Chapter 9 — Key Takeaways

Tirzepatide and Beyond

The core claims

Tirzepatide is a unimolecular dual agonist — 39 amino acids, ~4,800 Da, activating both the GIP and GLP-1 receptors with a ratio fixed by its structure. Its sequence is based on GIP, not GLP-1.

One molecule ≠ two drugs. One pharmacokinetic profile, one fixed activity ratio, one injection. A dual agonist is a bet on a specific ratio, and getting that ratio right is the design problem.

It works despite GIP having been reasonably written off — blunted insulinotropic effect in type 2 diabetes, apparent fat-storage promotion. Chapter 7 §7.5's four candidate explanations remain unresolved, and both GIP agonists and GIP antagonists are in development, which is the honest indication of that.

SURPASS (diabetes, approved 2022) and SURMOUNT (obesity, approved 2023) — the same two-brand-names-one-molecule pattern as Ozempic/Wegovy.

SURPASS-2 found tirzepatide superior to semaglutide 1 mg in type 2 diabetes at all three doses. Real, and narrower than "tirzepatide is better": different indication, comparator at the approved diabetes dose, manufacturer-funded.

The chapter's central skill: the two estimands

SURMOUNT-1, tirzepatide 15 mg, 72 weeks, adults with obesity without diabetes:

~−21% — the treatment-regimen estimand. "What happens if a doctor prescribes this?" Counts everyone as randomized, including those who stopped.

~−22.5% — the efficacy estimand. "What happens if you take it, for the full period?" Restricted to those who remained on treatment.

Same trial, same dose, same duration. Both pre-specified. Both correct.

The efficacy figure is systematically larger, because the people excluded from it stopped for reasons correlated with doing badly. An unnamed figure is therefore biased, not merely imprecise — and the size of the gap tells you about tolerability, which reporting only one figure discards.

"Up to 22.5%" performs three selections at once: the highest dose, the efficacy estimand, and the non-diabetic population.

The Phase discipline

  PHASE 2                              PHASE 3
  dozens to a few hundred              hundreds to thousands
  often shorter                        long enough for the real question
  frequently a favorable population    the population the drug is for
  multiple doses; the best is quoted   the dose that will be used
  detects a SIGNAL                     MEASURES an effect

  Five inflating mechanisms: small samples -> high variance · regression to the
  mean · best-dose selection · favorable populations · shorter duration.
  NONE requires anyone to do anything wrong.

A −24% Phase 2 figure and a −21% Phase 3 figure are not comparable quantities. Ranking them is a category error, not a close judgment call.

Evidence ratings issued in this chapter

Claim Rating Note
Tirzepatide for weight loss in adults with obesity/overweight-plus-comorbidity, without diabetes, while treatment continues ✅ ~−21% / ~−22.5% at 72 weeks; largest weight effect of any approved pharmacotherapy as of this writing
Tirzepatide for glycemic control in type 2 diabetes ✅ SURPASS; approved 2022; superior to semaglutide 1 mg head-to-head
Tirzepatide for cardiovascular outcomes NOT RATED Outcomes program ongoing. More weight loss does not substitute for a hard endpoint.
Retatrutide for weight loss ⚠️ Phase 2 ~−24% at 48 weeks. The book's standing example of a ⚠️ that reads like a ✅
Orforglipron for weight and glycemic control ⚠️ Oral small molecule, late-stage, not approved as of 2026
CagriSema ⚠️ Two molecules, two mechanisms, program ongoing
A striking Phase 2 result predicts a comparable Phase 3 result ❌ An inference, rated — see Case Study 2

Reading a pipeline announcement — six questions

  1. WHAT PHASE?          If unstated, assume 2 or earlier.
  2. WHICH ESTIMAND?      If unnamed, assume the larger.
  3. WHICH DOSE?          "Up to" = the highest studied.
  4. WHICH POPULATION?    Diabetes changes the answer by a third. Every time.
  5. WHAT ENDPOINT?       Weight and A1C are surrogates. Events are not.
  6. WHO, AND WHERE?      Press release != conference abstract != publication.

The habit worth building: when a pipeline compound's headline number exceeds an approved drug's, your first thought should be "what phase?" rather than "that's better."

Key terms

unimolecular dual agonist · polyagonist · SURPASS · SURMOUNT · head-to-head trial · non-inferiority · superiority · retatrutide · cagrilintide · amylin analog · orforglipron · Phase 2 optimism

What you can now evaluate

  • ✅ any weight-loss figure — by asking for dose, duration, population, and estimand
  • ✅ what a head-to-head trial does and does not establish
  • ✅ why a bigger Phase 2 number is not a better result
  • ✅ a comparison between two drugs, with the "what differs that isn't the molecule" row filled in
  • ✅ a pipeline announcement, in six questions
  • ❌ not yet: what else these drugs do. Chapter 10.

The one-sentence version

Tirzepatide does more than semaglutide on weight and has less evidence on outcomes; retatrutide's bigger number is a Phase 2 number; and the most consequential thing in the pipeline may be a small molecule that is not a peptide at all.

Next: Chapter 10 — the heart, the kidney, the liver, the brain, and addiction. One molecule, five ratings, two of them 🔬.