Interest in alumina-on-alumina ceramic bearings has increased as a result of the problem of wear debris generated by conventional metal-on-polyethylene articulations giving rise to osteolysis1,2. Alumina-on-alumina ceramic is a low-friction, highly wear-resistant bearing with chemically inert properties that are very attractive to the orthopaedic surgeon.
Ceramic articulations were first used in hip replacement by Boutin in 19703. Early prostheses had high failure rates as a result of poor acetabular fixation4,5, which subsequently was addressed with the use of metal-backed acetabular components. There is a low risk of fracture of the ceramic; a review of ceramic hip replacements by Hannouche et al. revealed only thirteen fractures, including eight of the femoral head and five of the acetabular socket, after 5500 replacements6. The ceramic material has also evolved with the release of a third-generation ceramic in 1994; this material is manufactured with use of hot isostatic pressing to reduce grain size, limit grain boundaries and inclusions, increase burst strength, and have better wear properties7. Ceramic components are now all proof-tested by loading, which sorts out bearings that are defective and have an increased risk of fracture7.
We evaluated a consecutive series of 301 primary cementless alumina-on-alumina total hip arthroplasties in order to determine the wear of the retrieved bearings, the rate and extent of osteolysis, and the clinical and radiographic outcomes after a minimum of five years of follow-up.
From June 1997 to February 1999, 301 consecutive primary cementless total hip arthroplasties were performed with use of an alumina-on-alumina ceramic bearing in 283 patients. The mean age of the patients at the time of the index surgery was fifty-eight years (range, twenty to seventy-six years); 222 arthroplasties were performed in patients under the age of sixty-five years, and sixty-five arthroplasties were done in patients under the age of fifty years. The study included 145 women (51%) and 163 right hips (54%). Bilateral total hip arthroplasty was performed in eighteen patients (6%), including six (2%) who had simultaneous bilateral total hip arthroplasty. The mean weight of the patients was 77 kg (range, 45 to 130 kg), and the average body mass index was 27 (range, 16 to 39). The most common diagnosis was primary osteoarthritis (Table I).
All of the operations were performed by the two senior authors (W.K.W. and B.A.Z.) at a single center, in a vertical laminar flow theater, through a posterior approach with an enhanced posterior repair8. The operative technique and postoperative protocols were identical for all patients, and all patients received the same prosthesis. The titanium-alloy shell has an arc-deposited pure titanium ingrowth surface with a hydroxyapatite coating (Osteonics ABC; Stryker Orthopaedics, Mahwah, New Jersey). The ceramic liner for this cup is recessed below an elevated metal rim to help prevent damage to the ceramic or scoring of the femoral neck if impingement occurs. The cementless femoral component used had a straight stem with either a tapered or a fluted distal geometry (Secur-Fit or Secur-Fit Plus; Stryker Orthopaedics). The stem is made of titanium alloy, and its proximal part has an arc-deposited pure titanium ingrowth surface with hydroxyapatite coating. All of the hips received the same type of alumina ceramic insert and alumina ceramic C-Taper head (Biolox Forte; CeramTec, Plochingen, Germany). If the acetabulum was not large enough for a 52-mm acetabular shell, which allows a 32-mm bearing, a smaller acetabular shell with a 28-mm bearing was used. We used 278 32-mm bearings and twenty-three 28-mm bearings.
Postoperative management included forty-eight hours of intravenous broad-spectrum antibiotics, low-molecular-weight heparin, thromboembolic deterrent stockings, and intermittent calf compression. Patients were routinely mobilized early with full weight-bearing as tolerated.
Clinical and Radiographic Follow-up Evaluation
Patients were assessed clinically and radiographically before the operation and at six weeks, one year, two years, and then every two years following the surgery. Seventeen patients who were unable to return for evaluation were assessed on the basis of telephone interviews and were asked to have radiographs made locally and sent to us.
At the time of the clinical review, the patients filled in a questionnaire based on the "Standard System of Terminology for Reporting Results," which included the modified Harris hip score9. An anteroposterior radiograph of the pelvis and a lateral radiograph of the hip were made at each review. Our prospective system of clinical and radiographic patient assessment with use of a relational database has been described elsewhere10. Radiographs were reviewed by an arthroplasty fellow who was not involved in the surgery. If there was any suspicion of osteolysis based on the plain radiographs, a computer tomography scan was performed.
