Bone Remodeling

The building block is the osteon. Cortical bone is built from Haversian systems: concentric lamellae, three to seven layers deep, around a central Haversian canal that carries a capillary and a nerve. Volkmann's canals run transversely and connect adjacent Haversian canals, so the vascular network is continuous across the cortex.

Old cycles leave a record. Interstitial lamellae are the remnants of previous osteons stranded between intact ones, and cement lines are the basophilic scalloped borders marking the edge of an earlier resorption cavity. A cortical section is therefore a history of remodelling as much as a structure.

Porosity is the clinical variable. Excessive cortical remodelling (hyperparathyroidism, Paget disease) increases cortical porosity by creating more Haversian canals, and porosity over 20% significantly weakens bone against a normal of 5-10%. This is why primary hyperparathyroidism causes cortical bone loss at the radius in excess of trabecular bone loss at the spine.

The building block is the plate. Trabecular bone is an interconnected network of plates and rods 100-200 μm thick, separated by marrow spaces of 300-600 μm filled with haematopoietic or fatty marrow. Plate-like architecture (spine, ilium) is stronger than rod-like architecture (greater trochanter).

Surface is what remodelling needs. The surface-to-volume ratio is 20-30 m²/L, ten-fold higher than cortical bone, which provides a correspondingly large area on which BMUs can be activated. That single fact explains most of the difference in turnover rate between the two compartments.

Why the spine gives way first. Postmenopausal osteoporosis preferentially affects trabecular bone because oestrogen deficiency increases the remodelling rate, and trabecular bone has the surface area to express it as a large number of remodelling sites. Vertebral compression fractures (70% trabecular) therefore occur 5-10 years before hip fractures (50% trabecular).

Bone matrix is a store of growth factors, and resorption is what opens it. Osteoclasts release into the local microenvironment factors that were laid down with the matrix, producing a high local concentration exactly where bone has just been removed. That is why formation occurs precisely at the site of prior resorption.

Transforming growth factor-beta (TGF-β) is the most abundant growth factor in bone matrix, at a concentration of 200 micrograms per kilogram. Released during resorption, it is a chemoattractant for osteoblast precursors and draws osteoprogenitors to the resorption site; it stimulates preosteoblast proliferation but inhibits terminal differentiation.

Insulin-like growth factors I and II (IGF-I, IGF-II) are stored in matrix as complexes with IGF-binding proteins, and are released and activated during resorption. They stimulate osteoblast proliferation and collagen synthesis and enhance osteoblast differentiation and function.

The bone morphogenetic proteins BMP-2, BMP-4, BMP-6 and BMP-7 are present in the matrix as potent osteoinductive factors. They promote differentiation of mesenchymal stem cells towards osteoblasts and activate the Runx2 transcription factor.

The two cells talk to each other directly, and the conversation runs both ways. EphrinB2 on osteoclasts binds EphB4 on osteoblasts and stimulates osteoblast differentiation (forward signalling), while the same binding sends a reverse signal back into the osteoclast that inhibits its differentiation - coupling and negative feedback in one interaction.

Membrane-bound RANKL carries spatial information as well as instruction. Because osteoblasts present RANKL on the membrane and must contact RANK on the osteoclast precursor directly, the signal that activates the osteoclast also fixes where it is activated.

Semaphorins set the balance. Semaphorin 3A from osteoblasts inhibits osteoclast formation, and semaphorin 4D from osteoclasts inhibits osteoblast differentiation.

Geometry does some of the work. Osteoclasts and osteoblasts are physically colocalised within the BMU, share a blood supply that delivers both sets of precursors to the same place, and sit under a canopy that isolates the microenvironment, so an osteoblast arriving anywhere in that compartment arrives at the resorption site.

The cement line is both a full stop and a foundation. Reversal cells deposit a glycosaminoglycan-rich cement line that signals the end of resorption and the start of formation, and provides the surface to which osteoblasts attach.

The cavity itself guides deposition. Resorption leaves a concave defect that osteoblasts must fill, and the concavity favours osteoblast attachment, so the shape of the hole constrains where new bone goes.

The stimulus is fluid, not strain directly. Loading deforms the matrix, matrix deformation drives fluid through the lacunar-canalicular network, and the resulting shear stress acts on the osteocyte membrane and its processes. The fluid flow amplifies the underlying strain, which is how a cell buried in mineral becomes an exquisitely sensitive mechanosensor.

Sensing runs through four structures, which operate together:

• Primary cilium - a mechanosensitive organelle detecting fluid flow
• Integrins - connecting cytoskeleton to matrix, sensing deformation
• Gap junctions - connexin 43 channels coordinating the osteocyte network
• Mechanosensitive ion channels - allowing calcium influx

The signal then leaves the cell. Calcium influx triggers intracellular signalling, prostaglandin E2 is produced and nitric oxide released, and gene expression changes - sclerostin, RANKL and OPG.

Sclerostin is the loading-responsive brake on formation. The product of the SOST gene and expressed exclusively by osteocytes, it is a glycoprotein secreted into the lacunar-canalicular network that binds the LRP5/6 co-receptors on osteoblasts and inhibits Wnt/β-catenin signalling, reducing bone formation.

Loading releases the brake; unloading applies it. Mechanical loading reduces SOST expression, so there is less Wnt inhibition and more formation; unloading and immobilisation increase SOST expression, so there is more Wnt inhibition, less formation and bone loss.

