Bone Density and Lean Mass on GLP-1s: What the DXA Trials Actually Show

The short version

Two questions get blurred into one here, and they have different answers.

Does losing a lot of weight cost you bone? Yes, and that part is not controversial. Rapid or sustained weight loss of any kind — diet, surgery or drug — is associated with lower bone mineral density, deterioration of bone microarchitecture, and heightened fracture risk³.

Does a GLP-1 do something to your skeleton beyond the weight it removes? Unsettled — and the two strongest bodies of evidence disagree at the edges, which is the honest thing to say rather than picking the reassuring one.

Underneath both sits the body-composition fact that usually leads coverage: roughly a quarter to a third of what comes off is lean tissue rather than fat, and that proportion is broadly the same whether the weight leaves via an incretin drug or via a lifestyle program¹. So bone here is a weight-loss problem first and a drug question second. The muscle-and-function half of the story — protein targets, training, why it bites harder after 40 — lives on its own page: muscle loss on a GLP-1.

What a DXA scan actually measured

The most detailed body-composition data on any of these drugs come from a substudy of SURMOUNT-1: 160 participants — 73% of them women, mean weight 102.5 kg — had dual-energy X-ray absorptiometry at baseline and again at week 72².

Two rows deserve a slow read. The placebo column is the first: lean mass fell there too, because that group also lost weight. The second is that the bottom two rows are a different kind of number from the ones above them. Lean mass itself dropped about 11%; lean tissue accounted for about 25% of the weight that came off. Those two get swapped constantly in coverage of these drugs, and swapping them either halves or doubles the apparent problem. The substudy's own conclusion is the unglamorous one — the fat-to-lean split was about the same on tirzepatide as on placebo².

Why weight loss costs bone in the first place

Bone is not a passive scaffold. A 2025 mechanistic review in Bone Research traces weight-loss-induced bone loss to several interlocking causes: reduced mechanical loading as you carry less weight, increased bone-marrow adiposity, hormonal and endocrine shifts, nutritional shortfalls, and disrupted energy metabolism — which together impair bone remodeling³. Its framing is the one to hold onto: moderate weight reduction may be metabolically beneficial, but rapid or sustained loss is the pattern associated with skeletal harm³.

The one randomized trial that measured bone directly

Most bone talk about these drugs is inference. One trial is not. A pre-specified secondary analysis of a Danish randomized trial put site-specific bone mineral density at the hip, spine and forearm as its primary outcome, in 195 adults with obesity and no diabetes — 64% of them women — over a year after an initial low-calorie diet⁴.

The finding that matters: the drug arm lost more bone density than the exercise arm despite similar weight loss, and the combination arm lost the most weight of all while its bone density held level with placebo⁴. That is a real, randomized argument that how you lose the weight changes what happens to your skeleton — and the caveats are equally real. It is one trial, it used liraglutide rather than semaglutide or tirzepatide, and bone density is a surrogate, not a fracture.

The fracture question, where the evidence pulls two ways

A 2025 systematic review of 25 randomized trials in type 2 diabetes found no significant increase in fracture risk with GLP-1 receptor agonists (RR 0.80, 95% CI 0.47 to 1.36) and, if anything, better bone mineral density than control at the lumbar spine (+0.07 g/cm²), femoral neck (+0.05) and total hip (+0.06)⁵. Reassuring — but read the population. Those were diabetes trials, at diabetes doses, mostly short, in people who were not losing 20% of their body weight.

The prescribing information points the other way on one specific outcome. In the semaglutide cardiovascular outcomes trial, more hip and pelvis fractures were reported on drug than on placebo in women: 1% (24/2,448) versus 0.2% (5/2,424), and in patients aged 75 and over, 2.4% (17/703) versus 0.6% (4/663)⁶. That is an imbalance in one trial, not an established causal effect — but it is the one signal that lands squarely on the readers of this site.

Why this lands harder for women

Two things stack. The trials that quantified all of this were majority-female — 73% in the DXA substudy, 64% in the bone-density trial — so these are not male averages being extrapolated^2,4. And women tend to start losing weight at exactly the life stage when estrogen withdrawal is already accelerating bone turnover, which is why our guide to GLP-1s through perimenopause and menopause treats bone as a first-order concern. Layering rapid weight loss onto that transition³ is the scenario worth planning around, and the label's female fracture imbalance⁶ is a reason to plan rather than to panic.

What the evidence says to do

Train while you lose, not afterwards. That is the intervention with randomized support on both fronts: the combination arm preserved bone density while losing the most weight⁴, and adding resistance training to a weight-loss program cut the lean-mass share of the loss to 17.5%, the lowest of any group analyzed¹. Eat enough, since nutritional deficiency is one of the named mechanisms³. And if you are perimenopausal, postmenopausal, over 75, or already have a fracture history, ask your clinician about a baseline DXA and a follow-up rather than assuming the question has been settled for you.

How strong is the evidence, honestly

Mixed, and worth grading in pieces. That significant weight loss lowers bone density: high — consistent mechanistic and clinical evidence, and not specific to these drugs. That the drug arm loses more bone than an exercise arm at matched weight loss: moderate — one randomized trial with bone as its primary outcome, using an older GLP-1. That GLP-1s raise fracture risk: low and conflicting — 25 randomized trials in diabetes find no increase, while one cardiovascular outcomes trial shows a female hip-and-pelvis imbalance in the label. What would change the picture is what nobody has yet run: an obesity-dose trial of semaglutide or tirzepatide with bone density and fractures as pre-specified endpoints, in postmenopausal women, followed for years rather than months.