GLP-1 Muscle Loss: What the Trials Actually Show on Lean Mass

The claim, and why it needs a comparison group.

"GLP-1 drugs make you lose muscle" is now one of the most repeated claims in metabolic health, and the number attached to it is usually somewhere between 25% and 40% of total weight lost. That number is not fabricated. It appears in the published body composition literature, and it is worth understanding precisely.

But a percentage on its own is not a finding. It is half of one. The question that determines whether this is a drug problem or a weight-loss problem is: what proportion of weight lost is lean mass when someone loses the same amount of weight by other means? Without that comparison, the number is a fact floating in a vacuum, and a vacuum is exactly where alarming claims thrive.

This matters commercially as well as scientifically, because the muscle-loss concern has become the primary marketing wedge for selling additional peptides alongside GLP-1 drugs. The argument runs: these drugs cost you muscle, therefore you need this compound to protect it. The first clause is roughly true. The second does not follow from it, and the evidence for what actually preserves lean mass points somewhere much less exotic.

This article covers what the body composition literature reports, how it compares to non-pharmacologic weight loss, what the combined exercise and drug trials found, and why the evidence-backed answer is resistance training and adequate protein rather than an additional compound. It contains no doses or protocols.

What the body composition literature actually reports.

Two recent systematic reviews put hard numbers on this, and they are the right place to start because single trials in this area are small and heterogeneous.

A 2026 systematic review in Annals of Internal Medicine examined body composition changes with incretin therapies in adults with obesity (Batsis et al., PMID 41996180). It screened over 8,000 records and included 35 primary studies with a median duration of 26 weeks and a median of 78 participants. Within the incretin groups, the median proportion of total weight loss attributable to reductions in muscle-based indices was 28.3%, with an interquartile range of 15.9% to 39.9%. Among studies using bioelectrical impedance or DXA the median was about 29%, and among studies using CT or MRI it was about 25.3%. The review also noted that no included study reported objective physical function outcomes, which is a significant gap: nobody in that literature measured whether people could actually do less.

A 2026 meta-analysis in Diabetes, Obesity and Metabolism pooled 20 randomized controlled trials comprising 15,782 participants using DXA or MRI (Eisa and Barood, PMID 41877354). It found lean mass constituted 25% to 39% of total weight lost with incretin agonists, broken out as roughly 35.2% for semaglutide, 25.4% for tirzepatide, and 26.8% for liraglutide.

A third meta-analysis in the International Journal of Obesity, restricted to GLP-1 receptor agonists at obesity doses versus placebo, found an absolute lean mass reduction of about 1.74 kg alongside an increase in lean mass as a proportion of total body weight of about 1.81 percentage points, with high heterogeneity (Laverde et al., PMID 42321502). That combination is not a contradiction. It is the arithmetic of losing much more fat than lean tissue: absolute lean mass falls while the fraction of your body that is lean rises.

So the headline claim survives scrutiny in its factual form. Roughly a quarter to a third of the weight lost on these drugs is lean tissue. The next question is what that number means.

The comparison that changes the interpretation.

Both of the systematic reviews above did the thing that makes this evidence interpretable: they compared against non-pharmacologic weight loss.

The Diabetes, Obesity and Metabolism meta-analysis found that lifestyle interventions produced a proportional lean mass loss of 26.2%, statistically indistinguishable from the incretin groups. The comparison p-value was 0.42, which is about as clear a null result as this literature offers. Their conclusion was explicit: lean mass loss during significant weight reduction is substantial, and the proportion of weight lost as lean mass is broadly comparable between incretin-based pharmacotherapy and lifestyle intervention.

The Annals of Internal Medicine review reached a compatible conclusion from a different angle. It applied prespecified benchmarks for expected muscle-related loss, about 25% of total weight loss for fat-free mass measured by BIA or DXA, and about 15% for skeletal muscle measured by CT or MRI. Those benchmarks were exceeded in about two thirds of incretin-based interventions. They were also exceeded in nearly half of the nonpharmacologic interventions that produced weight loss. The review's framing is that this is a weight-loss phenomenon in which incretin therapies sit at the higher end, not a category of harm unique to the drugs.

This is entirely consistent with decades of obesity physiology that predates GLP-1 drugs. When you lose substantial weight through energy deficit, you lose fat and you lose fat-free mass. Fat-free mass includes skeletal muscle, but also the water it carries, glycogen and its associated water, connective tissue, and the mass of organs that shrink modestly during sustained energy restriction. A DXA scan cannot distinguish between those components, which is one reason the CT and MRI estimates of skeletal muscle specifically tend to run lower than the DXA-based fat-free mass estimates.

There is a legitimate counterargument worth stating fairly: these drugs produce far more total weight loss than lifestyle intervention does, so a similar percentage of a much bigger number is a bigger absolute loss of lean tissue. That is true and it is not trivial, particularly in older adults who start with less muscle to spare. But it is a different claim from the one usually made. "This drug is uniquely catabolic" is not supported. "Losing a lot of weight quickly costs lean tissue, and these drugs make losing a lot of weight much easier" is supported, and it points toward a different intervention.

