Patients
We reviewed the records on 123 consecutive total hip arthroplasties that had been performed by one of us (M. E. M.) between January 1, 1981, and December 31, 1986, for the treatment of coxarthrosis due to congenital dysplasia of the hip. A minimum of five years of follow-up was required for inclusion in the study. After the patients who had mild dysplasia (type I, according to the system of Crowe et al.4) were excluded, seventy-nine patients in whom ninety-eight total hip arthroplasties had been performed with use of an acetabular reinforcement ring were identified. Of these patients, two had died, four were unable to return for the study, and three could not be located. The study group thus consisted of eighty-seven reconstructions in seventy patients. Seventeen patients had a bilateral reconstruction; eleven of them had both procedures performed during the same operation.
All patients were examined by one of us (M. E. M.). The patients answered a questionnaire regarding their ability to perform a variety of activities of daily living, the presence or absence of pain, their functional status compared with the status at their previous visit, whether their expectations had been fulfilled, and how they rated the outcome of the operation. All patients had a radiographic examination at the time of the clinical evaluation. Clinical data were recorded prospectively with use of the International Documentation and Evaluation System (IDES-4).
The average age at the time of the index operation was fifty-two years. According to the classification of Crowe et al.4, eleven hips had type-IV acetabular dysplasia, sixty-five had type-III, and eleven had type-II. A previous femoral osteotomy had been done in thirty-three hips, and a previous arthrodesis had been done in two. Two hips had had a previous infection. Thirty-three hips were considered severely painful; forty-seven, moderately painful; and seven, mildly painful. Eighteen patients needed medication for pain regularly, and forty-three needed it occasionally.
Twenty-nine patients had advanced degenerative changes in the contralateral hip. A previous femoral osteotomy had been done in ten contralateral hips; a previous total hip arthroplasty, in eight; and a previous arthrodesis, in one. Fifty-three patients stated that their walking was restricted by both hips. Fifty patients were unable to walk without support, fifteen were able to walk less than thirty minutes without support, and five were able to walk more than thirty minutes without support. The range of flexion of thirty-five hips was less than 30 degrees preoperatively, and four other hips were ankylosed. Sixty-one patients had a preoperative limb-length discrepancy, which was more than three centimeters in twenty-five.
Operative Technique
The operation was performed through a transtrochanteric approach in every patient. Sixty-one chevron-type trochanteric osteotomies were performed to enhance exposure. Femoral hardware was removed at the time of nine of the arthroplasties. Acetabular reconstruction was performed with the Müller acetabular roof-reinforcement ring and cementing of a polyethylene cup in all eighty-seven hips. The most commonly used reinforcement ring had an outer diameter of forty-four millimeters. Autologous graft from the femoral head was used in forty-two hips.
Eighty femora were reconstructed with the Müller straight-stem component for congenital dysplasia of the hip, and seven were reconstructed with the standard Müller straight-stem component (both components, Sulzer). Eighty-five stems were cobalt-chromium, and two were titanium. The most commonly used stem was size ten. Eighty-six femoral-head components were twenty-two millimeters in diameter, and one, which was used early in the series, was thirty-two millimeters in diameter. Low-viscosity cement was used in fifty-seven reconstructions (66 per cent), and high-viscosity cement was used in thirty (34 per cent). Both the femoral component and the cup were cemented in eighty-six hips, and only the cup was cemented in one. The average operative time was two hours and fifteen minutes (range, seventy-one minutes to four hours).
Postoperatively, all patients received warfarin for prophylaxis against deep venous thrombosis. Since 1984, all patients received antibiotic prophylaxis preoperatively and for the first thirty-six hours postoperatively.
