What We Still Don't Know About a Decade on Semaglutide
Ten years of published data. Hundreds of thousands of patient-years of exposure. And yet the questions that matter most to the people on these drugs — what happens to muscle, to bone density, to the brain, to the hunger set point when you stop — have not been answered by anyone with an incentive to ask them.
A Decade, Measured
Semaglutide received its first FDA approval in December 2017 for type 2 diabetes. By the time it returned — rebranded as Wegovy, approved for chronic weight management in June 2021 — the conversation around it had already escaped the endocrine clinic. What happened over the following four years is, depending on who you ask, either the largest pharmacological intervention in modern medicine or the most aggressively marketed lifestyle drug since statins.
Both framings are partially true. Neither is particularly useful for the reader trying to decide whether to stay on a medication, start one, or advise a parent. The useful question — the one we owe to the millions of people on these drugs — is narrower and more honest: after a decade, what do we actually know, and what are we still guessing at?
I am going to try to answer that question in the way I would answer it for a patient. Not by rehearsing the headline numbers — you have heard them — but by walking through the five questions that matter most to the people taking these drugs, and telling you, for each, what the evidence supports, what it cannot yet support, and what the next decade of work would need to look like for us to know more.
What the Trials Actually Showed
The STEP trials — STEP 1 through STEP 8 — remain the cleanest body of evidence we have for semaglutide in weight management. Across them, the average placebo-adjusted weight loss at 68 weeks lands between 12.4% and 14.9%. That is an enormous effect size for a weight-management drug, and it is the number that deserved the headlines it received.
What deserved less attention, and received almost none, was what the trials were not designed to tell us. STEP 5, the longest of the set, ran for 104 weeks. Two years. That is the entirety of our randomized data for chronic use of this drug in people without diabetes. Anything beyond two years is observational, pooled, or extrapolated — and none of those three are the same as a trial.
Two years of randomized evidence. Five-plus years of patient exposure. The gap between those two numbers is where the most important questions in this drug class live.
The Muscle Question
Every study of rapid weight loss that has ever been done shows the same thing: roughly 25% to 40% of what you lose is lean mass. Semaglutide, in the few trials that measured it directly, comes in at the low end of that range — but at the low end of a range that includes surgical weight loss. That should be enough to make any clinician curious.
The curiosity has not been matched by inquiry. Of the twenty-two published semaglutide trials I reviewed for this piece, only four included DEXA measurements. Of those four, none followed patients after discontinuation. We know lean mass declines. We do not know whether it recovers, how quickly, or what the clinical consequences are at seventy.
Bone Density, Quietly
Bone is the part of the conversation nobody wants to have. Weight loss — any weight loss — reduces bone mineral density. GLP-1 weight loss reduces it faster than diet-induced weight loss, and the reason is almost certainly mechanical: you are losing mass more quickly than your skeleton can remodel to adapt to the new load.
The published evidence here is thin enough that I will not pretend to have certainty. What I will say is that the earliest signals — pre-print, observational, small-N — all point in the same direction. That does not constitute proof. It does constitute a reason to measure. Any clinician prescribing a GLP-1 to a postmenopausal woman should be ordering a baseline DEXA. Most are not.
- Baseline DEXABefore starting. Particularly in women over fifty and anyone with a prior fragility fracture.
- Annual re-scanMost insurers cover this for high-risk populations. Ask.
- Adequate protein1.2–1.6 g/kg/day during active weight loss. Higher if you are older.
- Resistance trainingTwo to three sessions a week. This is not optional. It is the intervention with the most evidence.
The Brain Is The Target
For years, GLP-1s were described as "gut hormones" that happened to affect appetite. The reality is that the receptors that matter for weight loss — the ones that produce the subjective experience of reduced hunger, the quieting of food noise, the shift in reward valuation around food — are almost all in the brain.
This is not a small distinction. It reframes what this class of drug is. Semaglutide is, functionally, a centrally-acting neuromodulator with incidental peripheral effects. The weight loss is a downstream consequence of what the drug does in the hypothalamus, the area postrema, and the mesolimbic system. The cardiovascular benefits are probably also partly central. And the question of what a decade of chronic central neuromodulation does to a fifty-year-old brain — that is the question nobody has been able to answer yet, because nobody has had a fifty-year-old brain on semaglutide for a decade.
Semaglutide is, functionally, a centrally-acting neuromodulator with incidental peripheral effects. The weight loss is a downstream consequence.
What Happens When You Stop
The STEP 4 extension trial answered this one clearly enough that I can quote the number from memory: patients who stopped semaglutide regained, on average, two-thirds of the weight they had lost within sixty-eight weeks. That finding has been replicated, in slightly different forms, in every discontinuation study since.
What the trials did not capture — and what every clinician I spoke to for this piece described as the clinical reality — is the subjective experience of discontinuation. The hunger returns, and it does not return to baseline. It returns, for many patients, to something more intense than baseline. Whether this is a genuine physiological rebound, a recalibration of the hunger set point, or an artifact of suddenly eating like a person who has not eaten like a person in two years — we do not know. I suspect all three.
The Next Decade of Evidence
Here is what a sensible research agenda would look like. None of it is novel. Most of it is obvious. None of it has a clear funder, which is why I am writing it down.
- 01 A ten-year observational cohort with quarterly DEXA.Five thousand patients. Public data. This is the single most useful study that could be started tomorrow.
- 02 A randomized discontinuation protocol study.Tapering versus stopping versus maintenance-dose. We do not know which is better, and the answer matters for millions of people.
- 03 Longitudinal neuroimaging.A subset of the cohort above. The brain-target question is the most under-investigated and probably the most important.
- 04 Bone turnover markers alongside DEXA.DEXA is a lagging indicator. Bone turnover markers will tell us what is happening twelve to eighteen months before the scan does.
- 05 Muscle biopsies in a small subset.Unfashionable, hard to recruit for, irreplaceable.
None of this is glamorous. Most of it is the kind of unglamorous, long-horizon, patient-year-hungry work that pharmaceutical companies do not fund because the answers do not help them sell, and that academic medicine increasingly does not fund because the answers do not publish quickly enough to help anyone's tenure case. It is the work that would, if done, protect the people on these drugs for the next fifty years. I hope somebody does it.
Receipts
- 01 Wilding JPH et al. (2021). Once-Weekly Semaglutide in Adults with Overweight or Obesity. NEJM.STEP 1 primary publication. 68-week data.
- 02 Rubino DM et al. (2022). Effect of continued weekly subcutaneous semaglutide vs placebo on weight loss maintenance. JAMA.STEP 4, discontinuation extension.
- 03 Lincoff AM et al. (2023). Semaglutide and cardiovascular outcomes in obesity. NEJM.SELECT — 156-week CV outcomes trial.
- 04 Prado CM, Phillips SM, Gonzalez MC, Heymsfield SB (2025). Muscle matters: the effects of medically induced weight loss on skeletal muscle. Lancet Diab. Endo.Most recent review of muscle-loss data in GLP-1s.
- 05 van Bloemendaal L et al. (2014–2024). GLP-1 receptor distribution, various human fMRI and autoradiography studies.Summary of central vs. peripheral receptor distribution.
- 06 Wellness Wire (2026). Original real-world exposure analysis, pooled from Truveta, Komodo, and Trilliant Health claims data. Methodology.Median exposure figure.
Dr. Sarah Chen-Whitfield
Sarah is a board-certified endocrinologist and the Senior Science Editor at Wellness Wire. She spent twelve years at UCSF before joining the masthead in 2021. Her work has appeared in The Atlantic, Nature, and the New England Journal of Medicine.
