Eighty-nine patients had a revision total knee arthroplasty at our institution between 1983 and 1994. Twenty-one patients received morselized allograft, and twenty-eight patients (thirty knees) who had massive bone defects due to a failed previous total knee arthroplasty had bulk allograft applied to the proximal aspect of the tibia or the distal aspect of the femur, or both. These twenty-eight patients form the basis of this study. The indication for the use of structural allograft was an uncontained defect that was more than three centimeters long. The average age of the eight men and twenty women at the time of the index procedure was 65.8 years (range, twenty-four to eighty-nine years) (Table I).
The index procedure was a first revision in eighteen knees, a second revision in ten, and a third revision in two. The revision was performed for aseptic loosening in fourteen knees, periprosthetic fracture (with or without loosening) in eleven, loosening due to infection in four, and instability with malalignment in one.
The grafts were procured according to the protocol of the American Association of Tissue Banks10. The bone was deep frozen at -70 degrees Celsius and was treated with 25,000 gray of radiation.
All of the defects were classified as uncontained. Sixteen knees had the bulk allograft placed in the distal aspect of the femur; ten, in the proximal aspect of the tibia; and four, in both.
There were fourteen circumferential defects and sixteen non-circumferential defects (Table I). The non-circumferential structural allografts were fixed to the host bone with cancellous-bone screws, and additional fixation was provided by the long stem of the prosthesis (Figs. 1-A, 1-B, 1-C, 4-A, and 4-B). Cement was used at the allograft-implant interface and at the implant-host bone interface on the surface in the metaphyseal regions but not at the stem-host bone interface in the diaphysis.
For the circumferential allografts, the allograft-host bone interface was stabilized with a step-cut osteotomy secured with cerclage wire and a long prosthetic stem (Figs. 2-A, 2-B, 3-A, 3-B, and 5). Residual host bone that had been retained with its soft-tissue attachment to serve as a vascularized bone graft was used at all allograft-host bone junctions. Also, if the residual host bone had major ligament origins and attachments, it was wrapped around the allograft with cerclage wire to provide ligamentous support (Fig. 5).
The implants consisted of fourteen press-fit condylar modular prostheses (PFC; Johnson and Johnson Orthopaedics, Raynham, Massachusetts), seven Insall-Burstein prostheses (Zimmer, Warsaw, Indiana), four porous-coated anatomic revision prostheses (PCA; Howmedica, Rutherford, New Jersey), and five Guepar hinged prostheses (Benoist Girard, Bagneaux, France). A long stem was used for all revisions that necessitated restoration of bone stock.
The postoperative management varied according to the extent of the operation. Early movement was encouraged, and the patients were allowed to bear partial weight after six weeks. Full weight-bearing was allowed when there was evidence of union at the allograft-host bone junction, usually at three months.
The clinical assessment was made on the basis of the modified Hospital for Special Surgery knee score29, which is based on subjective factors (pain, instability, use of walking aids, and distance walked) and objective factors (lack of extension, flexion, and effusion). The total possible score is 100 points (indicating a normal knee). Clinical information was obtained by examination or from the referring orthopaedic surgeon. Failure was defined as an increase of less than 20 points in the knee score postoperatively or the need for an additional operation related to the allograft.
Union at the allograft-host bone junction, periprosthetic radiolucent lines, migration of the prosthesis, and fracture or resorption of the allograft were analyzed radiographically.
The Kaplan-Meier method16 was used for survivorship analysis, and the 95 per cent confidence limits were calculated with the Greenwood formula for variance.
Clinical Results
The average duration of follow-up was fifty months (range, twenty-four to 132 months; median, thirty-six months). Over-all, the procedure was considered a failure for seven knees and a success for twenty-three knees (a 77 per cent rate of success) (Table I). The Kaplan-Meier16 probability of survival of the graft at five years was 67 per cent, with 95 per cent confidence limits of 41 and 83 per cent.
The average score for the knees for which the procedure was successful was 42 points (range, 20 to 64 points) preoperatively and 71 points (range, 40 to 93 points) postoperatively.
