Giant Cell Tumor of Bone

Giant cell tumour (GCT) of bone is a benign but locally aggressive neoplasm. The name describes the wrong cell: the abundant osteoclast-like multinucleated giant cells are reactive, and the neoplastic component is the population of mononuclear spindle-shaped stromal cells that recruits them. Despite benign histology it recurs locally in 10-50% depending on treatment and metastasises to the lung in 1-5%.

Definition. A primary bone tumour composed of three cell populations: mononuclear histiocyte-like cells, mononuclear spindle-shaped stromal cells carrying the H3F3A mutation, and multinucleated osteoclast-like giant cells with 20-100 nuclei per cell. It characteristically arises in the epiphysis of a long bone after physeal closure.

Why it matters. Five things make this tumour examinable: its local aggression demands meticulous surgical technique; its eccentric epiphyseal position threatens the joint; pathological fracture complicates 10-20% of cases; the RANK-RANKL pathway gives it a targeted therapy, denosumab, whose outcomes are controversial; and it can metastasise to the lung despite benign histology.

Who. Peak age is 20-40 years (range 15-65), after physeal closure, with a slight female predominance of 1.5:1 and a higher incidence in Asian populations. It is rare in children: under 2% occur before physeal closure.

How common. Annual incidence is about 1 per million. GCT makes up 5% of primary bone tumours, 20% of benign bone tumours, and 18% of all bone tumours biopsied.

Where. About 85% occur around the knee.

• Distal femur - 30%, the most common single site
• Proximal tibia - 25%
• Distal radius - 10%
• Sacrum - 5-10%

The neoplastic cell. The stromal cell carries an H3F3A G34W mutation (rarely G34L), a histone H3.3 alteration that drives histone modification and gene dysregulation, and the result that matters is overexpression of RANKL. Behjati showed the mutation is confined to the stromal cells and not detectable in the osteoclasts or their precursors, which is the molecular proof that the giant cells are a reactive population recruited by stromal RANKL.

RANKL and the giant cell. RANKL (receptor activator of nuclear factor kappa-B ligand) is the key cytokine for osteoclast differentiation. Secreted by the neoplastic stromal cells, it binds RANK on osteoclast precursors of monocyte-macrophage lineage and stimulates their fusion into multinucleated giant cells, creating a local microenvironment that favours bone resorption. The giant cells then resorb bone: the tumour expands eccentrically in the epiphysis, reaches the subchondral bone, and in grade III eventually breaks through the cortex.

View larger

Denosumab. A fully human monoclonal antibody against RANKL. By blocking RANKL it prevents RANK activation on osteoclast precursors, so giant cell recruitment and bone resorption stop, and the tumour ossifies and shrinks. It does not eliminate the neoplastic stromal cells, which may lead to recurrence when treatment stops.

Histology

Three cell populations. The giant cells have 20-100 nuclei, express RANK, and are distributed evenly through the tumour. The stromal cells are spindle-shaped, carry the H3F3A mutation, produce RANKL, and have ovoid nuclei that look like the giant-cell nuclei. The histiocyte-like cells are round to ovoid, of monocyte-macrophage lineage, and are the osteoclast precursors.

What the slide shows. Sheet-like growth with no fibrous stroma, and giant cells scattered evenly rather than clumped, which is the diagnostic feature. Mitoses are frequent in the stromal cells, not the giant cells. Haemorrhage is common and produces ABC-like areas and golden-brown haemosiderin; older lesions contain lipid-laden foam cells.

View larger

View larger

Telling it from osteosarcoma with giant cells. Four features separate benign GCT from a malignant giant-cell-rich tumour:

• Uniform nuclear features - the stromal nuclei match the giant-cell nuclei
• No osteoid or malignant bone production
• Sheet-like distribution without necrosis
• Mitoses in the stromal cells are normal in form, not atypical

Malignancy in GCT

About 4% overall, and the split matters. Primary malignant GCT (about 1.6%) is a sarcoma arising within or alongside a GCT at first presentation, with atypical stromal cells showing frankly malignant features; it needs wide resection and treatment as a sarcoma. Secondary malignant GCT (about 2.4%) is a sarcoma developing in a previously benign, previously treated GCT, and it arises mainly after radiotherapy. It is the larger of the two groups, and it is the iatrogenic one.

