Ozempic for type 2 diabetes: does muscle preservation work the same as for weight loss?

Semaglutide and tirzepatide were approved first for type 2 diabetes, years before Wegovy and Zepbound arrived for weight loss alone. Yet almost every large body-composition study cited on this site, including the SURMOUNT-1 and STEP 1 substudies that most articles about GLP-1 muscle loss lean on, deliberately excluded people with diabetes. That is a real gap if you are one of the millions of people taking these medications primarily to manage blood sugar, not to lose weight for its own sake.

Type 2 diabetes is not a footnote on the weight-loss story. It changes how much weight comes off, it changes the starting condition of your muscle before treatment even begins, and there is early evidence it may change the ratio of fat to lean tissue in what is lost. Here is what the T2D-specific research says.

Do people with type 2 diabetes lose as much weight?

No, consistently not. A 2025 systematic review and meta-analysis pooled 10 randomized trials of once-weekly semaglutide 2.4 mg, separating participants by diabetes status. Among 1,872 participants with type 2 diabetes, the weighted mean weight loss was 6.34 percent. Among 3,842 participants without diabetes, it was 11.57 percent, a difference of about 5.2 percentage points that was highly statistically significant.¹ The same gap shows up trial by trial: the T2D-specific STEP 2 trial produced 9.6 percent weight loss on semaglutide 2.4 mg versus 3.4 percent on placebo,⁷ well below the roughly 14.9 percent seen in the non-diabetic STEP 1 trial. Tirzepatide follows the same pattern: up to 15.7 percent weight loss in SURMOUNT-2, the T2D trial, versus up to 20.9 percent in SURMOUNT-1, the non-diabetic obesity trial.⁸

Nobody knows the exact mechanism for the gap, though slower gastric emptying changes, differences in how insulin resistance affects appetite signaling, and generally higher baseline medication burden in T2D are all plausible contributors. For the muscle question, the practical implication is straightforward: less total weight lost generally means less absolute lean mass at stake, even if the proportion lost stays the same.

What the T2D-specific body-composition data shows

Two dedicated substudies have actually measured fat versus lean tissue in people with type 2 diabetes on a GLP-1, and their pictures do not fully agree with each other.

The SUSTAIN 8 DXA substudy randomized a subset of people with type 2 diabetes on metformin to once-weekly semaglutide 1.0 mg or once-daily canagliflozin, an SGLT2 inhibitor, and measured body composition by DXA scan at 52 weeks. Semaglutide produced a 3.4 kg reduction in total fat mass and a 2.3 kg reduction in total lean mass, versus 2.6 kg and 1.5 kg respectively on canagliflozin; the difference between drugs was not statistically significant.² Based on those figures, lean tissue accounted for roughly 40 percent of the weight semaglutide participants lost, somewhat higher than the roughly 25 to 26 percent lean-mass share reported in the non-diabetic SURMOUNT-1 tirzepatide substudy.⁹ The comparator here was another diabetes drug, not a placebo, which limits how directly this generalizes.

The SURPASS-3 MRI substudy took a more detailed look, using MRI to separate muscle volume from intramuscular fat in people with type 2 diabetes on tirzepatide over 52 weeks. Fat-free muscle volume fell by about 6.9 percent in women and 5.5 percent in men, a real reduction. But the same substudy found that fat infiltration within the muscle, a marker linked to poorer muscle quality and function, actually decreased with tirzepatide, reversing the age-related increase normally expected over a year.³ That is a more encouraging signal about muscle quality than the raw volume number suggests on its own, though it does not erase the volume loss itself.

A real-world signal worth flagging

Trial data is one thing; a single-center, real-world observational study of 90 people with type 2 diabetes starting semaglutide found something that runs against the usual pattern. Using bioelectrical impedance analysis, a less precise method than DXA or MRI, the study found lean mass fell by 3.37 kg at 6 months and 3.53 kg at 12 months, while fat mass fell by only 2.1 kg and 1.08 kg over the same periods, meaning lean tissue accounted for more of the total loss than fat did.⁴ That inversion has not been replicated in a DXA-based trial, and bioelectrical impedance is known to be less reliable than DXA or MRI for tracking lean-mass change, particularly as hydration status shifts during weight loss. It is one study, not a consensus finding, but it is a real, published data point that cuts against the reassurance offered by the controlled trials above, and it is a reason not to assume the T2D muscle ratio automatically mirrors the non-diabetic one.