What the exercise plus GLP-1 trials found.

The most informative evidence in this whole discussion comes from trials that added structured exercise to the drug, because those trials tested the actual proposed solution rather than describing the problem.

The cleanest example is a randomized, head-to-head, placebo-controlled trial published in the New England Journal of Medicine in 2021 (Lundgren et al., PMID 33951361). After an 8-week low-calorie diet during which 195 participants lost a mean of 13.1 kg, participants were randomized for one year to one of four strategies: a moderate-to-vigorous exercise program plus placebo, liraglutide plus usual activity, exercise plus liraglutide, or placebo plus usual activity.

All three active strategies beat placebo on weight. The important finding is what happened to body composition. The combination strategy decreased body fat percentage by 3.9 percentage points, approximately twice the decrease seen in the exercise-only group (1.7 points) and the drug-only group (1.9 points). Only the combination strategy improved glycated hemoglobin, insulin sensitivity, and cardiorespiratory fitness. In other words, adding exercise to the drug did not merely preserve something. It changed the composition of the weight lost and produced metabolic improvements that neither intervention delivered on its own.

The meta-analytic evidence points the same direction. The Diabetes, Obesity and Metabolism analysis found that lifestyle intervention combined with resistance training produced the most favourable profile of any arm examined, with lean mass accounting for about 17.5% of total weight lost, compared with roughly 25% to 39% for the incretin agonists and 26.2% for lifestyle intervention alone. That is the single most useful number in this entire literature, because it is the only one attached to an intervention that measurably changes the outcome.

And the meta-analysis restricted to GLP-1 receptor agonists reached the same practical conclusion, stating that lean mass loss should not be considered a limitation on the use of these drugs in patients with obesity, while emphasizing that drug treatment should be accompanied by nutritional and physical exercise interventions to preserve or improve muscle mass.

Why adding a peptide is not the evidence-backed answer.

The commercial response to the muscle-loss discussion has been to sell additional compounds alongside GLP-1 drugs, typically growth hormone secretagogues, GHRH analogs, or research peptides marketed for body composition. It is worth being precise about why that does not follow from the evidence.

First, the evidence base is asymmetric. Resistance training and adequate protein intake during energy restriction have been studied for decades across many populations with consistent directional findings, and the combined-intervention trials above tested them directly in this exact context. The peptides being marketed for muscle preservation during GLP-1 therapy have not been tested in that context at all. There is no randomized trial in which people losing weight on a GLP-1 drug were assigned to a research peptide versus placebo with body composition as an endpoint. The proposed solution is not under-evidenced; it is unevidenced for this use.

Second, the reasoning is mechanistic rather than empirical, which is the recurring failure mode in this field. "This compound raises IGF-1, IGF-1 is anabolic, therefore it will preserve muscle during a caloric deficit" is a plausible chain with no measured endpoint at the end of it. Plausible chains fail routinely.

Third, the intervention with the strongest evidence is also the one that produces the outcomes people actually care about. The concern underneath "muscle loss" is usually about strength, function, and metabolic health. Resistance training improves those directly and measurably. Notably, the Annals of Internal Medicine review found that no study in the incretin body composition literature reported objective physical function outcomes at all, which means the entire debate has been conducted on imaging proxies rather than on whether anyone became weaker.

Fourth, adding an unapproved compound to an approved one introduces the full set of research-peptide risks: no verified identity or purity, no human safety data for the combination, and no monitoring framework. That is a real cost being incurred to address a problem for which a well-supported alternative exists.

None of this means lean mass during weight loss is unimportant. It means the evidence points squarely at resistance training and adequate protein, both of which are supported by trial data in this exact scenario, and away from adding compounds that have never been tested for it.

The honest summary.

Strip away both the alarmism and the dismissal and the picture is fairly clear.

Lean mass loss on GLP-1 and dual-agonist drugs is real, and the commonly quoted range of roughly 25% to 35% of total weight lost is supported by systematic reviews and meta-analyses using DXA, MRI, and CT. That is not a fringe claim and it should not be waved away.

It is also not specific to these drugs. Lifestyle-based weight loss producing comparable proportional lean mass loss is the direct comparison, and the meta-analytic test of that comparison was null. Losing weight costs lean tissue by whatever route you lose it.

The absolute magnitude is larger because the total weight loss is larger, and that is a legitimate consideration, especially in older adults and anyone starting with low muscle mass. It is an argument for pairing treatment with resistance training, not an argument against treatment.

What measurably changes the outcome, in the trials that actually tested it, is exercise. Adding structured exercise to the drug roughly doubled the reduction in body fat percentage compared with either alone, and it was the only strategy that improved glycemic markers, insulin sensitivity, and cardiorespiratory fitness. Lifestyle intervention plus resistance training produced the lowest proportional lean mass loss of any arm in the pooled analysis.

And the evidence gap is worth naming one more time: nobody in this literature has yet measured objective physical function. We are inferring consequences from imaging. That is a reason for humility in both directions, and a reason to be especially skeptical of anyone selling a solution to a problem whose functional significance has not been measured.