That asymmetry is the argument. Most malignancy in GCT is not something the tumour did; it is something the treatment did, in a young adult with a benign lesion and a normal life expectancy. Radiotherapy is therefore reserved for genuinely unresectable disease with no alternative, and the availability of denosumab and of serial embolisation for axial lesions has narrowed that group considerably.

Campanacci grading

Campanacci grades the tumour on its radiographs and is the most widely used classification for treatment planning. The grade is read from the cortex, not from the distance to the joint: all three grades extend to subchondral bone, which is why the epiphyseal, subchondral position is the characteristic pattern rather than a pathognomonic one and does not by itself separate GCT from its differential. Grades I and II are treated by extended curettage; grade III may need resection depending on what can be reconstructed.

Recurrence by grade. Recurrence rates increase with grade: 10-15% for grade I and 15-25% for grade II after extended curettage with adjuvants, and 25-50% for grade III treated by curettage alone, against 10% for grade III after resection. A grade III lesion that has fractured is managed as grade III once the fracture has healed, with a recurrence risk quoted at 20-30%, similar to grade III.

View larger

Enneking staging

GCT is a stage 3 (aggressive) benign tumour in Enneking's system: stage 1 is latent (asymptomatic, well contained), stage 2 active (symptomatic, progressive), stage 3 locally aggressive with high recurrence. The system's real use here is its vocabulary for margins, which is the vocabulary Campanacci's recurrence table is written in.

• Intralesional - curettage, the standard operation for grades I-II
• Marginal - shelling out through the pseudocapsule, inadequate for GCT
• Wide - 1-2 cm of normal tissue, the margin of a resection
• Radical - the whole compartment, not needed for GCT

View larger

Presentation. Progressive pain localised to the involved bone, present for several months before diagnosis (median 6-12 months), activity-related at first and then constant. Swelling may be visible when the bone is superficial, at the distal radius or proximal tibia. In 10-20% the presenting event is a pathological fracture with acute pain.

The joint. Because the lesion is epiphyseal, a knee effusion is common, motion is lost to pain, and weakness shows as an antalgic gait or muscle inhibition. Catching or locking suggests intra-articular extension. Instability is rare unless cortical destruction is massive.

View larger

Examination. The skin is normal, neither warm nor red, and the mass, if the bone is superficial enough to feel one, is firm and fixed to bone. The rest is the standard sequence.

• Look - swelling or a palpable mass if superficial (distal radius); bone expansion or angulation if fractured; quadriceps wasting with chronic knee pain; antalgic gait in the lower limb
• Feel - localised bony tenderness; normal temperature; knee effusion with a distal femoral or proximal tibial lesion; distal pulses and sensation
• Move - active and passive range equally limited by pain when the lesion is juxta-articular; strength reduced by pain inhibition or atrophy; crepitus if fractured
• Assess - signs of pathological fracture (point tenderness, abnormal mobility); a comprehensive distal neurovascular examination; lymph nodes for the rare metastasis; a full neurological examination for a sacral lesion

Differential diagnosis. The epiphyseal lytic lesion overlaps radiologically and histologically with several other giant-cell-rich and lytic lesions. Age, exact location, biochemistry and H3F3A status are the discriminators.

Radiographs. The first test and often highly characteristic: an eccentric epiphyseal lytic lesion reaching the subchondral bone, geographic and well defined, with no matrix mineralisation and no sclerotic rim (unlike an ABC). The eccentric epiphyseal position is the most useful single feature, but not a diagnosis on its own. There is no periosteal reaction unless the bone has fractured, in which case the film shows cortical disruption with angulation or displacement. The soap bubble appearance is multiple trabeculated compartments within the lucency; it is less specific than the location.

