GLP-1s and Muscle Loss: What the New Data Actually Show — and How to Protect Your Muscle and Bone While Losing Weight

Ozempic, Wegovy, Mounjaro, and Zepbound have been accused of “melting muscle.” The newest evidence tells a more interesting story: most of what patients lose is not what the headlines think it is — but the part that is real is almost entirely preventable, and almost nobody is being told how.

Few questions come up more often in our clinic than this one: “If I go on one of these drugs, am I going to lose my muscle?” It is exactly the right question to ask — muscle is the tissue that most determines how strong, independent, and metabolically healthy you will be in your seventies and eighties — and until recently the honest answer was “we’re not sure.” The alarm began with a single trial suggesting that 40% of the weight lost on semaglutide was lean mass, a number that spread through the media faster than the caveats attached to it. Several years and many studies later, we can give a far more precise answer. This review consolidates that evidence: how much lean mass patients actually lose on each drug, why the DEXA scan that produced those numbers systematically misrepresents muscle, what happens to bone and fracture risk (the genuinely underappreciated risk), why strength and real-world function usually improve anyway, what the early retatrutide data suggest, and — most importantly — the protein and resistance-training protocol that turns an unavoidable-sounding side effect into a largely avoidable one.

---

Where the Muscle-Loss Alarm Came From

The origin of the concern

The modern GLP-1 era began when semaglutide — marketed as Ozempic for type 2 diabetes and Wegovy for obesity — produced weight loss no previous drug had approached. Regulators and trialists, understandably, focused on the scale. Body composition was a secondary consideration, and that framing shaped everything that followed.

Then the body-composition substudy of the landmark STEP 1 trial reported that roughly 40% of the weight lost on semaglutide was lean mass — for every 10 pounds lost, about 4 pounds of something that was not fat. That figure landed hard, and it deserved attention: losing 40% of your weight as lean tissue is a very different proposition from losing 20%. Scientific scrutiny of this question was genuinely valuable, and it made both physicians and patients ask a better question than “how much did I lose?”

But two things were lost in the amplification. First, STEP 1 has turned out to be something of an outlier — later studies of newer agents report considerably lower figures. Second, clinical trials report averages from populations whose nutrition and exercise were largely uncontrolled and often undocumented. Trials generally do not tell us how participants were counseled on protein, whether they lifted weights, or whether anyone made a deliberate effort to protect lean tissue. That matters enormously, because those are precisely the behaviors that determine the outcome. A trial average is informative; it is not destiny.

---

Why “Lean Mass” Is Not Muscle

Established · a limitation of the technology

This is the single most important concept in the entire debate, and it is almost never explained to patients. A DEXA scan is a low-dose X-ray that sorts your body into three buckets: fat, bone, and everything else. That third bucket is what gets labeled “lean mass.” It is not muscle. It contains muscle — and also your organs, your skin, connective tissue, extracellular water, and stored glycogen (each gram of which holds roughly three grams of water with it).

So when a study reports “lean mass loss,” it may be describing genuine loss of contractile muscle fiber — or it may be describing the water and glycogen that leave early in any weight loss, the shrinkage of organs and connective tissue that accompanies a smaller body, or something considerably more interesting.

The IMAT problem: when losing “lean mass” is a good thing

Fat is not only stored under the skin. Intramuscular adipose tissue (IMAT) is fat woven through muscle — the marbling you see in a steak — and intramyocellular lipid (IMCL) is fat stored inside the muscle fibers themselves. Both are metabolically unfavorable, and both are hallmarks of poor muscle quality. Here is the problem: DEXA cannot reliably distinguish subcutaneous fat from visceral fat from fat stored inside muscle. Because IMAT is interwoven with muscle tissue, DEXA tends to classify it as lean.

The implication is striking. When a patient loses IMAT — which is exactly what you want to happen — the scan can report it as “lean mass loss” even though muscle quality has improved. Separating true muscle from IMAT requires MRI, which is not practical in routine care. On top of this, there is evidence that DEXA overestimates muscle mass in people with higher BMI, particularly in the arms and legs — so when heavier patients lose weight, the apparent muscle loss may be partly an artifact of the baseline measurement being inflated in the first place.

---

How Much Lean Mass Is Actually Lost — Drug by Drug

Meta-analysis · the best current numbers

The right comparison is not “GLP-1 versus nothing.” It is “GLP-1 versus other ways of losing the same weight.” When you lose weight by any method, some of it is lean tissue — a smaller body genuinely requires less of nearly everything. In conventional diet-and-exercise programs, roughly 25% of weight lost is lean mass. That is the benchmark. A 2026 systematic review and meta-analysis comparing incretin therapies against intensive lifestyle intervention gives us the cleanest side-by-side we have.

