A review consolidating the field's known production barriers finds that personalized neoantigen peptide vaccines still face peptide synthesis bottlenecks, costly small-batch GMP manufacturing gaps, and unclear quality standards. Batches of 25 to 40 unique peptides at 50 to 60 mg scale strain…
A review of the field consolidates the manufacturing and regulatory barriers that keep personalized neoantigen peptide vaccines from reaching their clinical potential: peptide synthesis bottlenecks, quality control difficulties, high costs, and regulatory gaps. The review reports no new experimental or clinical data. Its contribution is structural: it assembles the known constraints into a single, explicit problem statement.
The timing matters. Neoantigen vaccines have moved from proof of concept into clinical development across multiple tumor types, and with that movement the production question has shifted. It is no longer whether such vaccines can be made at all, but whether they can be made well enough, fast enough, and cheaply enough for routine use. The catalog of barriers is essentially a description of what stands between a validated immunology concept and a deliverable medicine.
The numbers define the problem. A personalized cancer peptide vaccine may contain up to 20 different neoantigen peptides reference 3 , and an individualized production batch is specified at 25 to 40 unique peptide sequences produced at a scale of 50 to 60 mg reference 6 . With 25 to 40 sequences in a 50 to 60 mg batch, the average peptide is present at roughly 1.2 to 2.4 mg per component. The entire product for one patient, a few tens of milligrams divided among dozens of sequences, would be negligible output for a conventional peptide plant. By the standards of industrial peptide manufacturing, these are tiny runs. By the standards of pharmaceutical quality assurance, they are extraordinarily complex, because every batch is a mixture of sequences made for one patient and for no one else.
That combination of small scale and extreme heterogeneity is the central tension. Existing vaccine production facilities are designed for traditional vaccines and lack GMP manufacturing suited to small batches of 25 to 40 unique peptides at 50 to 60 mg scale. The product definition is itself patient-specific: the composition is not fixed before the patient's tumor is sequenced and analyzed, but is selected by prediction algorithms and validated by laboratory assays. The field has reached a point where the biology of individualized vaccines runs ahead of the chemistry required to deliver them.
A consolidating review has a function beyond summarizing. It gives peptide chemists, process engineers, and regulators a shared problem statement for a set of difficulties that otherwise sit in separate literatures: peptide synthesis, analytical chemistry, immunology, and drug regulation. Its authority is therefore the completeness of the primary literature it assembles, and its limitations are the limitations of that literature. Every barrier it names is a claim about the current state of manufacturing science, and each deserves scrutiny against the primary studies rather than acceptance on the strength of the review alone.
Current large-scale peptide synthesis platforms are the first bottleneck. The review notes that these platforms show low efficiency, high cost, and variable quality when asked to produce complex sequences, particularly long-chain polypeptides. Solid-phase peptide synthesis is mature for short peptides, but yields and purity erode as chain length grows, and the failure sequences that accumulate in a long synthesis become harder to remove in purification.
The reason is stepwise accumulation of error. Each coupling of a protected amino acid to the growing chain is efficient but never perfect. Side reactions truncate a fraction of the chains at every cycle, and a fraction of the full-length chains carry a deletion, an insertion, or a racemized residue. Over 20 or 30 cycles, even a small per-step failure rate means the desired full-length product is a diminishing fraction of the crude material. The impurities are not random: a chain missing a single internal residue resembles the target closely in length, charge, and hydrophobicity, which is exactly what makes it difficult to separate by preparative chromatography. For long-chain polypeptides, the target and its closest deletion impurities can be nearly co-eluting, and pure product is recovered only at the price of yield.
The mechanics of failure are specific to solid-phase chemistry. Each cycle couples one protected amino acid to the growing chain, and the coupling is never quantitative; a fraction of chains fails to react and is capped, producing a permanently truncated impurity. Incomplete deprotection leaves a protected amine that can mis-fire in the next cycle and scramble the sequence. Chains that fold into secondary structure on the resin shield their reactive termini from the incoming amino acid, so coupling becomes slower and less complete as the chain lengthens. The compounding losses are the reason yields fall with length, and why the most immunogenic formats are the least tractable.
Long peptides are not an optional format in this field; they are a preferred one, precisely because of their immunogenicity. That places the chemistry problem at the center of the development pathway. The format choice cuts across the chemistry in both directions. A long peptide immunogen carries multiple epitopes in one chain, which means fewer separate sequences to synthesize, purify, and release; fewer components would simplify the analytical burden. But each long chain concentrates the synthesis difficulty: longer syntheses accumulate more failure sequences, and the deletion impurities of a long chain are harder to separate from the full-length product than those of a short peptide. The field therefore faces a trade-off between analytical complexity and synthetic difficulty, and the optimum point of that trade-off is not settled. The review notes that long peptide-based vaccines are often more immunogenic, but it does not resolve whether that immunogenicity advantage survives the manufacturing penalty.