View larger

CT. CT provides superior delineation of cortical integrity, trabecular destruction and the soft-tissue extension through the cortex, which is why it is particularly useful for assigning the Campanacci grade and planning the cortical window. Acquire thin slices (1 mm or less) with bone and soft-tissue windows, multiplanar coronal and sagittal reconstruction, and 3D reconstruction for planning. Read it for cortical thickness, trabecular architecture, soft-tissue extension, proximity to neurovascular structures and the joint, and the bone stock available for reconstruction.

Chest CT. A baseline chest CT is mandatory to look for pulmonary metastases, since 1-5% have lung nodules at presentation, and serial imaging follows. Nodules may be benign implantation metastases.

MRI. MRI is the test for soft-tissue extent, intra-articular extension, skip lesions and neurovascular involvement, and for surgical planning. The lesion is low to intermediate on T1, heterogeneously bright on T2, and enhances intensely.

Fluid-fluid levels. Seen in 10-15% of GCTs, from secondary aneurysmal bone cyst change after intratumoral haemorrhage, and they do not make the diagnosis a primary ABC. The discriminator is a solid enhancing tumour component with superimposed cystic areas; a true ABC enhances only as a thin rim.

View larger

Biopsy. Histological confirmation is mandatory before definitive treatment, and the biopsy is done by the treating surgeon or under their guidance so that the tract can be excised during the curettage. CT-guided core needle biopsy (11-14 gauge) is preferred because it contaminates less; take 3-5 cores from different areas because the tumour is heterogeneous; and plan the tract for excision, which for the knee means an anterior approach. Avoid a transarticular, transneural or transgastrocnemius trajectory; crossing the joint risks contaminating it.

What the pathologist needs. Review by an expert musculoskeletal pathologist is mandatory, to exclude brown tumour, ABC and osteosarcoma with giant cells and to look for the malignant features of atypical stromal cells and necrosis. Send tissue for H3F3A mutation testing if the diagnosis is uncertain. Frozen section can confirm GCT but cannot grade it, and histological grading is unreliable in any case, since all GCT is treated alike.

Algorithm

View larger

The decision. Grade and location decide what is technically possible, and the joint decides what it is worth. Grade I and II lesions receive extended curettage with the triple adjuvant, which preserves the joint at a recurrence cost of 10-20%. A resectable grade III lesion is treated by curettage or en bloc resection; an unresectable one, in the sacrum or spine, receives neoadjuvant denosumab for 2-4 months and then curettage. Resection with wide margins and reconstruction brings recurrence down to 5-10% at the price of the joint.

Joint preservation comes first. Extended curettage with adjuvants is the preferred operation for grade I-II because the triple adjuvant (burr, phenol, PMMA) brings recurrence down to an acceptable rate while keeping the joint. En bloc resection is reserved for recurrence after curettage (especially a second recurrence), grade III with massive soft-tissue extension, expendable bones such as the proximal fibula, and malignant GCT; its lower recurrence rate comes with higher functional morbidity.

Pathological fracture. A fracture does not contraindicate curettage, provided it is allowed to heal first. Immobilise for 6-12 weeks, then curette; acute surgery increases recurrence because tumour is seeded into the fracture haematoma.

Denosumab's place. It is reserved for unresectable tumours (sacrum, spine, pelvis) and as a neoadjuvant to downsize grade III lesions. It is controversial because incomplete removal of the ossified tumour periphery raises recurrence to 20-50%, and it is stopped 4-6 weeks before surgery.

Extended curettage with the triple adjuvant

The goal is complete tumour removal with the joint preserved and recurrence held to 10-20%. Curettage alone recurs in about 50%; each adjuvant attacks the residual tumour by a different mechanism, and the combination is what brings the rate down.

Planning. CT to plan the cortical window, anterior or lateral according to the lesion; MRI for soft-tissue extent and neurovascular proximity; an approach that excises the biopsy tract; a decision between cement and bone graft; and consent for the 10-20% recurrence risk.

Exposure. An extensile approach, with the biopsy tract excised en bloc with the cortical window. The window must be large, one-third to one-half of the circumference, because a small window means inadequate visualisation; keep an intact cortical buttress to contain the cement. With the cavity open, identify the tumour's extent and any soft-tissue breakthrough.

