Massive Bone Allograft Reconstruction in High-Grade Osteosarcoma

Abstract

From 1986 to 1994, 112 bone allograft reconstructions were performed in patients with high-grade osteosarcoma in whom neoadjuvant chemotherapy was administered. The allograft reconstruction was used in arthrodesis in 44 cases (41 knees, three ankles), as an intercalary graft in 39 (28 femurs, 11 tibias), as an osteoarticular graft in 22 (three proximal and/or distal humeri, six distal femurs, 13 proximal tibias), and as an allograft and prosthesis composite in seven (two proximal humeri, one proximal femurs, four proximal tibias). In 20 patients an autologous vascularized fibula was used to augment the allograft. Functional results were excellent or good in 74% of the patients after the primary surgery, and in 83% of the patients after secondary surgery. Complications include delayed union (49%) and fracture (27%), although there were no cases of deep infection. The incidence of delayed union, but not infection or fracture, was increased by the use of chemotherapy.

Clinical use of massive bone allografts was reported by Lexter¹⁴ in 1908 using fresh allografts retrieved from amputated limbs. Ottolenghi^17,18 and Volkov,²¹ reported successful results in approximately 75% of patients using anatomic specimen frozen allografts. Parrish¹⁹ in 1966 and Mankin et al¹⁵ in 1976 reported good results in 80% of allograft reconstruction performed primarily for tumor resections. In the authors' institute, massive frozen allografts have been used as a means of reconstruction after segmental resection for bone tumors since 1984. Debate continues^{6,9,11,12,22,23} relative to the effectiveness of massive allograft reconstruction compared with other techniques (endoprostheses), particularly for patients receiving intensive chemotherapy.

MATERIALS AND METHODS

From August 1986 to December 1994, 112 patients with high-grade osteosarcoma underwent preoperative chemotherapy, limb-sparing resection, reconstruction with a massive allograft, and postoperative chemotherapy. There were 107 patients with Enneking Stage II lesions and five with metastatic disease (Enneking Stage III) at presentation. Patients who had resections of the proximal femur reconstructed with an iliofemoral arthrodesis and patients treated with immediate resection and postoperative chemotherapy were excluded from the study.

There were 63 male and 49 female patients, with ages ranging from 4 to 48 years (mean, 15 years).

Resection length varied from 8 to 32 cm (average, 15.6 cm). The tumor locations and osteosynthesis device are shown in Table 1.

TABLE 1: Site of Resection, Type of Reconstruction, and Osteosynthesis Devices

The type of reconstruction was an arthrodesis in 44 cases (41 knees [Figs 1, 2, 3, 4], three ankles), an intercalary allograft in 39 (28 femurs, 11 tibias) an osteoarticular allograft in 22 (three humeri, six distal femurs, 13 proximal tibias), and an allograft and prosthesis composite in seven (two proximal humeri, one proximal femur, four proximal tibias). In 16 patients, corticocancellous bone chips (13 autogenous, three allogenous) were applied primarily around the osteotomy site, whereas in another 20 patients (eight femurs, 12 tibias) a vascularized fibula transplant was used primarily to augment the allograft (Fig 5). The average length of the surgical procedure was 4 hours 20 minutes (range, 2.5-7 hours), excluding procedures with vascularized fibula graft.

Fig 1: Anteroposterior plain radiograph of osteosarcoma of the right distal femur in an 11-year-old boy.

Fig 2: Soft tissue involvement of the anterior compartment of the distal femur as seen on a magnetic resonance imaging scan.

Fig 3: Postoperative anteroposterior radiograph showing knee arthrodesis. A long fluted nail and a short antirotation plate at the cortical osteotomy site was applied for fixation.

Fig 4: Coronal magnetic resonance imaging scan showing osteosarcoma of the proximal tibia in an 11-year-old boy. There is no evidence of tumor beyond the physeal plate.

Fig 5: Anteroposterior radiograph showing reconstruction with shell intercalary allograft and a central free vascularized fibula after intraepiphyseal transverse cut.