The clinical correlations run in both directions. Exercise and loading decrease sclerostin and increase bone mass; bed rest and immobilisation increase it and cause loss; microgravity in spaceflight increases it and causes rapid loss; and sclerosteosis, in which SOST is mutated and no sclerostin is made, produces very high bone mass.

In a high-strain region the osteocyte turns both dials the same way. Sclerostin falls and Wnt signalling rises, while OPG rises and RANKL falls: formation is increased, resorption is suppressed, and mass increases. In a low-strain region the same two dials turn the other way, and bone is lost.

Microdamage is targeted by the death of the cells that sit in it. Fatigue loading creates microcracks, osteocytes at the damage site undergo apoptosis, and the dying cells signal to their neighbours to activate a BMU exactly there - targeted remodelling that removes the damaged bone.

The turnover state decides the drug class. Suppress excessive resorption in a high-turnover state, stimulate formation in a low-turnover state, and in every case correct the underlying cause - vitamin D deficiency, hyperparathyroidism, thyrotoxicosis - before or alongside bone-specific therapy.

Antiresorptive therapy

Anabolic therapy

Non-pharmacological management

Calcium and vitamin D are prerequisites, not treatments. Target 1000-1200 mg of calcium daily from diet plus supplement - dairy such as milk, yoghurt and cheese, leafy greens such as kale and bok choy, fortified foods such as soy milk and orange juice, and sardines with bones - and supplement the shortfall if dietary intake is under 700 mg daily, with calcium carbonate (40% elemental, taken with food) or calcium citrate (20% elemental, taken at any time and better absorbed in achlorhydria). Absorption saturates above a 500 mg single dose, so split larger requirements.

Vitamin D targets 25OHD greater than 75 nmol/L for bone health, PTH suppression and fracture reduction: maintenance is 1000-2000 IU daily, and deficiency (25OHD less than 25 nmol/L) is loaded with 50,000 IU weekly for 8-12 weeks first. Check 25OHD at baseline, 3 months after loading and then annually. Toxicity is rare and needs more than 10,000 IU daily for months, presenting as hypercalcaemia, hypercalciuria and nephrocalcinosis. None of this is optional: antiresorptive therapy reduces fracture risk only in a calcium and vitamin D replete patient, and anabolic therapy in a calcium-deficient patient causes hypocalcaemia and secondary hyperparathyroidism.

Exercise works through the osteocyte, and then through the fall. Mechanical loading reduces sclerostin expression, increases Wnt signalling and increases formation; unloading does the reverse. Walking, jogging, resistance training (squats, lunges, weightlifting) and impact activities (jumping, tennis) load the skeleton - swimming and cycling do not. The dose is 30-60 minutes on 3-5 days a week, with resistance training on 2-3 days. The BMD effect is modest, 1-3% over 12 months in postmenopausal women, and mostly prevents loss rather than building bone; the larger effect is on muscle strength, balance and falls, and exercise reduces fracture risk independent of BMD through better balance and coordination, periosteal apposition improving resistance to bending, and improvements in microarchitecture that BMD does not capture.

Fall prevention is the other half of fracture prevention. Remove tripping hazards, light the house, fit grab rails in the bathroom and non-slip mats; train balance (tai chi reduces fall risk by 40-50% in meta-analyses, along with yoga and single-leg stance work); correct vision annually and operate on cataracts; and review the drug chart, minimising sedatives and hypnotics, since more than four medications increases fall risk.

Smoking, alcohol and diet. Smoking increases remodelling - nicotine and cadmium stimulate osteoclasts and impair osteoblast function - and smokers carry a 30-40% higher fracture risk independent of BMD, through raised cortisol, reduced oestrogen, impaired calcium absorption and oxidative stress; after quitting, fracture risk falls towards non-smoker levels over 5-10 years and density improves by 1-2%. Alcohol should stay under 2 standard drinks daily for men and 1 for women, a standard drink being 12 oz of beer, 5 oz of wine or 1.5 oz of spirits and containing 14 g of ethanol; above 3 drinks daily there is direct osteoblast toxicity, reduced calcium absorption, secondary hyperparathyroidism, more falls and poorer nutrition, and a 2-fold fracture risk. Protein intake of 1.0-1.2 g/kg daily matters, and the elderly need the higher end of that range to prevent sarcopenia (below 0.8 g/kg is associated with bone loss and fracture), while high sodium (greater than 5000 mg daily) drives urinary calcium loss and more than 400 mg of caffeine daily - four cups of coffee - may do the same, while 1-2 cups daily is safe with adequate calcium intake.

Disease-specific management

Drug holidays

A bisphosphonate holiday is possible because the drug stays in the bone. Bisphosphonates bind the matrix with a half-life measured in years and go on suppressing remodelling after the last dose; denosumab is an antibody with a half-life of weeks whose effect is gone within 6 months, which is why it has no holiday and must be handed over to a bisphosphonate instead.

Treatment failure

Combination therapy

Simultaneous antiresorptive plus anabolic is not standard practice. It is expensive, has BMD rather than fracture data behind it, and adding a bisphosphonate to teriparatide is actively counterproductive - the suppressed remodelling space blunts the anabolic response. The one combination with supporting data is teriparatide plus denosumab, and it is reserved for very high fracture risk (T-score less than -4.0, multiple fractures, imminent risk). Sequential therapy remains the recommended approach for everyone else.

Transiliac bone biopsy

The operation is small; what makes it succeed or fail is everything around it. An unlabelled core answers half the question, a crushed core answers none, and a decalcified specimen destroys the labels that took three weeks to place. The indications are set out under Investigations; what follows is how it is done.