Curettage. Remove the bulk piecemeal with large curettes, then extend under direct vision to every margin, using angled curettes for the recesses. Take the soft-tissue component with it if the cortex has been breached, preserve the articular cartilage, and send representative tissue for final pathology.

Burr. A high-speed burr takes a 1-2 mm layer off the entire cavity wall: a mechanical margin beyond the curettage plane that removes microscopic tumour in the trabecular bone while preserving the cortical shell. Irrigate copiously to clear the debris and inspect for complete trabecular removal.

Phenol. Pure phenol on soaked gauze against the cavity wall for 2-3 minutes, with the soft tissues and neurovascular structures protected by moist packs. Irrigate with alcohol, which neutralises phenol, then with copious saline, and make sure none remains: phenol on skin or soft tissue causes necrosis.

Cement. PMMA polymerises at 70-80°C, so pack it in while liquid to maximise the thermal effect and make sure it reaches every recess; the heat kills cells beyond the curettage margin, and an incomplete fill leaves residual niches. It also gives immediate structural stability, which is why the patient mobilises the next day. Bone graft is the alternative in a young patient with a small defect. Keep cement out of the joint, where it causes arthrofibrosis.

Closure. Haemostasis, a drain for 24-48 hours, layered closure, and immediate mobilisation. Protect weight-bearing if the mechanical defect exceeds 50% of the cortex.

View larger

The wider adjuvant menu

The principle behind every local adjuvant is the same, to extend the cytotoxic kill zone a few millimetres beyond the visible curettage plane; the agents differ in mechanism, depth and complications. No single adjuvant is proven superior, which is why they are combined. Phenol penetrates bone more deeply than argon beam coagulation, so the more superficial argon beam is favoured next to neurovascular structures or thin cortex where a phenol spill would be dangerous. Cryotherapy achieves the deepest kill, roughly 1-2 cm, but historically carried higher rates of post-operative fracture, skin necrosis and nerve injury, so many units prefer burr, phenol and PMMA.

En bloc resection

Indications. Recurrence after multiple curettages; grade III with a soft-tissue mass too large to reconstruct after curettage; an expendable bone (proximal fibula, distal ulna); the rare primary malignant GCT; and patient preference after counselling about recurrence risk.

Planning. MRI for soft-tissue extent and neurovascular involvement, CT angiography where the tumour lies against vessels such as the popliteal, a planned wide margin of 1-2 cm of normal tissue that includes the biopsy tract, and a reconstruction chosen in advance from allograft, endoprosthesis or arthrodesis.

Resection. The tumour comes out en bloc with 1-2 cm of normal bone, the entire pseudocapsule and the soft-tissue mass. Margins go for frozen section to confirm they are clear, the neurovascular bundle is preserved if uninvolved, and the skeletal defect is assessed for reconstruction.

• Allograft - osteoarticular or intercalary, with plate fixation
• Endoprosthesis - tumour prosthesis at the distal femur or proximal tibia
• Arthrodesis - where sacrificing the joint is acceptable (wrist, ankle)
• No reconstruction - expendable bones (proximal fibula, distal ulna)

The trade. Resection recurs in 5-10% against 10-20% after adjuvant curettage, but it costs the joint, adds reconstruction complications and lengthens recovery. That is why most grade I-II and selected grade III lesions are curetted, and resection is kept for recurrent disease, malignant GCT and expendable bones.

Neoadjuvant denosumab

Role. Downsizing an unresectable tumour of the sacrum, spine or pelvis, or converting a grade III to a grade II so that curettage becomes feasible. The indications are an unresectable grade III lesion (sacrum, spine, skull base), a grade III needing downsizing for joint-sparing curettage, recurrent disease after multiple operations as palliation, and metastatic disease. It is not for routine grade I-II, where it makes the operation harder.

Dosing. Loading doses of 120 mg subcutaneously on days 1, 8, 15 and 29, then 120 mg every 4 weeks.

