Discussion Guide — Chapter 28

Six prompts for seminar use, with what to listen for. Each is designed to be answerable from the chapter alone but to reward students who bring in earlier material.

1. "The same measurement, four claims, and they do not all get the same rating. Walk me through them."

What to listen for. The four: diagnosis (✅), titration to a target (❌), prognosis (mentioned, not formally rated), and prevention/screening (⚠️ at best, flagged as a fifth claim in §28.8). Strong students name the study design for each without prompting — diagnostic accuracy study for the first, randomized trial of a management strategy for the second, cohort data for the third, randomized prevention trial for the fourth.

Listen for the failure mode: a student who says "so the biomarker is good for some things and not others" has restated the conclusion without grasping that these are different kinds of question. Push with: "Could a bigger diagnostic accuracy study have settled the titration question?" The answer is no, in principle, and being able to say why is the objective.

Bring in if it stalls: Chapter 5 §5.6. The titration claim is the surrogate fallacy applied to a management strategy rather than to a drug.

2. "Nesiritide was the exact human peptide, made correctly, given to exactly the patients whose physiology it was designed to correct. Why didn't it work?"

What to listen for. Students should reach at least two of: the setting was acute and short (no room for a disease-modifying effect); patients were already on treatments producing the same physiological change; the system was already resistant and downregulated (§28.2); a flat infusion is not the signal the body produces (Chapter 3).

The move that shows real understanding: a student who says "we don't actually know" and then lists the candidates as candidates. That is the correct epistemic state, and it is uncomfortable for students who want mechanism to explain outcomes.

Push back on anyone who sounds too confident, in either direction. Someone who says "obviously it was never going to work" is doing hindsight; ask them what they would have said in 2001, and whether their reasoning would also have rejected sacubitril/valsartan.

3. "Sacubitril/valsartan is a combination product. Make the case that it's a marketing bundle."

What to listen for. This is a deliberate trap and students should fall into it briefly. Let them build the cynical case — patent life, price, the general pattern of combination products — and then ask what neprilysin does to angiotensin II.

The turn: neprilysin degrades angiotensin II, so inhibiting it alone raises the peptide the whole treatment is trying to suppress. The ARB is not an addition; it is what makes the neprilysin inhibitor usable. And it must be an ARB rather than an ACE inhibitor because both neprilysin and ACE degrade bradykinin, and the drug that blocked both produced angioedema often enough to be abandoned.

What this teaches beyond the case: justified skepticism is a heuristic, not a verdict. The students who argued hardest for the cynical reading are the ones most likely to remember why it was wrong here. Say that out loud.

4. "On this drug, BNP goes up while the patient gets better. Explain it, and then tell me why it's the best summary of the chapter."

What to listen for. Mechanically: neprilysin degrades BNP, so inhibiting it raises measured BNP by reducing clearance, independent of cardiac status. NT-proBNP is not a neprilysin substrate, so it still tracks wall stress and falls with improvement. NT-proBNP is the measurement to use.

The second half is the harder half. Look for: the same gene product occupies two roles — diagnostic marker and therapeutic target — and the therapy aimed at one role corrupts the other. The corruption falls precisely on the fragment the enzyme happens to recognize, which is a molecular accident with clinical consequences.

If nobody gets the second half, ask: "What would happen to a patient whose clinician ordered the wrong test?" The answer — escalating treatment in someone who is responding — usually unlocks it.

5. "Preserving beat supplying. That's backwards. Chapter 7 said preserving has a ceiling that supplying doesn't. What happened?"

What to listen for. First, that the ceiling argument is correct as far as it goes: a degradation inhibitor is capped by endogenous production, which is why DPP-4 inhibitors are modest and GLP-1 agonists are not. Then the four candidate explanations from §28.9: timing and disease phase, physiological pattern (Chapter 3), the multi-substrate/network argument, and the difference in comparator and endpoint.

What distinguishes a strong answer: noticing that only one of the four is really about peptides. The rest are about disease biology and trial design. Students who spot that have understood something about how this book works — Chapter 5's machinery does more explanatory work than any amount of molecular detail.

Optional extension: ask them to design a study that would discriminate between their top two explanations. Most will find they cannot, easily. That is a real finding about the limits of clinical inference, not a failure of the exercise.

6. "'Targeting a system is not a strategy.' Take that sentence somewhere outside this chapter."

What to listen for. The sentence generalizes: identifying the right biological system is treated as the hard part, and the specific intervention as an implementation detail, when the evidence says the reverse. The natriuretic peptide system was correctly identified in 1981; the forty years since have been about how, and the how produced one success, one failure, and a diagnostic test more valuable than either.

Good places students take it: the GLP-1 system (supplying worked spectacularly, preserving worked modestly — the mirror image of this chapter); any Part III compound where the pitch is entirely about the system and silent on the intervention; the general pattern in biotech press releases of naming a pathway rather than a mechanism of action.

The closing beat, if you want one: the therapy that worked in this chapter contains no peptide at all. Sacubitril and valsartan are both small molecules. Ask what that implies for the coherence of "peptide therapeutics" as a category — and note that this book's own thesis, from Chapter 1, is that the word peptide carries no evidentiary weight. This chapter is the strongest evidence for that thesis, and it comes from the chapter with the best-supported peptide science in the book.