Tetracycline double-labelling

Labelling is mandatory for dynamic histomorphometry. Tetracycline binds calcium at mineralising surfaces and fluoresces yellow-green under ultraviolet light on unstained sections, so two courses separated by a known interval leave two lines whose separation is the mineral apposition rate. Without labels only the static parameters are available and the remodelling rate cannot be assessed at all.

1. Tetracycline 250 mg orally four times daily for 3 days (or doxycycline 100 mg twice daily)
2. Drug-free interval of 10-14 days, typically 12, long enough for measurable separation between the labels - and the interval must be recorded precisely, because an unknown interval invalidates the MAR calculation
3. Second course, tetracycline 250 mg four times daily for 3 days, starting exactly 10-14 days after the first course finishes
4. Wait 3-5 days after the second course, allowing the label to form fully, then biopsy

Total time from the first dose to the biopsy is 16-24 days. Where tetracycline cannot be used, the alternative fluorochromes are calcein green 15 mg/kg IV and alizarin red 20 mg/kg IV, given intravenously with a controlled interval; tetracycline is preferred because it is oral, widely available and strongly fluorescent.

The operation, step by step

Position and site. Lateral decubitus with the biopsy side up and hips and knees flexed to 90 degrees, which relaxes the abdominal wall and opens the access to the crest. Mark the site 2 cm posterior and 2 cm inferior to the anterior superior iliac spine: it targets trabecular-rich iliac bone with a high remodelling rate that is representative of the axial skeleton, while staying behind the dense anterior cortex and in front of the superior gluteal vessels. Either side is acceptable; use the opposite side to any previous biopsy or surgery, and in Paget disease biopsy affected bone if the disease is unilateral or a representative site if it is polyostotic.

Anaesthesia. Local anaesthesia suffices for most: lidocaine 1% with adrenaline, 15-20 mL infiltrated through skin, subcutaneous tissue and periosteum, with 5-10 minutes allowed for full effect. Conscious sedation (midazolam 1-2 mg IV, fentanyl 50-100 μg IV) is optional with saturation and respiratory monitoring and reversal agents (flumazenil, naloxone) available; general anaesthesia, propofol-based or inhalational, is reserved for severe anxiety, patient preference or children, and the procedure takes 20-30 minutes.

Exposure. A 5-8 mm stab incision with a number 11 blade perpendicular to the skin, deepened through fat to periosteum, then a periosteal elevator to clear soft tissue from a 1-2 cm circle of outer cortex. If the trephine will not engage an intact cortex, drill a 3-4 mm pilot hole and widen it to 7-8 mm with a larger drill or rongeur; in osteoporotic bone with a thin cortex the trephine can usually be advanced directly, rotating with firm steady pressure rather than jabbing, which fractures the cortex.

Taking the core. A Jamshidi trephine of 7-8 mm internal diameter and 10-15 cm length is standard (Minnesota and Islam needles are alternatives). Rotate clockwise with downward pressure, keeping the needle perpendicular to the bone surface so it does not exit the lateral cortex, and advance through outer cortex, trabecular bone and inner cortex until the give of the far cortex is felt. Rotate a full 360 degrees to shear the base, withdraw, and push the core out of the lumen with the obturator from the proximal end. The core should be 1-2 cm long and contain both cortices with the trabecular bone between them.

Handling the specimen. Place it immediately in fixative - 70% ethanol is preferred for histomorphometry, formalin is acceptable - and never let it dry, which is an artefact. Label with the patient identifier, the side, and the orientation, marking the anterior or superior end with a suture or ink.

Haemostasis, closure and aftercare

Haemostasis. Gauze soaked in 1:1000 adrenaline held with firm pressure for 5-10 minutes handles most bleeding. Bone wax, a pea-sized amount pressed into the exposed cortical edges, seals bleeding vascular channels; Gelfoam or thrombin-soaked gauze can be packed into the cavity and left in situ if oozing persists. Release the pressure, watch for a minute or two, and only then close.

Closure. A single interrupted 3-0 or 4-0 nylon suture closes the stab incision, or skin glue (Dermabond) or adhesive strips (Steri-Strips); no subcutaneous layer is needed at 5-8 mm. Dress with sterile gauze and an adhesive film (Tegaderm, Opsite), and wrap a compression bandage around the pelvis for 4-6 hours to reduce haematoma. Sutures come out at 7-10 days; glue sloughs at 7-14 days and needs nothing.

Aftercare. Observe for 30-60 minutes for bleeding or haematoma, check the dressing and observations, and confirm the patient can walk safely before discharge without syncope from the sedation or a vasovagal response. Avoid strenuous activity - heavy lifting, running, jumping - for 24-48 hours; walking and light activity are fine. Analgesia is NSAIDs (ibuprofen 400 mg six- to eight-hourly) or paracetamol 1000 mg six-hourly for 24-48 hours, with ice for 20 minutes every 2-3 hours on the first day; opioids are rarely needed. Warn about expanding haematoma greater than 5 cm, wound breakdown, and signs of infection.

Complications. Pain is the commonest, haematoma occurs in 5-10%, infection in less than 1% with sterile technique, and fracture is very rare even in osteoporotic bone.

