Allograft-Prosthetic Composite (APC) Reconstruction

An allograft-prosthetic composite combines a massive structural allograft (proximal femur, proximal tibia, proximal humerus or an intercalary segment) with a conventional or long-stemmed prosthesis. The prosthesis provides the articulation; the allograft provides the bone and the soft-tissue anchor points that a megaprosthesis cannot.

View larger

Bone stock. A failed APC usually leaves more bone than a failed megaprosthesis. That is critical in young patients, who will require future revision.

Soft-tissue attachment. This is the decisive functional advantage over a megaprosthesis, and it is site-specific:

• Proximal femur: host abductor tendon repaired to the allograft's retained abductor tendon and greater trochanter, for better gait, less Trendelenburg lurch and potentially lower dislocation risk
• Proximal tibia: host patellar tendon sutured to the allograft's retained patellar tendon, a biological extensor mechanism reconstruction that is superior to tendon-to-metal fixation
• Proximal humerus: host rotator cuff, capsule and remaining tuberosity tissue repaired to allograft cuff stumps, restoring stability and some active motion

Two philosophies in one construct. The junction and the soft tissues heal biologically; the articulation is a mechanically reliable prosthesis. That avoids the cartilage degeneration and joint instability of osteoarticular allografts, which APCs have largely superseded at the knee and shoulder.

The trade-offs. Operating time is longer, the operation depends on bone bank availability and size matching, and the junction must unite before full loading, so rehabilitation is delayed compared with a megaprosthesis. The allograft also brings its own late complications: fracture, resorption and infection of dead bone.

APC buys bone stock and biological soft-tissue attachment at the cost of union time and infection risk; a megaprosthesis buys immediate function. The decision matches that trade-off to life expectancy, site and adjuvant therapy.

View larger

The defect. Segmental bone loss after primary sarcoma resection (osteosarcoma, chondrosarcoma, Ewing), failed arthroplasty with a Paprosky IIIB/IV femoral defect, an unreconstructable periprosthetic fracture, a failed osteoarticular allograft or a failed megaprosthesis. Confirm first that the joint cannot be reconstructed by conventional revision means and plan the resection level; only then is the patient a candidate for an APC or a megaprosthesis.

Who gets an APC. The younger and more active the patient, the stronger the case, because bone stock is preserved for the lifetime of revisions ahead. The proximal humerus and proximal tibia depend on soft-tissue reattachment (rotator cuff, extensor mechanism, abductors) for function, so favour APC where tendon-to-tendon reattachment materially changes function; the proximal femur is genuinely debated, because modern megaprostheses with dual mobility narrow the gap. A failed osteoarticular allograft with a united shell that can be retained and converted is a further indication.

Who gets a megaprosthesis instead. Each of these tips the decision away from allograft:

• Active or recent deep infection at the site. Dead allograft placed in an infected bed fails. Stage the reconstruction, eradicate sepsis and reassess; if sepsis cannot be cleared, do not use an APC
• Metastatic disease or limited life expectancy. The bone-stock advantage is never realised and no soft-tissue reconstruction is achievable in the timeframe available
• Immediate full weight-bearing required. A megaprosthesis needs no wait for union
• Age over roughly 65 to 70 years, or low functional demand
• Planned or ongoing high-dose local radiotherapy across the junction. It sharply reduces union rates. Heavy chemotherapy, by contrast, delays union but does not preclude APC; it lengthens the protected-loading period
• Inability to comply with protected loading while the host-graft junction unites, typically 6 to 12 months
• Very poor host bone or soft tissues, heavy smoking, poorly controlled diabetes, immunosuppression. All raise nonunion and infection risk; optimise what is modifiable
• No suitably size-matched allograft. Default to megaprosthesis if no matched graft exists

Source. An accredited musculoskeletal tissue bank, with the cadaveric donor screened by serology (HIV, hepatitis B and C, HTLV, syphilis) and culture per national tissue regulations: Human Tissue Authority licensing in the UK, FDA/AATB standards in the USA, EU tissue directives elsewhere.

