Long-Term Followup of Proximal Femoral Allografts

Abstract

Between 1972 and 1999, the orthopaedic service at the authors’ institution treated 137 patients with proximal femoral bone lesions by resection and reconstruction with frozen allografts from cadavers. The data were entered into a computer program allowing a comprehensive analysis. Thirty-eight patients had proximal femoral osteoarticular allografts, 69 had allograft-endoprosthesis composites, 22 had intercalary allografts, and eight had an allograft arthrodesis. There were 74 complications in 54 of the 137 patients with allografts, which included 26 allograft fractures (19%), 15 infections (11%), and 20 nonunions (15%). The overall success rate for the patients with these procedures was 103 of 137 (75%) at a mean followup of 7.9 years ± 5.6 years. If the eight patients with tumor recurrences (surgical failures rather than allograft failures) are not included, the success rate is 103 of 129 (80%). Eighty-three of these patients (55%) without any allograft complications who have been treated and followed up for at least 2 years had a 92% success rate. It is concluded that although allograft reconstruction after resection of the proximal femur for patients with aggressive or malignant tumors has a high complication rate, approximately 80% of the patients have a successful outcome.

Since the advent of tumor stage classification, improved radiographic imaging, better adjuvant chemotherapy and radiation, and advanced surgical techniques, wide resection of sarcomas and aggressive benign tumors of the proximal femur currently provides an effective method for limb salvage and tumor control. However, these resections create a large tissue defect sometimes generating a difficult skeletal reconstruction problem. A successful reconstruction yields a more acceptable functional result than that with amputation.

In addition, in a series involving the distal femur, amputation has shown no difference in disease-free interval or long-term survival when compared with limb salvage and it can be presumed that the same is true for proximal femoral resections. ²⁸

Various methods have been described for restoration of the proximal femur after tumor resection, including osteoarticular allografts, allograft arthrodeses, intercalary allografts, endoprostheses, and allograft-endoprosthesis composites. Allografts have an advantage over endoprostheses in that they allow some bony host-to-allograft incorporation, preservation of bone stock facilitating future joint revision surgery, and provision of soft tissue attachments for the host limb and joint reconstruction. The end result theoretically is better limb function. The disadvantages in allograft reconstruction are osteoarthritis (in osteoarticular grafts), allograft fracture (necessitating revision for lack of callus), host to allograft nonunion, possible transmission of infectious disease (such as hepatitis C or human immunodeficiency virus [HIV]), and a risk of graft infection. ^{1,4,7,10,17,18,21–23,30,31} However, endoprosthetic devices also share some of these disadvantages. ^{2,5,11,15,24,25,29} In addition, endoprostheses have the additional disadvantages of instability, ^{2,11,15,24,25} and prosthetic loosening and wear. ^29,31 What has been seen in a past study is that although each reconstruction method has its own inherent advantages and disadvantages, no one method has been superior statistically in terms of longevity, function, and revision. ³¹ Adjuvant therapy (chemotherapy and radiation), has led to improved survival of patients with cancer, allowing them longer lifespans, which inherently has challenged the reconstruction longevity.

There have been few investigations directly dealing with the long-term survival of proximal femoral allograft reconstructions with patients with cancer as the main patient population. ^12–14,22 The purpose of the current study was to analyze the long-term survival, complications, and function of proximal femoral allografts based on type of graft (osteoarticular versus intercalary versus arthrodeses versus allograft-endoprosthesis composites).

MATERIALS AND METHODS

Between December 1972 and December 1999, 137 proximal femoral allograft implantations were done at the authors’ institution, mainly for tumors of bone. Patient information was gathered retrospectively from charts and from a computer database. The data were recorded in a computer system and then analyzed using Kaplan-Meier graphs and chi square analyses.

All patients received a thorough oncologic workup including physical examination and radiographic analysis of the extremities and chest, computed tomography (CT) scans of the chest, pelvis, and extremity, a technetium bone scan of the entire body, and magnetic resonance imaging (MRI) of the extremity. All patients were followed up for 2 to 28 years with an average followup of 7.8 years ± 5.6 years. There were 79 males and 58 females in the series with an average age of 43.1 ± 19.5 years (range, 5–79 years). The diagnosis for which the procedures were done included chondrosarcoma (45 patients), osteosarcoma (25 patients), failed total joint replacements or allografts (13 patients), malignant fibrous histiocytoma (10 patients), metastatic carcinoma (nine patients), Ewing’s sarcoma (eight patients), giant cell tumor (seven patients), fibrous dysplasia (five patients), soft tissue sarcoma (four patients), Gaucher’s disease (four patients), lymphoma (three patients) and osteoblastoma, myeloma, osteochondroma, and massive osteonecrosis (one patient each). In 38 of the patients the devices were osteoarticular, and in 69 patients the bone graft was augmented by a total joint replacement (usually a bipolar device). Twenty-two of the patients had intercalary grafts and eight had allograft arthrodeses.

