Semaglutide: The Deepest Record, On Treatment and Off It

The short version

Semaglutide (Ozempic, Wegovy, Rybelsus) is a laboratory-made copy of GLP-1, a hormone the gut releases after a meal. Natural GLP-1 clears quickly. Semaglutide is rebuilt to survive far longer, producing an extended signal rather than a brief one.

That signal reaches appetite circuits in the brain. Hunger falls, meals end sooner, and food intake drops. The weight change follows from eating less rather than from burning more.

It has the largest evidence base of the three compounds on this desk, including trials that counted cardiovascular and kidney events rather than only body weight. It also has the clearest record of what happens when treatment ends: an extension that kept following participants reported substantial regain, while cardiometabolic improvements drifted back toward baseline.

Nothing on this page is advice, and no quantity described here is a recommendation for any person.

What it is

Semaglutide is an acylated analogue of human glucagon-like peptide-1. Carefully chosen backbone substitutions resist enzymatic cleavage, while a fatty di-acid side chain attached through a spacer binds reversibly to serum albumin. Together those changes turn a short natural hormone signal into a long-acting compound. The safety review in this desk's source set describes the resulting clinical pharmacology and tolerability profile ^[5].

Regulatory status is unusually broad. Semaglutide is approved for type 2 diabetes mellitus, chronic weight management, reduction of major adverse cardiovascular events in adults with established cardiovascular disease and overweight or obesity, and metabolic dysfunction-associated steatohepatitis. Injectable and oral formulations are approved. Under current anti-doping rules GLP-1 receptor agonists are not specifically prohibited, though athletes subject to testing verify the current list rather than relying on a research summary.

The development codes NNC0113-0217, NN9535 and OG217SC appear in the literature alongside the general descriptions long-acting GLP-1 analogue and acylated GLP-1 analogue.

How it works

Semaglutide is a long-acting agonist at the GLP-1 receptor. By activating those receptors it potentiates glucose-dependent insulin secretion from pancreatic beta cells, suppresses inappropriate glucagon release from alpha cells, and slows gastric emptying. The glucose-dependent part matters: insulin release is amplified when glucose is high rather than driven unconditionally, which is why the compound carries a lower intrinsic hypoglycaemia risk than agents that push insulin regardless of glucose.

The weight effect, though, is primarily central. A rodent study mapping where the compound actually goes found that it reached the brainstem, the area postrema, the hypothalamic arcuate nucleus and the parabrachial nucleus, reduced food intake and shifted food preference — and did so without lowering energy expenditure ^[6]. That is a mechanistically clean result and it has a direct bearing on the cessation question. If the entire effect is an intake effect produced by continuous occupancy of receptors in feeding circuits, then removing the occupancy removes the effect, and the appetite that was suppressed is the appetite that returns.

The receptor map runs wider than appetite: beta-cell and alpha-cell receptors for the glycaemic effects, gastric smooth muscle and vagal afferents for delayed emptying and much of the nausea, and cardiovascular and renal receptors implicated in the pleiotropic effects that the outcome trials went looking for.

What the research shows

Weight. In the STEP 1 randomised trial, once-weekly subcutaneous semaglutide 2.4 mg produced a mean body-weight change of -14.9% from baseline to week 68, against -2.4% with placebo, in adults with overweight or obesity and without diabetes ^[4]. In the SURMOUNT-5 head-to-head trial in 751 adults with obesity, semaglutide reached -13.7% at week 72 against tirzepatide's -20.2%, a statistically significant difference (P<0.001) ^[1].

Cardiovascular outcomes. SELECT enrolled 17,604 adults with preexisting cardiovascular disease and overweight or obesity but without diabetes; once-weekly semaglutide 2.4 mg reduced major adverse cardiovascular events against placebo (HR 0.80; 95% CI 0.72-0.90; P<0.001) ^[3]. Earlier, SUSTAIN-6 tested once-weekly semaglutide at 0.5 or 1.0 mg in type 2 diabetes at high cardiovascular risk and reduced the composite of cardiovascular death, nonfatal myocardial infarction or nonfatal stroke (HR 0.74; 95% CI 0.58-0.95) — while rates of diabetic-retinopathy complications were higher in that trial (HR 1.76; 95% CI 1.11-2.78) ^[7].

Kidney outcomes. FLOW enrolled 3,533 participants with type 2 diabetes and chronic kidney disease and found that once-weekly semaglutide 1.0 mg reduced major kidney-disease events — kidney failure, a decline in eGFR of 50% or more, or kidney or cardiovascular death — against placebo (HR 0.76; 95% CI 0.66-0.88) ^[2].

