GLP-1 Medications and Bone Density: The DXA Monitoring Playbook

The headline from the GLP-1 trials is fat mass. The DXA substudy in STEP 1 reported that body weight fell 15.0 percent on semaglutide against 3.6 percent on placebo, with fat mass down 19.3 percent and lean body mass down 9.7 percent.¹ SURMOUNT-1 showed the same shape with tirzepatide and a larger total drop.² Every popular summary stops at the fat-to-lean ratio and calls it good. The skeleton is not in that summary.

Bone density usually moves with body mass. Postmenopausal women lose roughly 1 to 2 percent of lumbar spine bone mineral density per year during the menopause transition without any medication.³ A 12- to 18-month course of a GLP-1 sits on top of that baseline trajectory, drops mechanical loading, restricts intake, and changes the hormonal environment that bone responds to. The trial DXA substudies are small and short. They do not tell anyone whether a specific patient will lose enough bone to matter. They make it clear that the question deserves a real answer for the people most exposed to it.

This piece is the monitoring playbook. Who needs a baseline DXA before starting a GLP-1, what the follow-up scan can and cannot tell you, what to change when the numbers move the wrong way, and what happens to the skeleton when the medication stops. It pairs with the GLP-1 micronutrient floor on the nutrient side and the men's muscle protection guide on the training side.

01What the GLP-1 DXA Substudies Actually Showed

The body-composition data sit inside the larger weight-loss trials. The DXA arms are subsets, not the full enrolled population, which is part of why headline coverage usually skips them.

The bariatric data are a different population and a different mechanism. They are the closest available analog for what large, sustained weight loss does to a skeleton over one to two years. The pattern in the bariatric literature is real and consistent. The femoral neck loses faster than the lumbar spine. Younger patients with surgery still show measurable BMD decline.⁴⁵ GLP-1 weight loss is meaningfully smaller in magnitude than surgical weight loss, but the duration of pharmacologic therapy is now stretching into years rather than months, which collapses the difference over time.

The honest framing is this. The DXA substudies confirm that weight loss with semaglutide and tirzepatide follows the expected fat-greater-than-lean pattern. They do not establish what the bone trajectory is for an individual on chronic therapy, and they are not powered to answer that question. The clinical literature on weight loss in general points at meaningful bone changes when total mass falls quickly. That is enough to justify a structured monitoring plan for higher-risk patients, not enough to recommend universal screening.

02Why Bone Loses With Weight Loss

The skeleton is a load-responsive tissue. Three mechanisms run in parallel when body weight falls quickly on a GLP-1, and they each push in the same direction.

Mechanical unloading. Bone formation is driven by strain on the matrix. Lower body mass means less force per stride, less compressive load on the spine, less floor reaction force during daily activity. Wolff's framing applies in reverse. Less load means less signal to keep building bone. The Frost mechanostat framing places a daily minimum effective strain below which formation cannot keep up with resorption.⁶ A patient who loses 25 to 35 kg has lost a significant fraction of the load the skeleton was adapted to. The training prescription has to replace that signal deliberately.

Reduced intake of the substrate. Calcium balance is intake minus losses. At 1,200 to 1,500 calories per day on chronic therapy, total dietary calcium often falls below 600 to 800 mg. Vitamin D status is already lower in higher-BMI patients due to volumetric dilution.⁷ Protein intake at 1.2 to 1.6 g/kg adjusted body weight is the protective floor for both muscle and bone, but it is harder to hit when total food volume is suppressed. Bone gets the leftovers of an already-tight micronutrient budget unless the food plan is designed around the floor.

Hormonal environment. Estrogen loss across the menopause transition is the single largest risk factor for accelerated bone loss in women, with lumbar spine BMD falling roughly 10 percent across the 5 years before and 5 years after the final menstrual period, and 7.4 percent of that decline concentrated in transmenopause itself.³ Layering a GLP-1 weight loss program onto a woman in her 40s or 50s without controlling for transition stage means the medication-driven loss adds to a baseline that is already declining. Men are not immune, particularly older men with low testosterone, chronic glucocorticoid use, or both.