Radiographic Examination
All of the radiographic examinations were performed by one of us (T. J. G.). Because standard methods of radiographic evaluation of the acetabular component are not applicable when an acetabular reinforcement ring has been used, we developed a system for the evaluation of cups with such rings. A cup was considered definitely loose (type III) if screws used to fix the ring were broken; if there was a complete, progressive radiolucent line medial or superior to the ring or around the screws; or if there was evidence of migration of the cup or of fracture of the cement. Probable loosening (type II) was considered to have occurred if there was an incomplete progressive radiolucent line medial or superior to the ring. A cup was possibly loose (type I) if there was non-progressive radiolucency that did not involve the screws. The presence and degree of wear of the cup, the grade of heterotopic ossification according to the system of Brooker et al.2, and the presence and progression of any acetabular osteolysis were also recorded.
Loosening of the femoral component was defined according to the system of Barrack et al.1. A stem was considered definitely loose if there was evidence of subsidence, debonding, or fracture of the component. Stems that were surrounded by a continuous radiolucent line but had not subsided were considered probably loose, and non-progressive radiolucent lines of less than two millimeters in thickness were considered an indication of possible loosening. Areas of osteolysis and the degree of any calcar resorption were recorded as well. Areas of femoral radiolucency were defined according to the system of Gruen et al.11.
Radiographic analysis of the bone graft was performed on radiographs that were made postoperatively and at each follow-up examination. The follow-up radiographs were compared with the immediate postoperative radiograph to identify the development of radiolucency or resorption of the graft.
Clinical Evaluation
Function and Deformity
The average duration of follow-up was 9.4 years (range, five to fifteen years). Sixty hips (69 per cent) were described as having an excellent result; twenty-three (26 per cent), a good result; two (2 per cent), a fair result; and two, a poor result. Thirty-four patients (49 per cent) reported no limitations in walking due to the hips, and fifty-nine (84 per cent) needed no supportive devices for walking. Fifty-two patients (74 per cent) had a normal or nearly normal capacity for work or recreation. The ability to climb stairs in a reciprocal manner was restored to fifty-nine patients (84 per cent). Four hips (5 per cent) were described as causing "start-up pain," one hip (1 per cent) was painful after the patient had walked for thirty minutes, and one hip was painful at night. Two patients (3 per cent) needed regular medication for pain, and fourteen (20 per cent) needed it occasionally. No association was detected between the age of the patient and the clinical or radiographic result, with the numbers available.
The lengths of the limbs, as measured clinically, were equal in forty-four patients (63 per cent), differed by one centimeter or less in eighteen (26 per cent), differed by more than one to two centimeters in seven (10 per cent), and differed by four centimeters in one (1 per cent).
Revisions
There were nine revisions (10 per cent). One revision was performed because of aseptic loosening of the acetabular component; one, because of aseptic loosening of the femoral component; one, because of aseptic loosening of both components; and six, because of infection.
The failures of the acetabular components were analyzed according to the severity of the dysplasia. None of the eleven hips that had type-II dysplasia had a failure of the acetabular component. Of the sixty-five hips that had type-III dysplasia, none had failure of only the acetabular component and one had aseptic loosening of both the acetabular and the femoral component. Of the eleven hips that had type-IV dysplasia, one had aseptic loosening of only the acetabular component.
There was a highly significant difference between the rate of revision in the hips in which an acetabular reinforcement ring and cement had been used and the rate in the hips in which an acetabular reinforcement ring had been used alone or with bone graft (p < 0.001). Six (17 per cent) of the thirty-six hips in which cement had been used to fill osseous defects in the acetabulum were revised (Figs. 1-A, 1-B, 1-C and 1-D). One revision was done because of aseptic loosening of the acetabular component; one, because of aseptic loosening of both components; and four, because of infection. Three (6 per cent) of the fifty-one hips in which an acetabular reinforcement ring had been used alone or with bone graft to fill osseous defects (Figs. 2-A, 2-B, and 2-C) were revised. Of note, two of the revisions in this group were due to infection, one was due to aseptic loosening of the femoral component, and none were due to aseptic loosening of the acetabular component.