All allografts were procured in a sterile fashion and stored at −80°C until used. When the graft was removed from the freezer and opened in the operating room, it was cultured for aerobic and anaerobic organism and thawed in warm saline solution with antibiotics. It was cleaned of unneeded soft tissues and bone marrow.^1,9,16 The patients were given amikacin (500 mg for two doses) on the day of surgery and teicoplanin (lepetit S.p.a., Anagni, Italy) throughout the first postoperative week. Oral antibiotics were recommended for the ensuing 3 months. Cast immobilization was used for 1 to 3 months (average, 2.3 months), depending on the type of fixation. A brace was used thereafter, and the patient was kept partially weightbearing until the osteotomy sites healed. For osteoarticular grafts this ranged from 10 to 12 months, for intercalary grafts 14 to 23 months, and for arthrodeses 14 to 23 months. When an allograft and prosthesis composite was used, no postoperative bracing was used.

The 112 patients were treated with five different chemotherapy protocols. In the first protocol (1983-1986; four patients) preoperative chemotherapy included intravenous methotrexate and intraarterial cisplatin. Methotrexate dose was randomized to high dose versus moderate dose. Surgery was scheduled approximately 2 weeks after the last preoperative drug administration. Postoperative treatment included methotrexate, cisplatin, doxorubicin, bleomycin, cyclophosphamide, and dactinomycin according to the tumor necrosis evaluated on the resection specimen. In the second protocol (1986-1989; 29 patients), preoperative treatment included high dose methotrexate, doxorubicin, and cisplatin. After surgery, ifosfamide and etoposide were added. The third (1990-1991; 33 patients), fourth (1991-1992; 14 cases), and fifth (1993-1996; 32 patients) protocols differed from the previous two in that ifosfamide was used after surgery in the third protocol, and before and after surgery in the fourth, whereas in the fifth protocol, the chemotherapy regimen duration was changed from 12 to 45 weeks, according to the tumor necrosis evaluation.^3-5

Clinical evaluation of results was based on a modification by Gebhardt et al¹¹ of the Mankin scale. An excellent result was defined as patients returning to their previous recreational and occupational levels with no pain; a good result was assigned to patients who returned to their preoperative occupation but had restrictions in recreational activities; a fair result was assigned to patients with mild pain or significant functional limitation requiring aids or supports and who were unable to return to an appropriate work status; and a poor result was considered to be less than this level of function, if the graft was still present. A result of failure was assigned if a patient required amputation or graft removal.

Ninety-two patients had a followup from 24 to 125 months (average, 60 months), and 20 patients had followup less than 24 months (19 died of disease; one was lost to followup).

RESULTS

Twenty-six patients died of disease (including six patients with local recurrence) and two patients are alive with disease. Among the remaining 84 patients, 17 had one or more pulmonary metastasectomies and now are free of disease.

Complications

Nerve palsies occurred in 14 patients (13 peroneal nerves; one median nerve) (Table 2). Nine of these patients had complete recovery.

TABLE 2: Summary of Complications Divided by Type of Reconstruction

Four patients had postoperative wound necrosis, but only one was treated with a rotational medial gastrocnemius flap. One patient with wound hematoma required wound drainage, and one patient required revision surgery using an intraarticular screw without sequelae. There were no deep infections in these patients.

Delayed Union

Delayed union was defined as no radiographic signs of union by 1 year after surgery (Figs 6, 7). Fifty (49%) of 101 evaluable patients with more than 1-year followup experienced delayed union. Nine of these incidences of delayed union occurred despite placement of corticocancellous bone chips at the osteosynthesis site at the time of the initial procedure, and another nine occurred in patients treated primarily by vascularized fibula autograft augmentation.

Fig 6: Anteroposterior plain radiograph obtained after a distal femur reconstruction with osteochondral allograft. Good contact was achieved, but the osteotomy line still is not united after 1 year.

Fig 7: Lateral radiograph of the same patient shown in Figure 6. A posterior bone pedicle is evident; however, the posterior cortex is not united.

In 15 (30%) patients delayed union healed after a mean of 26 months without any supplementary surgery. In 27 (54%) cases, secondary application of corticocancellous autograft chips from the iliac crest was performed, and in 10 of these the osteosynthesis device was revised. In four patients the delayed union was treated with a vascularized fibula. The four remaining patients had the allograft removed (in two because the delayed union was associated with marked allograft resorption and in two because of fracture of the allograft). These patients underwent a second reconstructive surgery: a new allograft was replaced in three patients, and a modular prosthesis (HMRS, Howmedica Modular Resection System, Howmedica Italy, Rome, Italy) was used in one. Thus, eight (16%) patients had a major salvage operation.

