Osteoporosis treatment has quietly become a question of order rather than a question of which single drug is best.
A meta-analysis published in Osteoporosis International in 2026 pooled 39 studies and found that the same two medicines produce very different bone density gains depending on which one comes first.[1]
Osteoporosis treatment is now a sequence, not one prescription
Most people with osteoporosis will take more than one drug over their lifetime. Some medicines are licensed for a fixed period, others lose effect, and some cannot be stopped safely. So the practical question is not which drug is best but which drug follows which.
Until recently that question had no pooled answer. Individual trials compared one sequence against placebo or against continuing the first drug, and the results were scattered across two decades of literature.
What the 2026 meta-analysis of 39 studies found
Nayak and Greenspan searched PubMed, Embase and the Cochrane Library and included 39 primary studies. They reported the BMD change produced by the second medicine in each sequence, which is the number a patient actually experiences when switching.[1]
The largest gain came from bisphosphonates followed by one year of romosozumab: 10.1% at the lumbar spine (95% CI 9.9 to 10.4), 3.1% at the femoral neck (2.9 to 3.4) and 3.1% at the total hip (2.7 to 3.5).[1]
Romosozumab followed by one year of denosumab produced moderate gains of 4.0% at the spine (1.2 to 6.7), 2.2% at the femoral neck (0.5 to 3.8) and 2.8% at the total hip (0.8 to 4.9). Teriparatide switched to denosumab gave similar numbers.[1]
Bisphosphonates followed by denosumab gained 3.5% at the spine (2.8 to 4.2), 1.5% at the femoral neck (1.2 to 1.9) and 2.1% at the total hip (1.8 to 2.3). Bisphosphonates followed by teriparatide gained 5.1% at the spine (4.4 to 5.9) but showed no significant hip gain.[1]
Why the hip result matters more than the spine result
Spine density responds to almost every osteoporosis drug, so a large spine number is not by itself impressive. Hip fractures cause most of the disability and mortality, and hip density is much harder to move.
That is why the bisphosphonate-to-teriparatide sequence stands out in the wrong direction. It produced a respectable 5.1% at the spine while leaving the hip unchanged, which is a reminder that a single headline percentage can hide the site that matters most.
Stopping denosumab is the move that can backfire
Denosumab is not a drug that can simply be stopped. When injections end, bone turnover rebounds above the original baseline and density falls quickly, and multiple vertebral fractures have been reported in that window.
For that reason denosumab is treated as an endpoint rather than a stepping stone. Transitioning from long-term denosumab to an anabolic drug also carries this rebound risk, which is one reason the 2026 papers keep separating treatment-naive patients from switched patients.[2]
What the 2026 Yonsei study adds about individual response
A multicentre retrospective study published in the Yonsei Medical Journal in September 2026 followed patients who received 12 monthly romosozumab injections and then moved to denosumab. It asked which patients respond well.[2]
Three factors predicted the density response: baseline bone density, prior osteoporosis treatment, and nutritional status as reflected by total lymphocyte count. Patients who had never been treated before gained more than patients who were switched from another drug.[2]
Eight fractures occurred during follow-up, one in the treatment-naive group and seven in the switched group. The authors state plainly that this difference did not reach statistical significance, so it should be read as a signal to study rather than a finding.[2]
Who should not receive romosozumab
The Endocrine Society guideline recommends romosozumab for up to one year in postmenopausal women at very high fracture risk, and specifically advises against it for women at high cardiovascular risk, including anyone with a prior heart attack or stroke.[3]
This matters in Korea, where romosozumab reached 7.4% of the osteoporosis drug market with sales rising 59% in 2023, and denosumab grew 957.6% between 2018 and 2023 to overtake bisphosphonates.[4] Rapid uptake makes the contraindication easier to overlook.
Bone loss also has a nutritional floor that no injection replaces. Protein intake in particular has been shown to matter more after fifty, which we covered in a separate article on protein intake per day.
At which number does osteoporosis start
Osteoporosis is hard to notice. Bone thins without pain, and for most people the diagnosis arrives only after a fall or a compression fracture has already happened.
Diagnosis uses the T-score, which compares measured bone density against the average for a young adult of the same sex. Above -1.0 is normal, -1.0 to -2.5 is low bone mass, and -2.5 or below is osteoporosis. A previous fracture raises risk sharply on top of that.
The T-score alone does not decide when to start medication. Age, prior fracture, steroid use and a parent with a hip fracture are combined into a ten-year fracture risk estimate, and the decision follows from that.
The three drug classes act at different points
Bisphosphonates bind to the bone surface and are taken up by the osteoclasts that dissolve bone, reducing their activity. Because the drug stays in bone, some effect persists after dosing stops.
Denosumab is an antibody that captures the signalling protein telling the body to make more osteoclasts. It does not accumulate in bone, so injections must continue every six months, and stopping releases the suppressed signal all at once.