Structural Bone-Grafting
Structural grafting with use of autologous bone from the femoral head was performed in all eleven hips with type-IV dysplasia and in thirty-one (48 per cent) of the sixty-five hips with type-III dysplasia. One of the eleven hips with type-IV dysplasia and a structural graft had aseptic loosening of the acetabular component, and one (3 per cent) of the thirty-one hips that had type-III dysplasia and a structural graft had aseptic loosening of both components.
Radiographic Examination
All hips were reduced to within one centimeter of the true acetabulum, as measured on anteroposterior radiographs. As a result, the relationship between the results of acetabular reconstruction and a high hip center could not be analyzed in this study.
Acetabulum
Of the seventy-eight unrevised acetabular components, five (6 per cent) were classified as definitely loose; two (3 per cent), as probably loose; and thirteen (17 per cent), as possibly loose. Three hips (4 per cent) had evidence of superior migration of the acetabular component (less than five millimeters), and three had evidence of slight medial migration; one of the three had mild protrusio. Two hips (3 per cent) had a continuous radiolucent line around the acetabular construct. Non-progressive, incomplete radiolucent lines were present medially in thirteen hips (17 per cent). No hip had progressive tilting of the cup or radiolucency around the screws.
Twelve (33 per cent) of the thirty-six acetabula that had been reconstructed with an acetabular reinforcement ring and cement to fill the osseous defects had evidence of non-progressive, incomplete radiolucent lines. (The risk of radiolucency was especially high when cement had been used medial to the ring.) However, only one (2 per cent) of the forty-two acetabula that had been reconstructed with an acetabular reinforcement ring alone or with bone graft had evidence of radiolucent lines. This difference was significant (p < 0.0005).
All bone grafts had fully healed, with no visible resorption. There was no lateral uncovering or lack of osseous support of the ring in the hips in which bone graft had been used.
Six (7 per cent) of the eighty-seven hips had grade-III or IV heterotopic ossification, according to the system of Brooker et al.2.
Femur
Of the seventy-nine unrevised femoral components, three (4 per cent) were classified as definitely loose; one (1 per cent), as probably loose; and three (4 per cent), as possibly loose. Two stems (3 per cent) had subsided less than three millimeters. One hip (1 per cent) had a one-millimeter-thick radiolucent line in zone 7 of the bone-cement interface, three hips (4 per cent) had a radiolucent line in zone 1 of the bone-cement interface, and one hip (1 per cent) had a complete, non-progressive radiolucent line at the bone-cement interface. Two hips (3 per cent) had evidence of progressive varus tilt of the stem, and one of them had marked calcar resorption to the level of the lesser trochanter.
Six of the unrevised hips had evidence of endosteal osteolysis. Two hips (3 per cent) had symptomatic non-union of the greater trochanter. There were no mechanical failures of a femoral prosthesis.
Complications
Intraoperative complications included three fractures of the greater trochanter and one fracture of the femoral shaft. A sciatic neurapraxia developed postoperatively in two patients who had had a high dislocation. Systemic complications included three pulmonary emboli and three cases of myocardial ischemia. Local complications included one hematoma, which was treated with operative decompression, and six late infections. Two patients had postoperative dislocation, but neither needed revision. There were no mechanical failures of the components.
Total hip arthroplasty can be a technically challenging procedure in a patient who has congenital dysplasia of the hip. Our study supports the findings of Crowe et al.4, who reported that the rate of complications after thirty-one reconstructions in patients who had congenital dysplasia was 19 per cent, which was higher than the rate after reconstruction in a group of osteoarthrotic patients. The complications that are most commonly associated with reconstruction of hips with congenital dysplasia include palsy of the sciatic nerve, dislocation, subluxation, fracture of the femoral shaft, myocardial infarction, and pulmonary emboli. In series ranging from twenty-one to sixty-six total hip arthroplasties in patients who had congenital dysplasia, the rates of revision or of hips needing revision ranged from 14 to 43 per cent6,9,20,21,23; the rate of revision in the present series was 10 per cent (nine hips). The main reason for revision in almost all reported series has been aseptic loosening of the acetabular component, and some authors have stated that it is the single greatest factor in the long-term survival of a reconstruction23. However, only two of the nine hips that were revised in the present study had aseptic loosening of the acetabular component, and cement instead of morseled bone graft had been used in both of those reconstructions.