Preservation. Fresh-frozen at minus 80 degrees Celsius is the standard for structural grafts: freezing reduces immunogenicity while largely preserving mechanical strength. Freeze-dried grafts are brittle when rehydrated and unsuitable for major structural roles.

Irradiation. Gamma irradiation above roughly 25 kGy significantly weakens cortical bone, with a dose-dependent loss of bending and torsional strength and fatigue life. Avoid high-dose irradiated grafts for structural APC segments; low-dose or non-irradiated aseptically processed grafts are preferred.

Size matching. Match donor and host on calibrated radiographs or CT: canal diameter (to accept the stem plus a 2 mm cement mantle), cortical diameter at the planned junction, and segment length. A slightly larger allograft cortex can be telescoped or step-cut over host bone; a smaller graft is harder to salvage.

Soft-tissue retention. Request the graft with retained tendon stumps: abductor tendon and trochanter for the femur, patellar tendon for the tibia, rotator cuff insertions for the humerus. These are the anchor points for tendon-to-tendon repair.

Theatre preparation. In order:

1. Thaw in warm antibiotic saline
2. Take culture swabs before implantation
3. Ream and lavage the allograft canal to remove marrow elements, which reduces the immunogenic load and prepares the canal for cement
4. Fashion the osteotomy on the back table against trial components

Preparing the junction. The size-matched allograft is trimmed to restore length and version, and the cut surfaces are cleaned of marrow and debris. The junction is packed circumferentially with autogenous iliac crest or reamer aspirate graft. Compression, autograft and rotational control across the junction are what deliver union.

View larger

The cut. Transverse or step-cut. The trade is contact area and rotational control against ease of matching and of applying compression:

The plate. A long lateral plate with bicortical screws in both host and allograft, applied perpendicular to a transverse cut, delivers excellent interfragmentary compression and neutralises the torsion a stem controls poorly.

The bridging stem. A long cemented stem crosses from the prosthesis through the entire allograft into host bone, so the junction is splinted and the graft protected against fracture. Cement is confined to the allograft canal, and to the host if that is the chosen host fixation.

Both. A long bridging stem plus a junction plate is biomechanically the most stable construct, and increasingly the default for proximal femoral and tibial allograft-prosthetic composites in the lower limb.

View larger

Junction nonunion. Seen at the osteotomy interface at any site in 10 to 30 percent, declared at 9 to 12 months. The mechanisms are inadequate compression, a gap at the cut, cement interposed at the junction, and irradiated or devitalised graft ends; the risk factors are a transverse osteotomy without compression, poor host bone contact, chemotherapy, an irradiated graft and inadequate fixation. Prevention is the junction construct described above, with a non-irradiated graft. An established nonunion is treated by autogenous bone grafting plus revision compression plating: take down the fibrous tissue, remove any interposed cement, refresh the cut surfaces and upgrade to a longer or dual-plate compression construct with generous autograft. Union is usually achievable.

Deep infection. 10 to 15 percent in tumour series, early or late, and highest at the proximal tibia. The drivers are a large avascular graft mass, a long operating time and a thin soft-tissue envelope, and the allograft is dead bone that antibiotics cannot sterilise. Prevention is gastrocnemius flap coverage at the tibia, antibiotic-loaded cement, meticulous soft-tissue handling and a short operating time. Debridement rarely salvages the graft: established deep infection almost always requires removal of allograft and prosthesis and staged reconstruction, described under revision below.

Allograft fracture. Roughly 5 to 20 percent, typically beyond 2 years, mid-graft or at a stress riser. The mechanisms are a graft that never fully revascularises, screw and drill holes acting as stress risers, and an unsupported graft segment. A long cemented stem spanning the whole graft is protective; minimise unfilled drill holes and unicortical hardware clusters in the graft, and select a graft of adequate cortical thickness and diameter match. With the stem intact and the fracture minimally displaced, plate and graft; segmental failure means revision to a new APC (young patient, allograft available) or conversion to megaprosthesis.