Two lessons fall out of this table. First, percentages and absolute amounts are different things. Tirzepatide produces more total weight loss than semaglutide, so a patient may lose slightly more total pounds of lean tissue on it — while losing a smaller share of their weight as lean tissue, because the fat loss is so much greater. Both facts are true and they point in opposite directions; which one matters depends on the patient in front of you.

Second, the drug you choose appears to matter. The signal that tirzepatide preserves lean mass better than semaglutide is not yet definitive, but it is consistent, and it is one of several reasons that dual-agonist therapy has become the more common choice in our practice.

---

Is It the Drug, or Is It the Weight Loss?

Emerging · converging lines of evidence

This is the question that determines whether GLP-1 drugs are uniquely dangerous to muscle or simply very effective at causing weight loss — with all the consequences that follow from weight loss by any route.

A retrospective analysis in JAMA Network Open compared more than 1,000 bariatric surgery patients with nearly 2,000 patients on semaglutide or tirzepatide, all treated within the same health system, over roughly two years. Surgery produced substantially greater weight loss, greater fat loss, and greater absolute lean mass loss. But the ratio was the finding: lean mass accounted for approximately 18% of total weight loss in both groups. Nearly identical. Surgery did not turn out to be more “muscle-destructive” — it simply removed more weight.

The cleanest test would be a metabolic-ward study in humans with pair feeding — one group on the drug, one group eating precisely the same calories without it — and no such long-term study exists. But it has been done in animals. Two independent studies in diet-induced obese mice compared semaglutide against placebo and against calorie-matched pair-fed controls. In both, muscle outcomes were broadly similar between the drug-treated and pair-fed animals, with at most a slight excess in the semaglutide groups — and grip strength losses tracked together, even when muscle wasting was deliberately accelerated by immobilizing a limb.

Taken together: the best current interpretation is that lean mass loss on these drugs is largely a function of how much weight is lost, not a direct pharmacologic attack on muscle. That is genuinely reassuring — but it is not permission to ignore the problem. It relocates the responsibility. If the loss is driven by an energy deficit and reduced mechanical loading, then the tools that counteract an energy deficit and restore mechanical loading are the answer, and they work here just as they work anywhere else.

One important complication

When people lose weight through diet and exercise, they are by definition doing something — exercising, often lifting, often deliberately eating more protein. Those behaviors protect lean mass. Drug-induced weight loss can happen with no lifestyle change at all. That is precisely why the average trial result looks worse than what we see clinically in counseled patients: the comparison group has a built-in muscle-preservation program, and the drug group frequently does not.

---

The Metric That Actually Matters: Function

Consistent across multiple studies

If patients were losing large amounts of true contractile muscle, we would expect to see it: weaker grip, worse mobility, deteriorating physical performance. That is not what the data show. This is the disconnect that should reframe the entire conversation.

The SEMALEAN study followed 106 patients with a mean BMI of 46 for one year on semaglutide. They lost about 13% of their body weight and 18% of their fat mass. They also lost about 6.5 pounds of lean mass — roughly 40% of total weight loss — before it stabilized. On the lean-mass number alone, this looks like the STEP 1 nightmare. And yet: grip strength increased by about 10 pounds, and the prevalence of sarcopenia went down.

That is not an isolated result. At least four additional studies have found grip strength maintained despite substantial weight loss (Heymsfield 2026, Xiang 2023, Volpe 2022, Langer 2026), with one also showing no loss of knee extension strength. And on patient-reported function, the STEP 1 and STEP 2 trials found clinically meaningful improvements in physical function in more than 50% of semaglutide patients versus about 33% on placebo in STEP 1, and 43% versus 31% in STEP 2.

The most objective functional result comes from heart failure. In a prespecified analysis of STEP-HFpEF, patients with obesity and heart failure with preserved ejection fraction improved their six-minute walk distance by 21 meters on semaglutide versus 1.2 meters on placebo — a large, measured, real-world gain in what a person can physically do.

Why function improves while the scan says you lost muscle

Three reasons, all of them sensible. First, a lighter body demands less force — standing from a chair, climbing stairs, and walking all get easier when you are carrying 40 fewer pounds, even with no change in muscle capacity. Second, some of the “lost” lean mass was never functional — it was water, glycogen, and intramuscular fat. Third, some muscle loss is appropriate. A 220-pound body genuinely requires more muscle to move than a 175-pound body. Losing some of the muscle that existed only to carry the weight you no longer carry is not pathology; it is proportionality.

