Body Composition on a GLP-1: Why the Scale Only Tells You Half the Story

A bathroom scale can tell you that you lost twelve pounds. It cannot tell you what those twelve pounds were made of, and on a GLP-1 that distinction matters more than the number itself.

Every pound you lose comes out of some combination of fat, lean tissue, and water. The ratio isn't fixed, it's partly influenced by what you do while you're losing, and it's the single biggest reason two people can drop the same weight and end up in completely different metabolic shape. Here's what the DXA data from the major trials actually found, why a chunk of lean mass loss is normal in any kind of weight loss, and what we track at Defiant instead of just the scale.

What the Scale Can't See

Your scale measures the sum of everything: fat, muscle, bone, organs, glycogen, and the water attached to all of it. It reports one number and gives you no idea which parts moved.

That's a real problem in the first few weeks of a weight loss program, when the number tends to fall fast and a good portion of the early drop is water and glycogen rather than fat. It's an even bigger problem six months in, when the scale slows down and people assume nothing is happening. Body composition can keep improving during a stretch where scale weight barely moves, because you can be losing fat and holding or building lean tissue at the same time. On the scale that looks like a plateau. On a body composition scan it looks like progress.

The reverse is also true, and it's the scenario worth catching early. Someone losing weight quickly with very little protein, no resistance training, and a large calorie deficit can post excellent scale numbers while giving up more lean mass than they should. You cannot see that happening without measuring it.

What the Trial Data Actually Found

Two body composition substudies from the major GLP-1 trials give us the clearest picture. Both used DXA, which is the research standard for splitting body mass into fat, lean tissue, and bone.

Two things stand out. Fat mass fell substantially harder than lean mass in both, which is the outcome you want, and lean tissue ended up making up a larger share of total body weight than it did at baseline. That's a body composition improvement even though absolute lean mass went down.

The visceral fat number from STEP 1 deserves its own mention. Regional visceral fat mass fell 27.4%, more than the 19.3% drop in total fat mass.¹ Visceral fat is the deep abdominal fat packed around your organs, and it's the fat most closely tied to metabolic dysfunction and chronic inflammation. Losing it preferentially is arguably the most useful thing happening in that dataset, and it's completely invisible on a scale.

Why the Two Trials Landed on Different Numbers

The two substudies also reported different things. SURMOUNT-1 published an explicit fat-versus-lean split, about 75/25. STEP 1 emphasized how much total and visceral fat came off and reported that lean mass rose as a share of body weight, without publishing a clean percentage for lean mass lost. That difference is worth understanding rather than glossing over, because it tells you something about how to read this kind of data.

These were separate substudies inside separate trials, with different participants, different baseline body compositions, different scan timing, and different sample sizes. Both were exploratory analyses of a few hundred people pulled from trials that enrolled thousands. Substudies that small carry wide margins, and the STEP 1 body composition analysis in particular was presented as an exploratory conference analysis rather than a fully powered outcome.

So the honest read is a range, not a precise figure. The cleanest GLP-1 DXA data, from SURMOUNT-1, puts lean tissue at about a quarter of the weight lost, and broader weight-loss research suggests the figure can run higher depending on your starting composition, your protein intake, whether you're doing resistance training, and how fast you're losing.² What these trials do agree on is direction: fat mass falls considerably faster than lean mass, and body composition ratios improve.

We're not going to tell you one medication preserves muscle better than the other based on this. Comparing across two separately designed substudies isn't a valid head-to-head, and nobody has run the trial that would answer that question properly.

Lean Mass Loss Isn't a GLP-1 Problem

This is the part that gets lost in most of the online conversation about "GLP-1 muscle loss," and the SURMOUNT-1 substudy addresses it directly. Roughly 25% of the weight lost was lean mass in the tirzepatide group. It was also roughly 25% in the placebo group.²

Losing some lean tissue alongside fat is what happens in essentially any significant weight loss, whether it comes from a GLP-1, a calorie deficit, bariatric surgery, or illness. When your body sheds mass, some of that mass is muscle. That has been true since long before these medications existed.

What's different about GLP-1s is the pace and the appetite suppression. When food noise quiets down and eating becomes genuinely easy to under-do, protein intake is often the first thing to slip, and protein is the input your body needs most to hold onto lean tissue while losing weight. The medication isn't attacking your muscle. It's making it much easier to accidentally underfeed the tissue you're trying to keep, which is a fixable problem if someone is paying attention.

If you want the practical side of that, our guide on how to protect muscle while losing weight on a GLP-1 covers protein targets and resistance training. This post is about seeing what's happening. That one is about doing something with the information.

The Muscle Function Question

Lean mass on a scan is one thing. Whether you're actually getting weaker is another, and a 2026 study went looking specifically at that.

The SEMALEAN study followed 106 patients with obesity through 12 months on semaglutide 2.4 mg, measuring body composition by DXA alongside handgrip strength and resting energy expenditure.³ Fat mass fell 18% at 12 months. Lean mass dropped about 3 kg in the first 7 months and then stabilized rather than continuing to decline. Handgrip strength, which is a standard proxy for overall muscle function, improved by 4.5 kg over the year. The prevalence of sarcopenic obesity in the group, meaning low muscle mass combined with excess fat, fell from 49% at baseline to 33%.³

That's a single-center study in a population with a mean BMI of 46, so it doesn't generalize cleanly to everyone. Still, it's a useful counterweight to the assumption that losing lean mass automatically means losing function. In this cohort, people were carrying less tissue and squeezing harder at the end of it.

What We Measure at Defiant, and How Often

Every weight loss protocol at Defiant starts with a Styku 3D body composition scan as the baseline, and repeats it monthly. The scan takes about 30 seconds, you stay fully dressed, and it produces body fat percentage, circumference measurements, and a visual model you can compare side by side month over month.

Worth being straight about what it is. Styku is a 3D optical scan, not DXA, so it isn't the research-grade tool used in the trials above. What it's genuinely good at is consistency and repeatability, which is what tracking actually requires. Scanning the same person on the same device every month gives you a reliable trend line, and the trend is what changes decisions.

That data feeds directly into your bi-weekly titration check-ins. If the scan shows fat coming down and lean mass holding, we stay the course. If it shows lean mass sliding faster than we'd like, that's a conversation about protein, training, and pace, and it's a conversation we can have in week six instead of month six. Programs that check in monthly and track nothing but scale weight are flying blind on exactly the variable that determines whether you like your results.

Our medically supervised weight loss program starts from $295/mo and includes the medication, provider oversight, bi-weekly titration check-ins, monthly Styku scans, and ongoing nurse access. Single scans are also available on their own at $50, or $25 for new clients, if you just want a baseline. We're at 5100 Lincoln Ave in Lisle, serving Naperville, Downers Grove, Wheaton, and Chicago's western suburbs.

Frequently Asked Questions

Last updated August 14, 2026.