Semaglutide: Deep Outcome Evidence, Shallow Composition Evidence

The short version

Semaglutide (Ozempic, Wegovy) copies a natural gut hormone called GLP-1, which the body releases after eating. Chemical changes make the engineered version persist much longer than the natural hormone, which is why the cited trials used once-weekly administration ^[10][11].

Most of its weight effect happens in the brain. It reaches the appetite centres and turns down the drive to eat. The published description of its mechanism adds an important detail: it lowers how much food is eaten without lowering how much energy the body burns.

Its trial record is the deepest here. In STEP 1, average weight change at 68 weeks was -14.9% against -2.4% on placebo ^[10]. Separate trials of 17,604 and 3,533 people measured heart and kidney outcomes rather than weight ^[9][8].

What none of those headline papers reports is how much of the lost weight was fat and how much was muscle. That question was handled in smaller substudies, and the answer is less flattering than the weight number.

What it is

Semaglutide is a 31-amino-acid acylated analogue of human glucagon-like peptide-1, sharing roughly 94% sequence homology with the native hormone. Two backbone substitutions give it protease resistance: alanine at position 8 is replaced by alpha-aminoisobutyric acid, which blocks cleavage by dipeptidyl peptidase-4, and lysine at position 34 is replaced by arginine. The remaining lysine carries a C18 fatty di-acid side chain that binds albumin, extending circulation sufficiently for the once-weekly trial schedules ^[11].

Regulatory breadth is the distinguishing feature. Approvals span type 2 diabetes, chronic weight management, reduction of major adverse cardiovascular events in adults with established cardiovascular disease and overweight or obesity, and metabolic dysfunction-associated steatohepatitis. Both a once-weekly subcutaneous injection and a once-daily oral tablet exist; the oral form is co-formulated with the absorption enhancer SNAC and has very low oral bioavailability, which is why its administration conditions are unusually strict ^[11].

Material from outside the approved-product channel sits outside the evidence base summarised on this page. That is a statement about what the trials studied, not a sourcing recommendation, because this site makes none.

How it works

The receptor is the same GLP-1 receptor the native hormone uses, but the drug reaches tissues the native hormone barely does, because it survives long enough to get there ^[11].

In the pancreas it potentiates glucose-dependent insulin secretion from beta cells and suppresses inappropriate glucagon release from alpha cells. Glucose dependence is the safety-relevant property: the insulin signal scales with circulating glucose, which is why the class behaves unlike exogenous insulin.

In the gut it slows gastric emptying, which contributes to post-meal glucose control, to fullness, and to most of the gastrointestinal adverse effects that dominate its tolerability profile ^[11].

In the brain, where the weight effect actually lives, it reaches hypothalamic and brainstem appetite circuits — the arcuate nucleus and the area postrema in particular — activating anorexigenic POMC and CART neurons and inhibiting orexigenic NPY and AgRP neurons. Food intake falls and food preference shifts.

The published mechanism specifies that this occurs without lowering energy expenditure. For the muscle-sparing question that sentence cuts both ways. Expenditure not falling is favourable, because a preserved metabolic rate is part of what protects lean tissue during weight loss. But nothing in the described mechanism acts on skeletal muscle, on protein synthesis, or on the signalling that would bias tissue loss toward fat. This is an intake-side drug, and an intake-side drug produces the tissue split that a large energy deficit produces, modified by whatever the person eats and does while taking it.

What the research shows

STEP 1. In adults with overweight or obesity and without diabetes, once-weekly subcutaneous semaglutide 2.4 mg produced a mean body-weight change of -14.9% from baseline to week 68, against -2.4% on placebo ^[10]. This is the weight-management headline, and the endpoint is body weight.

SELECT. In 17,604 adults with pre-existing cardiovascular disease and overweight or obesity but without diabetes, once-weekly semaglutide 2.4 mg reduced major adverse cardiovascular events against placebo, with a hazard ratio of 0.80 and a 95% confidence interval of 0.72 to 0.90 ^[9]. The endpoint is an event count, not a tissue measurement.

FLOW. In 3,533 people with type 2 diabetes and chronic kidney disease, once-weekly semaglutide 1.0 mg reduced major kidney-disease events — kidney failure, a fall in estimated glomerular filtration rate of 50% or more, or death from kidney or cardiovascular causes — with a hazard ratio of 0.76 and a 95% confidence interval of 0.66 to 0.88 ^[8]. Again an event count.

SURMOUNT-5, from the other side. In the head-to-head against tirzepatide in 751 adults with obesity, semaglutide produced a mean weight change of -13.7% at 72 weeks against -20.2% for the comparator ^[1]. Reported as body weight.

Safety review. A dedicated review concludes the overall risk-benefit profile is favourable 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 including cholelithiasis, and pancreatic and thyroid-cancer signals for which definitive conclusions cannot yet be drawn because of low incidence ^[11].