The combination is what creates the risk profile. Any one of these effects on its own can be small. All three running together for 18 to 24 months on a patient who is already approaching osteopenia at baseline is the worst-case shape.

03Risk Stratification: Who Needs a Baseline DXA

Universal pre-treatment DXA overscans the population at lowest risk and pulls resources away from the patients who need real monitoring. The decision is whether the result would change management.

The principle is risk-stratified utility. Scan the people whose follow-up scan will be readable against a real baseline and whose treatment plan would change based on what the scan shows. Skip the people for whom the result is unlikely to drive any action.

For women specifically, the women's recomposition guide covers the cycle and menopause overlay in more depth. The pattern is that perimenopause is where the most aggressive monitoring matters because the patient is on the steepest part of the loss curve regardless of the medication.

04What a Follow-Up DXA Can and Cannot Tell You

The scan does not measure what most patients hope it measures. It measures areal bone mineral density, in grams per square centimeter, at specific skeletal sites. Trabecular microarchitecture, cortical thickness, and bone turnover happen below the measurement. The number still tracks fracture risk well at the population level, with a one standard deviation decrease in BMD raising overall fracture risk by 1.5, vertebral fracture risk by 2.3 at the spine, and hip fracture risk by 2.6 at the hip.⁹ The interpretation framework lives in the bone-mineral-density glossary entry, with three numbers that matter most.

The first number is the least significant change at the scanning facility. The 2023 International Society for Clinical Densitometry positions set minimum acceptable precision targets for central DXA at 1.9 percent for lumbar spine, 1.8 percent for total hip, and 2.5 percent for femoral neck, which correspond to least significant change values of 5.3, 5.0, and 6.9 percent respectively.¹⁰ Any change smaller than the site-specific least significant change is below the noise floor of the measurement. A 2.5 percent drop at the lumbar spine on a follow-up scan does not mean the patient lost 2.5 percent of bone. It means the measurement is not precise enough to distinguish a real change from scan-to-scan variation.

The second number is the rate of change adjusted for time. A 4 percent drop at the lumbar spine over 18 months is a different signal than the same drop over 36 months. The annualized rate of change is what compares against background aging-driven loss. Drift faster than 2 percent per year at the spine in an adult under 65 is a signal that warrants both a clinical workup and a treatment reconsideration. Drift faster than 3 percent per year at the hip is more concerning because hip BMD predicts hip fracture better than any other site.

The third number is the absolute T-score in postmenopausal women and men over 50. A T-score that crosses from above to below -2.5 is the diagnostic threshold for osteoporosis at a valid site.¹⁰ That crossing changes the clinical conversation from monitoring to treatment. A T-score still above -1.0 at follow-up with a small decline that is below least significant change is not a clinical event. Pretending it is one creates unnecessary anxiety and unnecessary medication trials.

Most patients will land in the first or second row. The plan rarely needs to change beyond reinforcing the nutrition floor and the resistance training prescription. The bottom three rows are where the medication choice becomes a clinical discussion that includes weighing cardiometabolic and weight-loss benefits against active skeletal risk.

05The Four Nutrition Levers That Actually Move Bone

Most of the supplement marketing aimed at GLP-1 patients is noise. The levers that have published evidence at the relevant doses are short.

Protein at 1.2 to 1.6 g/kg adjusted body weight. The lower bound is the floor for muscle preservation during weight loss in adults, and bone follows muscle more closely than most coverage admits. The 2024 multi-society advisory on nutrition during antiobesity medication use treats this range as a practical protein target during pharmacologic weight loss.¹¹ The leucine threshold piece covers the per-meal distribution that makes this dose actually land.

