Hip replacement is fortunately rarely required in children, but it may become necessary in young patients with a tumor of the proximal part of the femur who need a proximal femoral replacement after tumor resection. Few previous reports have described the effects of the procedure on the acetabulum and on the hip reconstruction itself following proximal femoral resection in this age-group.
In children, growth of the triradiate cartilage is responsible for the increase in the height and width of the acetabulum. The deepening of the socket requires appositional growth in the periphery and the presence of a round femoral head within the acetabulum1. There is normally a rapid decline in the acetabular angle during the first three years of life, followed by a further decrease of only 5° until the adult value is reached at around eight years of age. In the normal hip, there is no further acetabular growth after this age, but there is a potential for further acetabular development through the accessory ossification centers2,3.
Although endoprosthetic proximal femoral replacement in children is an acceptable method to treat a malignant bone tumor and has considerable advantages over radiation therapy or hip disarticulation, the best way to restore hip function in young children remains uncertain4.
Little is known about the development of the acetabulum without a natural femoral head to mold it, and there is little information regarding the development of the pelvis or of the hip itself after replacement of the joint in children. Both Bucholz et al.5 and Hallel and Salvati6 reported that premature closure of the triradiate cartilage following trauma leads to subsequent subluxation of the femoral head.
There have been few reports about prosthetic replacement of the hip in children, although some authors have included children in large series of hip replacements for tumors or juvenile arthritis. The use of a bipolar hemiarthroplasty in children who are between six and fourteen years old following tumor resection has been reported to lead to superior and lateral migration of the femoral head in all patients, resulting in a 20% revision rate by five years7. This migration has been attributed to three factors: lack of acetabular remodeling, increased wear of the acetabulum from the metal head, and weakening of the hip abductor muscles, which are then no longer able to oppose the stronger hip adductor muscles. This contrasts with the normal situation in adults in which bipolar prostheses usually show medial migration8,9, although other reports have not found this in younger patients10,11. An isolated case report of a patient who had a femoral hemiarthroplasty at the age of six years documented a satisfactory outcome after seventeen years of follow-up12,13.
Reports of hip replacements for juvenile chronic arthritis have included many young adults, with some under the age of sixteen years, but, on the whole, these patients have shown encouraging results with low revision rates14-16.
Previously, we documented the durability of proximal or total femoral replacements in children and identified acetabular complications leading to revision. This study represents a more detailed investigation of these complications4,17. Our aim was to assess the durability of the implant in children and adolescents (under the age of sixteen years) who underwent a hip replacement as part of the treatment for a tumor in the proximal part of the femur.
Between 1982 and 2002, forty children who were between two and fifteen years old (mean, 11.2 years) underwent proximal femoral resection for a primary malignant bone tumor or other destructive condition. Twenty-four patients had Ewing sarcoma, fourteen had osteosarcoma, one had a malignant peripheral nerve sheath tumor, and one had Gorham (disappearing bone) disease. Nineteen patients died of tumor-related disease. Of the twenty-one disease-free survivors, two were lost to follow-up after three and eighteen years. The remaining nineteen survivors were followed for three to twenty-three years (mean, 12.6 years for survivors and 2.9 years for those who died) (see Appendix). Institutional review board permission was obtained to complete this study.
All patients underwent intra-articular resection of the proximal part of the femur, including the greater trochanter, and received custom-made femoral replacements. The length of the bone resection was assessed with use of computed tomography or magnetic resonance imaging, and clear distal resection margins were confirmed by histology. Thirty-six patients had a proximal femoral replacement (twenty-four were fixed, and twelve were extendable), and four patients had total femoral replacements (two of which were extendable). In all cases, the abductor muscles, which had been detached from the greater trochanter, were attached to the fascia lata rather than to the implant. The psoas muscle was divided and left to reattach to scar tissue without being formally repaired. The extent of the soft-tissue resection was largely based on the extent of residual tumor involvement after neoadjuvant chemotherapy, and it varied from patient to patient. The hip joint was replaced with one of three types of reconstruction: (1) Nearly all children who underwent surgery before 1991 received a cemented high-density polyethylene acetabular cup; (2) After 1991, a unipolar hemiarthroplasty component, sized to match the removed femoral head, was used in most patients. This was done because the unipolar components became available at that time and other patients who had a proximal femoral replacement with a small head and a cemented cup had a high rate of dislocation; and (3) Since 1997, uncemented acetabular components were used if suitably sized implants were available.
