Chapter 26 — Key Takeaways

The one sentence

The immune system cannot perceive a virus or a tumor — only short peptide fragments displayed in grooves on cell surfaces — which makes vaccination a branch of peptide pharmacology, and makes every difficulty in this chapter a consequence of that single fact.

The mechanism

• T cells read peptides, not organisms. A T-cell receptor engages a short peptide held by an MHC/HLA molecule on another cell. There is no route by which a T cell perceives a whole pathogen.
• Class I presents 8–10-residue peptides from inside the cell, on nearly every nucleated cell, to CD8 cytotoxic T cells. This is how a killer T cell detects what is happening inside a cell it cannot enter.
• Class II presents 13–25-residue peptides from outside the cell, on professional antigen-presenting cells, to CD4 helper T cells, whose help is required for durable memory.
• Antibodies read shapes; T cells read sequences. Most antibody epitopes are conformational. This asymmetry is why short peptides are natural for T-cell-directed vaccines and awkward for antibody-directed ones.
• A vaccine's unit of effect is a changed repertoire and memory — not receptor occupancy. The material is cleared in days; the intended effect lasts years. Half-life is nearly irrelevant.

HLA: the fundamental obstacle

• HLA genes are the most polymorphic in the human genome. Two people shown the same protein display different peptides from it.
• A peptide vaccine that works in one person may not present at all in another. That is not a weak response; it is a mechanism that never engaged.
• This drove a generation of trials to enroll only carriers of one common class I allele — capping the addressable population and skewing which patients got studied.
• Individualized approaches improve this by selecting epitopes against your HLA, but prediction quality across all HLA backgrounds is not established.
• Predicted binding is not predicted response. Generation, density, an available T-cell clone, and activation rather than tolerization all stand in between.

Adjuvants

• A short peptide alone is a poor immunogen: small, rapidly cleared, and carrying no danger signal.
• The immune system responds to foreignness plus danger, detected by pattern recognition receptors.
• Peptide without costimulation produces anergy, deletion, or regulatory conversion. You can vaccinate someone into unresponsiveness.
• The adjuvant is not an optional additive — for peptide vaccines it is frequently the entire difference between a response and nothing, and adjuvant choice determines what kind of response you get.
• Location beats dose: the response is made in the draining lymph node. And a persistent depot can backfire — preclinical work shows depots retaining and deleting the T cells they induced.

Neoantigens — the chapter's central idea

• Most historical cancer vaccine targets were self proteins, against which central tolerance had already deleted the best clones. That is the structural reason therapeutic cancer vaccines failed for thirty years.
• Neoantigens are peptides carrying mutation-derived amino acid changes absent from the normal proteome. No tolerance was ever established against them, so the high-affinity repertoire is intact — and they are tumor-specific, so a response cannot damage tissue lacking the mutation.
• The pipeline: sequence tumor and matched normal → call somatic mutations → type the patient's HLA → predict which resulting peptides will bind and be immunogenic → select → manufacture peptides or mRNA encoding them → administer with adjuvant, usually alongside a checkpoint inhibitor.
• Indirect support: checkpoint inhibitor response is associated with high tumor mutational burden — an association, not proof, but real and reproducible.
• Design points that matter: long peptides beat minimal epitopes (they require professional processing); clonal beats subclonal; some hotspot neoantigens are shared, enabling off-the-shelf products for HLA-matched patients.

mRNA

• An mRNA vaccine delivers instructions to make the peptide, not the peptide. It is therefore not a peptide drug — but it is unambiguously a peptide-antigen vaccine, because the message is still a peptide sequence.
• Individualized programs reached for mRNA because one construct encoding many epitopes is one manufacturing process, where individualized peptides would be many.
• Same distinction Chapter 22 drew for CGRP: the peptide can be the target or the message without being the drug.

Therapeutic versus prophylactic

• Prophylactic: healthy recipients, intact immune systems, no time pressure, enormous populations, near-zero tolerable risk.
• Therapeutic: patients whose immune systems have already failed against this target, often suppressed by the disease or its treatment, under time pressure, facing a hostile tumor microenvironment.
• The long history of therapeutic cancer vaccine failure is largely a history of underestimating that difference.
• The most successful cancer vaccines in existence are prophylactic antiviral vaccines — hepatitis B and human papillomavirus — and neither is a peptide vaccine.
• Immunological success and clinical failure occurred in the same trials, repeatedly. That dissociation is the field's defining lesson.

Allergy

• Peptide immunotherapy inverts the goal: tolerance instead of immunity, using peptides too short to cross-link IgE and fire mast cells.
• Real randomized human trials exist in cat, grass pollen, dust mite, and bee venom allergy. Early results were encouraging; the most prominent late-stage program failed to separate from placebo, in a field where placebo responses are notoriously large.

The ratings, verbatim

Two rules the ratings enforce here:

1. A mechanistic ✅ licenses nothing clinical. That T cells recognize peptides is settled. That vaccinating with peptides extends anyone's life is a separate claim with separate evidence.
2. The ❌ to ⚠️ rating is about the historical class, not about current neoantigen approaches, which differ in specific and potentially decisive ways.

Why 🔬 and not ⚠️

Randomized data exist, so the reasoning must be precise:

• The randomized evidence is early and narrow — one tumor type, one setting, recurrence endpoints, confirmatory trials still accruing.
• The intervention is a process, not a molecule. Every patient receives a different product; what a trial validates is a pipeline, and its transferability is genuinely open.
• Attribution against an active comparator is unresolved in practice.
• Nothing is approved anywhere.
• Mechanism cannot upgrade a rating (rule 3).

What would move it up: completed, adequately powered randomized confirmatory trials showing durable recurrence-free or overall survival benefit in defined populations, with the vaccine's contribution separable from the checkpoint inhibitor's; replication in a second tumor type; approval; evidence that benefit is not confined to high-mutational-burden tumors and favorable HLA backgrounds; reproducible manufacturing outside specialized centers.

What would move it down: confirmatory trials failing to reproduce the early signal; benefit attributable to the checkpoint component; benefit so narrow the addressable population is negligible; manufacturing too slow for the diseases that need it.

Dossier

Field 3 gets an immune variant. For an intervention that trains a system rather than binding a receptor, record: what is administered, what the message is, who presents it and under which HLA, which cells are taught, the direction (immunity or tolerance), the danger signal, where it acts, the unit of effect, the immune readout, the clinical endpoint, duration logic, and what would defeat it.

The rule that matters: never let an immune readout occupy the line reserved for a clinical endpoint. An empty clinical line is information.

If someone you love has cancer

The chapter's honest position, in three sentences: this is a serious approach being tested seriously; access today is through clinical trials, not prescription; and your oncologist is the person who can say whether a trial is a reasonable option in your particular situation. Neither "a vaccine is coming" nor "it is all hype" is accurate. Chapter 39 is about having that conversation well.