Why type 2 diabetes might start from a different baseline

Independent of any GLP-1, sarcopenia, the age-related loss of muscle mass and strength, is already more common in type 2 diabetes. A meta-analysis found sarcopenia prevalence of 28.4 percent in people with type 2 diabetes versus 18.7 percent in matched controls, an odds ratio of 1.63.⁵ The proposed mechanisms are specific to diabetes: chronic high blood sugar promotes advanced glycation end-products that stiffen tissue and are linked to weaker grip strength, insulin resistance blunts the hormone's normal muscle-building signal, and low-grade inflammation common in diabetes independently favors muscle breakdown.⁵

That matters for how you read every number above. A given percentage of lean-mass loss lands on a population that, on average, already has less muscle reserve to begin with. The case for prioritizing resistance training and protein early is, if anything, stronger in type 2 diabetes, not weaker.

Does the drug itself act differently on muscle in diabetes?

A 2021 review focused specifically on glucose-lowering drugs and sarcopenia in type 2 diabetes concluded that the evidence for GLP-1 receptor agonists is "controversial and inconclusive," and the trial-by-trial picture explains why. In one small trial, dulaglutide reduced skeletal muscle mass by about 3.8 percent over six months in patients on hemodialysis, a population already prone to sarcopenia. In two liraglutide trials, muscle mass changes were minimal or not statistically significant. A short exenatide trial found no significant change over 12 weeks.⁶ The same review also describes a plausible upside specific to this drug class: GLP-1 receptor activation may increase GLUT4 expression and glucose uptake in skeletal muscle cells and improve insulin sensitivity directly in muscle tissue, a mechanism that has nothing to do with appetite suppression and is not present in simple caloric-restriction weight loss.⁶

The honest summary is that no consistent, drug-class-wide effect on muscle has been demonstrated in type 2 diabetes specifically, in either direction. The lean-mass loss seen in the DXA and MRI substudies above appears to track with weight loss generally, the same as in non-diabetic trials, rather than showing an obvious diabetes-specific pattern on top of it, with the real-world signal above as a notable exception worth watching as more data accumulates.

What still protects muscle, regardless

Nothing about a type 2 diabetes diagnosis changes the two levers with the strongest evidence behind them. Resistance training and adequate protein intake are established independently of diabetes status, and both remain the most defensible response here. Our guide to the best exercises to preserve muscle covers the movement patterns that matter most, and our breakdown of how much protein you actually need on a GLP-1 covers the target and how to hit it on a suppressed appetite. The muscle and protein calculator can give you a personalized starting number based on your own weight and activity level.

One diabetes-specific note: protein needs can interact with kidney function, and chronic kidney disease is more common in long-standing type 2 diabetes. Anyone with reduced kidney function should set a protein target together with their clinician rather than applying a general population number. Our Protein Playbook mini-guide covers the general target and practical food strategy for a suppressed appetite; it is not a substitute for that conversation if kidney function is a factor for you.

What the evidence does not say

Several limits are worth stating plainly. No trial has directly compared lean-mass loss between matched T2D and non-diabetic groups on the same GLP-1 at the same dose in the same protocol; every comparison in this article is across separate trials with different designs, which limits how precisely the numbers can be compared. The SUSTAIN 8 substudy compared semaglutide against another active diabetes drug, not a placebo, so it does not isolate the medication's effect on lean mass in an untreated baseline. The real-world study showing lean mass exceeding fat mass loss used bioelectrical impedance, a less accurate method than DXA or MRI, in a single center with 90 participants, and has not been replicated. And no trial specific to GLP-1 users with type 2 diabetes has tested whether resistance training and protein interventions restore the same roughly 90-percent protection against lean-mass loss seen in broader weight-loss populations; that evidence, cited throughout this site, comes from non-GLP-1, non-diabetes-specific research and is applied here by extension, not direct demonstration.

The bottom line

Muscle preservation on a GLP-1 does not work identically for type 2 diabetes and for weight loss alone, but the differences are in degree, not in kind. People with T2D generally lose less total weight, which likely means less absolute muscle at stake, while starting from a population already more prone to sarcopenia. The T2D-specific composition data confirms real lean-mass loss, alongside at least one encouraging signal on muscle quality and one discouraging real-world signal on the fat-to-lean ratio that deserves more research before it is dismissed. Until head-to-head trials close that gap, resistance training and adequate protein remain the best-supported response, arguably with more urgency here than in a population that started with more muscle to lose.

Frequently asked