Safety synthesis. A dedicated safety review concluded that semaglutide carries an overall favourable risk-benefit profile in type 2 diabetes, with mostly mild-to-moderate transient gastrointestinal effects — nausea in roughly one-third of patients — an increased risk of biliary disease, and pancreatic and thyroid-cancer signals on which definitive conclusions cannot yet be drawn given low incidence ^[5].

One structural feature runs through all of it. Every result above was accrued under continuing treatment. The cardiovascular and kidney benefits are on-treatment findings in exactly the way the weight figure is, and the trials did not follow participants through a period off drug to see whether the curves held.

Reported effects, cautions & safety

The first block below is anecdotal, not clinical evidence. It comes from patient reviews and community discussion, is unverified and unadjudicated, and carries no doses. It is recorded because it describes texture that trial tables do not.

Frequently reported as benefits. The description that recurs most is that the constant background chatter about food goes quiet; reporters describe filling up faster and no longer circling back to the next meal. Reduced cravings for sugar and rich food follow, sometimes with those foods becoming actively off-putting. Weight loss itself is described as steady, with the pace slowing after an early period. Commonly reported: improved blood-sugar readings among people using it for type 2 diabetes. Occasionally reported: a fading interest in alcohol.

Frequently reported as adverse. Nausea leads, sometimes escalating to vomiting, peaking early and again after escalation before easing. Commonly reported: sulfur-smelling burps, disrupted bowels in both directions, and tiredness after administration. Occasionally reported: reflux, food aversions, altered taste, smell sensitivity, headaches and light-headedness. Sometimes reported: hair shedding, facial gauntness after rapid weight loss, and minor injection-site redness or itching.

The documented cautions come from trials, labelling and pharmacovigilance.

• Gastrointestinal intolerance. Nausea, vomiting, diarrhoea and constipation dominate the adverse-effect profile and are the leading cause of discontinuation. The safety review found these effects mostly mild to moderate and transient, with nausea affecting roughly one-third of patients ^[5].
• Thyroid and pancreatic warnings. Thyroid C-cell tumours in rodents produced a boxed warning, while a clear human thyroid-cancer signal has not been established. Acute pancreatitis remains a class warning; the safety review describes pancreatic and thyroid-cancer signals as unresolved because incidence is low ^[5].
• Gallbladder and biliary disease. Cholelithiasis is increased against placebo, attributed largely to the rate and magnitude of weight loss rather than direct toxicity ^[5].
• Pre-existing diabetic retinopathy. Complications were more frequent in SUSTAIN-6 among participants with pre-existing disease undergoing rapid glycaemic correction (HR 1.76) ^[7].
• Lean-mass loss. Body-composition substudies show that weight loss includes lean tissue as well as fat, raising questions about sarcopenia risk.
• Weight regain after discontinuation. The direct withdrawal record reports substantial regain and movement of cardiometabolic improvements back toward baseline.
• Pregnancy and compounded material. Approved labelling treats pregnancy as a contraindication. Compounded or non-pharmaceutical material falls outside the approved-product evidence base and carries separate quality and safety concerns.

Where it sits in the discontinuation record

Semaglutide is the compound on this desk that has been studied from both ends, and that is what makes it the reference case for the whole theme.

The randomised withdrawal. A dedicated withdrawal study gave every participant active treatment during a run-in and only then randomised them, some to continue and some to switch to placebo. Because both groups arrived at randomisation having already lost weight, the design answered a narrower question than a standard placebo comparison: not whether the compound works, but what continuation contributes. The switched group regained weight.

The off-treatment extension. A separate extension followed participants after drug and lifestyle intervention were withdrawn. Participants regained a substantial share of the weight they had lost, and cardiometabolic improvements reverted toward baseline alongside it. The second half of that observation matters: the blood-pressure, lipid and glycaemic movement was not banked separately from the weight.

Why the shape is unsurprising. The mechanism is an occupancy mechanism. The compound suppresses appetite while it is present in the feeding circuits it reaches ^[6]. The albumin-bound design that supports prolonged administration also means exposure fades gradually rather than ending at the final injection. The direct studies report the outcome; mechanism helps explain its shape without proving that every participant follows the same course.

What that implies about the evidence, and what it does not. It implies that a -14.9% figure ^[4] describes an on-treatment state. By the same logic, SELECT ^[3] and FLOW ^[2] describe risk reduction accrued while treatment continued; those papers do not establish that the same curves persist after treatment ends.

What it does not imply is anything about what any particular person should do. This desk does not advise starting, continuing, tapering or stopping any medicine, and describes discontinuation only as an experimental condition that investigators deliberately created in order to measure it. A decision about a prescription belongs with the clinician who wrote it.