The clinically important question is therefore not “did the lean-mass number go down?” It is “can you still do the things you want to do — and are you getting stronger?” That is why our body composition program pairs serial DEXA with objective functional testing — grip strength, gait speed, sit-to-stand, and where appropriate cardiorespiratory fitness. The scan alone will mislead you in both directions.

---

Bone: The Risk That Deserves More Attention Than Muscle

A real signal · hard clinical outcomes

While the public conversation fixated on muscle, the more concrete finding may involve bone. Weight loss on GLP-1 therapy is associated with reductions in bone mineral density — as it is with lifestyle-induced weight loss and bariatric surgery. Both semaglutide and liraglutide have been shown to reduce BMD.

The evidence that matters most is not a scan but a fracture. The SELECT trial was large enough to count them. Among adults with obesity and established cardiovascular disease, hip and pelvic fractures occurred in 1.0% of semaglutide patients versus 0.2% on placebo — a roughly five-fold difference. In adults aged 75 and older, it was 2.4% versus 0.6%. Fractures are not a surrogate marker; they are the outcome that ends independence.

There is no evidence that GLP-1 drugs directly damage bone tissue. The likeliest explanation is mechanical: bone remodels in response to load, and a lighter body loads bone less. The same phenomenon appears after weight loss from any cause. Which points to the same countermeasure — and this is the elegant part of the whole story. Resistance training is simultaneously the intervention that preserves muscle and the intervention that preserves bone. One behavior, two problems. For older patients, or anyone with existing low bone density, this moves resistance training from “recommended” to “non-negotiable,” and makes a baseline DEXA bone density measurement before starting therapy a reasonable step rather than an optional one.

---

Timing, Risk Groups, and What Happens When You Stop

Limited data · act on the conservative assumption

When does lean mass loss happen?

This matters practically: if the loss is front-loaded, training and protein must start immediately — ideally before the first injection. Unfortunately, almost no study measures body composition more than twice. Weight trajectories are tracked in fine detail; lean mass is typically measured at baseline and at the end, and nowhere in between. Two controlled liraglutide studies (18-day and 35-day) suggest a substantial share of very early weight loss is lean tissue, while several real-world observational studies report patients barely losing lean mass, or even gaining it. The honest answer is that we do not know. The correct practical response to not knowing is to assume you are at risk from day one and start training and protein before you start the drug — not after the weight comes off.

Who is most at risk?

Less clear than you would expect. We anticipated that older adults would fare worse because of anabolic resistance — the reduced sensitivity of aging muscle to protein and training stimuli — but meta-analyses have not shown a meaningful difference in the proportion of weight lost as lean mass between adults over 65 and younger adults. That said, older adults are underrepresented in these trials and no study has been designed to answer the question. For sex, there is no direct evidence that women lose more lean mass; there is evidence that women lose slightly more total weight, with the gap largest for retatrutide (more than 9 pounds greater at some doses). Since lean loss scales with total loss, more lean loss in women is plausible — but it remains an inference, not a finding.

And if you stop the drug?

The widespread belief that all the weight comes straight back is not supported by the data. A systematic review and meta-regression of 48 studies found that about 40% of maximum weight loss was still maintained one year after stopping, with long-term projections suggesting roughly 25% persists. And these figures likely represent a worst case — participants typically stopped abruptly with little support around tapering, nutrition, or exercise.

More interesting is what comes back. In a semaglutide study, participants lost 9 pounds of lean mass on treatment (34% of their weight loss) and regained nearly 7 pounds of it after stopping — lean tissue accounted for 72% of the weight regained. In a liraglutide study, participants lost 22 pounds including 4 pounds of lean mass, and regained about 4.5 pounds of lean mass a year later — essentially all of it recovered. The same caveat applies in reverse, of course: regained “lean mass” is not necessarily regained muscle, and may substantially reflect water and glycogen returning.

---

The Protocol: How to Protect Muscle and Bone on a GLP-1

Established · and dramatically underused

Here is the uncomfortable data. In a survey of GLP-1 users, only about 38% reported doing any resistance training, and only a little more than a third reported following a high-protein diet. (Worth noting: the survey was conducted by a company developing a muscle-preservation drug, so read the framing accordingly.) The nutrition picture is worse. Supermarket receipt data show GLP-1 users buying more yogurt but less meat, seafood, and eggs. And one self-reported dietary study estimated average protein intake at just 33 grams per day — a fraction of what is needed to protect lean tissue during rapid weight loss.

That gap, not the pharmacology, is where the muscle is being lost. And it is fixable.

