GLP-1 receptor agonists reduced lean mass, without measurable effects on bone or joints

Design

Systematic review and meta-analysis (46 RCTs, 13 real-world studies, 1 pharmacovigilance study)

Population

1,250,717 people treated with GLP-1 receptor agonists

Primary outcome

Bone, muscle and joint outcomes

Effect

Lean mass SMD ≈ −0.52 (k=28, I² 88%); no effect on BMD, fractures or WOMAC

This meta-analysis included 60 studies — 46 randomised trials, 13 real-world studies and one pharmacovigilance study — covering 1,250,717 people treated with GLP-1 receptor agonists, including semaglutide, liraglutide, exenatide, dulaglutide and tirzepatide.

Lean or fat-free mass fell consistently (standardised mean difference about −0.5, 28 studies), mainly with liraglutide and semaglutide against placebo, and the finding held in every sensitivity analysis. There was no effect on bone mineral density or fractures in the most adjusted models, and no change in osteoarthritis pain, function or stiffness. Certainty was low, heterogeneity high (I² 88%), and the lean mass loss appeared largely proportional to weight lost.

Whether the lean mass loss matters for strength or function is unknown. It is reasonable to pair GLP-1 treatment with resistance exercise and adequate protein, particularly in older or sarcopenic patients, while recognising that this advice is precautionary rather than proven.

• Advise resistance exercise and adequate protein for patients starting GLP-1 receptor agonists, especially older adults.
• Lean mass loss was consistent but largely tracked total weight loss.
• No effect on bone density or fractures was seen in the most adjusted models.
• Consider assessing strength or function in frail patients before and during treatment.
• Whether lean mass loss impairs function remains unproven.

Why it matters

Concern about muscle loss on GLP-1 drugs is common, and this quantifies it while showing bone and joints were unaffected.

Don't overread it

Certainty was low, and lean mass is not the same as muscle strength or function, which were not well measured.

The statistics, in plain English

A standardised mean difference of about 0.5 is a moderate effect across studies that used different measurement methods. I² of 88% means the size of the effect varied a lot between studies, even though its direction was consistent.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.