Fracture

Fracture occurred in 28 (30%) of 92 evaluable patients with followup of 24 months or greater who were evaluated. In three patients the fracture was caused by significant trauma (motor vehicle accident). The location of the fracture was diaphyseal in 17 patients, metaphyseal in eight patients, and subchondral in three patients. The time of fracture varied from 4 to 75 months (average, 31 months).

Treatment was nonsurgical in 12 patients, of whom three are awaiting operative repair and nine healed with rest and cast or brace immobilization. Four of the patients who achieved healing had vascularized allografts at the time of the initial procedure.

Four patients were treated by open reduction, internal fixation, and iliac bone grafting at the fracture site. Three of these patients also had associated delayed unions.

Vascularized fibular graft was used in two patients, and the allograft was removed in an additional 10 patients. In six of these 10 patients a second allograft was applied. The HMRS was used in three patients and in the last a vascularized fibula alone was used.

More recently, it was decided to fill the medullary canal of the allograft with polymethylmethacrylate cement. Of 19 patients surgically treated with this technique, 13 have a followup of 2 years and no patient has sustained a fracture, although delayed union occurred in eight (61%) patients.

Functional Results

Of 92 patients who could be evaluated, the final functional outcome was rated as excellent in 41 (45%), good in 27 (29%), fair in seven (8%), and poor in three. Fourteen (15%) allografts failed: two patients had limb amputation for local recurrence, and 12 grafts were removed (two for nonunion and 10 for fracture).

Seventy-four percent of the patients achieved a satisfactory result (good or excellent) at last followup, and this percentage increased to 83% when patients who had a successful second reconstruction surgery were considered. Of seven of the patients who had a fair result, one was lost to followup, two died of disease (at 39 and 46 months, respectively), one had a cerebrovascular accident with resulting hemiplegia, and three are awaiting additional surgery. The one poor result involved a patient with a knee arthrodesis and fracture who is being treated nonoperatively for the fracture.

DISCUSSION

Anatomic specimen bone allografts have numerous advantages when used to reconstruct large bone defects after resection for high-grade extremity bone tumors. As shown in the current series, allografts may be used in various patients with different ages and different disease sites and resection lengths. In addition, an allograft can be used with a vascularized fibula and with an endoprosthesis (composite).

Chemotherapy has been reported in numerous studies^{6,10,12,13,22,23} to produce a consistent reduction in bone remodeling activity in a dose dependent manner.²² This is presumed to be because of the effect of the drug on osteoblast, osteoclast,^9,10 and endothelial cells, limiting the vascular invasion of the graft.⁶ However, the chemotherapy effect is reversible and probably ceases 1 week after the last cycle of administration.²³

Thus, chemotherapy may delay the osteointegration of the graft but does not represent an absolute contraindication to the use of allografts. In fact, intense and rapid revascularization may increase the risk of fracture.

Osteointegration occurs by a process of creeping substitution whereby cutting cones invade through preexisting vascular channels (Aversian system and Volkmann's canal) in the allograft leading to matrix resorption followed by laying down of new bone. This process occurs slowly and primarily at the allograft and host junctions and at the periosteal surface of the graft. Graft resorption tends to be more apparent in the second than in the first year after implantation. Complete bone integration has not been proved.^1,2,7,8

In the current series there was a considerable incidence (49%) of nonunion that did not appear to be influenced by the length of post-operative immobilization or by limitations in weightbearing. Failure of the allograft is low and is caused primarily by fracture or marked resorption. These complications appear to be influenced favorably by the use of more adequate internal fixation, the association of vascularized fibular autografts, and the use of long-stemmed prosthetic components in allograft and prosthesis composites.

In most cases of delayed union at 1 year, the osteosynthesis sites eventually healed, and progressive weightbearing was achieved by 2 years after the reconstruction. However, it is preferable to treat delayed union sites with autograft augmentation by 12 to 16 months after the primary procedure to avoid the risk of osteosynthesis device failure and possible loss of the graft. In some cases adding an additional plate may be appropriate, particularly in intercalary grafts.