Wear
Seven ceramic bearings that had been retrieved more than six months after implantation were analyzed for signs of wear. Five additional bearings that had been in situ for less than six months were excluded from the wear analysis as very slight wear is difficult to assess accurately. We replaced ceramic liners and femoral heads that showed evidence of edge-loading wear; we placed new components on used tapers that were macroscopically normal.
To assess wear, each bearing was first examined with the naked eye under directional lighting. Any surface damage was highlighted with a surgical marking pen, and the excess ink was wiped away from the undamaged surface. Equatorial and true polar two-dimensional roundness traces were recorded with a Roundtest RA300 machine (Mitutoyo; Andover, United Kingdom) that has an accuracy of 0.01 µm and records both full and partial circles11.
Statistical Methods
Survivorship analysis was performed with the Kaplan-Meier method12 with revision for any reason as one end point and revision due to mechanical failure (clinical or radiographic evidence of aseptic loosening) at the time of follow-up as the other end point. Statistical analysis was performed with use of nonparametric tests. Significance was defined as a p value of <0.05.
Ten patients (ten hips) died with the prosthesis in situ from causes not related to the surgery. Twenty-two patients (twenty-two [7.3%] of the 301 hips) were lost to follow-up. There were nine revisions (in nine patients) of one or both components, leaving 260 implants in 242 surviving patients with more than five years of follow-up. Of these patients, 241 (259 hip arthroplasties) had five-year clinical follow-up data available and 224 (241 hip arthroplasties) had five-year radiographic follow-up data available. The mean duration of follow-up in this study was 6.5 years (range, five to 9.2 years).
Complications
Complications are outlined in Table II. Five (1.7%) of the 301 hips sustained a traumatic periprosthetic femoral fracture. Three of these fractures occurred in the first three weeks after the surgery, before the stem was ingrown, and none were noted at the time of surgery. Two of the three patients with an early fracture had normal findings on anteroposterior and lateral radiographs made in the first week after the surgery, and the third patient fell prior to imaging. One fracture occurred eight months after the surgery, when the patient was knocked to the ground. These four patients were all treated successfully with femoral revision surgery. One patient sustained a trochanteric fracture in a fall eighty months following the index procedure, when she was eighty-one years of age. She had a good result with nonoperative management.
One femoral component required revision ten months after the index surgery as a result of aseptic loosening. The patient fell two months after the initial surgery and subsequently had thigh pain. Although there was no evidence of a fracture or loosening on the radiographs, a subsequent bone scan suggested aseptic loosening, which was confirmed at the time of the operation.
One patient with a proximal focal femoral deficiency had had femoral lengthening in the past. A sciatic nerve palsy developed immediately after the index total hip arthroplasty. He was taken back to the operating room the next day, and the sciatic nerve was found to be intact. He was treated with a femoral shortening osteotomy with a revision of the femoral component. He had made a full recovery by eight months postoperatively.
One patient presented with an infection at the drain site ten days after the surgery. He subsequently underwent hematoma evacuation and extensive débridement of the hip joint. There were four superficial infections, which did not require additional surgery; all were treated with antibiotics without sequelae.
One active patient complained of squeaking in the left hip six years after the surgery. The hip occasionally squeaked when he lunged forward, flexing the hip to 140° with internal rotation. He was advised to avoid that movement. The hip otherwise functioned well.
One patient had two episodes of dislocation; one occurred at twelve months and one, at five years. On both occasions, the hip was reduced with a closed method.
There was one chip fracture of a ceramic liner at the time of insertion; this required intraoperative replacement without removal of the acetabular shell. There were no fractures of the ceramic head.