Processing undecalcified bone

Staining and microscopy

Goldner trichrome is the workhorse. Weigert's haematoxylin stains nuclei dark purple-black, Ponceau-acid fuchsin stains mineralised bone red-orange and light green stains osteoid blue-green with pale green cytoplasm. On that section, osteoid appears as blue-green seams on the bone surface, osteoblasts as cuboidal cells with dark nuclei on those seams, and osteoclasts as large multinucleated cells in Howship's lacunae. It is the stain from which OS/BS, O.Th, ES/BS, Tb.Th and BV/TV are measured.

Von Kossa answers one question precisely: mineralised or not. Silver nitrate under ultraviolet light binds calcium and is reduced to black metallic silver, with nuclear fast red as counterstain, so mineralised bone is black and osteoid is pink and the mineralisation front is the sharp demarcation between them. In osteomalacia there are wide pink seams over 15 μm thick with a minimal black border; normally the seams are thin and the transition from pink to black is gradual.

Ultraviolet fluorescence microscopy measures the labels, and must use unstained sections because staining destroys tetracycline fluorescence. Excite at 365-400 nm and the labels fluoresce yellow-green. Two parallel lines separated by a dark space are the normal double label; a single line means either that the patient did not take the second course or that MAR is so low the labels have merged. Measure the interlabel distance with a calibrated ocular micrometer and divide by the interval: 8 μm over 12 days is a MAR of 0.67 μm/day, which is normal. BFR/BS follows as MAR multiplied by the labelled proportion of the perimeter and by 3.65 to convert to per cent per year.

Both ends of the turnover spectrum break bones, for opposite reasons: too much remodelling removes bone faster than it is replaced, and too little leaves microdamage unrepaired.

Medication-related osteonecrosis of the jaw

The definition is exposed bone in the maxillofacial region persisting beyond 8 weeks in a patient on antiresorptive or antiangiogenic therapy, without previous radiotherapy to the region.

The incidence tracks the dose, not the drug class. Oral bisphosphonates at osteoporosis doses carry a risk of 0.01-0.1% (1 in 1,000 to 1 in 10,000 patient-years) and denosumab at osteoporosis dose 0.04%, comparable to oral bisphosphonates. At oncology doses the picture changes: intravenous bisphosphonates run at 1-15%, ten-fold higher than osteoporosis dosing, and denosumab 120 mg monthly at 1-2%.

Why the jaw. Antiresorptives suppress turnover, so an extraction socket or traumatised mandible cannot heal; the exposed jawbone meets an exceptionally high oral bacterial load, infection supervenes, and the wound does not close. The potency and long half-life of zoledronate, around 10 years, contribute. The risk factors follow from that mechanism: invasive dental procedures (extraction, implants, periodontal surgery), duration of therapy beyond 4 years of bisphosphonates, glucocorticoids, diabetes, smoking and poor oral hygiene.

Prevention is dental, and it happens before the first dose. Examine and clear the mouth before starting an antiresorptive, complete necessary extractions and allow 4-6 weeks of healing before the drug is given, maintain scrupulous oral hygiene during therapy, and avoid elective invasive dentistry - endodontic treatment rather than extraction where the choice exists. If extraction is unavoidable, minimise trauma, close primarily, cover with antibiotics, and consider timing around the drug: stop a bisphosphonate 2 months before and resume 2 months after healing, or schedule the extraction before the next denosumab dose and restart once healed.

Recognise it early, from the symptoms rather than the exposed bone. Patients report a non-healing socket beyond 4 weeks with persistent pain and no mucosal cover, spontaneous bone exposure without any preceding procedure, persistent jaw pain, swelling or numbness from inferior alveolar nerve involvement, loose teeth or dentures that no longer fit, and purulent discharge from the gingiva. On examination look for exposed yellow-grey necrotic bone in the maxilla or mandible, mucosal ulceration over bone, gingival swelling and erythema, an intraoral or extraoral fistula, and - rarely, and late - pathological fracture of the mandible.

AAOMS (2014) staging drives the treatment, stage by stage:

• Stage 0 (at risk): no exposed bone but symptoms such as pain or altered sensation. Conservative management, oral hygiene, analgesia, close monitoring.
• Stage 1: exposed bone, no symptoms or infection. Chlorhexidine rinses, education, monitoring.
• Stage 2: exposed bone with pain and infection. Antibiotics (amoxicillin-clavulanate 875 mg twice daily for 2-3 weeks), chlorhexidine rinses, analgesia, superficial debridement of loose sequestra only.
• Stage 3: exposed bone with pain and infection plus pathological fracture, extraoral fistula or extensive bone involvement. Prolonged antibiotics, debridement, and teriparatide may be considered off-label.

Treat conservatively wherever possible: chlorhexidine rinses, antibiotics such as amoxicillin-clavulanate or doxycycline, pain control, and removal of loose sequestra only. Extensive surgical debridement worsens the non-healing. Stopping the drug may not help if it is a bisphosphonate, given the half-life. Teriparatide has enhanced healing in case reports and hyperbaric oxygen has been used as an adjunct on limited evidence.

Atypical femoral fracture

The ASBMR criteria require all five major features, and they describe a fracture that behaves nothing like an osteoporotic one:

• Location: subtrochanteric (below the lesser trochanter) or femoral diaphyseal
• Fracture line transverse or short oblique, less than 30 degrees from transverse
• Minimal or no comminution
• Complete fractures involve both cortices; incomplete fractures involve the lateral cortex only
• No trauma, or low-energy trauma equivalent to a fall from standing height

The minor features support the diagnosis without being required: a lateral cortex spike with periosteal beaking, prodromal thigh or groin pain for weeks to months, bilaterality in 20-30% (complete or incomplete on the other side), delayed healing, and generalised cortical thickening of the femoral diaphysis.