That is five sources and five different endpoint families: percent body weight, cardiovascular events, kidney events, comparative body weight, adverse-event incidence. Fat mass and lean mass are in none of them.

Reported effects, cautions and safety

The first paragraph here is anecdotal, not clinical evidence — a summary of what people describe in patient reviews and research communities, carrying no quantities, schedules or protocols. The dominant reported benefit is that food stops occupying mental foreground: reviewers describe eating a fraction of former portions and, for the first time, not thinking about the next meal. Sugar and sweet cravings are widely described as dropping sharply, with fried and high-fat foods becoming actively unappealing. Weight loss itself is reported by the overwhelming majority, usually described as steady over months and slowing after the early period, and attributed by most reporters to eating far less rather than to any change in activity. People using it for type 2 diabetes commonly describe markedly better glucose and long-term average readings. A recurring secondary observation is a fading interest in alcohol. On the burden side, nausea is the single most reported effect, sometimes escalating to vomiting and flaring after large or fatty meals; sulfurous burping, bloating and a sense of food sitting too long are distinctive complaints; bowel habits swing between constipation and diarrhoea; reflux and heartburn track with dose increases; fatigue is common after an injection and early in treatment; taste changes, smell sensitivity and over-suppressed appetite appear in a minority, occasionally to the point of struggling to eat enough; and hair shedding along with a hollower, gaunter face is reported later and attributed by reporters to the speed of the weight loss rather than to the drug. Headaches, lightheadedness and mild injection-site reactions round out the list. All of it is self-report, unverified, and not a clinical finding.

The documented cautions read differently. Gastrointestinal intolerance during dose escalation is a clinical-trial finding, predominantly mild to moderate and transient, concentrated around titration, and the leading cause of discontinuation ^[11]. A boxed warning for thyroid C-cell tumours derives from rodent studies at supratherapeutic exposures; a dedicated assessment concluded that available human data do not establish a clear increase in medullary or other thyroid cancer, so the signal is unconfirmed in humans, while a personal or family history of medullary thyroid carcinoma or multiple endocrine neoplasia type 2 remains a contraindication on the strength of the rodent finding ^[11]. Acute pancreatitis is a recognised class warning and treatment is conventionally stopped if pancreatitis is suspected ^[11]. Increased biliary disease, including cholelithiasis, is a real trial and pharmacovigilance finding, attributed largely to the rate and magnitude of the weight loss rather than to direct drug toxicity ^[11]. In SUSTAIN-6, diabetic-retinopathy complications were significantly more frequent, with a hazard ratio of 1.76 and a 95% confidence interval of 1.11 to 2.78, concentrated among people with pre-existing retinopathy undergoing rapid correction of glycated haemoglobin; the leading interpretation is early worsening driven by the speed of glycaemic correction rather than retinal toxicity ^[11].

Further documented cautions in the compound record include pregnancy, weight regain after discontinuation, and the lean-mass concern discussed below. Their supporting publications sit outside this site's numbered source set, so this review does not reproduce their numerical estimates.

Where it fits in the muscle-sparing question

Semaglutide is the compound on this desk where the mismatch between evidence depth and evidence type is starkest. Two of its trials enrolled more people than the other three compounds' entire cited literature combined, and both measured clinical events ^[9][8]. The weight trial that anchors its reputation measured body weight ^[10]. The head-to-head measured body weight ^[1]. Nothing in that list separates fat from lean tissue.

The compound record does not leave the question blank, but what it supplies is qualitative rather than quantified. A body-composition substudy run within the STEP programme, using DXA, reported that the weight lost with semaglutide comprised both fat mass and a meaningful proportion of lean mass. The record frames the downstream sarcopenia concern — particularly in older adults, for whom lean tissue is already declining — as an extrapolated, mechanistically reasoned worry rather than a measured outcome, and notes that this is what motivated research into protein intake and resistance training as preservation strategies. The lean-mass loss is described as an observed clinical-trial finding; the functional consequence of it is not.

Two things follow, and they are easy to conflate and should not be.

First, an intake-side mechanism with no muscle-directed signal is not expected to spare muscle, and nothing in the published mechanism claims that it does. The favourable half of the mechanistic story — reduced intake without reduced energy expenditure — is about metabolic rate, not about tissue partitioning.

Second, the absence of composition data in the headline trials is not evidence of a good result or of a bad one. It is an absence. The most defensible statement this review can make about semaglutide and lean mass is that a DXA substudy found a meaningful lean-mass component to the loss, that the magnitude of that component is not fixed by the sources numbered here, and that any confident percentage a reader encounters is coming from somewhere other than STEP 1, SELECT, FLOW or SURMOUNT-5 ^[10].

The tirzepatide page explains how the same evidence gap appears for the dual agonist; the side-by-side comparison places all four compounds against the same measurement question.