Proximal Femur Reconstruction by an Allograft Prosthesis Composite

Abstract

Twenty-seven patients who had resection of the proximal femur for bone tumors and reconstruction with an allograft prosthesis composite are reported. In most of the patients, the prosthesis was a long-stem revision type, cemented in the allograft and uncemented in the femoral shaft. The abductor muscles and iliopsoas were sutured to the corresponding tendons on the allograft. Implant-related complications and functional results were evaluated and are reported. Twenty-two patients achieved a minimum followup of 36 months (range, 36–126 months; average, 58 months). The implant was removed in two patients (one for infection, one for intraoperative fracture of the allograft). One patient experienced nonunion, whereas in the remaining 24 patients, the allograft eventually united to the host bone. A frequent late complication (17 patients) was fracture of the greater trochanter of the allograft. In the whole series, only four new operations were done for implant-related complications. In 22 patients who could be evaluated, the functional evaluation according to the Musculoskeletal Tumor Society System was excellent in 16 (73%) patients, good in four (18%), and fair in two (9%). These results compare favorably with those of megaprostheses for tumor resection of the proximal femur, where a Trendelenburg gait almost always is present.

Failure of a conventional prosthetic implant in the hip frequently leads to significant bone osteolysis in the femoral shaft. In addition to the problem of a new reliable prosthetic reconstruction, the need for bone stock restoration must be addressed. There are several methods that could be used to approach this situation. The choice usually is made based on the extent of bone to be replaced. ^2,6,19,24 An allograft prosthesis composite has been used as an alternative to other methods for revision of failed conventional total hip replacements with a massively reduced femoral bone stock. ^{1,7,13–15,21} In orthopaedic oncology, allograft prosthesis composites have been used since the late 1980s. (^8,16,17,26

The main advantage, as compared with a resection megaprosthesis, is the effective reattachment of the tendons of the hip abductors and iliopsoas muscles, thereby preventing dislocation and allowing better function. In addition, fusion of the allograft to the host bone and permanent stability of the cement-allograft interface avoid resorption of the host bone around the proximal part of the prosthetic stem because of stem stress-shielding.

An osteochondral allograft is not indicated in reconstruction of proximal femoral resections because fixation of the allograft is less sound than an allograft prosthesis composite and because the allograft almost always will fracture in the femoral head or neck. ^10,17,23 The allograft prosthesis composite is used in younger patients with a longer life expectancy, when substantial portions of the periacetabular muscles and tendons can be spared and it is not planned to use radiation therapy that can interfere with bone healing.

The allograft prosthesis composite also is preferred in cases of short resections (less than 12 cm, mainly in treating benign tumors) because some modular resection prostheses have a minimal length of 12 cm. A series of 27 consecutive patients who had allograft prosthesis composite reconstructions for bone tumors of the proximal femur is reported.

MATERIALS AND METHODS

From June 1989 to December 1996, 27 patients with a tumor of the proximal femur had an en bloc resection and reconstruction surgery with an allograft prosthesis composite. There were 13 male patients and 14 female patients, with an age range from 11 to 64 years (average, 32 years). The tumor diagnosis was osteosarcoma in six patients (four low grade), Ewing’s sarcoma in five patients, chondrosarcoma in five patients, malignant fibrous histiocytoma in one patient, metastasis in one patient, hemangioendothelioma in one patient, giant cell tumor in seven patients, and angioma in one patient. Eight patients received neoadjuvant antiblastic chemotherapy. No patients had adjuvant radiotherapy.

The resection length, measured from the tip of the greater trochanter, ranged from 6 to 22 cm (average, 14 cm). In all patients the rectus femoris was spared, the vastus lateralis was partially or mostly sacrificed, and the vastus medialis and adductor muscles only partially were removed. The tendons of the gluteus maximus, medius, minimus, and iliopsoas muscles were preserved as much as possible and sutured to the corresponding tendons of the allograft (mattress suture, tendon-on-tendon). However, the gluteus maximus tendon was not reattached to its anatomic place, and the fascia lata simply was closed around the reconstruction. The allograft was selected from the bone bank (fresh frozen and stored at −80° C) according to the diameter of the femoral shaft. A long revision stem endoprosthesis was cemented in the allograft, and the composite implant was press-fit inserted (uncemented) in the host femur through a transverse osteotomy line with no supplementary grafts. The prosthetic stem in the host bone was 14 cm long for all prostheses used (Fig 1). In four patients the stem was cemented in the host bony shaft because of age (one patient), diagnosis (metastasis in one patient), and insufficient experience by the surgeon (two patients). In 24 patients the acetabulum was left intact, using a bipolar prosthesis, whereas in three patients, two uncemented components and one cemented acetabular component were used (Table 1).

TABLE 1: Details of the Surgical Technique and Results in the 27 Patients

TABLE 1: (continued) Details of the Surgical Technique and Results in the 27 Patients

Fig 1.: Postoperative anteroposterior radiograph of an allograft prosthesis composite in a 15-year-old girl after resection of a Ewing’s sarcoma. The standard reconstruction is made by a Wagner-type femoral stem and the bipolar cup; the stem is cemented in the allograft and press-fit in the host femoral shaft.

