Two systematic reviews on GLP-1 receptor agonists and body composition published almost simultaneously in 2026. They reach what look like opposite conclusions.

Sawicka-Gutaj and colleagues, in International Journal of Obesity (April 2026), pooled 24 trials and concluded GLP-1 RAs deliver "quality" weight loss with relatively preserved lean tissue — reductions of approximately 1–4% in lean body mass across treatment durations from 3 to 12 months.

Batsis, Prado, and colleagues, writing in Annals of Internal Medicine, applied prespecified benchmarks for expected muscle-related losses — about 25% of total weight loss for FFM or lean soft tissue measured by BIA or DXA, and about 15% for skeletal muscle measured by CT or MRI. Their conclusion: muscle-based losses exceeded those benchmarks in 68% of incretin-based interventions.

Same drug class. Same time period. Two systematic reviews. Different answers.

The reviews aren't actually contradictory. They're answering different questions.

The IJO review reports the percentage change in lean body mass from baseline. Across 24 pooled trials with treatment durations from 3 to 12 months, reductions in lean body mass clustered in the 1–4% range — modest, by that metric.

The Batsis/Prado review reports lean mass loss as a proportion of total weight lost. If a patient drops 12 kg of body weight and 4 kg of that is lean tissue, that's 33% — exceeding the 25% benchmark.

These numbers move in the same direction. A 4% drop in absolute LBM accompanying a 12% drop in body weight gives roughly 20% of weight loss as 'lean tissue.' Batsis reports closer to 35% — and as Langer makes clear later in this piece, the gap between those numbers is partly explained by what DXA actually counts as 'lean.

The benchmark — that ~25% of weight lost is FFM under caloric restriction — comes from decades of work on diet-induced weight loss. It's the expected ratio when humans drop weight through energy deficit alone. The remaining ~75% comes from fat mass.

This ratio has been remarkably stable across hypocaloric diet studies. Not because anyone wants a quarter of weight loss to come from muscle, but because that's roughly what happens when humans go into negative energy balance without targeted intervention.

So when a meta-analysis finds GLP-1 patients losing 30–35% of total weight as lean mass, the question isn't whether GLP-1s are uniquely catabolic. The question is whether they're better, worse, or the same as comparator interventions producing similar magnitudes of weight loss.

The Batsis review answers this directly. Among 14 studies reporting lifestyle or placebo comparator data, median weight loss was only ~2.4%, but 50% of those interventions still exceeded the muscle-loss benchmark. The muscle-to-fat loss ratio with GLP-1s isn't categorically different from non-pharmacologic weight loss. The difference is the magnitude of weight loss, not the quality of it.

Here's where calling it "quality weight loss" runs into trouble.

A 45-year-old patient losing 4% of lean body mass while dropping 15% of total weight is in a fundamentally different position than a 75-year-old patient losing the same 4% LBM on the same percentage drop in total weight.

For the younger patient, that lean mass is largely recoverable with resistance training and adequate protein. For the older patient with baseline sarcopenia, osteoporosis, or pre-existing functional limitation, 4% of lean mass might be the threshold between independent function and not.

The IJO review acknowledges this in passing — they cite the Perna et al. study showing modest BW loss but stable LBM in elderly liraglutide patients, and they note BIA/DXA limitations in distinguishing skeletal muscle from organ mass and extracellular fluid. But the headline conclusion of "selective fat reduction with relative preservation of lean tissue" can't hold equally across age strata, and the meta-analytic average obscures the patients we should be most concerned about.

Yeah. That Arnold Schwarzenegger.

So do Codie Sanchez, Scott Galloway, Colin & Samir, Shaan Puri, and Jay Shetty. And none of them are doing it for fun. They're doing it because a list you own compounds in ways that social media never will.

beehiiv is where they built it. You can start yours for 30% off your first 3 months with code PLATFORM30. Start building today.

The most striking finding from the Batsis review: among 36 primary studies, none reported objective physical function outcomes.

Not grip strength. Not gait speed. Not chair-stand. Not stair climbing. Not VO₂ max. Not balance.

We have substantial data on tissue mass and almost none on what the muscle is actually doing. This matters because the relationship between lean mass and physical function isn't linear. Older adults can lose substantial muscle mass with preserved strength. They can also have preserved muscle mass with profound weakness — myosteatosis, anabolic resistance. What we care about clinically — falls, disability, frailty progression, all-cause mortality — depends more on function than on mass per se.

