GLP-1 Evidence

Durability

What happens when you stop taking a GLP-1 drug

Most of the weight comes back. This is the most important and least advertised finding in the field, and it reframes what these drugs are for.

Updated 3 min read 9 citations Evidence strength 5/5

The withdrawal experiment

STEP 1 ran for 68 weeks and then stopped the drug in everyone. The extension followed participants for a further year off treatment [5]. Having lost roughly 17% of body weight on average, participants regained about two-thirds of it within twelve months. The improvements in blood pressure, lipids and glycaemic markers moved back toward baseline on the same timescale.

This was not subtle, and it was not a subgroup. It was the central finding of the extension.

Weight regain after discontinuation

The continuation experiment

STEP 4 approached it from the other direction. Everyone ran in on semaglutide for 20 weeks; then participants were randomised either to continue or to switch to placebo [6]. Those who continued went on losing weight for the next 48 weeks. Those switched to placebo regained. Same population, same starting point, opposite trajectories, with the only difference being whether the drug continued.

Continuation versus withdrawal
TrialDesignContinued treatmentStopped treatment
STEP 1 extension [5]68 weeks on, 52 weeks off—Regained ≈ two-thirds of loss
STEP 4 [6]20-week run-in, then randomisedFurther weight lossProgressive regain
STEP 5 [4]104 weeks continuousLoss sustained to 2 years—
SELECT [3]~4 years continuousLoss sustained, events reduced—

Why this is not the scandal it is sometimes presented as

Nobody finds it damning that blood pressure rises when you stop an antihypertensive, or that cholesterol rises when you stop a statin. Obesity behaves like the chronic conditions it sits alongside: treat it, the numbers improve; withdraw treatment, they revert. The reason this lands differently is that weight is still widely understood as a character issue rather than a physiological one, so a treatment that must continue reads as failure rather than as normal chronic disease management [9].

The mechanistic reading agrees. If these drugs reset a set point, weight would not return this reliably. They suppress appetite for as long as they are present, and the appetite returns when they are not.

What this changes in practice

  • Cost is recurring, not one-off. Assess affordability over years, not over a course. This is the most common reason people stop, and stopping predicts regain.
  • Supply interruptions matter. Gaps are not neutral pauses; they are partial withdrawals.
  • Build the habits during treatment, not after. Resistance training and protein intake are far easier to establish while appetite is suppressed. Whether they meaningfully blunt regain has not been properly tested, but they are the only candidate levers available. See muscle loss.
  • Tapering has no evidence behind it. It is widely recommended and entirely untested. It may help; nobody has shown it.
  • The drugs still did something durable. Weight regain does not undo four years of reduced cardiovascular events in someone who stayed on treatment [3].

Common questions

Do I regain more than I lost?
The trial data shows regain toward baseline, not past it. Overshoot is not a finding of the withdrawal studies [5].
Does tapering the dose help?
It is plausible and completely untested. No trial has compared abrupt discontinuation with a taper.
Can I take it intermittently?
No trial has tested intermittent regimens. Given how quickly regain begins after withdrawal, there is no evidential basis for expecting it to work.
What about the lower maintenance dose?
Also untested as a distinct strategy. STEP 4 tested continuation at the full dose against placebo [6], not against a reduced dose.
Was it worth it if the weight comes back?
That depends what you were treating. If the goal was a wedding photograph, the answer may be no. If it was cardiovascular risk, the events prevented while on treatment happened and stay prevented [3].

References

Every citation below links to the original peer-reviewed record on PubMed or via DOI. Nothing here is a substitute for medical advice.

  1. Orforglipron, an Oral Small-Molecule GLP-1 Receptor Agonist for Obesity Treatment Wharton S, Aronne LJ, Stefanski A, et al. · The New England journal of medicine · 2025 · Randomised controlled trial DOIPubMed 40960239
  2. Once-Weekly Semaglutide in Persons with Obesity and Knee Osteoarthritis Bliddal H, Bays H, Czernichow S, et al. · The New England journal of medicine · 2024 · Randomised controlled trial DOIPubMed 39476339
  3. Semaglutide and Cardiovascular Outcomes in Obesity without Diabetes Lincoff AM, Brown-Frandsen K, Colhoun HM, et al. · The New England journal of medicine · 2023 · Randomised controlled trial DOIPubMed 37952131
  4. Two-year effects of semaglutide in adults with overweight or obesity: the STEP 5 trial Garvey WT, Batterham RL, Bhatta M, et al. · Nature medicine · 2022 · Randomised controlled trial DOIPubMed 36216945Full text
  5. Weight regain and cardiometabolic effects after withdrawal of semaglutide: The STEP 1 trial extension Wilding JPH, Batterham RL, Davies M, et al. · Diabetes, obesity & metabolism · 2022 · Randomised controlled trial DOIPubMed 35441470Full text
  6. Effect of Continued Weekly Subcutaneous Semaglutide vs Placebo on Weight Loss Maintenance in Adults With Overweight or Obesity: The STEP 4 Randomized Clinical Trial Rubino D, Abrahamsson N, Davies M, et al. · JAMA · 2021 · Randomised controlled trial DOIPubMed 33755728Full text
  7. Once-Weekly Semaglutide in Adults with Overweight or Obesity Wilding JPH, Batterham RL, Calanna S, et al. · The New England journal of medicine · 2021 · Randomised controlled trial DOIPubMed 33567185
  8. Semaglutide and Cardiovascular Outcomes in Patients with Type 2 Diabetes Marso SP, Bain SC, Consoli A, et al. · The New England journal of medicine · 2016 · Randomised controlled trial DOIPubMed 27633186
  9. Obesity Management in Adults: A Review Elmaleh-Sachs A, Schwartz JL, Bramante CT, et al. · JAMA · 2023 · Review DOIPubMed 38015216Full text