It is clear that acetabular coverage is a central concern when a reconstruction is performed in a patient who has a deficient acetabulum. It is generally believed, although not by everyone4, that at least 75 per cent of the cup must be covered by bone9,23,35 to prevent aseptic failure. Because of the altered acetabular anatomy (a shallow acetabulum with a roof that is deficient superiorly and laterally), a cup with a smaller-than-normal outer diameter frequently must be used in order to obtain adequate coverage9. In one series, cups with outer diameters ranging from thirty-three to forty-four millimeters were used in twenty-three (77 per cent) of thirty reconstructions performed in patients who had congenital dislocation or dysplasia of the hip4. Even with a twenty-two-millimeter femoral head, the polyethylene thickness of such cups would be marginally adequate, especially if a metal-backed or press-fit cup was used. Polyethylene wear, lysis, and loosening may all increase as a result of such thin polyethylene. Excessive reaming to obtain sufficient acetabular coverage reduces bone stock and creates the potential for axial migration of the cup, loss of the position of the cup, and fatigue fracture of the acetabular column4.
In order to avoid this problem, a variety of techniques have been employed to reconstruct the deficient acetabulum. Most techniques involve either realigning available host bone13,16 or augmenting the site of the cup with bone graft or cement. Bipolar prostheses have also been used for reconstruction in patients who have a deficient acetabulum, but complications such as acetabular stress fracture and loosening of the femoral component are common39.
Some authors have favored the use of a high hip center to improve coverage of the component and to avoid the difficulty of reducing a high subluxation or dislocation22,23,33. Others have reported increased risks of aseptic loosening of both the femoral and the acetabular component, higher rates of dislocation, and poor abductor-muscle mechanics in association with high hip centers and have stressed the importance of performing the reconstruction in the true acetabulum19,21,41. We believe that placement of the cup in the true acetabulum is essential, and all preoperative planning should be performed with this goal in mind.
The most common method of reconstructing deficient acetabula involves grafting with autologous bone from the femoral head. This technique is employed especially for hips that have type-III dysplasia4 or when more than 75 per cent of the cup cannot be covered by the host ilium9. Whether the cup should be fixed with or without cement depends in part on the amount of the cup that is covered by the graft37,41 and in part on the surgeon's preference. Use of autologous graft from the femoral head has been advocated by a variety of authors, and excellent short to mid-term results have been reported4,10,12,15,18,24,25,40. However, despite its theoretical advantages, not all authors believe that grafting reduces the rate of loosening of the acetabular component14,20,21,23,24,30,32. The weight-bearing portion of the bone graft tends to remodel over time, and any resorption or remodeling reduces the load transmitted through the graft and increases the stress at the bone-cement interface. Such a situation may actually lead to, not prevent, loosening at the bone-cement interface. Supporting this concept are reports of marked resorption of the graft14,32 and rates of loosening as high as 46 per cent (twenty-one hips)30 after longer-term follow-up.
The issue, then, becomes how best to obtain coverage of the acetabular component with excellent immediate fixation and without leading to long-term loosening of the cup or resorption of the graft. We have addressed this issue by using the acetabular reinforcement ring, with bone-grafting as needed, for reconstruction of deficient acetabula. In 1981, the senior one of us, extrapolating from his experience using the acetabular reinforcement ring in revision total hip arthroplasty for patients who had deficient pelvic bone stock, began using this ring in all primary total hip arthroplasties for patients with congenital dysplasia in whom bone graft was needed to cover the acetabular reconstruction. Because use of the reinforcement ring in patients who needed a graft yielded better results than did insertion of an all-polyethylene cup with cement and use of a graft (Figs. 3-A, 3-B, 3-C and 3-D), he began using the acetabular reinforcement ring in all reconstructions that involved an acetabulum with a deficient rim.
