The rapid clinical validation of mRNA technology during the COVID‑19 pandemic has powerfully accelerated its application in oncology, and this comprehensive review provides a state‑of‑the‑art assessment of mRNA cancer vaccines. It systematically covers the molecular design principles of synthetic mRNA, the diverse antigen‑targeting strategies (from conventional tumor‑associated antigens to patient‑specific neoantigens and non‑canonical sources), the major delivery platforms (lipid nanoparticles, lipoplexes, protamine complexes, and cell‑based systems), and the mechanistic pathways by which these vaccines activate both cellular and humoral antitumor immunity. The review then synthesizes preclinical and clinical evidence across solid tumors—melanoma, pancreatic ductal adenocarcinoma, non‑small cell lung cancer, prostate cancer, glioblastoma—and hematologic malignancies, including acute myeloid leukemia, myelodysplastic syndromes, and multiple myeloma. It also critically discusses current challenges, such as the immunosuppressive tumor microenvironment, delivery barriers, and manufacturing complexities, before outlining future directions that involve next‑generation delivery systems, artificial intelligence‑driven vaccine design, and combination strategies with immune checkpoint inhibitors and adoptive T‑cell therapies.

Synthetic mRNA vaccines are built from carefully engineered structural elements. A 5′ cap, 5′ and 3′ untranslated regions, an open reading frame, and a poly(A) tail function together to determine stability, translation efficiency, and immunogenicity. Nucleoside modifications—particularly pseudouridine and N1‑methylpseudouridine—greatly reduce unwanted innate immune recognition while enhancing protein output. Codon optimization and sequence engineering tools (e.g., LinearDesign, mRNAchitect) further improve expression. Beyond conventional linear mRNA, the review describes self‑amplifying mRNA (saRNA) and trans‑amplifying mRNA (taRNA), which carry replicase genes to enable dose‑sparing and prolonged antigen production, as well as circular RNA (circRNA) vaccines, whose covalently closed structure confers high stability and sustained translation. Each format has distinct trade‑offs, but all aim to maximize the therapeutic index.

A key decision in vaccine design is the choice of antigen. Tumor‑associated antigens (TAAs) are self‑antigens overexpressed on cancers; they are easy to manufacture as “off‑the‑shelf” products but face immune tolerance and potential on‑target/off‑tumor toxicity. Neoantigens, derived from somatic mutations, are truly tumor‑specific and highly immunogenic, making them the centerpiece of personalized vaccines (e.g., mRNA‑4157/V940 and BNT122/autogene cevumeran). Importantly, the review expands the antigen repertoire to non‑classical sources: cryptic antigens from non‑canonical open reading frames, aberrant splicing variants (notably in splicing‑factor‑mutant leukemias), and transposable element‑derived antigens. These novel targets can be shared across patients and provide opportunities for fixed, broadly applicable vaccines.

Effective delivery is the main technical hurdle. Lipid nanoparticles are the most clinically advanced system, comprising ionizable lipids, helper phospholipids, cholesterol, and PEGylated lipids. They protect mRNA, enable endosomal escape, and have been successfully used in approved COVID‑19 vaccines. Anionic lipoplexes preferentially target dendritic cells in lymphoid organs, as exemplified by BioNTech’s BNT111 and BNT122. Multilamellar “onion‑like” lipid particle aggregates activate RIG‑I in stromal cells, triggering a potent cytokine storm that reprograms the tumor microenvironment. Protamine–mRNA complexes were early carriers but have been largely superseded by LNPs due to consistency issues. Virus‑like particles and dendritic cell‑based ex‑vivo loading represent alternative strategies: DC vaccines offer personalization but are labor‑intensive, whereas in vivo targeting is now a major focus.