Allograft resorption. Late and progressive, in a minority. Micromotion and soft-tissue stripping devitalise the graft surface, so prevention is rigid fixation and avoiding circumferential stripping of the graft periosteal envelope. Observe if mechanically stable; graft or revise once cortical integrity is threatened.

Instability and dislocation. Proximal femoral APCs dislocate at a meaningful rate despite abductor repair, through loss of abductor tension and the capsular envelope after resection, or component malversion. Tendon-to-tendon abductor repair to the graft trochanter, capsular repair, dual mobility or large heads, and correct combined version are the defences. Closed reduction and address the cause; recurrent instability is revised to a constrained or dual-mobility bearing.

Extensor lag and cuff failure. The site-specific functional failures of the proximal tibia and proximal humerus: a patellar tendon or cuff attached to the graft heals by tenodesis and stretches out. Prevention is the tibial repair under tension in full extension with 6 weeks of protection, and a reverse design at the humerus. Salvage is revision tendon reconstruction, an allograft extensor mechanism, or acceptance of the deficit.

Proximal humerus. The deltoid and axillary nerve must be intact. A hemiarthroplasty megaprosthesis behaves as a spacer with poor active elevation, and osteoarticular allograft alone has been largely abandoned here. In Abdeen's series the failure modes were prosthetic loosening (3 of 36) and delayed union at the osteosynthesis site (4 of 36); dislocation was rare, one in 36; construct survival was 88% at 10 years, and all patients had mild or no pain and normal hand function. The case for APC is strong, but counsel on function by what must be resected: deltoid loss predicts reduced motion and extra-articular resection predicts a lower MSTS score.

Proximal tibia. Extensor lag, infection and wound breakdown bite before junctional nonunion at this site. Tendon-to-metal fixation with mesh or tube augments gives inferior extensor function, so the functional case for APC is strong where the soft tissues allow; the thin envelope means flap coverage is required either way.

Proximal femur. The construct is a long-stem hip prosthesis cemented into the allograft, with a dual-mobility or constrained head where the abductors are deficient, and the failure modes are dislocation, junctional nonunion (10% in Farid's APC arm, all of which united after grafting) and trochanteric escape if the abductor repair fails. The case against megaprosthesis is not survivorship: Kaplan-Meier implant survival was identical at 86% at 10 years for both. The difference is abductor power, 4.6/5 with APC against 2.8/5 with endoprosthesis, and MSTS 82% against 70%. Decide on abductor function and life expectancy, not on survival, and note the caveat in the source: APC follow-up was a median 76 months against 146 for endoprostheses, so the 10-year APC figure is the less mature of the two.

View larger

Aseptic loosening or stem failure with a united junction. The united allograft has become part of the host skeleton, so the prosthetic side alone is revised, with a longer stem through the retained allograft shell. This scenario is the payoff of the APC philosophy: bone stock has been banked.

Junction nonunion with an intact implant. Compression plating and autografting as described under complications; exchange to a longer bridging stem if the existing stem is short or loose.

Infection. Staged: remove prosthesis and allograft (dead bone cannot be sterilised), antibiotic spacer and targeted systemic antibiotics, then reconstruct with a megaprosthesis, a new APC in selected cases after confirmed eradication, or arthrodesis. Amputation is the salvage for uncontrolled sepsis or recalcitrant infection.

Failed osteoarticular allograft converting to APC. If the original allograft-host junction has united, resurface through the allograft with a stemmed prosthesis, a well-described salvage that preserves the united segment.

Massive allografts unite to host at the junction like a fracture, but the graft itself revascularises only superficially and slowly, with creeping substitution limited to the outer millimetres over years. The bulk remains dead scaffold indefinitely, hence the late fracture risk and the rationale for a protective intramedullary stem.