How common, and how it compares with the benefit. The incidence is 3-50 per 100,000 patient-years on bisphosphonates, rising with duration - rare before 5 years, around 100 per 100,000 after 10 years. Set against that, for every 100 hip fractures prevented by bisphosphonates approximately 1 atypical femoral fracture occurs, so the net benefit remains strongly positive.

The mechanism is the low-turnover mechanism. Chronic suppression beyond 5 years of bisphosphonates, or with denosumab, impairs targeted remodelling; microdamage accumulates in the lateral femoral cortex where tensile stress during gait is highest; a stress reaction progresses to an incomplete fracture of the lateral cortex, and then to a complete fracture if nothing is done. The cortical thickening that accompanies it is paradoxical - it is woven bone, not lamellar, and it does not make the femur stronger.

An incomplete fracture is an opportunity. Stop the antiresorptive immediately, protect weight-bearing with crutches, give calcium 1200 mg and vitamin D 2000 IU daily, consider teriparatide 20 mcg SC daily as an anabolic stimulus (off-label, limited evidence), and radiograph every 4-6 weeks to watch for progression. Nail it prophylactically if the radiographs progress, if pain persists despite conservative management, if the patient cannot comply with weight-bearing restriction, or if the lesions are bilateral - nail the symptomatic side and consider prophylactic nailing of the other.

A complete fracture needs an intramedullary nail - cephalomedullary for subtrochanteric, antegrade femoral for diaphyseal. Stop the antiresorptive, give teriparatide 20 mcg SC daily for 6-12 months (approved for this indication in some countries), consider bone stimulation as an adjunct on limited evidence, assess and treat the contralateral femur, and warn everyone that union takes 6-12 months against the 3-4 months of an ordinary femoral fracture.

Systemic complications of high-turnover remodelling

Sarcomatous transformation in Paget disease

Rare, in less than 1% of patients with Paget disease, and devastating. Pagetic bone transforms to osteosarcoma, fibrosarcoma or chondrosarcoma, the risk concentrated in long-standing polyostotic disease of more than 10 years, extensive skeletal involvement, and skull or facial bone involvement.

Suspect it when stable pagetic bone starts hurting. New severe bone pain in previously stable pagetic bone, a rapidly enlarging soft tissue mass, a pathological fracture through pagetic bone, or constitutional symptoms of weight loss and fatigue.

Investigate and treat it as the sarcoma it is. Radiographs show aggressive lytic destruction of pagetic bone with a soft tissue mass, cortical breakthrough and, for osteosarcoma, sunburst periosteal reaction; BSAP may climb further above the pagetic baseline; MRI shows a large soft tissue mass with marrow replacement and enhancement; and biopsy is required, showing malignant spindle cells in osteosarcoma, malignant cartilage in chondrosarcoma or malignant fibrous tissue in fibrosarcoma arising from pagetic bone. Treatment is wide resection where feasible, whether by limb salvage or amputation, with neoadjuvant and adjuvant chemotherapy on the MAP protocol (methotrexate, doxorubicin, cisplatin) as for primary osteosarcoma, and palliative radiotherapy if unresectable, while bisphosphonates continue to control the untransformed pagetic bone. The prognosis is very poor: 5-year survival is less than 10%, and metastasis to lung occurs early.

Immobilisation hypercalcaemia

Immobilisation uncouples remodelling the wrong way. Bed rest, spinal cord injury or prolonged casting increases resorption and decreases formation, and where turnover is already high - Paget disease, hyperthyroidism, an adolescent in a growth spurt - the additional resorption tips calcium release over the threshold.

The two classic scenarios. An adolescent in traction or a cast after a femoral fracture who becomes nauseated, confused and polyuric 2-3 weeks after injury; or a patient with Paget disease made bedbound by a hip fracture and its surgery.

Investigation and treatment. Calcium greater than 2.75 mmol/L (greater than 11 mg/dL) with suppressed PTH, markedly elevated CTX reflecting both the immobilisation and the pre-existing high turnover, and an elevated ionised calcium confirming true hypercalcaemia. Hydrate intravenously to restore volume and increase renal calcium excretion, and mobilise as early as possible - loading the skeleton is the most effective treatment there is. Add a bisphosphonate (pamidronate 60-90 mg IV or zoledronate 4 mg IV) if calcium exceeds 3.0 mmol/L or immobilisation will continue, with calcitonin as a bridge for rapid effect. Avoid thiazides, which reduce renal calcium excretion, and use furosemide only after adequate hydration. Prevention is early mobilisation, adequate hydration, and prophylactic bisphosphonate in the highest-risk patients.

Teriparatide and transient stress fracture risk

The paradox is that the drug that builds bone may weaken it briefly first. Teriparatide increases remodelling on both arms even though the net balance favours formation, and the enlarged remodelling space - the volume of bone in active resorption and formation - is mechanically weaker than quiescent bone. After 6-12 months new bone fills that space and strength rises.

The evidence is conflicting. Some randomised trials report a small increase in stress fractures in the first 6 months, a relative risk of 1.5-2.0; others show no increase. Overall fracture risk falls, with vertebral fracture risk reduced by 65% against placebo.

What to do with that. Warn the patient of a theoretical transient risk, avoid high-impact activity for the first 3 months while formation catches up, and take new bone pain seriously as a possible stress reaction. If a stress fracture does occur during treatment, continue the teriparatide - it promotes healing - because the overall benefit outweighs the transient risk.

