GLP-1 Weight Loss: How Much Is Lean Mass?

Weight loss on GLP-1 drugs does not automatically mean muscle wasting. But the body-composition numbers deserve attention: a 2026 meta-analysis found that semaglutide users lost 35.2% of their lost weight as DXA-measured lean mass, versus 26.2% with lifestyle intervention alone. Tirzepatide came in at 25.4%. The distinction matters, but so does what DXA actually measures.

The practical question is not whether every kilogram of "lean mass" on a scan was muscle fiber. It was not. The better question is whether a weight-loss plan protects strength and functional capacity while fat comes off. The best available answer still starts with structured resistance training.

The headline numbers: semaglutide differs; the drug class does not

Eisa et al.'s 2026 systematic review and meta-analysis pooled 20 randomized controlled trials involving 15,782 participants. It calculated lean-mass loss as a proportion of total weight loss. Semaglutide averaged 35.2% (95% CI, 31.5% to 38.9%); tirzepatide, 25.4% (22.8% to 28.0%); liraglutide, 26.8% (23.1% to 30.5%); and lifestyle-only interventions, 26.2% (24.1% to 28.3%).

The overall comparison between incretin-based pharmacotherapy and lifestyle intervention was not statistically significant (p=0.42). In other words, the pooled trial evidence does not support the claim that GLP-1-based treatment as a class causes a disproportionately larger share of lean-mass loss than intensive dieting.

Semaglutide is the notable within-class signal. Its confidence interval did not overlap with the lifestyle estimate, while tirzepatide and liraglutide tracked much closer to lifestyle. That finding warrants scrutiny, not a blanket verdict about every incretin drug. The meta-analysis also reported moderate-to-high heterogeneity across pooled estimates (I²=68%), so these percentages are useful approximations rather than a personal forecast.

What the trial subsets found

Smaller DXA subsets from major drug trials point in the same direction. In the STEP 1 semaglutide subset, 95 participants lost 6.9 kg of lean soft tissue and 10.4 kg of fat mass—about 40% of the weight lost as lean soft tissue. In the SURMOUNT-1 tirzepatide subset, 124 participants lost 5.6 kg of lean soft tissue and 15.9 kg of fat mass, or about 26% of weight loss as lean soft tissue. Tinsley et al.'s 2025 case series discusses both analyses and their limits.

These subsets were small and were not prespecified primary trial endpoints. They can support a signal; they cannot settle why the drugs differ. Greater total fat loss may lower the percentage of weight loss labeled lean mass simply because fat makes up more of the total loss. Dubin et al.'s 2024 review places these results in that physiological context: lean-mass change generally tracks the amount of total weight lost and a person's starting body composition. A lower lean-mass fraction is not proof that one drug directly protects muscle tissue.

"Lean mass" on DXA is not a synonym for muscle

This is the caveat that gets lost when a scan result becomes a social-media warning label. DXA divides body mass into fat mass, bone mineral content, and lean soft tissue. Lean soft tissue includes skeletal muscle, but it also includes organs, connective tissue, intracellular fluid, glycogen, and the water stored with glycogen. Each gram of glycogen carries roughly 3 grams of water.

Calorie restriction can reduce glycogen stores and associated water, and DXA will count that change as lean-mass loss. Changes in organ perfusion mass can contribute as well. Therefore, 35.2% does not mean that 35.2% of semaglutide-related weight loss was contractile muscle. MRI or CT can assess skeletal muscle more directly, yet none of the trials in the meta-analysis used either as a primary endpoint. That measurement gap applies to diet arms and medication arms alike.

DXA is still useful. It can track meaningful changes in body composition, and lean soft tissue is more informative than body weight alone. It just cannot answer the muscle-fiber question by itself. A scale cannot answer it at all; it is busy being a scale.

Why the concern remains real for older adults

Measurement uncertainty does not make function irrelevant. A 2026 review by Prokopidis et al. raised a separate concern: in older adults taking GLP-1 receptor agonists, muscle strength may decline more than DXA-measured mass changes would predict. The review does not establish that GLP-1 therapy causes a specific strength deficit, but it identifies an outcome that the mass-based meta-analysis did not measure.

That distinction matters most for people with low baseline muscle, frailty, previous falls, limited mobility, or sarcopenia risk. González-Luis et al.'s 2025 review makes the same clinical point: a proportion of lean-mass loss that looks typical at the population level may still matter greatly for an individual near a functional threshold. Losing a small amount of reserve is different when the reserve was already small.

Real-world outcomes add more uncertainty. A prospective observational study of tirzepatide in routine care measured short-term changes in fat mass, fat-free mass, total body water, and extracellular water, while a retrospective study of compounded semaglutide reported variable body-composition outcomes in a commercial wellness program. These studies are less controlled than randomized trials, so they cannot establish drug effects with the same confidence. They do show that food intake, adherence, and exercise behavior can vary widely outside a trial. See Adamidis et al. and Chun et al..

Resistance training is the strongest available countermeasure

In Eisa et al.'s meta-analysis, lifestyle interventions that included structured resistance training had 17.5% of total weight loss classified as lean mass (95% CI, 14.2% to 20.8%). That is lower than lifestyle intervention alone at 26.2% and lower than semaglutide's pooled 35.2% estimate.

It is an encouraging result with an important boundary: those resistance-training data came from lifestyle-plus-training arms, not completed randomized trials that paired resistance training with semaglutide or tirzepatide. Locatelli et al.'s 2024 review identifies resistance exercise as the primary available strategy to reduce lean-mass loss during incretin-based weight loss, while also noting the need for direct combination trials.

Protein is a sensible partner, though the GLP-1-specific evidence is thinner. In older adults generally, Kirwan et al.'s meta-analysis of randomized trials found that protein supplementation augmented resistance exercise effects on appendicular lean mass and handgrip strength. Those were not trials of GLP-1 users, and controlled trials have not established a specific grams-per-kilogram protein target during GLP-1 therapy. Treat precise online protein prescriptions accordingly.

The verdict

Semaglutide's 35.2% pooled lean-mass fraction is a legitimate signal, and tirzepatide's 25.4% estimate sits closer to the lifestyle benchmark. But the class-level comparison with lifestyle was non-significant, the evidence shows substantial variation among trials, and DXA cannot tell us how much of the measured loss was skeletal muscle rather than glycogen, water, and other lean tissue.

For people losing weight with medication or diet, resistance training has the strongest evidence for improving the body-composition profile of weight loss. Older adults and anyone at risk of sarcopenia should also pay attention to strength and daily function, not merely body weight or a single DXA category. The current evidence supports a cautious conclusion: protect muscle while you lose fat, and measure outcomes that matter when you stand up, climb stairs, and carry groceries.