
This dossier grades an absence. Women in the menopausal transition are among the heaviest users of incretin therapy, and the pivotal trials were not designed to answer anything about them. We ran the search: a PubMed query combining GLP-1 agonists, postmenopausal women, body composition, bone and lean mass returned ZERO indexed results — not few, none. A semantic search across the peer-reviewed corpus surfaced three papers, of which one is a narrative review of women over 65, one is a case report of a single patient, and one is a general lean-mass paper that is not menopause-specific at all. Broadening the query returns breast-cancer literature and a study in ovariectomised mice. What that thin base does support is that the drug still works here — 10–20% sustained loss in women 65+, with cardiometabolic improvement. What it does not support is any statement about the skeletal and muscular cost in a body already losing bone. The market has filled that vacuum with confident instruction. We are publishing the shape of the hole instead.

Two facts sit next to each other and almost nobody puts them in the same sentence. The first: women in the menopausal transition are among the heaviest users of incretin therapy — the class arrived precisely when the metabolic terrain shifts, visceral fat redistributes, and conventional approaches stop working. The second: the pivotal trials were not designed to answer anything about them. Menopausal status is rarely a stratification variable, oestrogen status is rarely reported, and bone and muscle endpoints in this specific population are largely absent. The drug is not the problem here. The reading is.
A PubMed search combining GLP-1 receptor agonists, postmenopausal women, body composition, bone and lean mass returned ZERO indexed results. Not few. None. We report this the way we would report a trial result, because in a field this crowded an empty query is a finding — it means the intersection every clinic markets to has not been assembled as a research question.
A semantic search across the peer-reviewed corpus surfaced three relevant papers. One is a NARRATIVE review — not systematic — covering women over 65 rather than the menopausal transition (Moscucci et al., Nutrients 2026). One is a CASE REPORT: a single 65-year-old woman on semaglutide (Ambrogini, JBMR 2026). The third is the general lean-mass review already cited in Dossier 021 and 029, which is not menopause-specific at all (Neeland et al., Diabetes Obes Metab 2024). Broadening the search returns mostly breast-cancer literature and a study in ovariectomised mice.
Industry commentary widely repeats that roughly 70% of GLP-1 users are women navigating perimenopause and menopause. We could not trace that figure to a peer-reviewed source — it appears in commercial and trade material. It may well be true, and it is the reason this dossier exists, but we will not print it as evidence. A number that supports our own thesis is exactly the number we are obliged to check hardest.
In phase 3 trials, women aged 65 and over achieved sustained weight reduction of 10–20% with consistent cardiometabolic improvement (Moscucci et al. 2026). Nothing suggests the drug stops working after the menopausal transition, and a 2025 analysis reported tirzepatide producing weight reduction in postmenopausal women comparable to premenopausal. This is the part of the picture that is reasonably solid — and it is also the only part the market talks about.
The menopausal transition is already a period of accelerated bone loss. Layering rapid weight reduction on top of it is a plausible compounding risk, and current commentary now argues that in older patients losing ~9% of body weight on semaglutide, monitoring bone remodelling markers and repeating DXA after one year is justified — rather than the usual two-to-three-year interval (Ambrogini, JBMR 2026). That recommendation rests on a case report and expert reasoning, not on trial data. We grade it as a signal worth acting on with a physician, and explicitly not as a demonstrated outcome.
Sarcopenia risk rises with age and with oestrogen withdrawal, and the potent incretins produce large, rapid weight loss. The concern is coherent. What does not exist is a trial showing that lean-mass loss is WORSE in this population than in any other — and Dossier 021's correction applies here too: the ~25% lean-mass fraction is the price of losing weight by any means, not a distinctive drug penalty. The honest statement is that the risk is plausible, the population is more vulnerable at baseline, and nobody has measured the interaction.
How does the largest user population end up the least studied one?
Registration trials are powered for weight and cardiometabolic endpoints in broad obesity populations. Menopausal status is a covariate nobody was required to stratify by, so the data may exist inside trial databases without ever having been analysed or published as a subgroup. This is the cheapest gap to close and the one most likely to close first.
Moscucci et al., Nutrients 2026 — narrative review of trial data in older women ↗Bone density and muscle function are slow endpoints requiring DXA, remodelling markers and strength testing over years. Weight is fast, cheap and on the label. A trial optimised to demonstrate weight loss has no reason to carry the instruments that would detect the skeletal cost, so the cost goes unmeasured rather than being found absent.
Ambrogini, J Bone Miner Res 2026 — the monitoring argument ↗The party with the resources to run the subgroup analysis is the party that sells the drug. Finding nothing is worth little to them; finding something is worth less. This is not an accusation of misconduct — it is the ordinary shape of a literature where the funder chooses the question. It is also precisely why an independent compendium exists, and why we say who paid for evidence in the same breath as we publish it.
See Dossier 029 — sponsor disclosure policy applied to SURMOUNT-4 ↗Where evidence is absent, content is not. The subject is served by an enormous volume of confident writing — protocols, dosing advice, hormone-stacking guidance — built on mechanism, extrapolation and anecdote. The absence of data has not produced an absence of instruction. It has produced instruction with nothing underneath it.
OCCABUZZ discard rule 01 — the outcome sold was never the outcome measured ↗The gap is structural, not accidental: nobody designed the trials to ask, the instruments that would answer are slow and expensive, and the only party who could fund the answer has no reason to want it. Which means it will not close on its own, and until it does the honest position is the uncomfortable one — the drug probably works here, the costs are plausible and unmeasured, and anyone selling certainty in either direction is selling something we cannot audit. We publish the shape of the hole because the shape is real, and knowing where the map ends is a usable fact.
Because the alternative was to publish nothing, and silence would have been read as absence of risk. Our grade ladder has an invisible rung — NOT PUBLISHED — for things that fail the audit. But this subject does not fail the audit; it was never given one, by anyone. Those are different states, and collapsing them would leave the most exposed population with the least information.
This is what a calibrated read looks like when the calibration comes back low. The certainty score on this dossier is deliberately among the lowest in the compendium, and it should be. A number in the forties is not a failure of the analysis — it is the analysis, reporting honestly on a base that cannot support more.
If you are in this window and considering an incretin, the usable conclusion is narrow and real: the drug will probably work, the questions your physician should be asking about bone and muscle do not yet have trial answers, and that is a reason for closer monitoring rather than for avoidance or for confidence. Take this dossier to the appointment. It is written to make that conversation better, not to replace it.
Peer-reviewed trials — sample size, effect size, stage.
Applies to healthy apex, not only to clinical deficit.
Survives a demanding calendar. Zero executive friction.
Signal is real but unproven. Watchlist, not protocol.
The score is a geometric mean — a single failed layer collapses it. Excellence in two cannot rescue a gap in the third. That is why hype scores low and proven, feasible, broadly-applicable work scores high. The restraint is the product.