How to Protect Muscle During Intentional Weight Loss

Two errors recur. The first is treating DXA "lean mass" as though it were skeletal muscle. The second is treating the risk as uniform across patients, when it varies enormously with age, baseline muscle, rate of loss and what the patient is doing while they lose.

Getting both right changes what you do about it.

What the data actually show

The numbers are worth stating precisely, because they are quoted loosely in both directions.

In the STEP 1 body composition substudy, semaglutide produced a mean weight reduction of 15.3 kg with a lean mass reduction of 6.92 kg — a fraction of weight lost from lean mass of about 45%. In SURMOUNT-1, tirzepatide produced a 22.1 kg weight reduction with 5.67 kg of lean loss, a fraction of about 26%. Liraglutide figures have run higher still. [1]

Three qualifiers matter, and omitting them is how this topic becomes alarmist.

Lean mass is not muscle. Fat-free mass on DXA includes skeletal muscle, but also organs, bone, connective tissue and body water. A patient losing 20 kg reduces liver volume, plasma volume, glycogen and its associated water, and the structural tissue supporting a smaller body. Some of that reduction is appropriate rather than harmful.

Relative composition usually improves. In STEP 1, lean mass as a proportion of total body weight rose by about three percentage points. Absolute lean tissue fell while the patient's composition got better, which is what you would expect from any successful weight loss.

Function has generally held up. Where handgrip strength and muscle function have been measured alongside DXA, they have been preserved or improved rather than degraded. [2] A 2026 meta-analysis of GLP-1 trials reached the same conclusion: proportional body composition improves, and the muscle-function signal does not show deterioration. [3]

None of this means the concern is invented. It means the headline percentage is the least informative number in the discussion.

Who is actually at risk

Risk is concentrated, not uniform, and identifying it takes a few minutes.

Older adults are the primary group, because they start with less muscle, rebuild it more slowly, and have less reserve before function is affected. Patients with sarcopenic obesity — high fat mass with already low muscle — are the clearest case, and they are easy to miss because their weight looks like the problem.

Beyond that: patients losing weight rapidly, those eating well below protein requirements, those doing no resistance training, those undergoing very large total reductions, and post-bariatric patients with restricted intake.

A patient who is 38, lifts three times a week, eats adequate protein and is losing 12% of body weight does not need a lean mass intervention. A patient who is 71, sedentary, eating poorly on a GLP-1 and losing 20% does.

The two interventions that work

Everything else is secondary to these.

Resistance training. This is the single most effective measure, and it is not optional for the at-risk group. Two to three sessions weekly, compound movements covering the major muscle groups, progressive in load over time. Progression matters more than volume: a patient doing the same weights in month six as month one is maintaining a habit rather than a stimulus.

Adequate protein. During energy restriction, requirements rise above maintenance. A working target of 1.2 to 1.6 g/kg of body weight daily is reasonable for most adults losing weight, with the upper end for older patients and those training hard. Trial evidence in energy deficit combined with training shows higher protein intakes producing better lean mass outcomes than lower ones. [4] Distribution across meals matters more than people expect; a single large evening serving is less effective than the same total spread across the day.

Neither of these is novel, expensive or proprietary, which is presumably why they get less attention than they deserve.

Rate and deficit

Faster weight loss takes a larger proportion from lean tissue. A rate around 0.5 to 1% of body weight per week is a reasonable ceiling for most patients, and slower is better tolerated in older adults.

With GLP-1 therapy this often requires actively slowing the process — holding a dose rather than escalating, if weight is coming off faster than function can be maintained. That is a legitimate clinical decision and it should be framed to the patient as such, rather than as failure.

The practical problem with appetite suppression

The difficulty with incretin therapy is not that protein is unhelpful. It is that appetite suppression makes hitting protein targets genuinely hard, and nobody warns patients about this at initiation.

Practical strategies: eat protein first in each meal, before carbohydrate or fat displaces capacity. Use smaller, more frequent meals rather than fewer large ones. Liquid protein sources are better tolerated than solid meat when early satiety or nausea is prominent. And set the expectation at the start, because a patient told at month four that they should have been eating more protein has already lost four months.

What to measure

Follow function. It is cheap, repeatable in clinic, and closer to what actually matters than any scan.

Grip strength with a dynamometer, five-times sit-to-stand, gait speed over four metres, and the patient's own report of stairs, shopping and getting off the floor. Record them at baseline, because without a baseline a follow-up measurement is uninterpretable.

DXA is useful where available, with the caveat above about what lean mass includes. Bioimpedance is less reliable in patients whose hydration is shifting, which describes most people actively losing weight.

If function is holding and composition is improving proportionally, the programme is working, whatever the absolute lean figure has done.

What not to reach for yet

There is a developing pipeline of agents targeting the myostatin and activin pathways with the explicit aim of preserving muscle during weight loss. Several are in active trials. None is approved for this purpose, and the outcome data that would justify routine use does not yet exist.

Predictably, a parallel market has appeared offering peptides and compounds marketed for muscle preservation on the strength of mechanism alone. The established interventions here are resistance training and dietary protein, both of which have decades of evidence, no regulatory ambiguity, and no sourcing risk.

When a patient asks whether they should add something to protect muscle, the accurate answer is that the two things that work are already available to them, and most people asking are not yet doing either consistently.

A note on selection

This article assumes weight loss is clinically indicated. That is worth confirming rather than assuming, particularly in older patients where unintentional weight loss carries different meaning entirely, and in anyone whose relationship with eating suggests that a structured weight-loss programme may not be the right intervention. Screening for that before starting is part of doing this well.

For monitoring patients on incretin therapy more broadly, see Managing GLP-1 therapy. For why a body composition scan is not the same as a health outcome, see Your biomarkers improved — did your healthspan?.