View larger

• Global epidemiology and practice variation: APC use tracks the availability of accredited musculoskeletal tissue banks. Well-resourced centres in North America, Europe, Australasia, Japan, Brazil and Argentina (which pioneered large bone-banking programmes) use APCs routinely; in regions without bone banks, megaprostheses, extracorporeally irradiated or pasteurised/liquid-nitrogen-treated autografts (recycled autograft techniques popular in parts of Asia) substitute for allograft.
• Tissue regulation: allograft procurement and processing are governed nationally — Human Tissue Authority licensing in the UK, FDA regulations and American Association of Tissue Banks standards in the USA, EU Tissues and Cells Directives across Europe, Therapeutic Goods Administration frameworks in Australia. Surgeons must know their bank is accredited and grafts serologically screened.
• Society guidance: no society issues an APC-specific guideline; principles are embedded in musculoskeletal oncology consensus practice (ISOLS — International Society of Limb Salvage — meetings and classification of reconstruction failure by Henderson and colleagues, which categorises failures into soft tissue, aseptic loosening, structural, infection and tumour progression and is the standard reporting framework for APC series). AO principles govern junction fixation. BOA/BOOS guidance in the UK directs sarcoma reconstruction to specialist centres, where bone banking is concentrated.
• Registry evidence: national arthroplasty registries (NJR, AJRR, AOANJRR, SHAR, Norwegian, NZJR) capture megaprosthetic and some composite revision constructs only coarsely; APC evidence therefore rests on institutional series and ISOLS multicentre reports rather than registry data — a recognised evidence gap.
• Resource-setting adaptation: where allograft is unavailable, alternatives include megaprosthesis, recycled tumour-bearing autograft (irradiated, pasteurised or frozen) combined with a prosthesis — an "autograft-prosthetic composite" applying identical junction and soft-tissue principles.

Related pages: Limb Salvage Surgery Principles sets the margins and the resection this reconstruction has to fill, and Osteosarcoma, Ewing Sarcoma, Chondrosarcoma and Giant Cell Tumour of Bone are the diagnoses that generate the defect - chemotherapy timing in the first two is what makes wound healing and union so much harder here than in a non-oncological reconstruction. Megaprosthesis in Non-Oncologic Salvage is the alternative construct at every site discussed above, and Metastatic Bone Disease is the setting in which it wins outright, because a construct that depends on union is the wrong operation for a patient who may not live to unite it. Periprosthetic Joint Infection behaves differently here - the allograft is dead bone and becomes a sequestrum, so eradication means removing the construct rather than exchanging a liner. Masquelet Induced Membrane Technique and Femoral Impaction Bone Grafting are the biological alternatives for segmental and cavitary loss respectively, Revision THA and Revision TKA Bone Loss: Cones and Sleeves cover the non-tumour route to the same problem, and TKA Extensor Mechanism Failure with Extensor Mechanism Ruptures explain why tendon-to-tendon repair at the proximal tibia is worth the wait for union. Hip Disarticulation and Forequarter Amputation is the salvage when the composite fails irretrievably.

• Proximal femur: APC or megaprosthesis? Comparable survivorship in most comparative series; dual-mobility megaprostheses have eroded the dislocation argument for APC, while APC retains the bone-stock argument in the young. No randomised evidence exists or is likely.
• Stem fixation in the host: fully cemented (immediate stability, easier in poor bone) versus press-fit host fixation (easier future revision, biological fixation) — practice varies by centre with no comparative superiority shown.
• Plate, stem, or both at the junction: combined constructs are biomechanically stiffest but add soft-tissue stripping and hardware; some series report equivalent union with a well-fitted compressed junction and long stem alone.
• Step-cut versus transverse osteotomy: theoretical advantages of step-cut are not consistently reflected in union rates; precision of contact and compression may matter more than geometry.
• Reverse shoulder APC: increasingly favoured over anatomical/hemiarthroplasty APC in the proximal humerus, but long-term data on glenoid survival in young tumour patients are limited.
• Role of APC in the era of custom 3D-printed implants: additive-manufactured endoprostheses with porous collars and soft-tissue attachment augments may replicate some APC advantages without allograft-specific risks; comparative data are immature.