Cemented acetabular cups were used after 1991 in patients in whom the surgeon suspected possible tumor involvement of the joint because it was our belief that complete removal of the hip capsule and ligamentum teres followed by reaming of the acetabular cartilage could improve the margin of excision. In children with extendable prostheses, we ensured that the affected limb never became too long, thus maintaining sufficient abduction of the hip to try to prevent lateral subluxation. There was no acute hip subluxation after any of the lengthening procedures.
All patients received chemotherapy according to the appropriate protocol for the tumor type. Four patients with Ewing sarcoma received radiation therapy, which was administered preoperatively in two of them and postoperatively in the other two.
The clinical records and serial radiographs of all patients were reviewed, and the clinical and radiographic outcomes of the different types of hip reconstruction were documented. The size of the outer diameter of the implanted acetabular components was recorded and plotted against the age at which the patients had the initial surgery (Fig. 1).
Survivorship of the patients and of the prostheses was estimated with use of the Kaplan-Meier method, with failure defined as either the death of the patient or revision of the acetabular component of the prosthesis. Because of the low numbers of patients, confidence intervals are also given, and, when used graphically, they are shown as actuarial survival.
Fifteen (56%) of the twenty-seven children who were between the ages of six and fifteen years when they received a cemented acetabular component were still alive, and six of the ten patients who were between the ages of two and fifteen years at the time of unipolar hemiarthroplasty survived to the time of follow-up. Uncemented acetabular components were used in three older children (twelve, thirteen, and fifteen years old), all of whom died from metastatic disease within four years of surgery.
Oncological Results
Seventeen patients died from metastatic disease, and two died from chemotherapy-related cardiotoxicity. One patient required a hindquarter amputation for radiation-induced osteosarcoma of the pelvis arising five years after initial treatment with surgery and radiation therapy for Ewing sarcoma. The overall patient survival rate was 58% at five years, 55% at ten years, and 53% at fifteen years (confidence interval, ±16%). Six patients had a local recurrence, and five of them had concomitant metastases. All of the local recurrences were in the soft tissues around the shaft of the prosthesis, and none were either in or around the hip joint itself. Three patients underwent additional surgery; one of them had excision and two had hindquarter amputation. The other three had palliative treatment with radiation therapy, but all six died of metastatic disease. One patient with uncontrollable infection also required amputation.
Results by Type of Hip Reconstruction
Cemented Acetabular Component
Of the twenty-seven children who had a cemented acetabular component, fifteen survived and two of these were lost to follow-up. In the ten who had their first limb salvage operation before the age of eleven, acetabular growth and development caused the implanted cups to become increasingly vertical, with loosening of the acetabular component (Fig. 2). Four of the surviving seven children required acetabular revision for implant loosening at four to nine years after implantation, and one had a hindquarter amputation for infection. Only two children in this group, who had their first operations at eight and nine years of age, remained well without further intervention at seventeen and twenty-three years of age, respectively.
Of the eight survivors who were over eleven years of age when they received a cemented acetabular component, only one needed revision for mechanical loosening at five years after implantation. The other cemented cups remained in situ from seven to twenty-one years. Two patients in this group had dislocations, which were treated by reduction with the patient under anesthesia, followed by the use of an abduction brace and physical therapy for six weeks; neither patient required revision surgery. There were three deep infections, all of which resolved with early aggressive treatment. The survival rate of these cemented hips without revision was 76% (95% confidence interval, ±29%) at ten years.
