For most of the history of obesity medicine the practical ceiling of drug treatment was a loss of a few per cent of body weight beyond what lifestyle advice achieved. The incretin-based agents changed that. Liraglutide at its obesity dose gave a moderate but dependable effect (Pi-Sunyer et al., 2015), once-weekly semaglutide roughly doubled it (Wilding et al., 2021), and tirzepatide, which acts on both the glucagon-like peptide-1 (GLP-1) and the glucose-dependent insulinotropic polypeptide receptors, produced average losses that approach those seen after some bariatric procedures (Jastreboff et al., 2022). Semaglutide has since been shown to lower the rate of major cardiovascular events in people with established cardiovascular disease and overweight or obesity but without diabetes (Lincoff et al., 2023).

Success on treatment raises an obvious question about what follows it. People stop these medicines for many reasons: nausea and other gastrointestinal effects, cost and insurance rules, shortages, pregnancy planning, or a belief shared by patient and clinician that a target weight marks the end of a course. The idea of a course is borrowed from acute illness and sits uneasily with the modern description of obesity as a chronic, relapsing disease process (Bray et al., 2017). Whether weight returns after withdrawal, how fast, and with what consequences are therefore questions of direct clinical weight.

This article reviews the published evidence on weight trajectories in adults with obesity after incretin-based therapy is stopped, and sets that evidence against what was already known about the biology of weight regain.

This is a narrative review of the published literature. It draws on the randomised trials that included a planned withdrawal or an off-treatment follow-up, on physiological studies of weight loss maintenance that predate the current drugs, and on reports of treatment persistence in routine practice. The focus is adults treated for obesity or overweight with weight-related complications; trials in type 2 diabetes are mentioned only where they clarify mechanism. No formal search protocol or quantitative synthesis was undertaken, and the selection reflects the author's judgement of which sources are most informative. Trial results are described in words and in approximate terms, and readers who need exact estimates should consult the original reports.

Regain after weight loss is not peculiar to drug treatment. A pooled analysis of structured weight loss programmes in the United States found that, on average, well over half of the lost weight had returned within a few years, with a small fraction maintained at five years (Anderson et al., 2001). The explanation offered by physiology is that body weight is defended. Leibel et al. (1995) showed in inpatient studies that maintaining a reduced body weight is accompanied by a fall in energy expenditure larger than the change in body mass predicts, and follow-up of participants in an intensive televised weight loss competition suggested that this adaptation can persist for years (Fothergill et al., 2016).

Appetite appears to matter more than expenditure. Sumithran et al. (2011) measured circulating mediators of appetite before and after diet-induced weight loss and found that leptin, peptide YY and cholecystokinin remained lower, and ghrelin and subjective hunger higher, a year after the diet ended. Polidori et al. (2016) used the response to a sodium-glucose cotransporter 2 inhibitor to estimate the strength of the feedback and concluded that the drive to eat rises in proportion to weight lost and outweighs the accompanying fall in energy expenditure several times over. A person who has lost weight is, in this account, biologically pressed to eat more than a person of the same weight who was never heavier (Hall and Kahan, 2018).

GLP-1 receptor agonists act on precisely this pressure. They reduce food intake through receptors in the hypothalamus and hindbrain, slow gastric emptying, and alter the reward value of food (Drucker, 2018; Müller et al., 2019). They do not, so far as is known, reset the defended weight. When the drug is cleared, the pharmacological brake on appetite is released at the moment when the endogenous signals favouring intake are at their strongest, because the person now weighs much less than before. On physiological grounds alone, regain after withdrawal is the expected outcome and sustained loss the one that would need explaining. Figure 1 groups the influences on the post-treatment trajectory that are discussed in this review.

Four bodies of trial evidence bear directly on the question. Their designs differ in ways that affect interpretation, and Table 1 summarises them before each is discussed.

Total body weight is a blunt outcome. Any substantial weight loss includes fat-free mass as well as fat, in proportions that vary with the size and speed of loss, sex, age, protein intake and activity (Heymsfield et al., 2014). A long-standing concern in the weight cycling literature is that regain may restore fat more readily than lean tissue, leaving a person at the original weight with a less favourable composition. Whether this happens after incretin withdrawal is not established. The published withdrawal trials report weight and waist circumference, and detailed body composition during the off-treatment period is scarce. The question is of particular concern for older adults, in whom loss of muscle carries functional consequences.

The fate of cardiometabolic benefit is clearer. In the STEP 1 extension the improvements in risk factors tracked weight back towards baseline (Wilding et al., 2022). The cardiovascular outcome benefit shown for semaglutide was obtained with ongoing treatment (Lincoff et al., 2023), and there is no trial evidence that it outlasts the drug. It would be unsafe to assume a durable legacy effect from a time-limited course.

Trials describe withdrawal under supervision in people selected for adherence. Routine care looks different. Analyses of dispensing records from health systems in the United States indicate that a large proportion of people who start an anti-obesity medicine are no longer collecting it a year later, with persistence varying by agent and related to the weight loss achieved (Gasoyan et al., 2024). Discontinuation is therefore the common experience, and much of it is unplanned. The trial populations, by contrast, stopped at a scheduled visit after reaching and tolerating a full dose. People who stop early because of side effects or cost will have lost less and may differ in other respects, so the proportion regained in trials cannot simply be transferred to them.

Practice has moved ahead of evidence on how to stop. Gradual dose reduction, longer intervals between injections, a switch to a lower maintenance dose, and intensified dietary and exercise support at the time of withdrawal are all in use. Each is plausible. None has been compared with abrupt cessation in an adequately sized randomised trial with body composition as an outcome, and reports of success with tapering come mainly from uncontrolled service evaluations. Guidelines written before the current agents already advised that pharmacotherapy be continued when it is effective and tolerated (Apovian et al., 2015), and nothing in the withdrawal trials contradicts that advice.

Several questions follow from the evidence reviewed. The first concerns heterogeneity. Average trajectories conceal individuals who maintain most of their loss after stopping, and it is not known whether they can be identified in advance by duration of treatment, behaviour change during treatment, starting weight or biology. The second concerns duration: the randomised withdrawal periods last about a year, and whether weight eventually settles below, at or above the pre-treatment level has not been observed. The third is the comparison of exit strategies, including tapering, lower-dose maintenance, and structured exercise, in trials that measure fat and lean mass, physical function and quality of life as well as the scale.

Two further matters deserve study. Repeated cycles of starting and stopping, driven by supply or payment, are now common and their metabolic and psychological effects are undescribed. And the experience of regain after a period of marked success, including its effect on mood, eating behaviour and willingness to re-engage with care, has received little systematic attention. Observational data from registries can describe these patterns, but confounding by the reason for stopping will limit causal inference, and randomised designs remain necessary for the questions about strategy.

The published trials agree with one another and with the older physiology. When semaglutide or tirzepatide is withdrawn from adults with obesity, weight rises progressively over the following year and the associated improvements in cardiometabolic risk factors recede. The effect is best understood as the removal of a treatment from a chronic condition whose underlying biology is unchanged. Supervised exercise during treatment is the one adjunct with randomised support for improving what follows, and that support rests on a single trial. Counselling before treatment starts should make the likelihood of regain after stopping explicit, and research should now turn from showing that regain occurs to finding out for whom, and by what means, it can be limited.

To be declared by the author(s) before publication.

To be declared by the author(s) before publication.