The evidence that this works is strong in the general weight-loss literature and, so far, encouraging but limited in GLP-1 users specifically. Small observational studies suggest lean mass losses are smaller when protein and resistance training are emphasized (Al-Badri 2024, Grannell 2021). Clinical experience in well-counseled patients — consistently hitting ~1.6 g/kg with a structured lifting program — is that very little lean mass is lost at all.

The training prescription

There is no special “GLP-1 workout.” The principles that build and preserve muscle in any context apply here unchanged.

Managing the side effects that get in the way

Nausea, dehydration, and fatigue in the first two to three months are the main reasons patients skip training and under-eat protein — which is exactly how the muscle gets lost. These are largely a dosing problem. Start at the lowest effective dose and titrate slowly. Aggressive escalation buys faster weight loss at the cost of the two behaviors that protect you. Hydrate deliberately, and if a dose level is making training impossible, that is a reason to hold or slow the titration, not to abandon the program.

---

Retatrutide — and a Serious Warning About the Gray Market

Investigational · not FDA approved

Retatrutide is a triple agonist still moving through the FDA approval process. Preliminary results from the TRIUMPH-1 trial reported weight loss of up to 28.3% at the highest dose — approaching bariatric surgery territory.

The body-composition data are thinner. The available figures come from a phase 2 substudy in type 2 diabetes: at the highest dose, participants lost about 34 pounds more than placebo, of which roughly 13 pounds was lean mass — about 38% of total weight loss. As a percentage, that is similar to semaglutide and somewhat worse than tirzepatide. But because the total weight loss is so much larger, the absolute lean mass loss could be substantially greater: a rough comparison would be ~12.5 pounds of lean mass with a 33-pound semaglutide weight loss versus ~23 pounds with a 62-pound retatrutide weight loss. These are DEXA-derived numbers, with all the caveats above.

TRIUMPH-3 will add lean-mass data, and some retatrutide trials will assess self-reported physical function. Conspicuously absent from the program: objective strength testing, performance-based functional assessment, and robust bone outcomes. Given the magnitude of weight loss involved, those omissions are hard to defend.

---

The Long View

We asked the wrong question about muscle — and got a useful answer anyway

For three years the field argued about a number produced by a machine that cannot measure the thing being argued about. DEXA sorts tissue into fat, bone, and everything else, and we took “everything else” to mean muscle. It never did. It meant muscle plus water plus glycogen plus organs plus connective tissue plus the fat marbled through the muscle itself — and it meant them all at once, inseparably, in a population where the baseline measurement was probably inflated to begin with. We then built a public panic on the difference between two of those numbers.

The argument was not worthless. It forced the right studies. But the answer that emerged is quieter and more useful than the question: lean mass loss on these drugs is mostly the ordinary arithmetic of losing weight, and the outcome that actually matters — what a person can physically do — consistently gets better, not worse. Grip strength rises. Walking distance rises. Sarcopenia prevalence falls. The patients in these trials are not becoming frail; most of them are becoming more capable than they have been in years.

This is a lesson that extends well beyond obesity medicine, and it is the one we keep relearning: the metric that is easiest to measure is rarely the metric that matters. Lean mass is easy — a scan, a number, a chart. Function is harder — it requires grip dynamometry, gait speed, a six-minute walk, a conversation about whether the stairs are easier this month. Bone density is easy; a fracture is what actually ends a life as its owner knew it. When medicine optimizes for the convenient measurement, it can spend years worrying about the wrong thing while a genuine signal — the five-fold fracture difference in SELECT — sits quietly in a safety table.

Which brings us to the bargain. These drugs are the most consequential development in metabolic medicine in a generation, and their benefits appear to extend well beyond the scale — semaglutide has now shown meaningful benefit in knee osteoarthritis, and there are reasonable grounds to suspect broader anti-inflammatory and healthspan effects still being mapped. But they are not free, and the price is not measured in dollars. The price is that you must do the work the drug cannot do for you: eat the protein, lift the weight, load the bone. The pharmacology removes the mass. It has no opinion about whether what remains is a stronger body or merely a smaller one.

That is the whole argument for medical supervision rather than a prescription and a refill reminder. Someone has to set the protein target, build the training program, watch the bone density in a 74-year-old, titrate slowly enough that nausea does not derail the gym, choose tirzepatide over semaglutide when lean mass preservation is the priority, and measure function rather than trusting a number that was never measuring muscle. Done that way, the goal stops being weight loss and becomes what it always should have been: keeping everything you want to keep, while losing what is hurting you.

---

If You Are On — or Considering — a GLP-1

---

Questions We Get About GLP-1s and Muscle