Duration depends on the plan. When the plan is surgery, keep the course short: 2 to 4 months, and no longer. This is a ceiling, not a floor. Van der Heijden's consensus limits the neoadjuvant course precisely because prolonged treatment ossifies the lesion so thoroughly that curettage becomes macroscopically incomplete, leaving viable peripheral tumour behind a hard rim; more denosumab before an operation makes the operation worse, not better. When the disease is unresectable, denosumab is the definitive therapy rather than a bridge, and it continues long-term with response reassessed periodically.

Monitoring and hazards. Supplement calcium and vitamin D throughout and correct any deficiency before the first dose: hypocalcaemia occurred in 5% and grade 3-4 hypophosphataemia in 3% in the Chawla trial. Osteonecrosis of the jaw occurred in 1% at this dose, which is 120 mg every 4 weeks, roughly twenty times the cumulative osteoporosis exposure, so the risk is not the one quoted in osteoporosis practice; arrange a dental assessment and complete invasive dental work before starting, as for high-dose bisphosphonates. Atypical femoral fracture is described with prolonged high-dose antiresorptive therapy, so ask about new thigh or groin pain at each review. On the other side is rebound hypercalcaemia after stopping, reported particularly in younger patients weeks to months after the last dose and sometimes severe, so a patient coming off denosumab needs a plan and a calcium check rather than simple discharge.

Response and timing. MRI every 2-3 months for size, CT for ossification (a rim of bone formation), and the patient's pain. Stop 4-6 weeks before surgery because of rebound growth, and operate 4-6 weeks after the last dose. Expect a dense, ossified tumour that may need osteotomes rather than curettes, with a higher risk of incomplete removal because the ossified periphery hides viable tumour; some surgeons avoid neoadjuvant denosumab for this reason.

View larger

The controversy. Proponents point to downsizing that allows joint-sparing surgery, conversion of unresectable to resectable, and less intraoperative bleeding. Critics point to ossification that makes complete curettage difficult, viable tumour persisting at the periphery, rebound growth after discontinuation, and recurrence after denosumab-treated curettage of 20-50% against 10-20% for adjuvant curettage without it; across the wider literature van der Heijden reviewed the range ran from 20% to 100%. The consensus is to reserve it for genuinely unresectable disease or downsizing, avoid it for routine grade I-II, and stop it 4-6 weeks before surgery.

Why the sacrum is different. The sacrum is the classic site where complete resection is limited by morbidity: en bloc sacrectomy sacrifices sacral nerve roots, and bilateral loss above S3 means loss of bowel, bladder and sexual function. It offers the lowest local recurrence and the highest neurological and visceral morbidity, and is reserved for selected or malignant cases.

The two options the long-bone pathway omits. Nerve-sparing intralesional curettage accepts a higher recurrence rate to preserve continence. Serial selective arterial embolisation, as a primary and definitive treatment or before surgery, controls growth over repeated sessions, induces ossification, relieves pain and reduces operative blood loss; it is valuable when resection is too morbid. Denosumab is now a mainstay for unresectable sacral disease, with recurrence on stopping, and radiotherapy is the last resort because of the malignant-transformation risk. These cases belong in a specialist sarcoma multidisciplinary team.

View larger

View larger

Local recurrence. The complication the whole operation is designed around. 90% occur within the first 2 years (Campanacci's pre-adjuvant series put 90% within 3 years), usually at the periphery of the cement, and MRI finds them before the radiograph does; early detection improves salvage outcomes. The risk factors are the ones the technique section predicts: no adjuvant (about 50% against 10-20% with adjuvants), a grade III lesion with a soft-tissue mass, incomplete curettage of that soft-tissue component, and a sacral or spinal location where complete removal is difficult.

Treating a recurrence. A first recurrence is treated by repeat extended curettage with adjuvants, which succeeds in 70-80%. At a second recurrence the success of repeat curettage falls to 50% while en bloc resection succeeds in 90%, so resection is considered. After multiple recurrences the options are en bloc resection or denosumab for palliation where resection is not feasible.