Chemotherapy could cause delay in wound healing, which may lead to deep wound infection. Only four patients experienced wound necrosis, which is low considering the magnitude of the surgical procedure and the intensity of chemotherapy. One of these four patients had revision surgery before deep infection occurred. Attention to adequate soft tissue coverage of the allograft, meticulous closure of the wound, and postoperative cast immobilization increase the chance of avoiding a catastrophic wound complication. In addition, using prophylactic antibiotics appears to be beneficial. Administration of a late generation glycopeptide (teicoplanin) and an aminoglycoside (amikacin) in this series seem to result in a low incidence of infection with organisms such as Staphylococcus species, Enterobacter, and Escherichia coli. This regimen also may be effective in preventing contamination of the allograft from bacteremia caused by indwelling catheters in the initial 3 months after surgery.

Another complication possibly influenced by chemotherapy is nerve palsy, but definitive studies are needed to document this possibility. This complication can be related to surgical technique, as shown by it occurring primarily after proximal tibia resection arthrodesis, during which it is necessary to dissect the peroneal nerve in an extremity in which it is not possible to flex the knee during the early postoperative course. However, it is likely chemotherapy has toxic effects on the peripheral nerves: 11 of 47 patients who had knee arthrodesis in this series had nerve palsies.

The incidence of fracture was less and occurred later than did delayed union. Fractures also were less common in an arthrodesis than in an osteoarticular allograft. The tibia had the highest incidence of fracture. It was impossible to correlate the incidence or type of fracture with chemotherapy. Fracture often led to failure of the graft, especially if the osteosynthesis had been achieved by plate fixation.

No fracture occurred in patients who had an allograft and prosthetic composite because of the long stem of the prosthesis. Using a vascularized fibula to augment the construct does not prevent allograft fractures, but in most cases these are fatigue fractures that will heal by conservative means. In only one such case was the fracture sufficiently displaced to require removal of the graft and revision to a new implant. Filling the medullary canal of the allograft with polymethylmethacrylate cement seems to result in a lower fracture incidence, but it may increase the likelihood of a nonunion. The use of intramedullary cement in osteoarticular allograft carries the theoretical disadvantage that if joint deterioration occurs, revision to a prosthetic arthroplasty technically will be difficult.

In the authors' experience the primary indication for an osteoarticular allograft is a young patient (age, 10-15 years) with a tumor of the knee. The advantage over prosthetic reconstruction at this age and site is the ability to preserve the growth plate on the opposite side of the resection and lessen the resultant limb length discrepancy.

The articular cartilage of the allograft does not appear to undergo rapid immunologic attack by the host, and glycerol cryopreservation appears to have a beneficial effect in blocking or reducing the lysosomal enzyme activity.^1,20 However, cryopreservation does not appear to be effective in preserving the viability of more than 50% of chondrocytes.¹ Other factors influencing the longevity of the joint are the size match of the graft, the storage time of the graft, and the effects of altered nutrition of the allograft articular cartilage in its new joint environment.^1,8,11,15,16 For these reasons and because of the relatively high incidence of subchondral fractures, in adult patients it is preferable to use allograft and prosthesis composites to reconstruct the proximal tibia and to use modular prostheses for the proximal femur.

The oncologic results for patients included in the current study are similar to those reported in another series,¹⁶ and it does not appear that limb salvage procedures are associated with a worse disease outcome. The functional results were satisfactory in 74% of the patients in the current series, and 47% of patients had excellent results. Thus, approximately 1/2 of patients with osteosarcoma can expect an excellent functional outcome if the indications are correct and no complications occur. Failures were common (13%), but with more experience and a better fixation technique, many complications can be avoided. However, currently approximately 1/2 of patients with osteosarcoma can expect to have at least one additional operation to manage a complication.

Massive bone allografts are an effective reconstructive method after resection for high-grade osteosarcoma of the extremities. Post-operative chemotherapy adversely affects osteointegration, as shown by the number of complications, particularly delayed union. More uncertain is the effect of chemotherapy on wound healing and associated deep infection, nerve palsy, and fracture. Fracture incidence appears to be related to the reconstructive technique. If the surgical indication is appropriate, the incidence of failures can be reduced. Functional results in this series were satisfactory in nearly 80% of the patients.

Acknowledgment

The authors thank Mark C. Gebhardt, MD, for advice in reviewing the article.