After surgery, all but two patients wore a long spica cast for 3 weeks and then had joint and muscle rehabilitation. Partial weightbearing was permitted at 3 months, and full weightbearing was permitted 6 months after surgery. All patients were followed up clinically and radiographically every 3 months during the first 2 years and then twice or more in 1 year, depending on the diagnosis. The general followup ranged from 11 to 126 months, with an average of 58 months. Twenty-two patients had a minimum followup of 36 months. The other five patients died of metastatic spread of the tumor within 26 months after surgery; one patient had local recurrence of tumor and experienced a hip disarticulation. Pulmonary metastases occurred, and the patient died 1 year later.

RESULTS

Complications

Numerous complications can be attributed to the learning curve of a new technique (Table 2). In one patient the allograft fractured during preparation, becoming unstable and leading to progressive loosening and resorption. The implant was retrieved after 1 year, and a modular megaprosthesis was substituted (fair functional result after 4 years).

TABLE 2: Complications and Outcome

Achievement of good contact at the osteotomy line during insertion of the allograft prosthesis composite is difficult. The conical and flanged shape of the stem can produce a longitudinal fracture of the femoral shaft, which occurred in three patients. In two patients a femoral cerclage was applied in the same operative stage. In all cases the fractures healed (range, 8–12 weeks) with no additional complications.

In five patients a good contact was not achieved and delayed union occurred in four of the patients (range, 13–18 months), whereas only one patient in the current study with cement in the osteotomy line interface had nonunion. The other patients achieved healing within 12 months (Fig 2).

Fig 2.: A followup radiograph of the same patient as in Figure 1 after 2 years. The well-healed osteotomy line is the basis of a successful implant.

Leg length discrepancy occurred in eight patients (range, 1–2 cm) and led to complications in two patients. The first complication was a sciatic nerve palsy that resolved after 6 months. In a 12-year-old boy (Ewing’s sarcoma), the reconstruction was 2 cm longer than the original femoral length. Progressive wear had developed in the superior and external part of the acetabulum. The implant was revised with a good result 6 months after the revision surgery.

Infection occurred 2 months after surgery in a patient with Ewing’s sarcoma treated with neoadjuvant chemotherapy. The infected graft was retrieved, the prosthesis was left in place, and the graft was substituted with antibiotic-impregnated cement spacer around the prosthesis. The infection healed, and the patient still is walking after 4 years (good functional result). In one patient in whom an acetabular prosthetic component with a metal-on-metal joint couple had been used, a chronic allergic reaction to cobalt manifested 40 months later. The prosthesis was revised by insertion of a ceramic acetabular component and a ceramic femoral head. The graft progressively resorbed, but the functional result now is excellent 3 years after revision surgery.

A frequent late complication was fracture of the greater trochanter, which occurred in 17 patients (63% of the series); however, if only the patients with the graft surviving more than 3 years are considered, the percentage increases to 85%. This complication occurred from 6 to 32 months after complete restoration of function (Fig 3). The complication was treated by cast immobilization in two patients and by simple rest in the remaining patients. Only six patients had mild pain during walking, which usually resolved with temporary reduction of weightbearing and motion of the hip. This complication did not lead to permanent restriction of the range of motion (ROM) of the hip or to significant reduction of the functional score.

Fig 3.: In the same patient as in Figure 1, the trochanteric tip detachment was evident 18 months after surgery but does not affect the excellent functional result.

Radiographic Results

Postoperative analysis of the implants was based on the correct placement of the prosthesis into the allograft and host bone. The allograft used was smaller than the host bone in six patients. All other selected allografts were equal to or larger than the original proximal femur. Cementation was considered insufficient (incomplete cement mantle around the prosthesis in the two radiographic projections) in three patients, but no loosening at the allograft-prosthesis interface was detected. The only loosening occurred in the case of graft fracture that eventually was retrieved. In all patients, the trochanteric tip was placed 3 to 32 mm more medial. The acetabular part of the prosthesis, bipolar or standard cup, always was in a correct position. Good contact at the junction site was achieved in 22 patients, and the other five patients had poor contact. Poor contact was defined as a gap wider than 2 mm in two or more cortices according to the two radiographic projections. The press-fit was considered in a correct position in all patients (at least 6 cm prosthesis-shaft contact).

At the longest followup, no loosening at the stem-host bone was evident, except in the case of the intraoperative graft fracture. In 22 patients, the allograft united to the host bone between 3 to 12 months (average, 7 months). Four patients with poor contact achieved union in 12 months, whereas one patient has not achieved union. However, surgery has not been considered because there were no evident signs of loosening between the prosthesis stem and the femoral host shaft. In two patients, marked resorption of the graft was evident: one patient had an allergic reaction, and the other patient had no evident reason detectable. Both patients had an excellent functional score.

There were no signs of acetabular wear in 23 of 24 bipolar prostheses, and no sign of loosening was detected in the three acetabular cups. Dislocation of the prosthesis did not occur, and no failure was detected in the prosthetic components.