Unipolar Prostheses
Ten patients, who were between two and fifteen years old at the time of the first operation, had a unipolar hemiarthroplasty (Fig. 3-A). Four children in this group died as a result of the malignancy, and one of them had required a revision for infection after twelve months. The six survivors in this group were followed for between three and fourteen years. All but one had femoral head subluxation within nine years of implantation (Fig. 3-B). In two patients, each two years old at the time of the first operation, shelf procedures were carried out in an attempt to contain the femoral head (Fig. 3-C). A triple osteotomy was done in two children, but none of these acetabular augmentation procedures were successful. One child had an acetabular reconstruction with bulk allograft (Fig. 3-D). All of these children had stable and pain-free hips following conversion to uncemented metal-on-metal components at the ages of ten, eleven, and thirteen years (Fig. 3-E). One patient who was fifteen years of age at the time of insertion of a unipolar implant retained the implant for almost eight years, but then she began to experience some pain, suggesting that hip revision may be needed in the near future.
Uncemented Prostheses
Of the three older children (twelve, thirteen, and fifteen years old) who received a primary uncemented acetabular component, two died of the disease within a year after the operation. The third had no problems in the hip but died four years later of tumor-related disease.
Premature fusion of the triradiate cartilage occurred in one eight-year-old child who had received a cemented cup when she was three years old and in one seven-year-old child with a cemented cup who underwent postoperative radiation therapy. This child subsequently had a deep infection that was treated by amputation.
The overall survival of all hip replacements, with failure defined as revision for any reason, was 74% at five years and 47% at ten years (95% confidence interval, ±21.8%) (Fig. 4). Better results were seen in the twenty-two children who were more than eleven years old at the time of hip reconstruction (a survival rate of 75% at ten years compared with 25% for those who were less than eleven years old), but, owing to the low numbers, this difference was not significant (Fig. 5). In this series, all unipolar femoral prostheses had failed by nine years, while 62% of the cemented cups were functioning well after ten years (Fig. 6). There were no failures in the patients who had uncemented acetabular reconstructions, but all of these patients died from tumor-related disease within four years after the implantation.
Results in Children with Extendable Prostheses
Fourteen children between the ages of two and twelve years had an extendable endoprosthesis implanted. Five of these children died before they had a lengthening procedure, but the others had between one and twenty-four lengthenings (median, five lengthenings). The amount lengthened each time was usually between 6 and 10 mm, and the total lengthening achieved ranged from 6 to 150 mm. Of the skeletally mature patients at the time of final follow-up, only two had a limb length difference of >2 cm. There were no acute dislocations of the hip following any lengthening procedure. One child had an infection develop and required an amputation, while seven patients subsequently had a revision of the acetabular component at a time ranging from twenty-two to 103 months after the initial operation. One child had a well-fixed cemented cup fourteen years after implantation of a total femoral extendable prosthesis and seven lengthenings. In view of the young age of these patients, it is impossible to determine whether the lengthening procedures themselves led to an increased risk of failure of the acetabular component.
Revision Surgery
Of the forty children in this series, thirteen had revision surgery and three had an amputation with one child later requiring amputation after multiple failed revisions. Sixteen died at a mean of twenty-six months after the initial operation without requiring revision or other major surgery, and nine survived for an average of twelve years after the original operation without the need for revision. There were thirteen first revision procedures, seven of which took place after lengthening of extendable prostheses.
On the whole, cemented cups were revised to another cemented implant while the unipolar prostheses were revised to uncemented metal-on-metal implants. The rationale for this management was based largely on the technology available at the time of surgery. At the time of the latest follow-up, nine patients had the original hip replacement in place, four had had one revision (three to a cemented cup and one to a metal-on-metal uncemented cup), two had had two revisions (one to an uncemented cup and one to a metal-on-metal cup), four had had three revisions, and one had had four revisions. The final patient remained alive following an amputation for infection.