Fresh-frozen allografts were obtained from the institutional bone bank of the authors’ institution after careful processing to reduce the risk of transmission of infection. ²¹ All allografts were sized preoperatively using special radiographs of the graft taken with rulers in place compared with similar imaging of the host bone. After tumor resection, surgical fixation was achieved either by plating (osteoarticular or intercalary grafts), or a long stem hip prosthesis (allograft-endoprosthesis composites). Postoperatively the limb was placed in a brace and the patient was nonweightbearing or partial weightbearing for as many as 8 weeks, followed by a progressive increase in activity and discontinuation of the brace once bony host-allograft union was observed radiographically. Most patients without complications had achieved full function by 4 months.

Determination of physical functional status was evaluated by computer data and review of the patient’s last visit that included physical and radiographic examinations. The grading system of Mankin et al ¹⁹ for functional status was used. A score of excellent meant no tumor recurrence, pain-free, and normal function; a score of good meant no tumor recurrence, pain-free, and moderate restricted activity; a score of fair meant no tumor recurrence, mild to moderate pain, and marked functional limitation; a failure meant a complication necessitating amputation or graft removal. A complication was defined as a condition leading to additional treatment after the primary reconstruction procedure, being either allograft or tumor related. Complications included allograft fracture, infection, nonunion, dislocation, or local tumor recurrence. None of the patients had a dislocation or significant loosening of the stem within the graft. In prior studies, the Mankin score was compared with the Enneking Musculoskeletal Tumor Society system and generally was concordant and slightly more critical. ^10,20,21

RESULTS

One hundred thirty-seven patients in the series have been followed up for more than 2 years and this group was studied carefully for outcome and complications. For the total series of 137 patients, 103 were classified as currently having good or excellent results (75%), five were considered to have fair results (4%), and 29 patients (21%) had failed results (Table 1). If the eight patients with failures as a result of local recurrence are not included, 103 of 129 (80%) patients currently are considered to have good or excellent results at a mean followup of 7.9 ± 5.6 years (Fig 1).

TABLE 1: Effect of Complications of Allograft Surgery

Fig 1.: A Kaplan-Meier plot shows the overall performance of 137 allografts in the proximal femur. The mean estimate of the success rate is 75% (103 of 137) at a mean followup of 7.9 ± 5.6 years (range, 2–28 years).

Eighty-one (59%) of the patients had high-grade tumors and of this group, 21 (26%) have died of disease, 33 (41%) have had metastases develop, and eight (10%) have had a local recurrence. Four patients required an amputation, two for local recurrence and two for uncontrollable infection. In terms of allograft complications, 15 patients (11%) had an infection and all of these patients had failed results (Table 1, Fig 2). Twenty-six of the 137 patients (19%) had an allograft fracture at a mean of 3 years after surgery and the graft success rate after surgical repair for these patients was 46%. Twenty of the patients had a nonunion (15%), which generally responded well to realigning hardware and autograft implantation (Table 1, Fig 2). The success rate for the entire group of 137 patients was 75%. Eighty-three of the patients (55%) had no allograft complications and these patients had a 92% success rate (Table 1, Fig 2). Dislocation did not occur in this series.

Fig 2.: The effect of complications on allograft success rate is shown by this Kaplan-Meier plot. All 15 patients whose allografts became infected were considered to have failed results. The rate of success for the 26 patients with allograft fractures is 46%. The 20 patients with nonunions did well and had an 85% success rate. For the 83 patients who did not have complications, the success rate was 92% (p < 0.00001) (Table 1). I = infection; F = fracture; N = nonunion,; O = no complications

Graft survival varied with the type of graft reconstruction used (Fig 3). The patients with osteoarticular allografts had the poorest results with a mean success rate of 23 of 38 (61%), and the patients with allograft arthrodeses had approximately the same outcome (five of eight (63%) good or excellent results). The patients with allograft plus prosthesis and intercalary grafts did much better with success rates of 56 of 69 (81%) and 19 of 22 (86%) respectively (p < 0.05 by chi square). Age, stage of the tumor, gender, chemotherapy exposure, or diagnosis (Table 2) did not materially alter the success rate. Fifteen of the 38 patients with osteoarticular grafts required total joint replacements at a mean of 5.7 ± 3.1 years (range, 1.4–11.6 years). Of the 137 allografts, 27 were removed between 1 and 150 months (mean, 34 months) after initial reconstruction. Twenty-three of these patients (17%) had a second allograft procedure, and four had an amputation (two for local recurrence and two for infection). An allograft removal regardless of the cause was considered a failure. Of the 23 patients who had a second allograft procedure, four had a third reconstruction procedure, two of which were modular prostheses.