Romosozumab blocks sclerostin, a protein that holds bone formation back. Removing that brake raises building while formation-independent resorption is suppressed, but the state lasts roughly a year, which is why dosing is capped at 12 months.
What the drugs cannot supply
No drug works well without raw material. If calcium and vitamin D are short, an anabolic agent has nothing to build with. The Yonsei finding that nutritional status predicted response points the same way.[2]
Weight-bearing movement belongs in the same column. Walking, stair climbing and resistance work load the skeleton, and they also build the strength and balance that prevent the fall in the first place.
Conditions that pull bone density down need attention too: smoking, heavy drinking, long-term steroid use and aggressive weight loss. These remain after treatment starts and quietly reduce whatever the drug achieves.
How long to treat, and when to pause
Bisphosphonates are commonly reviewed after five years of oral therapy or three years of infusions. Patients whose risk has fallen may pause; those whose risk remains continue.
A pause is possible because the drug stays in bone. Density does not drop immediately after stopping; it declines gradually over several years.
Denosumab does not allow this. As covered above, bone loss accelerates after stopping, so a decision to end it must include a plan to move onto another agent such as a bisphosphonate.[2]
Romosozumab ends at 12 months by design. If nothing follows it, much of the density gained during that year is given back, which is exactly why the meta-analysis calculated results for the second drug separately.[1]
A prior fracture narrows the options
Someone who has already had a spine or hip fracture is classed as very high risk. For that group the aim shifts from raising density slowly to lowering risk within one to two years.
That is why an anabolic agent or romosozumab first, followed by an antiresorptive, is the recommended order in this group. Starting with long-term bisphosphonate therapy and moving to an anabolic later has been reported to blunt the early response.[1]
The Yonsei finding that treatment-naive patients gained more fits the same picture. It was a retrospective observation, though, so it is not yet strong enough to set sequence on its own.[2]
Checks before and during treatment
Kidney function, serum calcium and vitamin D are checked before starting. Giving denosumab or romosozumab while calcium is low can cause hypocalcaemia, so that has to be corrected first.
Dental plans are worth mentioning in advance. Osteonecrosis of the jaw has been reported rarely in people on long-term antiresorptive therapy, so a planned extraction or implant may change the timing.
Response is usually checked with a repeat density scan about a year later. It should be done on the same machine, and repeating it too soon means measurement error exceeds the real change.
How the density scan works and what it shows
DXA uses two x-ray energies to separate bone from soft tissue. The scan takes five to ten minutes lying on a table, and the radiation dose is lower than a standard chest x-ray.
Measurement is usually taken at the lumbar spine and the femur. The two sites often disagree, and when they do the lower value is used for the diagnosis.
Compression fractures or advanced degenerative change push the spine reading falsely high. Radiographic findings are read alongside the number, and another site such as the forearm may be added.
Screening policies differ by country, but a previous fracture or a clear risk factor is generally reason to measure earlier than the routine screening age.
Does more density mean fewer fractures
Density and fracture risk move together but not one for one. A large share of the fracture reduction seen with treatment is not explained by the density change alone.
Bone strength also depends on microarchitecture and turnover rate. That is the usual explanation for antiresorptive drugs cutting fractures without producing large density gains.
So when reading trial results it is worth checking whether fracture rates were reported at all, rather than stopping at the percentage. Most sequential studies do not yet have that data.[1]
| Sequence | BMD change on the second drug (1 year) |
|---|---|
| Bisphosphonate then romosozumab | Spine 10.1%, femoral neck 3.1%, total hip 3.1% — the largest pooled gains |
| Romosozumab then denosumab | Spine 4.0%, femoral neck 2.2%, total hip 2.8% — moderate but consistent |
| Bisphosphonate then denosumab | Spine 3.5%, femoral neck 1.5%, total hip 2.1% — smaller, widely used |
| Bisphosphonate then teriparatide | Spine 5.1%, no significant gain at either hip site |
| Denosumab then nothing | Not a sequence. Turnover rebounds and vertebral fractures have been reported |
Limitations
Almost all of these numbers are bone density, not fractures. Density predicts fracture risk but does not measure it, and few sequential studies were powered for fracture outcomes.
The Yonsei study was retrospective and observational, so its predictors describe association rather than cause. Pooled estimates also mix populations with different fracture histories, ages and baseline treatment, which is why the confidence intervals for some sequences are wide.
References
- Nayak S, Greenspan SL. A systematic review and meta-analysis of sequential treatment strategies for osteoporosis. Osteoporosis International, 2026;37(1):1-13.
- Kuwabara A, Inage K, Yamashita M, et al. Impact of Prior Treatment and Predictive Factors in Sequential Romosozumab-Denosumab Therapy for Severe Osteoporosis. Yonsei Medical Journal, 2026 Sep;67(9):680-685.
- Endocrine Society. Pharmacological Management of Osteoporosis in Postmenopausal Women, guideline resources.
- Five-Year Sales Trends of Osteoporosis Medications in Korea: A Market Analysis Based on IMS Health Sales Audit Data (2018-2023).
