Stopping a GLP-1 and keeping the weight off: what SURMOUNT-4, STEP 4 and the STEP 1 extension actually measured

People stop GLP-1s for good reasons: they reached a goal, they are planning a pregnancy, the price or the supply changed, the side effects won, or they simply do not want to inject for life. The question they all face is the same and, unusually for this field, it has been answered by randomised trials rather than by opinion. This guide sets out the three studies that matter, what they do and do not show, and how to plan for the period after the last injection.

What the labels say

The Wegovy label's indication is for chronic weight management; the Zepbound label's is for weight reduction and long-term maintenance of weight reduction. Both describe a drug intended to be continued. Neither describes how to stop, whether to taper, or what to expect afterwards. The Zepbound label does summarise its randomised withdrawal trial (Study 4, which is SURMOUNT-4) in section 14, which is as close as a label comes to saying what happens when you stop.

SURMOUNT-4: tirzepatide

Aronne and colleagues (JAMA 2024, PubMed 38078870) gave 783 adults with obesity or overweight open-label tirzepatide (10 or 15 mg) for 36 weeks, during which they lost a mean of 20.9 percent of body weight. The 670 who reached week 36 were then randomised to continue tirzepatide or switch to placebo for a further 52 weeks, without knowing which.

Those who continued lost a further 5.5 percent. Those switched to placebo regained 14.0 percent. At week 88 the continuing group was 25.3 percent below their starting weight and the placebo group 9.9 percent below it. Put differently, the placebo group gave back roughly half of what they had lost in the first year off the drug, and were still losing ground at the end.

STEP 4: semaglutide

Rubino and colleagues (JAMA 2021, PubMed 33755728) ran the same design earlier with semaglutide. Adults with obesity or overweight took semaglutide for a 20-week run-in, escalating to 2.4 mg, and lost a mean of 10.6 percent. The 803 who completed the run-in were randomised to continue or switch to placebo for 48 more weeks.

From week 20 to week 68, those who continued lost a further 7.9 percent; those on placebo regained 6.9 percent. The difference between the arms at week 68 was 14.8 percentage points. Cardiometabolic measures (waist circumference, blood pressure, lipids, HbA1c) improved further with continued treatment and moved back toward baseline on placebo.

The STEP 1 extension: a year off, unsupervised

Wilding and colleagues (Diabetes Obes Metab 2022, PubMed 35441470) followed 327 participants from STEP 1 for a year after the trial drug and the lifestyle intervention both stopped. Those who had been on semaglutide had lost 17.3 percent over the 68-week trial; one year after stopping they had regained 11.6 percentage points, leaving a net loss of 5.6 percent. That is about two-thirds of the lost weight regained. The cardiometabolic improvements seen during treatment largely reverted toward baseline in parallel.

This is arguably the most relevant of the three for real life, because nothing was substituted for the drug: no placebo injections, no trial visits, no structured lifestyle programme.

What the trials do not show

They do not show that everyone regains. Averages hide a spread, and in each trial some participants maintained most of their loss without the drug. They do not identify who those people were in a way that lets you predict yourself.

They do not test a taper. Every one of these trials stopped the drug abruptly at full dose. Whether stepping down through the lower labelled doses before stopping changes the regain curve has not been measured.

They do not test intermittent use. On-and-off cycles, "maintenance doses" below the labelled range, or stretching the interval have not been studied, and the labels do not describe them.

They do not test what happens if resistance training and a protein target are added at the point of stopping. The closest evidence is the liraglutide trial by Lundgren and colleagues (NEJM 2021, PubMed 33951361), where after a low-calorie diet a year of supervised exercise maintained weight loss about as well as liraglutide alone, and the combination did best; when the year ended, the exercise group held its loss better than the drug-alone group. Liraglutide is a weaker, older drug, but the direction of the finding is the best available.

Why the weight comes back

The drugs reduce appetite and slow gastric emptying while they are present. When they are gone, both effects go with them, and the body's response to weight loss, which includes increased hunger signalling and a reduced energy expenditure relative to the new size, is still in place. The regain in these trials is not a rebound above the starting weight and not a sign that the drug did harm; it is the underlying condition reasserting itself once the treatment stopped, which is how the labels' phrase "chronic weight management" should be read.

Planning a stop

If the stop is elective, plan it like a dose change: with the prescriber, with a date, and with what comes next written down.

Decide what you are maintaining. A weight, a waist measurement, a set of blood results, a clothing size. Pick one you will measure monthly, because regain in the trials was gradual and is easy to ignore week to week.

Put the two evidence-based maintenance tools in place before stopping, not after. Resistance training two or three times a week, and a protein intake at the upper end of the range in the muscle guide. Both should already be running while you are on the drug; if they are not, start them and delay the stop.

Plan for appetite to return. It will, within weeks, and it is a physiological event, not a lapse in willpower. Meal structure, protein first, fibre, and no liquid calories are the boring things that work. The weight loss site goes further.

Set a threshold for restarting. Agree with the prescriber in advance what regain would prompt a conversation about resuming, so that the decision is made against a number rather than against a bad week. Restarting is a new escalation from the label's starting dose unless the prescriber decides otherwise; see the switching guide for how that judgement is made.

If stopping for pregnancy, the pregnancy guide covers the semaglutide two-month rule, and the maintenance plan above needs to survive a pregnancy.

If stopping for cost or supply, the insurance guide and the pricing site cover the options, including the lower labelled maintenance doses, which the dose escalation guide explains are on the label and were effective in the trials.

If you have already stopped

The trial curves are gradual. Weight regained in the first three months is a fraction of the year's total, and the tools above work at any point. A year on, the STEP 1 extension's participants were still 5.6 percent below where they began; that is not nothing, and neither is the training habit if you built one.

Compounded products

The trials above studied the brand products. Compounded semaglutide and tirzepatide are not FDA approved and are not interchangeable with them; there are no withdrawal data for any compounded product. The physiology of regain does not depend on the manufacturer. FormBlends describes what its pharmacies dispense on its semaglutide and tirzepatide pages, and its GLP-1 cost report is relevant to anyone stopping for price.