Monitoring exists to confirm the drug is working, to find the patient in whom it is not, and to catch the complications early. What is measured and how often depends on the drug class, the baseline fracture risk and the comorbidities: bone mineral density, turnover markers, calcium and vitamin D, incident fractures and adverse events.

Before starting an antiresorptive

Density and biochemistry first. Measure baseline BMD at lumbar spine, total hip and femoral neck with T-scores for postmenopausal women and men over 50, which sets the comparator for every later scan. Then the biochemistry: corrected calcium must be normal, since hypocalcaemia is a relative contraindication and is corrected before starting; 25OHD should exceed 50 nmol/L before the first dose, supplemented with a loading course if it does not; creatinine and eGFR decide the drug, because bisphosphonates are contraindicated below 30-35 mL/min/1.73m² while denosumab is safe in renal impairment but carries a higher hypocalcaemia risk; PTH if calcium is borderline or vitamin D very low, to exclude hyperparathyroidism; and phosphate, which is usually normal but low in osteomalacia.

Markers, teeth and previous fractures. A baseline CTX allows the biochemical response to be confirmed later, though P1NP is less often measured at baseline when an antiresorptive is planned. Dental examination and clearance come before the first dose, with any necessary extractions completed, and existing dental problems documented. Record the prevalent fracture burden - vertebral fractures on a lateral thoracolumbar radiograph, peripheral fractures by history - because that is the baseline against which treatment failure will be judged.

Follow-up on antiresorptive therapy

At 3 months check serum calcium (particularly after denosumab or an intravenous bisphosphonate) and vitamin D, optionally CTX, which should have fallen 50-70% from baseline, and review for adverse events (gastrointestinal symptoms with oral bisphosphonates, myalgias), new fractures and - with an oral bisphosphonate - whether it is actually being taken correctly.

At 6-12 months confirm that suppression is maintained if CTX is being followed, target under 400 pg/mL, and review fractures, adverse events and compliance. With denosumab the single most important thing is that the next dose is on schedule: it is not a drug that tolerates delay.

At 12-24 months repeat the DEXA, typically at 2 years. BMD should be stable or increased - expect an increase of 2-5% at the spine and 1-3% at the hip in the first 2 years - and a fall greater than the least significant change of 3-4%, or an incident fracture, defines the non-responder. Calcium and vitamin D are checked annually from here.

Years 3-5 need less: DEXA every 2-3 years while stable, annual clinical review for fractures, adverse events and medication changes, annual calcium and vitamin D, and no routine markers if density is stable and there have been no fractures.

Beyond 5 years the question becomes whether to continue, and the drug holiday algorithm under Management answers it. If therapy continues, keep the DEXA interval at 2-3 years with annual review, and watch specifically for thigh pain and dental problems. If a holiday is taken, markers every 6-12 months and DEXA at 1-2 years, resuming if either moves.

Monitoring anabolic therapy

Hypocalcaemia after denosumab or intravenous bisphosphonate

The risk is higher than with oral bisphosphonates because suppression of resorption is rapid and potent, and bone resorption is a major source of serum calcium. The patients who get into trouble are those with vitamin D deficiency (25OHD less than 30 nmol/L), chronic kidney disease with eGFR less than 45 and impaired vitamin D activation, hypoparathyroidism, malabsorption or previous parathyroidectomy. Calcium reaches its nadir 7-10 days after the dose.

The presentation runs from nothing to tetany. Mild hypocalcaemia at 2.0-2.1 mmol/L is asymptomatic; below 1.9 mmol/L expect perioral numbness, paraesthesiae of hands and feet, muscle cramps, carpopedal spasm and, at the extreme, seizures, with Chvostek's sign (a facial twitch on tapping the facial nerve) and Trousseau's sign (carpopedal spasm on inflating a blood pressure cuff) on examination.

Prevention is measurement before administration. Check vitamin D before denosumab or an intravenous bisphosphonate and, if it is under 50 nmol/L, load with 50,000 IU weekly for 4-8 weeks before the drug is given. Ensure 1200 mg of calcium daily. In CKD with eGFR less than 45 use calcitriol 0.25 mcg daily rather than cholecalciferol, which needs renal activation, starting 1-2 weeks beforehand. Then check calcium at 7-10 days in anyone high-risk.

Treatment scales with severity. Mild and asymptomatic: oral calcium carbonate 1-2 g of elemental calcium three times daily with calcitriol 0.5 mcg twice daily, rechecking in 3-5 days. Moderate, symptomatic or severe (below 2.0 mmol/L): intravenous calcium gluconate 10%, 10-20 mL in 50-100 mL of saline over 10-20 minutes, then an infusion of 1-2 mg/kg/hr of elemental calcium with oral calcium and calcitriol alongside, checking calcium every 4-6 hours and titrating to 2.0-2.2 mmol/L, continuing intravenous calcium for 24-48 hours before transitioning to oral. Check magnesium, since hypomagnesaemia impairs PTH secretion, and correct it below 0.7 mmol/L. Do not withhold the next denosumab dose: the hypocalcaemia is transient, and the answer is adequate calcium and vitamin D beforehand.