Results in Patients Who Had Radiation Therapy
Five patients with Ewing sarcoma received radiation therapy at the ages of seven years (one patient), twelve years (one patient), and fifteen years (three patients). The seven-year-old patient had an infection develop after several femoral lengthening procedures and eventually required an amputation to treat uncontrolled infection. One patient died after forty-seven months because of metastatic disease. The twelve-year-old patient had no complications from a unipolar hip replacement, but a radiation-induced sarcoma of the pelvis developed after 5.6 years and was treated by hindquarter amputation. He subsequently had metastatic disease develop and died. Of the remaining two patients, one was symptom-free and had had no additional surgery for 21.3 years, while the other had loosening followed by revision to an uncemented cup complicated by infection controlled, at the time of writing, with long-term antibiotics. He was pain-free and walking with a cane seventeen years after the initial treatment.
Hip replacement is seldom required in children, and thus no report on the subject has been published, as far as we know, except with regard to children with tumors. In a child with a tumor of the proximal part of the femur, no standard method has been recommended to preserve hip function. In very young patients, the use of a vascularized fibular graft inserted inside an allograft to replace both the proximal part of the femur and the femoral capital epiphysis has been reported18. A more aggressive alternative is a type-B-IIIa rotationplasty, in which the entire femur is resected and the upper tibial condyle is inserted into the hip joint19.
The challenge in children who need a hip replacement following resection of a tumor is to provide not only a functional and pain-free hip but also one that will last and will not cause frequent complications.
The types of hip reconstruction that we used were determined partly by the available technology at the time of the procedures. Our first proximal femoral replacement for malignant disease in a child was carried out in 1982, and during the following ten years, we exclusively used the small cemented acetabular cups that were then available, with the smallest having a 35-mm outer diameter and 25-mm internal diameter. In 1991, when treating a three-year-old child in whom we removed a 32-mm femoral head, we chose a unipolar implant with a head of the same size. We more recently used uncemented acetabular components in older children; however, because we were impressed with the simplicity and low dislocation rate of unipolar implants in adult patients who needed proximal femoral replacement in our own center, we increasingly used them in younger patients with tumors as well. Unipolar hemiarthroplasty is associated with acetabular protrusio in older patients who have had a hemiarthroplasty to treat a fractured neck of the femur. We have not found any reports of protrusio with such implants in children20 and we assume that the decreased abductor muscle force and the better bone quality in children were responsible for this.
Dislocation of the hip after proximal femoral replacement for a tumor is common and is related to the loss of muscle power around the hip21. Dislocation rates are decreased with use of a large femoral head, which is one reason why we used unipolar femoral components when they became available. While this led to a decrease in the risk of dislocation, it appears to have led to an increased risk of late hip subluxation. It is hoped that the use of a large femoral head-resurfacing type of prosthesis will resolve both of these problems of hip instability.
Our results show that, in children over the age of eleven years, uncemented acetabular components work well in the short term. Because we had no patients with an uncemented cup who survived beyond four years, we cannot comment about long-term results. The cemented cups also functioned well in this age-group, but we believe that loosening and wear become inevitable with the passage of time, as we have already seen this in the patients requiring revision surgery. Our results resemble those of Witt et al., who reported on a group of young patients requiring hip replacement for juvenile chronic arthritis15.
Hip hemiarthroplasty with use of unipolar proximal femoral replacements has been shown to work well in adults, but our experience with the procedure in children has been disappointing, with all prostheses failing by ten years. We have had insufficient experience with the use of bipolar femoral heads to predict whether these would fare better, but Manoso et al., using bipolar components, had results remarkably similar to ours with the unipolar implants7.
Children over the age of eleven years at the time of the first operation underwent revision arthroplasty primarily for pain relief. In the younger patients, revision was done mainly for prosthetic femoral head subluxation, similar to the patients in the report by Manoso et al.7.
The dilemma regarding the best treatment for children under the age of eleven years who need a hip replacement after tumor resection has not been resolved. In our patients, cemented acetabular components performed well in the short term, but in some patients the pelvis continued to grow and the cemented cup became almost vertical in the acetabulum. Revision of these implants to other cemented components has not been uniformly successful in our hands. It has been suggested that fusion of the triradiate cartilage at the time of hip replacement may prevent this problem; however, we did not use this technique in the present series, and we would be concerned about potential abnormal growth of the pelvis in younger patients.