TABLE 2: Tumor Complications by Diagnosis for Primary Tumors

Fig 3.: The outcomes for the patients with the four types of grafts are compared in this graph. The 23 patients with an osteoarticular graft (O) had a 61% success rate, whereas the 56 patients with allograft prostheses (P) had an 81% success rate. There were only eight patients with allograft-arthrodeses (A) and five of these patients were rated as having good results, making the success rate 63%. The 22 patients with the intercalary grafts (I) whose success rate was 86% had the best results (p < 0.03).

Graft survival also was affected markedly by complications, especially for infection, wherein all 15 grafts failed (Table 1, Fig 2). Fracture had a similar effect on graft survival, but less marked in that 46% of patients had a successful result after treatment. Patients with nonunions did well with an 85% success rate. Assessment of patient outcome with time was evaluated assuming all patients started with ratings of excellent or good. The graphed and tabular data suggest that patients with nonunions or no complications did the best long-term and that their functional rating remained stable. Most complications occurred within 2 to 4 years, and dramatically decreased the functional rating then and thereafter.

DISCUSSION

Various methods have been described for restoration of the proximal femur after tumor resection. These include osteoarticular allografts, intercalary allografts, allograft arthrodeses, endoprostheses, and allograft-endoprosthesis composites. Endoprostheses have the advantage of being a technically simpler reconstruction method and also enabling a shorter period to postoperative weightbearing. ^2,5,11,15,31 The rapid rehabilitation also may be ideal in patients with a limited lifespan. Allografts have an advantage over endoprostheses in that they allow some bony host-to-allograft incorporation, preservation of bone stock facilitating future joint revision surgery, and provision of soft tissue attachments which may be functionally important and helpful. ^{3,12–14,20,23} The disadvantages in allograft reconstruction are osteoarthritis (only in osteoarticular grafts), allograft fracture (necessitating revision for lack of callus), host-allograft nonunion, possible transmission of infectious disease (HIV or hepatitis C), a considerable risk of graft infection, and a longer rehabilitation period. ^{1,4,8,10,17,20,22,27,30} However, endoprosthetic devices share some of these disadvantages and also include dislocation as a problem. ^{2,5,11,16,24,25,29,31} Dislocation is uncommon for osteoarticular and allograft-endoprosthetic composite and was not a problem in the current series. Implant osteopenia and loosening have been reported as high as 43% and 45%, respectively, in patients with oncologically related proximal femoral endoprosthetic reconstructions who have been followed up for a mean of 2.5 years, and this is much less likely to occur with allografts. ^11,29,31

An analysis of the series of 137 proximal femoral resections done between 1972 and 1999 secondary to aggressive benign tumors or sarcoma revealed an average followup of 62 months, with 103 (75%) of the patients having good or excellent functional results, five (4%) of the patients having fair results, and 29 (20%) of the patients having failed results. This correlates with previously reported oncologically related proximal femoral allograft series. ^{12–14,20,31} The results also similarly compare with success rates for patients with massive allografts at all locations. ²⁰ If patients with failures secondary to recurrence are not included, 80% of patients could be considered as having good or excellent results, findings similar to those from earlier studies. ^12–14,20 Local recurrence can be thought of as a complication of the resection rather than of the allograft. Nevertheless, it is a cause of allograft removal, which by the current authors’ definition is a failure. Eighty-three of these patients (55%) do not have any allograft complications and have been treated and followed up for at least 2 years and as many as 28 years postoperatively with a 92% success rate.

A difference existed between the osteoarticular allograft and all other allograft reconstruction types (Fig 3). Sixty-one percent of patients with osteoarticular allografts were rated as having excellent or good results, whereas 81% of patients with allograft-endoprosthesis composites had good or excellent results. Only 68% of the eight patients with allograft arthrodeses had successful results. The 22 patients with intercalary procedures did very well with an 86% success rate. These data correlate with other study findings that show that patients with intercalary or allograft-endoprosthetic composites in any part of the body tend to have the best functional results compared with results in patients with osteoarticular allografts. ^{3,6,8,13,20,22} Fifteen of the 38 patients (39%) with osteoarticular grafts required total joint replacements at a mean of 5.7 ± 3.1 years (range, 1.4–11.6 years). Osteoarthritis is a known problem for patients with osteoarticular allografts, occurring at an average of 6 years after the procedure, necessitating a 16% total joint replacement rate in patients with distal femoral, proximal tibial, and proximal femoral grafts. ^21,31 The symptoms of osteoarthritis usually are less severe clinically than the radiographs would imply, and many patients only require observation. Biopsy specimens from these sites have shown abnormal but viable tissue, with abnormalities different from those of related osteoarthritis unrelated to allografts. ³ Fit, congruity, and quality of preserved cartilage may play a large factor in the decreased survival and success of osteoarticular allografts in patients with accelerated arthritis. ⁶