The manufacturing fleet is configured for the opposite task: existing large-scale platforms were built to make one sequence at a time in large quantity, not to run many distinct syntheses in parallel. A personalized workflow needs exactly that, parallel channels, each producing a different patient-specific sequence, under GMP, on a timeline dictated by the patient's disease. At GMP scale, parallel synthesis is not merely a matter of running more reactors. Each reactor is a separate product stream with its own documentation, segregation, and release records, so the operational cost of running 25 to 40 sequences in parallel scales with the number of sequences, not with the number of patients.
The facility gap compounds the chemistry gap. A GMP plant built around a fixed vaccine antigen has no obvious slot for a workflow that must reset itself for 25 to 40 new sequences every time a new patient enters treatment. The capital cost of building flexible small-batch capacity is high, and the operating cost of idling that capacity between patients is higher.
Neoantigens arise from somatic mutations in tumor DNA. The identification workflow begins with whole exome sequencing and RNA-seq to detect single nucleotide variations , frameshift insertions/deletions , and gene fusions , then applies bioinformatics prediction , immunopeptidomics , and in vitro immunological assays to rank the candidates. A figure from Zhang et al. 2023, published in an MDPI open-access journal, depicts this identification and validation workflow and is cited by the review as the reference architecture for the discovery phase.
The three mutation classes are not equivalent as vaccine targets. A single nucleotide variation changes one amino acid in a self-protein; the resulting peptide differs from its normal counterpart by a single residue, and whether that difference creates a new T-cell epitope depends on how it alters MHC binding and T-cell receptor recognition. Peptides of this class are often only weakly immunogenic. The underlying constraint is immune tolerance. T cells with high-affinity receptors for self-proteins are deleted or silenced during development, so a peptide that differs from self by a single residue is seen by a repertoire that has been partially purged of the strongest responders. The surviving T cells are those that avoided deletion because their affinity for the self version was low, and those are the cells a single-residue neoantigen must recruit. Frameshift insertions or deletions are different: they shift the reading frame and produce a long stretch of amino acid sequence that is entirely novel, absent from the normal proteome, with no self counterpart against which the repertoire was purged. Gene fusions generate chimeric proteins whose junction-spanning sequences are likewise unique to the tumor. This is why the choice of mutation type matters for manufacturing: not every predicted neoantigen is worth the synthesis capacity and quality-control effort required to put it in the vaccine.
Not every mutation produces a usable target. For a neoantigen to be useful in a cancer vaccine, it must satisfy three conditions in sequence: it must be presented by MHC molecules, it must be recognized by immune cells, and it must trigger an immune response. Only neoantigens that are recognized and induce a strong T-cell response are considered good candidates. Most passenger mutations fail at least one of these steps, which is why prediction and validation matter as much as sequencing. The workflow is a funnel with four layers. Whole exome sequencing enumerates the coding mutations; RNA-seq asks which of them are actually expressed, because a mutation that is not transcribed cannot yield a peptide. Bioinformatics prediction scores each expressed mutation for the likelihood that its peptide binds the patient's HLA molecules. Immunopeptidomics provides an empirical check: MHC-bound peptides are eluted from tumor or antigen-presenting cells and identified by mass spectrometry, which demonstrates real presentation rather than predicted binding. In vitro assays test the final condition, whether patient T cells recognize the peptide and are activated by it. Each layer removes candidates. Since synthesis capacity is committed only after the funnel, the stringency of the filters is itself a manufacturing variable: a permissive filter spends scarce synthesis and quality-control resources on peptides that fail later.
The mechanism of the vaccines themselves is illustrated in a figure from Fan et al. 2023, published under a Creative Commons BY 4.0 license, also cited by the review. The two formats, short peptide vaccines and long peptide immunogens , differ at the level of antigen presentation. Short neoantigen peptides bind directly to MHC-I and MHC-II molecules on the surface of antigen-presenting cells ; they do not require processing to be presented. Long peptide immunogens must first be taken up and processed by antigen-presenting cells, and after processing they can activate both CD4+ and CD8+ T cells.
That distinction has consequences for immunogenicity and for manufacturing. The two MHC classes impose different length constraints. MHC-I presents short peptides, typically around 8 to 10 residues, to CD8+ T cells; MHC-II presents longer peptides, commonly 13 to 25 residues, to CD4+ T cells. A short vaccine peptide competes for direct binding to either class if its sequence matches the patient's HLA molecules with sufficient affinity, which makes the short format mechanistically simple but tightly dependent on the accuracy of binding prediction for that patient's HLA type. A long peptide cannot bind without processing; it must be internalized and fragmented, after which the recovered epitopes are loaded onto both MHC classes. That dual loading is the mechanistic reason long…
Related reading: Neoantigen peptide vaccines face a purity-testing standards gap, Porton Advanced Secures China Class C Drug Manufacturing License, Donald Bachem Obituary Notice: Wilmington, North Carolina, Piramal Pharma adds spray drying suite at Turbhe peptide site.