Effect of Incretin-Based and Nonpharmacologic Weight Loss on Body Composition: A Systematic Review
Batsis JA, Gavras A, Gross DC, Cheever CR, Da Silva BR, Meira Filho LF
Annals of Internal Medicine (2026)
Annals of Internal Medicine systematic review of 36 randomised controlled trials of liraglutide, semaglutide, tirzepatide, and dulaglutide in adults with obesity, evaluating fat mass, fat-free mass, lean soft tissue, and visceral adiposity outcomes against prespecified benchmarks — the most rigorous synthesis to date of the lean-mass concern that has dominated public discourse about incretin therapy.
This Annals of Internal Medicine systematic review by John Batsis's group at University of North Carolina (UNC Chapel Hill geriatric medicine and nutrition) with collaborators at Verona and Alberta synthesises the randomised controlled trial evidence on body composition changes during incretin-based weight-loss therapy. The work addresses one of the most-discussed clinical concerns about GLP-1 receptor agonists and dual GIP/GLP-1 agonists — that the substantial total weight loss they induce includes disproportionate losses in fat-free mass and skeletal muscle, with implications for sarcopenia, functional status, and weight-loss durability.
The search covered Scopus, Embase, PubMed, CINAHL, PsycINFO, and ClinicalTrials.gov from January 2003 to February 2026. Eligible studies were English-language randomised controlled trials of liraglutide, semaglutide, tirzepatide, or dulaglutide in adults (≥18 years) reporting body composition outcomes. Primary outcomes were changes in fat mass, fat-free mass (FFM), lean soft tissue (LST), muscle-related indices, and visceral adiposity measured by bioelectrical impedance analysis (BIA), dual-energy X-ray absorptiometry (DXA), computed tomography (CT), or magnetic resonance imaging (MRI).
The authors prespecified benchmarks for contextualising expected muscle-related losses: approximately 25% of total weight loss expected as FFM or LST loss when measured by BIA or DXA, and approximately 15% as skeletal muscle when measured by CT or MRI. These benchmarks are drawn from the broader weight-loss literature and provide the reference for what "disproportionate" lean-mass loss would look like.
Among 8,102 titles and abstracts screened, 36 primary studies met inclusion criteria — median duration 26 weeks, median 71 participants per study. 41.7% were rated at low risk of bias; only 10 studies (27.8%) prespecified body composition as a primary outcome. Mean participant age ranged from 20 to 63.7 years; mean BMI 27.9 to 41.6 kg/m². Weight loss was consistently larger in the incretin arms than in placebo or lifestyle comparators.
This is abstract-only extraction — the headline result on the proportional fat-free-mass vs fat-mass loss split is not visible in the abstract excerpt available in the queue, only the methods. The 25%-FFM and 15%-skeletal-muscle benchmarks are the reference, but whether the included incretin trials cluster above, at, or below those benchmarks is the load-bearing result and must be read from the full paper.
Methodological caveats are real: median trial duration of 26 weeks is short relative to the multi-year weight-maintenance question; median per-trial sample size of 71 is small for body-composition primary endpoints; only 27.8% of included trials prespecified body composition as a primary outcome, meaning most of the synthesised data is from secondary or exploratory analyses with the associated multiple-testing and reporting-bias concerns. 41.7% low-risk-of-bias rating means the majority of underlying trials carry meaningful methodological limitations.
DXA, BIA, CT, and MRI measure different physical compartments and are not directly interchangeable — BIA is least reliable, DXA is the field standard for clinical research, and CT/MRI are most accurate but logistically constrained. The mixing of measurement methods across the included trials introduces between-study heterogeneity that the prespecified benchmarks partially address but do not eliminate.
Tirzepatide trial data are heavily weighted toward the SURMOUNT family, which uses DXA in subset analyses. Semaglutide data are weighted toward the STEP family. Liraglutide and dulaglutide trial body-composition data are sparser. The review aggregates across compounds; per-compound comparison is limited by the available data.
07·Member discussion
No member discussion yet.
Member-only conversation lives here — cycle notes, practitioner commentary, pattern-matching. Be the first paying member to start the thread.