SURMOUNT-4: what happened to weight when people stopped tirzepatide

SURMOUNT-4 is a randomised trial of what happens to weight when tirzepatide is replaced with placebo. Adults who switched to placebo gained a mean 14.0% relative to their week-36 weight over 52 weeks, while those who kept taking it lost a further 5.5% [1]. The result sits alongside other withdrawal trials of GLP-1-based drugs in a 2026 systematic review [2].

Why this study matters

Most obesity drug trials answer one question: how much weight comes off while people are on treatment. SURMOUNT-4 answers the next one. The authors state that the effect of continued tirzepatide on maintaining an initial weight reduction was unknown before the trial [1].

Tirzepatide is a once-weekly injectable peptide that acts on two gut-hormone receptors, GIP and GLP-1. We cover its main trials in our tirzepatide evidence guide, where SURMOUNT-4 appears as one section. This page takes the trial on its own, because the stopping question is the one most people have before they start.

What they did

The trial ran at 70 sites in 4 countries. It enrolled adults with a body mass index of 30 or more, or 27 or more with a weight-related complication, and it excluded people with diabetes [1].

Everyone began with 36 weeks of open-label tirzepatide at the maximum tolerated dose, either 10 mg or 15 mg weekly. At week 36, 670 of the 783 who entered were randomised one to one to carry on with tirzepatide (n = 335) or switch to placebo (n = 335) for 52 weeks. All participants were given lifestyle counselling to encourage a healthy diet and regular physical activity, in both arms [1].

The primary end point was the mean percentage change in weight from week 36 to week 88. A key secondary end point was the share of people who still held at least 80% of the weight they had lost during the lead-in [1].

What they found

People who completed the lead-in had lost a mean of 20.9% of their body weight by week 36 [1]. The randomised period then split the two groups in opposite directions.

The difference between groups in the primary end point was −19.4% (95% CI −21.2% to −17.7%; P below .001) [1].

Two readings of that table matter. First, stopping did not return people to where they began: the placebo group was still 9.9% below its starting weight at week 88. Second, 14.0% is an average regain measured against the week 36 weight, so it describes the direction and scale of the change for the group, not a figure any one person should expect.

The authors concluded that withdrawing tirzepatide led to substantial regain of lost weight, whereas continued treatment maintained and added to the initial reduction [1].

Safety reported

The abstract reports that the most common adverse events were mostly mild to moderate gastrointestinal events, and that they occurred more often with tirzepatide than with placebo [1]. In the double-blind period (week 36 to 88), any adverse event was reported by 60.3% of the tirzepatide group and 55.8% of the placebo group. Serious adverse events were 10 (3.0%) in each group. One person in each group died (heart failure on tirzepatide, colon adenocarcinoma on placebo; neither judged drug-related). Stopping because of an adverse event was 1.8% on tirzepatide and 0.9% on placebo. Diarrhoea was 10.7% vs 4.8%, nausea 8.1% vs 2.7% and vomiting 5.7% vs 1.2%. No pancreatitis was adjudicated. Acute cholecystitis occurred in 0 vs 3 people, and hypoglycaemia below 54 mg/dL in 2 vs 0 [1].

The lead-in carried more side effects: 81.0% reported at least one, including nausea (35.5%), diarrhoea (21.1%), constipation (20.7%) and vomiting (16.3%). In all, 113 people stopped the drug during the lead-in, most because of an adverse event or withdrawal [1].

Caveats

• Sponsor. Eli Lilly and Company, which makes tirzepatide, sponsored the study. Its employees are among the authors. The paper states that the sponsor was involved in study design and conduct, data collection, management, analysis and interpretation, manuscript preparation, review or approval, and the decision to submit. It also states that the sponsor had no veto over publication or the choice of journal, and that final decisions rested with the authors, who included sponsor employees [1].
• Who was randomised. Only people who completed the 36-week lead-in were randomised, 670 of 783. The authors note that people who tolerated initial treatment may represent a subgroup. Of the 783 who entered, 113 stopped the drug in the lead-in, and 92.7% of those randomised were on the 15 mg dose, so the result describes people who tolerated the drug, not everyone who starts it [1].
• Who was excluded. People with diabetes were left out, so the trial does not speak to them [1].
• Duration. The withdrawal period lasted 52 weeks. What happens to weight in later years was not tested [1].
• Design. The trial compared continuing against switching to placebo. Dose adjustment was not permitted after randomisation, so it did not test tapering or a lower dose. It did not evaluate intensive behavioural therapy, and the authors note it was not designed to represent racial diversity [1].

What it means in practice

A 2026 systematic review of 18 reports from 16 study families looked at regain after stopping GLP-1-based drugs. It cites SURMOUNT-4's +14.0% over 52 weeks, and +6.9% over 48 weeks after semaglutide was switched to placebo in STEP 4 [2]. In the STEP 1 extension, people regained 11.6 percentage points over a year, about two-thirds of their prior loss [2].

The same review found that real-world estimates were smaller and more variable, because stopping was often inferred from prescription records, follow-up weights were missing and many people restarted or switched drugs. It also found that cardiometabolic measures tended to move back as weight returned, and that clinical cardiovascular events were not evaluated [2].

On ways to limit regain, the review found one randomised study supporting supervised exercise after liraglutide, and no controlled trial of a specific diet after stopping. It judged the evidence on tapering and other transition strategies to be limited [2]. Treatment decisions belong with a prescriber who knows the person's history. This article reports what the trials measured and does not give guidance on starting, stopping or dosing.

What is still open

• Whether any tapering schedule changes the regain pattern, since SURMOUNT-4 tested only continuing against switching to placebo [1,2].
• How the picture looks beyond one year off treatment, and in people with diabetes, who were excluded [1].
• Whether the same pattern holds outside a trial population that had already completed 36 weeks of treatment [1].

What this means

We read SURMOUNT-4 as one randomised trial, run by the drug's sponsor, in which much of the on-treatment weight loss was lost after switching to placebo. In the 335 people switched to placebo, the average direction was regain relative to week 36, and in the 335 who continued, it was further loss [1]. A randomised withdrawal in semaglutide, STEP 4, points the same way [2]. These are group averages under a treatment-regimen analysis, so they do not show what any one person who stops will see. The open question is size for an individual, and tapering.

HPR treats obesity as a chronic condition, and a weight-loss drug as a treatment people may need to plan around over the long term, with the decision to stay on or come off made openly with a clinician. Equally, a 14.0% regain is not a failure of the drug. The placebo group was still 9.9% below its starting weight at week 88, so the effect did not disappear [1]. To disagree with this reading, a reader would need a randomised comparison showing weight stays off after stopping, and we know of none in this trial or in the 2026 review [1,2].

Educational content only — not medical advice. Many peptides discussed on HPR are not approved for human use. Talk to a qualified clinician before making any decision about your health.