Once the first hip replacement fails, further procedures have had varying degrees of success. On the whole, additional revisions with a cemented acetabular component have not worked, but revisions with an uncemented component have done well, provided that a good fit and osseous integration within the pelvis could be achieved. Our most successful revisions thus far have been those in which we used uncemented metal-on-metal components (Birmingham Hip; Smith and Nephew, Memphis, Tennessee), which have a modular taper on the femoral side to provide immediate and secure fixation. The hydroxyapatite-coated acetabular implant relies on bone integration for fixation, and the main advantage of the implant is its large head size, which reduces the risk of dislocation. The success of these revisions has led us to use an uncemented metal-on-metal type of replacement in all children over the age of eleven years who undergo proximal femoral excision and in whom the procedure is technically possible, although long-term data on this approach are lacking. The smallest size of most uncemented acetabular components is 44 mm (outer diameter), and this is likely to be too large for most children under the age of twelve years.
Concerns remain about the safety of any artificial implant in a child, and there have been particular concerns recently about metal debris after metal-on-metal hip replacements22,23. All of these children, however, already have in situ a large metallic endoprosthesis, which can itself produce elevated serum metal ion levels24. The risk of possible carcinogenicity from these metal ions also has to be offset against the sixfold to tenfold increased risk these children have of a second malignancy as a result of the treatment they have undergone with chemotherapy and sometimes radiation therapy25-27. Thus, the risk of using a metal-on-metal prosthesis in younger patients remains unknown, and the patient and his or her parents need to be warned of this potential additional risk.
There remains the question of what is the best procedure for very young children requiring a hip replacement. The simplest and least destructive solution initially is to use a hemiarthroplasty with a unipolar or bipolar head, but to revise it to an uncemented acetabular component when pelvic growth is complete. We agree with Manoso et al.7 that there is benefit to an increased initial varus position of the prosthesis, but we doubt that soft-tissue reconstruction would improve abductor strength. We have found other options, such as a shelf procedure or a triple osteotomy to prevent subluxation of the femoral component, to be unsuccessful, and we no longer recommend these procedures.
We accept the very substantial shortcomings of this investigation, with a small number of patients and the long time during which the study was carried out, but these outcomes will allow others to make more reasoned treatment decisions when faced with similar problems.
We have not reported functional outcomes in this group of patients as they were not collected prospectively, and such outcomes would to a large extent depend on the amount of soft-tissue resection required to completely remove the sarcoma. The relative strength of the abductor compared with the adductor muscles in contributing to failure may be a relevant point and is worthy of further investigation. Interestingly, Unwin et al. found in adult patients that the more bone that was resected in proximal femoral replacements, the less the risk of failure, and they postulated that this was due to the reduced offset of the tip of the stem28. They did not explore the possibility of altered muscle dynamics on the prosthesis. The main point of this study has been to demonstrate survivorship of the implants. Similarly, we have not commented in further detail about the oncological outcomes, which have been extensively reported elsewhere29,30. The decision as to whether to carry out limb salvage as opposed to amputation for children with a tumor around the hip can be an emotional one, and the development of local recurrence is disastrous. Whether the eventual outcome can be improved by carrying out more amputations remains controversial31. In this series, however, all but one of the children who had a local recurrence develop already had concomitant distant metastases and thus a poor prognosis32.
In conclusion, restoration of hip function in a growing child after removing the proximal part of the femur is a challenging problem for which there is no clear-cut solution. We recommend that at the initial hip surgery an attempt should be made to cause as little damage as possible to the acetabulum and that hemiarthroplasty with a unipolar or bipolar head should be used in children under eleven years old, although eventually all of our patients required revision. We also recommend that the affected limb should be kept slightly shorter than the contralateral limb to maintain the hip in relative abduction. The patient and family should know that revision may be needed as the child gets older. Biological reconstruction of the hip should be considered in very young children, but in older children we recommend arthroplasty with uncemented components although the long-term results remain uncertain. In children over the age of eleven, both cemented and uncemented acetabular components appear to perform well.