An incident fracture on therapy

Not every fracture on treatment is a treatment failure. A low-energy fragility fracture - a fall from standing height or less, at the vertebrae, hip, distal radius or proximal humerus - suggests inadequate treatment and requires escalation. A high-energy fracture from a motor vehicle accident or a fall from height does not, and therapy continues unchanged. Timing matters too: a single fracture after 6-12 months may reflect microarchitectural damage that predated treatment, while multiple fractures, or a fracture within the first 3 months, point to severe disease or non-response.

Investigate before changing anything. DEXA if it has not been done within 12 months, CTX and P1NP for the biochemical response, a secondary osteoporosis screen, an adherence assessment, and a lateral thoracolumbar radiograph, which frequently finds additional asymptomatic vertebral fractures and reclassifies the patient as very high risk.

Then escalate by where the patient already is. A fracture on an oral bisphosphonate in an adherent, vitamin D replete patient moves up to a more potent antiresorptive - zoledronate 5 mg IV annually or denosumab 60 mg SC six-monthly. A fracture on intravenous bisphosphonate or denosumab, with the same provisos, moves across to anabolic therapy and then back to denosumab for consolidation. Multiple fractures or very high risk (T-score less than -3.5, prevalent vertebral fractures) go straight to anabolic therapy. Where the fracture is explained by non-compliance, vitamin D deficiency or a secondary cause, fix that first rather than escalating.

Surveillance for atypical femoral fracture

Prodromal pain precedes the complete fracture in 70% of cases, which is the opportunity for prevention.

Educate proactively at year 5. Explain that the risk is real but rare - approximately 1 in 1,000 with more than 5 years of therapy - and ask the patient to report new thigh, groin or hip pain immediately: dull, aching, insidious, worse with weight-bearing and better with rest. Bilateral thigh pain deserves particular suspicion.

Ask the question at every annual review from year 5. If the answer is yes, palpate the lateral femoral shaft for tenderness, perform a single-leg stance test, confirm that hip range of motion is full - which distinguishes the pain from hip arthritis - examine the other thigh, and radiograph the entire femur in two planes from hip to knee, looking for lateral cortex thickening, periosteal beaking and a lucent line in the lateral cortex. Radiograph the other femur too, since 20-30% are bilateral. If the radiographs are negative and suspicion remains high, MRI shows the marrow oedema of a stress reaction before any cortical line appears.

Radiographic surveillance of the asymptomatic patient is not standard, though some recommend femoral radiographs at year 5 and then every 2-3 years on continued bisphosphonates, particularly in the higher-risk groups: Asian ethnicity, glucocorticoid use, more than 10 years of therapy, and the femoral geometry of a high neck-shaft angle with anterolateral bowing.

If an incomplete fracture is found, stop the bisphosphonate, refer to orthopaedics that week, protect weight-bearing with toe-touch on the affected side, give calcium 1200 mg and vitamin D 2000 IU daily, discuss teriparatide as an adjunct, and radiograph every 4-6 weeks while the decision between prophylactic nailing and conservative management is made on the extent of the lesion, the severity of pain and the patient. Bilateral incomplete fractures raise the question of bilateral prophylactic nailing. Do not restart an antiresorptive until the fracture has healed at 6-12 months, and prefer denosumab thereafter, though atypical fractures have been reported with denosumab too.

Continuing or stopping long-term therapy

Fracture risk reduction is the outcome that matters; BMD is the surrogate that correlates with it, the turnover markers confirm the drug is working, and tolerability decides whether any of it continues.

Antiresorptive therapy

Anabolic therapy

What the head-to-head trials show

Romosozumab against alendronate (ARCH). Vertebral fracture reduction at 12 months favoured romosozumab, a 75% relative reduction against alendronate, and hip fracture at 24 months was reduced 38% by romosozumab followed by alendronate against alendronate throughout. Spine BMD rose 13% with romosozumab against 5% with alendronate at 12 months. Romosozumab is therefore superior for high-risk patients, with the cardiovascular safety signal as the qualifier.

Romosozumab against teriparatide was compared head-to-head in STRUCTURE, in which romosozumab gave the greater BMD gain.

Teriparatide against alendronate has no large randomised comparison, only indirect comparison through meta-analysis: 65% vertebral RRR against placebo for teriparatide and 47% for alendronate suggests superiority but does not prove it. BMD at 24 months is similar at the spine (+10% against +8%), and alendronate is the better of the two at the hip (+6% against +4%). Teriparatide is therefore preferred for severe vertebral osteoporosis and alendronate is adequate for moderate disease or hip-predominant disease.

Denosumab against alendronate has limited direct evidence. DECIDE showed spine BMD +5.3% with denosumab against +2.2% with alendronate at 12 months, and there is no adequately powered head-to-head fracture trial; indirect comparison suggests similar vertebral reduction and possibly better hip protection with denosumab, given the larger hip BMD gain. Denosumab is preferred where greater BMD gain is needed or oral bisphosphonates are not tolerated, with the rebound risk as the qualifier.

Zoledronate against alendronate has no head-to-head trial either. Vertebral reduction is similar (70% against 47%, but in different trial populations), zoledronate may be better at the hip (41% RRR against no significant reduction for alendronate in primary prevention), and spine BMD is higher at 3 years (+12% against +8%). Zoledronate is preferred where compliance is a concern, higher potency is wanted, or oral dosing is not tolerated.

Who responds, and who does not

The patient factors that predict a good response. High baseline turnover - there is more remodelling to suppress, so a high CTX or P1NP predicts a better antiresorptive response and low baseline turnover predicts a poor one. Younger age, since osteoblast function declines and patients under 65 gain more than those over 75. Recent menopause, within 5 years, when turnover is highest. Adequate vitamin D, above 75 nmol/L, against less than 50. And primary rather than secondary osteoporosis.

