirus-derived replicase elements. These systems amplify the RNA template intracellularly, increasing antigen production by one to two orders of magnitude while preserving the transient and non-integrating nature of mRNA vaccines [114]. Higher antigen expression can enhance immune priming without requiring higher vaccine doses.

Structural features can also direct antigen processing toward specific immune pathways. Incorporation of proteasomal targeting motifs, such as ubiquitin tags or degron sequences, promotes rapid cytosolic degradation and improves peptide generation for MHC class I presentation, thereby strengthening CD8⁺ cytotoxic T-cell responses. In contrast, antigens intended to stimulate CD4⁺ T-cell responses are often fused to lysosomal-targeting domains such as lysosome-associated membrane protein 1 (LAMP1), which redirects proteins toward the endosomal–lysosomal pathway and enhances MHC class II presentation [67, 115, 116]. The cytoplasmic tail of LAMP1 contains a tyrosine-based sorting motif (YXXΦ) that interacts with adaptor protein complexes (AP-1 and AP-3), guiding the fusion protein from the Golgi apparatus to late endosomes and lysosomes where antigen processing occurs. In preclinical models, LAMP1–antigen fusion constructs significantly increased CD4⁺ T-cell proliferation and cytokine production compared with cytosolic antigen expression [116].

Hybrid vaccine constructs present additional structural complexity. In these systems, TSAs and TAAs may be encoded within separate expression modules using polycistronic architectures or internal ribosome entry site (IRES) elements. Alternatively, 2 A peptide linkers can be used to produce multiple antigen products from a single transcript [117]. Maintaining appropriate epitope spacing is essential to prevent proteasomal interference and preserve efficient antigen processing. Experimental studies using hybrid constructs in glioma models demonstrated that separating TSA and TAA epitopes into distinct expression modules enhanced both CD8⁺ and CD4⁺ immune responses by more than 50% [118].

Shared neoantigen or off-the-shelf platforms also rely on multi-epitope arrays incorporating recurrent driver mutations or conserved tumor antigens. These sequences must be carefully arranged to ensure efficient antigen processing and broad HLA coverage across diverse patient populations [119]. Synthetic linkers such as AAY or GPGPG are frequently used to separate epitopes and minimize structural interference during proteasomal cleavage [120].

In some cases, structural design is complemented by heterologous prime–boost strategies, where an initial viral-vector priming vaccination is followed by mRNA boosting to reinforce recall responses and improve T-cell durability, as explored in clinical trials such as Gritstone’s adenoviral–mRNA platform (NCT03639714↗) [36, 37, 121]. Across these diverse approaches, effective structural architecture must preserve physiological antigen processing while minimizing immunodominance and maximizing the breadth of tumor-specific T-cell responses.

Efficient delivery systems are a central determinant of the clinical performance of neoantigen mRNA vaccines because they influence antigen biodistribution, cellular uptake, immune activation, toxicity, and manufacturing scalability [122]. Delivery design becomes particularly important when targeting TSAs, which are highly individualized and typically expressed at low levels. In these cases, effective immune priming requires precise delivery of the mRNA payload to APCs, especially dendritic cells that initiate T-cell responses [123, 124]. LNP systems have therefore become the dominant delivery platform for mRNA vaccines because they protect RNA from degradation while enabling efficient cellular uptake and endosomal release.

To improve immunological specificity, LNPs are increasingly engineered with targeting elements that promote uptake by dendritic cells. Surface ligands such as mannose or antibodies directed against receptors like DEC205 can guide nanoparticles toward APC populations in lymphoid tissues [125]. In experimental TSA vaccine models, mannose-modified LNPs have been shown to enhance accumulation in lymph nodes and improve antigen-specific CD8⁺ T-cell responses [126]. Some formulations also incorporate innate immune stimulators such as STING or RIG-I agonists within the same nanoparticle, enabling localized activation of antiviral signaling pathways that amplify T-cell priming and enhance neoantigen immunogenicity [127].

For TAAs or shared driver mutations such as KRAS^G12D and TP53^R175H, broader delivery strategies may be preferable. PEGylated LNPs and cationic nanoemulsions can distribute more widely across tissues and are often combined with immune-stimulating adjuvants, including TLR3 or TLR7/8 agonists, to help overcome immune tolerance mechanisms [128]. Alternative delivery technologies are also being explored. Intradermal electroporation and microneedle array systems, for example, enable direct targeting of skin-resident antigen-presenting cells. These approaches may be particularly useful for hybrid vaccine designs that combine TSAs and TAAs and therefore require coordinated activation of multiple immune pathways.

Another emerging strategy involves sequential or layered delivery approaches. In this framework, an initial priming dose is designed to reach lymph node–resident APCs, followed by a booster administration that promotes broader antigen exposure throughout the immune system. Such spatial and temporal coordination of antigen presentation can increase the magnitude and durability of vaccine-induced T-cell responses [129].

In parallel with advances in targeting strategies, improvements in ionizable lipid chemistry have significantly expanded the capabilities of LNP delivery systems. One recent approach introduced a modular “plug-and-play” platform for synthesizing biodegradable ionizable lipids, which form the structural core of LNPs. Using a simple combinatorial reaction scheme, researchers generated large libraries of candidate lipids that could be rapidly screened for efficient mRNA delivery [130]. From these libraries, several formulations demonstrated improved delivery efficiency compared with earlier benchmark lipids used in mRNA therapeutics [131]. These modular lipid systems are designed to enhance endosomal escape and improve intracellular RNA release while maintaining low toxicity.

Additional innovations have focused on achieving organ-specific delivery. Siloxane-incorporated lipid nanoparticles (SiLNPs) represent one such development. By introducing siloxane groups into ionizable lipid structures, researchers generated a range of nanoparticle formulations with distinct tissue distribution patterns [132]. Some variants preferentially delivered mRNA to liver tissue, whereas others demonstrated enhanced delivery to pulmonary endothelial cells or immune cells in the spleen. These findings suggest that modifications to lipid structure can influence tissue targeting and may enable more precise delivery of mRNA therapeutics to specific organs or immune cell populations. Improved membrane interaction and endosomal escape properties are thought to contribute to the enhanced delivery efficiency observed in these systems.

Together, these advances considerably expand the delivery toolkit available for mRNA-based cancer immunotherapies. Modern LNP platforms can now be engineered to optimize biodistribution, enhance immune activation, and support tissue-specific targeting, enabling delivery strategies tailored to different antigen classes and therapeutic goals [132].

Finally, practical considerations such as manufacturing scalability and global distribution are becoming increasingly important for the development of neoantigen vaccines. Technologies that reduce dependence on ultra-cold storage conditions are therefore receiving growing attention. Lyophilized or spray-dried LNP formulations, as well as self-amplifying RNA platforms capable of maintaining stability at moderate temperatures, offer promising approaches for improving storage and transport logistics [133]. Recent initiatives have also explored dry powder formulations for mucosal delivery of mRNA vaccines, further expanding the potential routes of administration and simplifying vaccine deployment [134].

The relative advantages of these emerging delivery systems, including modular ionizable lipids, siloxane-based nanoparticles, targeted surface modifications, and thermostable formulations, are summarized in Table 2. Together, these technologies provide a framework for selecting delivery strategies that align with the immunological requirements, antigen types, and clinical deployment scenarios of neoantigen-based cancer vaccines.