Functional Results

Functional results were evaluated in 22 patients having a followup of at least 3 years. The Musculoskeletal Tumor Society scoring system ⁴ was used for evaluation. There were 16 (73%) excellent and four (18%) good results. The two patients who had fair results were the patient with failure of the implant salvaged by a resection prosthesis and the patient with nonunion. No patient had a poor result. There were 20 (91%) satisfactory (excellent or good) results. No patient had pain or an unstable joint; passive ROM of the hip was similar to normal. All patients had good strength of quadriceps, iliopsoas, and glutei muscles, and active abduction of the hip exceeded 30° with the patient lying on the contralateral side and standing. Fourteen patients had no Trendelenburg gait, whereas six patients had a slight Trendelenburg gait. Only the two patients with fair results had a marked Trendelenburg gait and needed a cane for walking indoors. The functional percentage ranged from 75% to 100%, with an average of 92%. Patients with excellent results have the functional ability to do trekking and light but not competitive sports.

DISCUSSION

There are few reports regarding the use of allograft prosthesis composite in tumor surgery. ^8,16,17,26 Some patients experienced symptoms after revision arthroplasty, and other patients experienced symptoms after failure of an osteochondral allograft. In both cases, the results are worsened by the need to minimize surgery and leave bone around the failed procedure. More satisfactory results could be achieved if complete resection of the proximal femur is done. In addition to the risk of infection, which in the current series was 3.7%, some of the complications are related to a learning curve and do not affect the final result. Only two patients had significant loss of the functional result. The choice of an uncemented revision prosthesis coupled with the graft seems to be the better solution for this kind of reconstruction. Avoiding the use of plates while achieving good fusion at the osteotomy line is the basis for a durable implant without the risk of incoming fracture in the graft and material. For joint reconstruction, using a bipolar implant provides good immediate stability and minimizes the socket failure. ²⁰ However, in young patients bipolar arthroplasty can be combined with a valgus necked prosthesis. ¹⁷ The method of gluteus medius and minimus reattachment is a matter of debate. ^9–12,20 Fascia lata insertion is not advisable when a good graft tendon is available, ^9,26 and reinsertion through the host trochanteric bone is not feasible in patients with tumors because of the risk of local recurrence. In addition, with an allograft, the iliopsoas insertion can be replaced, which may be an advantage in early stability and late functional results.

The current authors prefer a tendon mattress suture without the use of drill holes in the graft ⁷; however, trochanteric tip treatment occurred frequently (63%). This complication is related to two factors: the excessive stress caused by the valgus neck-shaft angle of the prosthesis ¹⁷ and the magnitude of trochanteric curettage to allow prosthetic insertion. The prostheses used by the current authors were developed for conventional hip stem revision. In this type of surgery, a valgus neck-shaft angle allows more axially direct weightbearing. Usually the greater trochanter is left in place and in the original position, whereas in tumor surgery, the trochanter often is medialized, consequently magnifying the gluteus vector force in traction. The possible solution is to broaden the prosthetic offset or use an oversized graft. In the first case, the prosthesis would be smaller in width to allow better insertion and to avoid excessive reaming in the graft. However, no disadvantages were seen in the functional results after trochanteric tip detachment. In most patients, the gait performances were comparable to the gait performances of patients who had conventional hip replacement. Many of the patients were able to regain the ability to do usual daily activities with movement and strength.

Few recent reports are available to document functional activity with the use of endoprosthetic replacement in the proximal femur. ^3,18,22,25 Some of these series report that most patients have good functional results (according to the Musculoskeletal Tumor Society system ⁴), whereas with allograft prosthesis composite, 73% of patients had excellent results. A study with gait analysis is being done to better assess the objective performances in these patients in comparison with patients who have endoprosthetic replacement.

There still are some problems using this type of reconstruction. Virus and bacterial transmission are possible, and a good graft selection is needed. In addition, if compared with the modular resection prosthesis, delayed weightbearing and a more complex surgical technique must be considered. However, the advantage of direct tendon suture to the allograft has been addressed, but providing a reliable attachment onto the modular prosthesis, thereby reducing early instability of the system and allowing a better functional pattern with time, has not been clinically proved. ^11,12,20

The current authors use an allograft prosthesis composite in the proximal femur providing that every time after the resection of the proximal femur, the iliopsoas and glutei tendons can be spared enough to be sutured to their counterpart of the graft. Modular prostheses are used in wider resections, when radiotherapy is used, or when a vascular reconstruction is needed.

In the current series, excellent functional results in a high number of patients with relatively few complications were seen. The design of the allograft prosthesis should have sufficient offset and allow for minimal removal of bone from the intertrochanteric area of the graft, thereby reducing the incidence of late fracture of the greater trochanter. The prosthesis should be cemented in the graft and uncemented in the host bone with press-fit insertion because no stress-shielding effect or periprosthetic radiolucent lines were seen using these directions. The head of the prosthesis should be bipolar with the possible exception of patients who are elderly or skeletally immature.