Failures
There were three failures (in three patients) because of deep infection. One was in a seventy-one-year-old diabetic man (Case 16) who had had loosening due to infection around a primary total knee prosthesis. This was treated with an excisional arthroplasty, which was followed, after several months, by reconstruction with bulk tibial and femoral allografts and a Guepar prosthesis. An above-the-knee amputation was performed three years later because of recurrent infection. The second failure was in a seventy-one-year-old woman (Case 7) who had had loosening due to infection around a primary total knee arthroplasty. The knee was revised in two stages with use of a bulk allograft to reconstruct the proximal tibial defect. Infection recurred one year later. A cement spacer was inserted after débridement. Six months later, the patient had a successful revision with a press-fit condylar modular prosthesis. The third failure was in an eighty-four-year-old man (Case 28) who had had a loose and unstable previous total knee prosthesis. The knee was revised with bulk tibial allograft to reconstruct a proximal tibial defect. The knee was revised again two years later because of deep infection.
The procedure failed because of loosening of the tibial component in two patients (Cases 17 and 25) in whom the allograft had been placed in the tibia. One of these patients was managed successfully with revision. The allograft was intact and solidly united to the host bone at the time of this revision. The other patient was scheduled for a revision at the time of this review.
Another failure was in a seventy-eight-year-old woman (Case 10) in whom the bulk femoral allograft had been applied to the femur at the second stage of a revision of a total knee prosthesis that had loosened because of infection. Fracture of the graft necessitated additional reconstruction with allograft bone.
The remaining failure was in a sixty-two-year-old diabetic woman (Case 14) in whom the bulk allograft had been used to reconstruct a distal femoral defect during a second revision. An additional operation was necessary for autogenous bone-grafting at the site of a symptomatic non-union at the allograft-host bone junction. Union was obtained with this procedure.
Radiographic Results
Fracture of the bulk femoral allograft was noted in one knee in which the procedure had failed, and loosening of the tibial component was noted in two, as mentioned previously. An additional operation was indicated for all of these knees. One femoral allograft did not unite to the host bone. This was treated with autogenous bone-grafting at the allograft-host bone interface, and union was eventually obtained, as already described.
The distal femoral allograft in one knee in which the procedure was successful fractured six months after the operation; however, it subsequently healed without operative intervention. Periprosthetic radiolucent lines were noted in twelve knees but were incomplete, less than two millimeters wide, and non-progressive.
Union occurred in all but one of the knees in which the procedure failed. Four knees (two that had tibial allograft, one that had femoral allograft, and one that had both) had less than three millimeters of migration of the allograft-implant composite. The migration was not progressive or associated with symptoms in any knee.
Complications
Complications, in addition to those associated with the failed procedures, included necrosis of the skin in one knee and avulsion of the patellar ligament in another; both were treated successfully with an operation.
Encouraging medium-term results have been reported after the use of bulk allograft for reconstruction of defects at the time of a revision total hip arthroplasty2,4,12,22. However, there have been few reports on the use of structural grafts to reconstruct defects around the knee11,13,17,24,25,32,33. Most reports have dealt with the use of morselized allograft bone in the tibia1,8,18,24,26,32,36. The operative technique is extremely important. Proper selection of stemmed components21,32,34 and step-cut osteotomy at the allograft-host bone junction can provide primary stability29 (Fig. 5). Cement must be kept out of allograft-host bone interfaces, and residual autogenous bone graft, preferably with its soft tissue intact, should be wrapped around the allograft-host bone junctions as a vascularized graft29,35 (Figs. 2-B, 3-B, and 5). Residual host bone containing ligament origins and insertions should also be wrapped around the allograft.
We do not recommend the use of plates for the fixation of bulk allografts11,21 because multiple drill-holes produce channels within the allograft that facilitate revascularization and possible failure of the graft6.
A major concern is late infection of the allograft31. Non-vascularized allografts are an excellent nidus for the growth of organisms30. However, the reported rates of infection after major revision operations performed with allograft are similar to those associated with such operations performed without it19,31. Lord et al. reported a 12 per cent rate of infection (thirty-three of 283) in a series in which massive allografts had been used20. In the present series, two of the three knees in which the procedure failed because of infection had been infected before the revision with the bulk allograft.
The results of the present study are relatively short-term with regard to the assessment of structural allografts. However, they are encouraging, particularly when the difficulty of these reconstructions is considered.
We believe that properly applied allograft can be useful in the reconstruction of massive bone defects, can produce stability and support for implants, and can restore bone stock should an additional operation be necessary.