Case Study 2 — The Drug for a Few Dozen People

Precision medicine when the population is countable

Type: Real, public, ongoing · Tier 1 approval facts, Tier 2 economics · Relevance: §13.5

Background: the opposite of a blockbuster

Everything in Part II so far has been about very large populations. Semaglutide is prescribed to millions. The commercial logic of pharmaceutical development assumes it: development costs are enormous and fixed, so they are recovered across a large number of patients.

Setmelanotide inverts every part of that.

It treats obesity caused by specified rare genetic deficiencies in the leptin–melanocortin pathway — POMC deficiency, leptin receptor deficiency, and certain related conditions. The eligible population worldwide is small enough to be plausibly countable, and many affected individuals have never received a molecular diagnosis at all.

The development program therefore faced a problem that has nothing to do with pharmacology: how do you run a trial when you cannot find the patients?

The identification problem

To enroll a patient you must first know they have the condition. To know that, someone must have suspected a monogenic cause and ordered genetic testing.

Most did not. Severe childhood obesity is common; monogenic obesity is a small fraction of it; and for most of the period during which these children were growing up, testing was expensive, rarely ordered, and not obviously actionable — because there was no treatment.

This produces a circularity worth naming: the treatment made the diagnosis worth making. Before setmelanotide, identifying POMC deficiency changed management at the margins. Afterward it changed everything. The drug did not merely treat a diagnosed population; it substantially created the diagnostic practice that identifies that population.

THE PRECISION MEDICINE LOOP

  no treatment ──▶ testing rarely ordered ──▶ patients not identified
       ▲                                              │
       │                                              ▼
  no trial population  ◀──── cannot recruit ◀─── diagnosis not made
       │
       └── (broken by) ── a sponsor willing to fund both the trial
                          AND the diagnostic effort that makes it possible

  For rare disease, THE TRIAL AND THE DIAGNOSTIC INFRASTRUCTURE ARE THE
  SAME PROBLEM. This is not true for common conditions and is easy to
  underestimate from outside.

What the trials showed, and what "small" means for evidence

The pivotal studies enrolled small numbers of genetically confirmed patients and reported substantial weight reduction and — emphasized by patients and families as more important — marked reduction in hyperphagia, the relentless, distressing hunger that defines these conditions.

Now apply Chapter 5 honestly. A trial of a few dozen participants would normally be a serious limitation. Small samples produce imprecise estimates; §5.10 lists tiny sample size as a red flag.

Why the concern applies with much less force here, and this is the case study's methodological point:

The effect size is enormous relative to the variability. Chapter 5's concern about small trials is that a modest true effect is hard to distinguish from noise. When the effect is dramatic, the sample required to detect it is small — the same reason insulin needed no trial in 1922 (Chapter 11 §11.1). The strength of evidence required scales with the plausibility of alternative explanations, and for a genetically defined deficiency corrected by an exact-mechanism agonist, the alternatives are thin.

The mechanism is confirmed rather than proposed. These patients have an identified molecular defect and the drug acts precisely downstream of it. This is not mechanism used to substitute for evidence (which Chapter 2 forbids); it is mechanism used to make a small trial interpretable, which is legitimate.

And withdrawal designs are informative. Trials in which treatment is withdrawn and hyperphagia returns provide within-patient evidence that a parallel-group design in a large population would need many more participants to match.

What remains genuinely limited: long-term safety data in a small, young population; effects across the full range of genotypes; and durability over decades. These are real gaps and small numbers make them harder to close.

⚠️ Hype Check — "a cure for genetic obesity"

What's true: for the specified deficiencies, this is as close to a mechanism-corrective treatment as obesity pharmacology has produced. The effect on hyperphagia in particular is reported as life-altering by families, and that report is credible.

What "cure" overstates: the drug does not repair the genetic defect. It bypasses it, and only while it is being taken. Stop it and the deficiency is still there — which makes it chronic therapy in exactly the sense Chapter 8 §8.9 defined, and which raises the same access questions Chapter 12 raises, for a population too small to have any market power.

What "genetic obesity" overstates: the approval covers specified deficiencies. Many people with severe early-onset obesity have no identified monogenic cause, and some have variants of uncertain significance. "Genetic obesity" is broader than the approved indication, and the gap between them is where inappropriate use would occur.

The honest version: a genuinely transformative treatment for a small, molecularly defined population, which must be continued indefinitely, and whose indication is narrower than the phrase people reach for.

The economics, stated without villains

A drug for a few thousand patients worldwide, at most, cannot recover development costs the way a blockbuster does. Orphan drug frameworks exist precisely for this — offering extended exclusivity, fee reductions, and development incentives in exchange for pursuing conditions that would otherwise be commercially irrational.

The consequence is a price per patient that looks extraordinary next to a common-disease drug and that reflects arithmetic rather than opportunism: fixed development costs divided by a very small denominator.

Two things are simultaneously true, and Chapter 12's discipline applies here as much as it does to semaglutide.

Without the incentive structure, this drug would not exist. No rational developer pursues a population this small on ordinary commercial terms. The framework worked as designed.

And the resulting prices strain health systems and produce genuinely difficult coverage decisions for individual patients whose alternative is a condition with no other treatment. That difficulty is not resolved by noting that the incentive was necessary.

The uncomfortable structural observation: the smaller the population, the stronger the case for public or philanthropic development rather than market-based development — and the weaker the political constituency to argue for it.

What this case teaches

Precision medicine is real and it is narrow. Setmelanotide is what it looks like when a mechanism is genuinely understood, a population is molecularly defined, and a drug corrects the specific defect. It is the field working. It is also a treatment for a population smaller than a large secondary school.

A proven mechanism does not generalize. §13.5 made this point and this case sharpens it: the mechanism here is not merely plausible but demonstrated, and it still does not extend to common obesity, because the premise it depends on — a broken upstream step — is absent there.

Small trials are not automatically weak evidence. They are weak evidence for modest effects. For a dramatic effect against a confirmed mechanism in a defined population, a small trial can be close to definitive — which is Chapter 11's principle appearing in a modern setting rather than a historical one.

And the diagnostic infrastructure is part of the treatment. For rare disease, finding patients and treating them are not separable problems.

Discussion questions

1. The treatment made the diagnosis worth making. Describe the circularity, and name another area of medicine where the same loop operates.

2. Chapter 5 lists tiny sample size as a red flag, and this case study argues the concern applies with much less force here. Reconstruct the argument. Is it special pleading? What would make it so?

3. Compare setmelanotide's evidence base with insulin's in 1922 (Ch 11 §11.1). What do they share, and what does that suggest about when small or uncontrolled evidence is sufficient?

4. "Genetic obesity" is broader than the approved indication. Describe concretely how inappropriate use could arise from that gap, and what would prevent it.

5. Orphan drug incentives made this drug exist and produced a price that strains health systems. Is there an arrangement that achieves the first without the second? Argue for a specific alternative and identify who would object.

6. Compare across the book. Leptin for congenital deficiency, setmelanotide for POMC deficiency, mecasermin for severe IGF-1 deficiency, insulin for type 1 diabetes. State what all four share, and what that pattern predicts about where the next genuinely dramatic peptide treatment will come from.