Four patients (four arthroplasties) reported notable groin pain and presented with clinical symptoms and signs of iliopsoas tendinitis more than six months after the index surgery. These patients had a painless passive range of motion, but active straight-leg raising reproduced the symptoms. All patients had bone scans, which excluded the possibility of loosening or fracture. In three patients, Marcaine (bupivacaine) and hydrocortisone were injected into the iliopsoas tendon sheath and provided dramatic temporary relief of pain. All four patients had physiotherapy, hydrotherapy, and analgesia for at least six weeks, but the symptoms persisted. Two patients were treated successfully with psoas tendon release and revision of the acetabular component, which had been prominent anteriorly. The other two patients were treated with release of the iliopsoas tendon at the lesser trochanter and exchange of the bearing surfaces. This treatment was successful in one patient, but the other patient, a very active thirty-two-year-old, presented later with groin pain on deep flexion. A computed tomography scan (Fig. 1) demonstrated osteolysis in association with the apical and posterior holes in the acetabular shell. Revision surgery was performed eighty-two months after the index surgery, revealing metallosis. There were multiple sites of impingement with associated edge-loading, and the metal rim of the cup had been worn away, allowing the femoral component to impinge on the ceramic insert (Figs. 2-A and 2-B). The acetabulum was revised with use of morselized allograft bone and an ABG II cup without holes (Stryker Howmedica Osteonics, Allendale, New Jersey) with a 32-mm ceramic liner. A cup without holes was selected to prevent reoccurrence of osteolysis in association with holes in the shell13,14. The femoral component was revised to an SROM stem (Joint Medical Products, DePuy, Warsaw, Indiana) to improve femoral anteversion. This moved the so-called safe envelope of hip movement to a more functional position to prevent impingement. Histological analysis of the periarticular tissue revealed fibrovascular tissue lined by granulation tissue that contained a heavy inflammatory infiltrate composed of lymphocytes, histiocytes, and neutrophils. Tissue cultures were negative for bacteria.
Bearing Wear Data
Seven bearings were retrieved after more than six months in situ. Two of these bearings had no macroscopic wear, and the other five had evidence of edge-loading wear. The average wear rate of the retrieved femoral heads was 0.3 mm3 per year (median, 0.2 mm3 per year). The maximum femoral wear rate was 0.7 mm3 per year, and this was measured in both of the femoral heads retrieved from the hip with osteolysis.
Clinical Results
The mean Harris hip score improved from 56 points (range, 17 to 89 points) preoperatively to 95 points (range, 61 to 100 points) at the time of the latest follow-up. The result was rated as either excellent or good (a score of 80 to 100 points) for 256 (98.8%) of the 259 hips, as fair (a score of 70 to 79 points) for two hips, and as poor (a score of <70 points) for one hip. Two hundred and thirty-seven patients (255 hip arthroplasties; 98.5%) reported that they had no or mild postoperative groin pain, and only four patients (four arthroplasties; 1.5%) had moderate pain. Similar results were seen for thigh pain, with 232 patients (250 hip arthroplasties; 96.5%) having mild or absent thigh pain and the remaining nine patients (nine hip arthroplasties; 3.5%) having moderate pain. After the surgery, all patients reported increased function and decreased pain and stated that they were satisfied with the result of the operation. The average visual analogue score for overall satisfaction was 9.6 (range, 5 to 10) of a possible 10.
Radiographic Results
All 241 femoral stems had radiographic evidence of stable bone ingrowth at the time of the most recent follow-up, with no evidence of subsidence or migration15. Radiolucent lines were not observed proximally (in Gruen zone16 1 or 7), but a radiolucent line was seen in one distal zone in two hips. There was no femoral osteolysis. Stress-shielding (loss of trabecular bone density or corticocancellization) was observed in 131 (54.4%) of the 241 hips, but severe stress-shielding (with cortical thinning) was not detected in any hip. No radiolucent lines were observed on the acetabular side, and focal periacetabular osteolysis was not seen. There was also no cup migration.
According to the classification system of Brooker et al.17, heterotopic ossification was Grade I or II in sixty-seven hips (27.8%) and Grade III or IV in twelve hips (5.0%).
Revisions and Survivorship
Nine of the 301 hips had a revision of one or both components, so that the rate of revision for any reason was 3.0% (Table III). With revision of either the acetabular or the femoral component for any reason considered to be a failure, the rate of survival of the prostheses was 96% (95% confidence interval, 92% to 98%) at seven years and six months (Fig. 3). When revision of either component due to aseptic loosening or osteolysis was the end point, the survival rate of the prosthesis was 99% (95% confidence interval, 95% to 100%) at seven years and six months. Survival with revision of the femoral component because of a fracture as the end point was 99% at seven years and six months. In the worst-case scenario, in which it is assumed that all patients who were lost to follow-up had a revision, the survival rate with revision of either component as the end point is 89% (range, 85% to 93%) at five years18. However, we do not believe that this is a reliable estimate as all patients who were lost to follow-up had been doing well at the time of their last examination. The nine patients (nine arthroplasties) had a good clinical score when they were seen at the time of follow-up at two years, and three of these patients said that they were doing well when they were contacted after five years, although no clinical scores or radiographs were available.
Note: The authors acknowledge Sharon Wales and Lyn McDonald for their assistance with collection of the data. They also acknowledge Michael Tuke for the measurement of ceramic wear and Rajmoran Murali and Fiona Bonar for the histological analysis.