Of the 137 allografts, 27 were removed between 1 and 150 months (mean, 34 months) after the initial reconstruction. Twenty-three patients (17%) had a second allograft procedure that was successful. An interesting comparison of allograft-endoprosthesis composites and endoprostheses is their temporal relation to failure. As shown in the current study for allograft-endoprosthesis composites, if they fail, they do so by 3 years, but remain stable thereafter. However, in some studies with oncologic endoprostheses, the constructs have been shown to continue to fail after 3 years without plateau. ^11,24,31

Infection was a devastating complication, leading to failure in all grafts. This correlated with the findings of other studies. ^{12,13,17,18,22,31} Such a high infection rate (11%) is concerning, given that proximal femoral endoprostheses have a reported infection rate of 6% or less. ^{2,11,15,24,31} Certain factors intrinsic to tumor surgery, such as wide resection, skin slough, adjuvant radiation and chemotherapy, and multiple operations, all potentiate a high risk of infection. ¹⁷ Prophylactic measures against allograft infection include meticulous hemostasis, appropriate drain use, and postoperative parental and oral antibiotics. ²³ The avascular nature of allografts may make them more susceptible to infection. ²³ In addition, it has been suggested that a host-versus-graft reaction may occur to a variable degree with each patient, effectively walling off the graft and inhibiting a good blood supply. ¹³ However, this high complication rate may be outweighed by the benefits of allograft over endoprosthesis mentioned above. Subsequent allograft implantation after a failed infected allograft has not been shown to affect union rates or provide an increased risk of subsequent infection in the new graft. ²⁶ Chemotherapy did not seem to appreciably alter the outcome of allograft survival. ²⁴

Allograft fracture occurred in 19% of patients at an average of 3 years after surgery. Notwithstanding the known fact that allograft fractures generally form little or no callus despite a viable cortex, 26 of these patients were able to achieve a 46% success rate after surgical repair. ^1,9,20 Revision surgeries after such a complication have been shown to be successful. ^1,30 It is known that the creeping substitution that takes place in these grafts is incomplete for the lifetime of the host, and that this process also weakens the graft because it undergoes osteoclastic resorption during revascularization. ³⁰ Because of this, Mnaymneh et al ²³ advised that the rigid internal fixation never should be removed. Allograft fracture often occurs at cortical defects most commonly created by fixation screw holes, and there is an increased allograft fracture rate when using plating versus intramedullary methods of fixation. ^1,3,30 The fracture rate for the current study is higher than those of other studies (0%–16%), ^1,23,31 but it compares similarly with earlier studies for proximal femoral allografts done at the authors’ institution. ^{1,12,13,14,20}

Nonunion was the least devastating to allograft survival of all the complications, accounting for 15% of cases, but yielding an 88% success rate after surgical revision of hardware and autografting. This nonunion rate is higher that that of other studies (3%–12%). ^13,20,23 Rigid internal fixation and close host-allograft bone opposition (< 3 mm) is a known prerequisite for successful allograft union. ^25,30 An allograft in the proximal end of the femur must derive its blood supply from invasion of the vessels and creeping substitution from the recipient site, which may have less vascularity as compared with other anatomic sites. ⁶ A combination of these factors may have been the cause of a high nonunion rate. In addition, nonunions have been correlated with a higher allograft fracture and infection rate, which may account for the numbers seen in the current study. ^1,10 There were two infections in patients with fractures (both patients had failed results), and three nonunions in patients with fractures (one was designated as having a failed result) (Table 1).

The patient’s age, tumor type, diagnosis, gender, or chemotherapy did not materially alter the success rate. The fact that chemotherapy did not affect the success rate corresponds with other published findings. ^4,7 In addition, one large series showed that at least for osteoarticular allografts of the lower extremity, chemotherapy has no effect on infection, fracture rates, or even union rates for proximal femurs and proximal tibias; only distal osteoarticular allograft femur union rates were selectively adversely affected by chemotherapy for unknown reasons. ⁷

It may be concluded that allograft transplantation after resection of proximal femoral malignant or aggressive tumors has a moderate success rate, not dissimilar to that reported for the modular or custom prostheses at the same site. This is a finding similar to earlier studies of proximal femoral allografts done at the authors’ institution. ^12–14 One has to be cautioned when comparing these results with those of other studies of proximal femoral allografts, which are primarily joint revision and nontumor related cases. ^6,27 The host abductor mechanism is largely spared in joint revision surgery which is not always so in resections of tumors of the proximal femur. This would bias the functional results with patients having revision surgery having more favorable results. In addition, tumor resections require greater dissections and sacrifice of tissue structures, potentiating a higher complication rate.