For hips with mild dysplasia (type I and sometimes type II4), the ring alone is sufficient to provide coverage of the cemented polyethylene cup. The ring provides immediate fixation through multiple screws that are directed toward areas of the best host bone stock and that preferably are in the direction of the weight-bearing forces of the acetabulum. Excessively long screws should be avoided; our experience with other patients has shown that such screws are more likely to break than are shorter screws. Long screws also have a longer lever arm, which theoretically could cause motion between the ring and the acetabulum.
The ring helps to minimize iatrogenic bone loss as well. The holes of the ring provide a suitable interface for fixation with cement. Excessive reaming and removal of bone to obtain a porous cancellous surface for cement interdigitation, or to provide a bed of bleeding subchondral bone for osseous ingrowth, can thus be avoided.
In our experience, reconstruction with use of an acetabular reinforcement ring in patients who have a small, deficient acetabulum has been technically easier to perform than reconstruction with insertion of a cup without cement. Because the ring and the polyethylene cup are placed independently, the ring can be positioned in the area of best fit to maximize contact with the host bone, virtually regardless of version. Once the reinforcement ring has been placed, the cup can be cemented in almost any position to optimize its orientation in the acetabulum. The need for bone-grafting can thus be avoided in some dysplastic hips as the ring provides support for the cup. Furthermore, the acetabular reinforcement ring has essentially no effect on the thickness of the polyethylene of the cup, since the ring and the cement combined are no thicker than most components that are inserted without cement.
Bone-grafting is indicated when more than 25 per cent of the acetabular reinforcement ring remains uncovered after restoration of the anatomical hip center and reaming of the bed for the cup. The graft is secured by placement of screws through the reinforcement ring and the graft and into the host bone. In this study, the acetabular reinforcement ring seemed to protect the bone grafts from resorption and contributed to their rapid incorporation into the host bone. None of the grafts resorbed and there were no non-unions in this series, not even in the hips that had type-III or IV dysplasia4. Furthermore, since there is no bone-cement interface when an acetabular reinforcement ring is used, the cemented polyethylene cup is protected from the remodeling that takes place at a microscopic and macroscopic level in all grafts. The ring sits against the graft and helps to distribute stresses to the host bone. The ring does not rely on osseous ingrowth for stabilization, so remodeling should not cause loosening of the acetabular construct. The absence of such a bone-ingrowth or bone-cement interface may also be the reason that no major osteolysis was seen around the acetabulum at an average of 9.4 years after the operation.
In this series of reconstructions performed with the use of an acetabular reinforcement ring, a risk factor for loosening of the acetabular component was the filling of osseous defects with cement: of the thirty-six hips that were treated with this method, six (17 per cent) were revised and twelve (33 per cent) had a radiolucent line. When an acetabular reinforcement ring is used, the only role of cement is fixation of the cup to the metal ring. In addition to increasing the number of late radiolucent lines and failures, use of cement between the ring and the osseous support negates a major advantage of the acetabular reinforcement ring—that is, the virtual elimination of the bone-cement interface that is so essential to the biological mechanism for loosening of cemented acetabular components34. Similar problems with loosening of acetabular components in defects filled with cement were reported by MacKenzie et al.23. Osseous defects, especially medially, should be filled with morseled or structural bone graft as needed. There was no difference between the performances of allografts and autologous grafts in this series.