View larger

Pulmonary metastases. They occur in 1-5% despite benign histology, are usually asymptomatic and found on surveillance, and the risk factors are grade III, recurrent disease and pathological fracture. A few nodules (under 5) are resected; multiple or unresectable nodules are treated with denosumab. Lung nodules do not mean the tumour has become malignant.

View larger

The cement is load-bearing from the moment it sets, so a curetted patient mobilises on day one and the early weeks are about the wound, the cement and the joint. Surveillance then runs for five years, because the recurrence risk is front-loaded and the lungs need watching as well as the bone.

Local surveillance. Radiographs every 3 months in year 1 and every 6 months in years 2-5, with MRI at 6 months, 1 year and 2 years because it detects recurrence earlier than a radiograph, and CT if a radiograph is suspicious. At each visit ask about pain and swelling, check motion and function, and feel for a mass if the bone is superficial. On the films, a thin uniform lucent or fluid-signal rim around the cement can be the stable postoperative clear zone; a lucency that is progressive, focal or deepening is different and should trigger MRI.

View larger

Pulmonary surveillance. Chest CT at diagnosis and then annually for 5 years; a chest radiograph is the less sensitive fallback where CT is not available.

Discharge. Five years disease-free, with no local recurrence and no pulmonary metastases, is the discharge point: 90% of recurrences have declared themselves by 2 years and 98% by 5 years, so the risk of a late recurrence after 5 years is under 2%. Discharge to primary care with instructions to return for new symptoms. Three groups stay under surveillance beyond 5 years: malignant GCT (lifelong, as for a sarcoma), pulmonary metastases (annual chest CT lifelong), and denosumab-treated cases, whose higher recurrence risk extends follow-up to 10 years.

What predicts a good result. A grade I-II lesion, complete curettage with adjuvants, no pathological fracture (better bone stock for reconstruction), a resectable location that allows complete removal, and a young patient with better functional recovery and remodelling.

What predicts a poor one. Grade III, soft-tissue extension that is hard to clear completely, a sacral or spinal location with inadequate access, recurrent disease with its lower success rate at repeat curettage, and a pathological fracture that has seeded tumour into the haematoma.

Global Epidemiology

Giant cell tumour of bone (GCT) is consistently reported as roughly 4-5% of all primary bone tumours and around 20% of benign bone tumours worldwide, with an estimated incidence of approximately 1 per million population per year. A long-recognised geographic pattern is a relatively higher proportion in East and South Asian series (where GCT can account for up to 20% of primary bone tumours) compared with Western series, with a peak in the third to fourth decade and a slight female predominance. Single-institution and pooled series (Rizzoli, Mayo, UK and Asian sarcoma centres) consistently localise about 85% of lesions to around the knee, with the distal radius and sacrum the most common non-knee sites.

Side-by-Side Guidance & Consensus

Registry & Real-World Evidence

There is no dedicated international GCT registry; population-level estimates derive from national bone-tumour and sarcoma registries (e.g. the Netherlands Cancer Registry, the UK National Cancer Registration and Analysis Service, and the Rizzoli and Mayo institutional databases) and from the pooled denosumab trial programme (NCT00680992). These data sources underpin the convergent ~1 per million incidence figure and the recognition that recurrence after curettage following denosumab may exceed that of curettage with adjuvants alone, as summarised by van der Heijden et al (Curr Opin Oncol 2020).

Practice Variation & Access

Practice variation is greatest around denosumab (duration of neoadjuvant therapy, and whether to use it at all in resectable disease) and around reconstruction after resection (endoprosthesis vs allograft vs arthrodesis), reflecting differing drug access, reimbursement and surgical resources. Internationally, denosumab is reserved for unresectable, recurrent-inoperable or metastatic GCT with histological confirmation and specialist management; loading doses are given on days 1, 8, 15 and 29 followed by 120 mg subcutaneously every 4 weeks, with calcium and vitamin D supplementation to prevent hypocalcaemia, though funding and access pathways differ markedly between health systems. Across all systems, outcomes are best when GCT is managed within a specialist sarcoma multidisciplinary team with structured long-term surveillance.