Mechanistically, after administration, mRNA is taken up by antigen‑presenting cells—especially dendritic cells—via endocytosis. Once in the cytosol, it is translated into antigens that enter both the MHC class I and class II presentation pathways, activating CD8⁺ cytotoxic T lymphocytes and CD4⁺ helper T cells. Simultaneously, the mRNA itself can act as an adjuvant through Toll‑like receptor and RIG‑I signaling. Beyond encoding tumor antigens, the review discusses mRNA that encodes immunomodulators: cytokines (IL‑12, OX40L, IL‑23, IL‑36γ), adjuvants (TriMix: CD70, CD40L, caTLR4), and STING agonists. Another breakthrough is mRNA for in vivo cell engineering: targeted lipid nanoparticles carrying chimeric antigen receptor or T‑cell receptor mRNAs can generate CAR‑T or TCR‑T cells directly inside the body, potentially eliminating the need for ex vivo manufacturing.

Clinical applications have advanced most in melanoma. The phase IIb KEYNOTE‑942 trial showed that mRNA‑4157 plus pembrolizumab significantly improved recurrence‑free survival compared with pembrolizumab alone, leading to a phase III trial. BNT111, encoding four melanoma TAAs, demonstrated durable responses in checkpoint‑resistant patients. In pancreatic cancer, the personalized vaccine BNT122 induced long‑lived neoantigen‑specific CD8⁺ T cells in half of the patients and delayed recurrence; a fixed KRAS G12V mRNA vaccine also showed early clinical benefit. For non‑small cell lung cancer, fixed‑antigen vaccines (CV9201, CV9202) and personalized approaches (mRNA‑4157, BNT116) are in various trial phases. In glioblastoma, RNA‑lipid particle aggregates (RNA‑LPAs) have produced rapid cytokine release, T‑cell trafficking, and radiographic pseudoprogression in first‑in‑human studies. For hematologic malignancies, the review notes that the immune dysfunction in acute leukemia and myelodysplastic syndrome is a major barrier. Nevertheless, WT1‑ and PRAME‑loaded dendritic cell vaccines have shown safety and reduced relapse risk in AML. Fusion‑derived neoantigens (e.g., CBFB::MYH11) and mis‑splicing‑derived neoepitopes in SRSF2‑mutant leukemias represent emerging targets. In multiple myeloma, a BCMA‑mRNA LNP vaccine demonstrated preclinical efficacy, and a first‑in‑human in vivo CAR‑T trial (ESO‑T01) has shown early activity.

Safety data from multiple trials indicate that mRNA vaccines are generally well tolerated. Intravenously administered formulations commonly cause transient, grade 1‑2 flu‑like symptoms (fever, chills, fatigue) reflecting cytokine release. Local intradermal or intranodal injections produce mostly injection‑site reactions. When combined with CAR‑T cells (e.g., BNT211), cytokine release syndrome can be more pronounced but remains manageable.

The review candidly addresses remaining challenges. The immunosuppressive tumor microenvironment—rich in regulatory T cells, myeloid‑derived suppressor cells, and inhibitory cytokines—limits monotherapy efficacy. Solid tumors also pose physical barriers: dense extracellular matrix and elevated interstitial pressure hinder nanoparticle penetration. For hematologic cancers, profound host immune impairment and rapid disease progression complicate vaccine strategies, especially those requiring prolonged manufacturing. Scaling up high‑quality mRNA with efficient capping and removal of double‑stranded RNA impurities remains a manufacturing challenge. Finally, the review offers forward‑looking perspectives. Next‑generation delivery systems will incorporate active targeting ligands and stimulus‑responsive (“smart”) release mechanisms. Artificial intelligence and machine learning are poised to enable end‑to‑end vaccine design—from neoantigen prediction to sequence optimization and lipid formulation. Expanded combination therapies with checkpoint inhibitors, conventional chemotherapy, and adoptive cell transfers are expected to unlock the full potential of mRNA platforms. Collectively, this review solidifies the view that mRNA technology, driven by interdisciplinary innovation, is transforming cancer immunotherapy.
DOI
10.1007/s11684-026-1210-6