The 17 per cent rate of complications (fifteen of eighty-seven hips) in this series was similar to the rates reported in the literature4,6,24. There was only one revision due to loosening of the acetabular component alone, one due to loosening of the femoral component alone, and one due to loosening of both components. There were no revisions due to mechanical failure of a component. Careful preoperative planning is an essential element of a good rate of success. Both a standard straight-stem and a congenital dysplasia model must be available so that the prosthesis that best fits the dysplastic femur can be chosen. Custom femoral components are not necessary.
Deep infection developed between one and eight years postoperatively in six (7 per cent) of the eighty-seven hips in this study. This rate is probably due to several factors. Two of the patients had had previous episodes of hip infection. In addition, total hip arthroplasty for a deficient acetabulum often requires extensive exposure, with more stripping of the ilium needed if grafting with bone from the femoral head must be performed. In addition, bone-grafting itself is associated with a higher risk of infection. Most importantly, antibiotics were administered perioperatively to only forty-nine patients in this study. Since 1984, we have routinely used antibiotic prophylaxis for all patients who are being managed with a total hip arthroplasty.
Sciatic neurapraxia developed in two patients in this study. One neurapraxia fully resolved and one partially resolved. Both patients had had reduction of a high dislocation into the true acetabulum. It has been recommended that limb lengthening not exceed two centimeters in order to minimize the risk of injury due to stretching of the sciatic nerve9. We believe that, in patients who have a high dislocation, the sciatic nerve should be exposed in order to allow direct assessment of the amount of tension in the nerve. The femur should be shortened as needed in order to allow placement of the acetabulum in the true hip center without overlengthening of the sciatic nerve.
The results of this study compare favorably with those in the literature. MacKenzie et al.23 recently reported the results of arthroplasty with cement in fifty-nine hips with congenital dysplasia. Four (7 per cent) of the hips were revised because of aseptic loosening of the acetabular component, and twelve (32 per cent) of the thirty-seven unrevised acetabular components for which radiographs were available were radiographically loose. In a series of sixty hips with congenital dysplasia that had been treated with a Charnley low-friction acetabular component, Sochart and Porter38 reported a long-term rate of revision of 37 per cent (twenty-two hips) and a long-term rate of aseptic loosening of 37 per cent. In the present study, only one (1 per cent) of eighty-seven hips was revised because of aseptic loosening of only the acetabular component. Seven (9 per cent) of the unrevised acetabular components were classified as probably or definitely loose. If the hips in which cement instead of bone graft was used with the acetabular reinforcement ring are excluded, the rate of aseptic loosening of the acetabular component is only 2 per cent (one of forty-two hips).
We recommend that reconstructions of deficient acetabula include restoration of the anatomical hip center with the use of an acetabular roof-reinforcement ring and insertion of a polyethylene cup with cement. Bone graft should be used medially and superiorly as needed, since filling acetabular defects with cement increases the risk of aseptic loosening. No hip that was treated with our recommended method was revised due to aseptic loosening of the acetabular component, and only 2 per cent (one of forty-two) had radiolucency.
At the time of follow-up (average, 9.4 years; range, five to fifteen years) in our study of the results of reconstruction with use of the acetabular reinforcement ring, there was a low rate of acetabular osteolysis (six [8 per cent] of seventy-eight unrevised hips). With longer-term follow-up, less osteolysis should lead to a decreased rate of aseptic loosening. Also, the improved bone stock makes revision easier. The use of either a standard or a congenital dysplasia straight-stem femoral component as well as so-called newer-generation cementing techniques also resulted in a low rate of aseptic loosening at the time of follow-up.
Perhaps the most essential element of a successful reconstruction of a hip with congenital dysplasia is meticulous preoperative planning. Only with such planning can the appropriate reconstructive technique, prosthesis, and bone graft be chosen. In addition, investigators must continue to strive for accurate, comprehensive documentation of their clinical results. In this way, there can be meaningful comparisons of reconstructive techniques, which will allow both future technical advances and, more importantly, the delivery of optimum patient care.