The early biochemical response predicts the later densitometric one. On anabolic therapy, a greater P1NP rise at 1-3 months predicts a greater BMD gain at 12-24 months, and a rise under 50% should prompt investigation. On antiresorptive therapy, greater CTX suppression at 3-6 months predicts greater BMD gain, and a reduction under 30% suggests non-response.

The factors that predict fracture despite treatment. A baseline T-score below -3.5, where risk persists through therapy and anabolic treatment may be required; more than two prevalent vertebral fractures, which predict further fractures even on antiresorptives and argue for anabolic therapy first-line; age over 80, with higher fracture risk, lower BMD response and more falls; and a history of more than two falls a year, where the risk is driven by falling rather than by density and needs a falls strategy alongside the drug.

Genetic prediction is not yet clinical. LRP5 polymorphisms in the Wnt pathway have been associated with BMD response to bisphosphonates and vitamin D receptor variants with response to supplementation and antiresorptives, but neither is used outside research.

Long-term bisphosphonate extension trials

FLEX - the FosamaX Long-term Extension - asked whether 10 years beats 5. Women who completed 5 years of alendronate in FIT were randomised to continue or to placebo for 5 more years. Continued alendronate held lumbar spine BMD stable with no further increase, while the discontinuation group lost 2-3% of spine BMD over the 5 years but stayed above their original baseline; hip BMD was stable in both, on the residual effect of the first 5 years. Clinically, continued therapy reduced clinical vertebral fractures by 55% (2.4% against 5.3%, NNT 34), with no significant difference in non-vertebral fractures and no difference in morphometric vertebral fractures - which suggests 5 years is enough for low-risk patients. Continue beyond 5 years if fracture risk is high (prior vertebral fracture, very low BMD); consider a holiday if it is not.

HORIZON extension asked the same of zoledronate. Women who completed 3 years were randomised to 3 more or to placebo. Continued zoledronate maintained spine BMD while the discontinuation group lost 1-2%, less than after alendronate because the residual effect is stronger. Morphometric vertebral fractures were reduced 50% by continuing (14% against 25%, NNT 9 in the high-risk subset), again with no difference in non-vertebral fractures. Continue to 6 years in very high-risk patients; stopping at 3 years is reasonable in lower-risk patients, who remain protected by the residual effect.

The principle underneath both. Bisphosphonates accumulate in bone and provide a residual anti-fracture effect for 2-5 years after discontinuation, longer for the more tightly bound drugs, which is what makes a holiday possible at all.

Denosumab: benefit that does not stop, and does not persist

FREEDOM extension followed patients to 10 years, and unlike bisphosphonates the BMD did not plateau: lumbar spine rose to +21% from baseline at 10 years, against +13% at 3 years, and total hip to +9% against +6%, a continued near-linear increase. Fracture rates stayed low throughout, with annual vertebral fracture rates of 1-2% and non-vertebral 1-3%, and no loss of anti-fracture efficacy over time. Safety was favourable: atypical femoral fracture at 0.8 per 10,000 patient-years and osteonecrosis of the jaw at 5.2 per 10,000 patient-years, slightly above bisphosphonates.

The off-treatment data are the other half of the story. Participants who stopped after 2-7 years lost about 6% of bone mineral density, 50-100% of their gains, within 12 months, their turnover markers rebounded above baseline, and 10-15% sustained multiple vertebral fractures 7-20 months after the last dose. Denosumab is therefore effectively lifelong therapy unless it is handed over to a bisphosphonate.

Sequence changes the outcome

Teriparatide then antiresorptive is the sequence that works. After 24 months of teriparatide, alendronate holds spine BMD and the hip continues to gain slightly (+1-2% over 2 years), while denosumab increases both spine and hip further (+3-5% over 2 years), and fracture protection is sustained.

Bisphosphonate then teriparatide is the sequence that does not. Patients pretreated with a bisphosphonate gain about half as much: spine +6% against +12% at 18 months in bisphosphonate-naive patients. Suppressed turnover leaves no remodelling space for teriparatide to build into, and the effect is worse with longer exposure and with the more potent drugs, zoledronate blunting more than alendronate. If a patient on a bisphosphonate needs escalation, either stop it 6-12 months before starting teriparatide, or use romosozumab, which is less affected by prior bisphosphonate exposure.

Romosozumab then denosumab is the optimal sequence for severe osteoporosis and the one FRAME tested: spine BMD +17% at 12 months with romosozumab rose to +18% at 24 months on denosumab, with fracture reduction maintained and enhanced. Romosozumab then alendronate is not: in ARCH, spine BMD of +13% at 12 months fell back to +11% at 24 months after the switch, by a mechanism that remains unknown, possibly because romosozumab's antiresorptive effect occupies the bisphosphonate binding sites.

Simultaneous teriparatide and denosumab outperforms either alone on BMD. In DATA, spine BMD rose 9.1% with the combination against 6.2% with teriparatide alone at 12 months and hip BMD 4.2% against 0.8%; over 24 months spine BMD rose 10% with teriparatide, 6% with denosumab and 17% with both, an additive effect explained by teriparatide driving formation while denosumab prevents the concomitant rise in resorption. It is expensive, has no fracture data, and is not standard practice - reserved for research or extreme cases such as pregnancy-associated osteoporosis with multiple fractures.

Weighing benefit against harm