Calcium 1,000 to 1,200 mg per day, food first. A patient eating 1,400 calories per day rarely clears 1,000 mg of calcium without targeted food choices. Greek yogurt, dairy in general, sardines and salmon with bones, and tofu set with calcium are the highest-yield anchors. A 500 mg calcium supplement closes the gap when food cannot, and should be taken with food at a time of day separate from any iron supplement to avoid absorption interference. Calcium without vitamin D and without mechanical loading does not protect bone. Calcium plus the other two does.

Vitamin D managed by lab status. Higher-BMI patients have lower baseline 25(OH)D due to volumetric dilution rather than sequestration.⁷ Practical dosing on a GLP-1 should be guided by baseline 25(OH)D and retesting after dose adjustment rather than a fixed maintenance dose. Monthly bolus dosing has a clear harm signal at high single doses in older adults and should be avoided.¹²

Magnesium 200 to 400 mg per day from food or split-dose glycinate. Magnesium is a cofactor for vitamin D activation and for parathyroid hormone secretion. A low-volume GLP-1 diet built on shakes and lean meats often falls below 250 mg per day. The magnesium piece covers form, dose, and timing. Glycinate or citrate are the cleanest choices for patients with GI sensitivity to oxide.

The supplements that do not earn a default place on this list include collagen for bone, K2 alone, boron, strontium, and any branded bone formula. Collagen has some signal for cartilage and tendon outcomes but bone evidence is thin. K2 has plausible mechanism but inconsistent clinical outcome data outside of warfarin-treated patients. Strontium is not the same as strontium ranelate, which had its own regulatory withdrawal in Europe.

06The Resistance Training Prescription That Moves Bone

Bone responds to load that exceeds the daily minimum strain. Walking does not exceed it in trained adults. Light resistance training does not exceed it. The training prescription that has actually moved BMD in postmenopausal women is the LIFTMOR protocol from Watson and colleagues, which showed that 8 months of twice-weekly high-intensity progressive resistance training plus impact loading improved lumbar spine BMD by 2.9 percent compared with a 1.2 percent decline in controls, and femoral neck BMD by 0.3 percent versus a 1.9 percent decline.¹³

The protocol elements that matter:

• Heavy compound lifts above 85 percent of one-rep max for 5 sets of 5 reps, with deadlift, overhead press, and back squat as the primary lifts.
• Supervised progression, since the safety profile of heavy load training in postmenopausal women requires technique that does not come from a video.
• Impact loading, which in LIFTMOR was jumping chin-ups with controlled drop landings. Patients who cannot tolerate impact can substitute weighted step-downs and box step-ups at sufficient height to generate ground reaction force.
• Twice per week, with at least 48 hours between sessions to allow remodeling.

The 8-month duration is the floor for measurable BMD changes at central sites. Patients in the LIFTMOR control arm lost 1.9 percent of femoral neck BMD across the trial window, which is meaningful regression even in a supervised research setting. The training arm did not just hold steady. It reversed direction at a measurable rate.

For GLP-1 patients, the practical translation is twice-weekly progressive heavy resistance work that escalates as the patient adapts, paired with impact loading once symptoms and joint health allow it. Frail older patients without prior resistance training experience need supervised progression. Patients who have already been training for years and are simply adding a GLP-1 to a stable program need to maintain training intensity through the weight-loss phase rather than dropping load alongside the body weight drop.

A GLP-1 patient who stops training during titration because energy is low and appetite is suppressed is removing the strongest available signal to retain both muscle and bone. The GLP-1 muscle preservation guide for women and muscle protection guide for men both treat this as the central protective intervention, and the bone case is the parallel rationale.

07Decision Tree: BMD Has Declined on Therapy

A follow-up scan shows a real change above the least significant change. The next step is not panic. It is a structured workup.

First, audit the nutrition floor. Most patients with measurable BMD decline on a GLP-1 are not hitting protein at 1.2 g/kg, not clearing 800 mg of calcium, or have a 25(OH)D below 25 ng/mL. The audit panel from the GLP-1 micronutrient floor catches most of these. If the patient is missing one or more of these targets, fix the floor first and rescan in 12 to 18 months.