That gap is finally starting to fill in. Langer and colleagues, in a March 2026 Cell Reports Medicine paper combining four preclinical mouse studies with a proof-of-concept clinical trial, performed what may be the first published direct measurement of muscle function alongside body composition in patients on a GLP-1 RA. Twelve weeks of semaglutide in patients with obesity and type 2 diabetes produced a significant reduction in vastus lateralis cross-sectional area on ultrasound — and yet maximum voluntary contraction (knee extension) and hand-grip strength were both unchanged. Absolute and relative muscle strength held flat despite the decrease in muscle size.

Two larger trials cited within the Langer paper point in the same direction. STEP 9 in patients with obesity and knee osteoarthritis showed that semaglutide-induced weight loss improved physical function despite the expected lean mass changes. The SLIM LIVER study showed trends toward improved gait speed even with decreased psoas muscle volume. The picture forming across these is that mass loss and function loss are dissociable on incretin therapy — and may be moving in opposite directions for many patients.

There's a measurement problem buried inside the entire muscle-loss debate, and the Langer paper makes it impossible to ignore.

DXA — the technology underlying nearly every body composition substudy in the GLP-1 trial literature, including STEP 1 and SURMOUNT-1 — cannot distinguish skeletal muscle from other lean tissues, and cannot distinguish lean tissue from fat that resides within predominantly lean tissues. Intrahepatic triglyceride, intramuscular fat, liver glycogen, muscle glycogen, extracellular fluid — DXA reads all of these as some combination of "lean" and "fat" mass without separating them.

In Langer's preclinical mouse work, this turned out to be enormous. Across multiple incretin agents (semaglutide, tirzepatide, the GLP-1/glucagon dual agonist survodutide), liver mass decreased by 20–55%. With survodutide and immobilization combined, livers shrank by more than half — consistent with the robust reductions in liver fat now well-documented in MASH trials of GLP-1/glucagon co-agonists. Liver mass tracked tightly with body weight (r² = 0.71). Muscle mass barely tracked with body weight at all (r² = 0.09).

Translation: a meaningful fraction of what the clinical trial literature has called "lean mass loss" on GLP-1s isn't muscle. It's hepatic fat being mobilized, hepatic glycogen being depleted, intramuscular triglyceride disappearing, extracellular fluid shifting. These are the therapeutic targets of incretin therapy. They are not pathological. They are showing up as "lean mass loss" because the technology can't see the difference.

This connects to the second insight from the Langer work. Across mice and humans, the consistent pattern was that absolute muscle parameters — mass and strength — mildly declined, while relative parameters (muscle-to-body-weight ratio, strength-to-body-weight ratio) improved or held flat. The mice ran longer and faster on treadmill VO₂ max testing. The patients held their grip strength and knee extension force despite smaller cross-sectional areas.

Whether absolute or relative muscle mass is the right outcome is a clinically meaningful question. A 70-year-old who loses absolute muscle mass while preserving the muscle-to-weight ratio is in a different functional position than the same loss in someone who never had excess weight to lose. For patients with obesity — the population GLP-1s are prescribed to — relative muscle mass and the strength-to-weight ratio may matter more than the absolute numbers that drive the Batsis benchmark.

The strongest evidence for what changes the body composition equation comes from a randomized, head-to-head, placebo-controlled trial published by Lundgren and colleagues in NEJM in 2021.

After an 8-week 800 kcal/day low-calorie diet that produced a mean 13.1 kg weight loss, 195 adults with obesity (BMI 32–43, no diabetes) were randomized for one year to one of four arms: placebo, structured exercise, liraglutide 3 mg/day, or exercise plus liraglutide. The exercise program was moderate-to-vigorous intensity (78–79% max heart rate), averaging 118 minutes per week across roughly 2.4 sessions, with about a third of sessions supervised group cycling and circuit training.

The body composition data are the punchline.

Total weight loss from baseline to week 52 was 6.7% in the placebo group, 10.9% with exercise alone, 13.4% with liraglutide alone, and 15.7% with combination therapy. The combination produced approximately twice the fat-mass and waist-circumference reduction of either single intervention. Only the combination delivered all three together: HbA1c reduction, insulin sensitivity improvement, and cardiorespiratory fitness gains.

A clinically important secondary finding: liraglutide alone produced a sustained increase in resting heart rate, consistent with prior trials. The combination did not. Aerobic conditioning appears to neutralize the chronotropic effect that has been a persistent concern with chronic GLP-1 therapy.

Resistance training is the most potent non-pharmacologic stimulus for preserving and building muscle. The Lundgren protocol used moderate-to-vigorous aerobic-dominant exercise (cycling, running, circuit training) and still produced lean preservation; the strongest practical recommendation in 2026 is to combine that aerobic base with at least 2–3 dedicated resistance sessions per week. Older adults and patients with baseline sarcopenia risk benefit most.

