In studies that measure lean mass from baseline, the number describes the total change within a group. In the three-trial analysis below, 0.78 kg describes the difference in change between the treatment and comparison groups.
What this analysis did
A search of five databases to the end of 2024 turned up 2,776 articles, of which three randomized trials covering 171 participants met the criteria: adults with obesity, without type 2 diabetes, on a GLP-1, with body composition reported. [1] On average, treatment groups lost 0.78 kg more lean mass and 3.43 kg more fat mass than comparison groups. The paper gives a slightly different lower confidence limit for lean mass in the abstract and results section. Heterogeneity was substantial for both outcomes, with I² of 63.2% for lean mass and 78.2% for fat mass.
What the split looks like
Taking the two point estimates at face value, the extra weight lost on the drug was about 18.5% lean and the rest fat. That division is our arithmetic rather than the paper’s — it uses only the central estimates, ignores both confidence intervals, and ignores the substantial heterogeneity. Treat it as a shape, not a measurement.
Our summary of lean mass as a share of total weight lost covers studies with different comparison groups and follow-up periods. The roughly one-third share discussed there cannot be compared directly with a ratio of two treatment-versus-control estimates.
Why the evidence base is this small
Three trials met the eligibility criteria, which required randomization, body-composition measurements and participants without type 2 diabetes. With so few studies and substantial differences between estimates, the pooled numbers may change as more trials become available.
What the findings can tell you
The authors discuss physical activity and nutrition alongside treatment. Our review of lean mass and resistance exercise covers a separate trial designed to test exercise during tirzepatide treatment. The three-trial meta-analysis did not establish a training plan or protein target.