Second, audit the training. A patient doing 30 minutes of walking and a yoga session per week is not loading the skeleton at a meaningful intensity. The clinical question is whether the patient is doing supervised progressive resistance training at least twice per week with heavy loads or whether they need a referral to a clinical exercise physiologist or a trained strength coach. The LIFTMOR-style protocol is the model.

Third, screen for secondary causes. Vitamin D deficiency, hyperparathyroidism, hyperthyroidism, celiac disease, glucocorticoid exposure, alcohol intake above 2 drinks per day, smoking, and undiagnosed multiple myeloma in older patients are the common findings on a workup. A patient losing bone faster than expected has often been losing bone for reasons other than the GLP-1, and the medication is the visible trigger that prompted the scan.

Fourth, weigh continuation. A patient with measurable BMD decline and a high cardiometabolic benefit from continued therapy is in a different position than a patient using the medication for cosmetic weight loss with marginal medical benefit. The continuation discussion includes the magnitude of the bone change, the patient's baseline fracture risk, the patient's response to the nutrition and training plan, and the availability of alternative weight-management strategies. There is no clean rule. The conversation is real medicine, and the GLP-1 prescriber and bone-density clinician have to share the decision.

Fifth, consider pharmacotherapy referral. Patients who cross into the osteoporosis range on T-score or who present with a fragility fracture during or after therapy belong with an endocrinologist, rheumatologist, or osteoporosis specialist. Pharmacotherapy choices for osteoporosis are outside the scope of this piece. The structural point is that the decision to start bisphosphonates, denosumab, romosozumab, or teriparatide does not belong with the GLP-1 prescriber alone. It belongs with the specialist who manages bone disease on a daily basis.

08After Discontinuation: What Happens to Bone

The off-ramp from GLP-1 therapy gets discussed almost entirely in terms of weight regain. The bone side gets less attention because the data are thinner. The mechanism prediction is that mechanical loading rises as body weight rises, intake usually rises, and the substrate available for remodeling improves. Whether bone density follows weight back up depends on what regained mass actually is, since fat-driven weight regain restores mechanical load without restoring lean tissue or training response.

The honest summary of the discontinuation literature is that long-term skeletal outcomes after GLP-1 cessation have not been well characterized in prospective DXA cohorts. The GLP-1 off-ramp piece covers the body-composition side. The bone-specific guidance is the same shape as everything else in this playbook: keep the protein floor, keep the resistance training, retest BMD 18 to 24 months after discontinuation if the patient had measurable decline on therapy, and treat the post-cessation period as a continuation of the structured plan rather than a return to baseline.

A patient who regains 60 percent of lost weight after stopping the medication has a different mechanical-load environment than the same patient at the lowest weight on therapy. That is partial relief from the unloading mechanism. It is not equivalent to restoring the skeleton, because the lean-tissue and training adaptations developed during weight loss are what carry forward. A patient who maintained training and protein through discontinuation is in a meaningfully different position than a patient who let both lapse alongside appetite restoration.

09Six-Line Operating Plan

Get a baseline DXA before starting therapy if you are postmenopausal, a man over 50 with any fracture risk factor, have a prior fragility fracture, are on chronic glucocorticoids, or have a secondary risk profile. Hold protein at 1.2 to 1.6 g/kg adjusted body weight, calcium at 1,000 to 1,200 mg per day food first, and vitamin D dosing guided by baseline 25(OH)D plus retesting. Train heavy resistance with progressive load twice per week, ideally with supervised technique, and add impact loading as joint health allows. Rescan at 18 to 24 months on therapy, with the least significant change at your facility as the noise floor below which apparent changes are not real. If BMD declines above the noise floor, audit nutrition, audit training, screen for secondary causes, and have the continuation conversation with the prescribing clinician before reaching for pharmacotherapy. After discontinuation, keep the floor and the training in place, because bone recovery follows lean tissue and mechanical load, not the scale.