Protein intake matters. Most clinicians use 1.2–1.6 g/kg of adjusted body weight daily during active weight loss, recognizing that GLP-1-induced satiety can suppress protein-rich food consumption. This needs to be a deliberate intervention, not an assumption. Patients on incretins eating "less of everything" usually need targeted counseling to maintain protein.

Function monitoring matters more than mass monitoring. Hand-grip dynamometry takes 90 seconds and costs nothing once the device is in the office. Gait speed over 4 meters takes about the same. DXA before and during therapy in higher-risk patients (older adults, lower baseline BMI, prior weight cycling, baseline frailty markers) provides the mass data, but the function data is what tells you whether the mass changes matter.

The Lundgren trial used liraglutide. Modern incretin therapy is a different drug class — semaglutide and tirzepatide produce roughly 2–3 times the weight loss of liraglutide. Whether the lean mass preservation observed with combination liraglutide-plus-exercise translates to higher-magnitude weight loss on more potent agents is a question the published RCT literature has not yet definitively answered.

A case series published in late 2025 in SAGE Open Medical Case Reports by Tinsley (Texas Tech University) and Nadolsky (Vineyard Health) provides the first detailed published body composition trajectories from a virtual obesity medicine practice that emphasizes structured resistance training and protein targets alongside incretin therapy.

Three patients (two female, one male; baseline BMI 32.9 to 51.9 kg/m²) on semaglutide or tirzepatide were tracked with serial DXA scans across 39 to 139 weeks of treatment. All three engaged in intentional exercise 4–7 days per week, with resistance training 3–5 days per week. Protein intakes ran 0.7–1.7 g/kg of body mass, or 1.6–2.3 g/kg of fat-free mass. All three patients took a multivitamin; two took creatine; one took omega-3.

The results:

In two of three patients, lean soft tissue increased during substantial weight loss on incretin therapy. Case 2 dropped 26.8% of body weight in 39 weeks while gaining 1.2 kg of LST. Case 3 lost 13.2% over 139 weeks while gaining 3.8 kg. The third patient lost 33% of body weight with only 8.7% of that loss attributable to LST — well under the 25% benchmark.

The caveats need stating plainly. N=3. Self-selected, highly motivated patients in a virtual obesity medicine practice. Self-reported nutrition and exercise data. DXA performed at fee-for-service centers rather than standardized research labs. Exercise volume (4–7 days per week, with 3–5 sessions of resistance training) substantially above population norms. A case series cannot establish causation, and these patients are not representative of typical incretin users. (As Vineyard Health is the practice that produced this case series and where I now serve as Chief Medical Officer, I have a direct interest in this work; readers should weigh accordingly.)

What the case series does establish is that the 25–35% of weight loss as lean mass observed across meta-analyses is not a biological inevitability of GLP-1 therapy. Combined with the Lundgren RCT data on liraglutide, the trajectory points in one direction: when patients implement structured exercise — particularly resistance training — alongside adequate protein, lean tissue preservation appears achievable across the incretin class.

All three sources are correct on their own terms.

The IJO review correctly observes that absolute lean mass losses on GLP-1 therapy are modest — typically 1–4% from baseline.

The Batsis/Prado review correctly observes that as a proportion of total weight lost, muscle losses commonly exceed traditional benchmarks. But that framing is built on DXA data and the assumption that DXA-detected lean mass loss is, in fact, muscle. Langer and colleagues make a serious case that it isn't — that hepatic and intramuscular fat mobilization, glycogen depletion, and fluid shifts are being misclassified as muscle loss in the trial literature, with the mismeasurement potentially substantial.

When you actually measure function, function holds up. Grip strength preserved. Knee extension preserved. Treadmill performance maintained or improved in mice. Physical function scores improving in patients with osteoarthritis on semaglutide. Gait speed trending upward in HIV/MASH patients despite measurable psoas changes.

The Lundgren NEJM RCT establishes that exercise plus liraglutide preserves lean mass and doubles fat loss versus either alone. The Tinsley & Nadolsky case series suggests the principle extends to current-generation GLP-1s when patients implement structured resistance training and meet protein targets. The Langer work suggests that even without those interventions, the function picture is better than the mass numbers alone would imply.

The clinical work doesn't wait for the literature to fully resolve. Resistance training, protein, and functional monitoring — do all three. The patients who get the most out of GLP-1 therapy are the ones whose physicians built the muscle-preservation framework in from the start.

REFERENCES