Discussion Guide — Chapter 26

Six prompts, each with what to listen for. These are designed for a seminar of 8–25. Every one has been written so that the wrong answers are productive.

Prompt 1 — "Why did thirty years of cancer vaccines fail, given that they worked?"

Pose it exactly that way, as a contradiction. Give the room the fact that many of these trials demonstrably induced antigen-specific T-cell responses and still missed their primary endpoints.

Listen for: students reaching for trial-design explanations first (underpowered, wrong endpoint, wrong patients). Let that run — it is a reasonable hypothesis and it is mostly wrong here, which makes the correction memorable. Then listen for whoever gets to the responses were the wrong ones. Push them to say why: high-affinity clones against self antigens were deleted during development.

Steer away from: "the immune system is complicated" and "the tumor is immunosuppressive." The second is true and is a genuine contributing factor, but if the room settles there they will miss tolerance, which is the decisive one.

Close with: the observation that every intermediate measurement said the approach was working. That is the transferable lesson, not the cancer biology.

Prompt 2 — "A vaccine works in one trial participant and does nothing in another. What are the possibilities, and which one is this chapter about?"

Listen for: the full list — non-presentation, no available clone, tolerized clone, assay failure, compliance, timing. Reward breadth first.

Then push hard on the distinction between weak response and no engagement. Students almost always frame non-presentation as a quantitative difference. It is categorical. A peptide that does not fit the groove is not a low dose; it is a message in a language the recipient cannot read.

Listen for the follow-on: someone should notice that this is why whole-antigen approaches dodge the problem — they supply many candidate peptides and let each person's grooves choose. If nobody does, ask directly.

Escalation if the room is strong: ask what this implies about the ethics and the statistics of enrolling only carriers of one HLA allele. That moves you into the equity material without moralizing.

Prompt 3 — "Is 'the intervention is a process, not a molecule' a reason for caution or a reason for optimism?"

This is the best prompt in the chapter for an advanced group, and it is genuinely two-sided.

Listen for the chapter's case: what a trial validates is a pipeline whose performance depends on inputs — mutation count, prediction accuracy for a given HLA background, manufacturing quality — that vary in ways a fixed molecule's do not. External validity therefore works differently.

Listen equally for the counter-case: a process that adapts to each patient might transfer better across populations than a fixed epitope-restricted product, precisely because it does not assume a shared target.

What you want: students holding both, and then identifying what evidence would discriminate — replication across tumor types, performance stratified by HLA background, multi-site manufacturing comparability.

Watch for: the student who concludes "so we can never know." Push back. The trials are designed to answer this; uncertainty is not unanswerability.

Prompt 4 — "Rewrite this sentence so it is honest: 'Patients developed neoantigen-specific T cells, and most remained recurrence-free at follow-up.'"

Put it on the board. Give them two minutes to write.

Listen for: rewrites that separate the two claims into distinct sentences with no causal connective, and that label the first as an immune measurement.

The deeper move, which some rooms reach unprompted: the original sentence is not false. Every clause is accurate. The dishonesty is entirely in the juxtaposition. That is a more sophisticated observation than "it overstates the evidence," and it is worth naming explicitly when it appears.

Follow up with: why does this construction survive peer review? Good answers involve the difficulty of reporting two true findings from one study without implying a relationship, and the fact that authors often believe the relationship — which is a hypothesis, not a result.

Prompt 5 — "The most successful cancer vaccines in existence are not cancer vaccines. Discuss."

Hepatitis B and HPV vaccination prevent large numbers of cancers. Neither treats a tumor. Neither is a peptide vaccine.

Listen for: recognition that these succeeded by solving a categorically easier problem — preventing a chronic infection in a healthy person with an intact immune system and unlimited time.

Listen for the uncomfortable inference: the field's biggest wins came from not attacking the tumor. Some students find this deflating. It should instead sharpen the therapeutic-vs-prophylactic distinction that §26.8 is built around.

Good extension: ask what other cancers would become preventable if their causes were infectious and vaccinable. This is where the public health students in the room come alive.

Prompt 6 — "Your friend sends you a headline: 'Personalized cancer vaccine wipes out tumors in trial.' Their mother is in treatment. Write the reply."

Give them five minutes, silently, then read three aloud without attribution.

Listen for accuracy in both directions. The most common failure is over-correction: a technically accurate reply that is cold, condescending, or that leaves the friend with nothing. The second most common is capitulation — agreeing that it sounds promising and saying nothing about what is actually known.

What a good reply contains: the progress is real; the trials are randomized and ongoing; nothing is approved and access is through trials; and a concrete next step, which is the oncologist, not the internet.

What no reply should contain: the word "hype," any assessment of the mother's prognosis, or a recommendation about her care.

Close the session here. This prompt collapses the whole chapter into one act of communication, and students remember the discomfort of writing it far longer than they remember the class I processing pathway. That is the correct outcome.