The Oxford unicompartmental knee replacement (UKR) was designed to minimise wear utilising a fully-congruent, mobile, polyethylene bearing. Wear of polyethylene is a significant cause of revision surgery in UKR in the first decade, and the incidence increases in the second decade. Our study used model-based radiostereometric analysis to measure the combined wear of the upper and lower bearing surfaces in 13 medial-compartment Oxford UKRs at a mean of 20.9 years (17.2 to 25.9) post-operatively.
The mean linear penetration of the polyethylene bearing was 1.04 mm (0.307 to 2.15), with a mean annual wear rate of 0.045 mm/year (0.016 to 0.099). The annual wear rate of the phase-2 bearings (mean 0.022 mm/year) was significantly less (p = 0.01) than that of phase-1 bearings (mean 0.07 mm/year).
The linear wear rate of the Oxford UKR remains very low into the third decade. We believe that phase-2 bearings had lower wear rates than phase-1 implants because of the improved bearing design and surgical technique which decreased the incidence of impingement. We conclude that the design of the Oxford UKR gives low rates of wear in the long term.
Wear of polyethylene in unicompartmental knee replacement (UKR) has been well documented1,2 and leads to revision in cases of catastrophically worn bearings.3 In 2009 the Swedish Knee Arthroplasty Register reported that 22% of all UKR revisions were performed for polyethylene wear.4
The Oxford UKR was designed to incorporate a fully-congruent mobile bearing with large contact areas and low contact stresses which should, theoretically, reduce polyethylene wear5 without increasing the risk of loosening. A radiostereometric analysis (RSA) study by Price et al6 showed a rate of linear penetration of 0.02 mm/year at ten years in well-functioning bearings in the phase-2 Oxford UKR. Studies of polyethylene wear in vitro have been published7,8 which have highlighted the low-wear benefits of mobile bearings in total knee replacement (TKR). There have also been retrieval studies for both mobile congruent and fixed non-congruent bearings in both TKR and UKR9–12 which have shown reduced wear in mobile bearings, but acknowledge the potential risk of bearing dislocation. Retrieval studies from mobile-bearing UKRs have shown that the wear rate is increased if the bearing impinges on bone or cement.12,13 If the bearing is functionally normal without impingement the wear rate is less than 0.03 mm/year.12
The low penetration rate of mobile-bearing UKR is particularly relevant when it is viewed as a definitive treatment rather than as an intermediate measure before inevitable TKR. Although there has been debate on the role of UKR, including the exact indications and contraindications, our opinion is that UKR can be used as a definitive treatment for patients with anteromedial osteoarthritis of the knee. Good wear rates have been shown at ten years,6 but it is important to establish the penetration rate over a longer period of time, in order to support its use as a definitive treatment for anteromedial osteoarthritis. This is of particular relevance in regard to the increasing number of young patients undergoing knee replacement.14
Patients and Methods
All patients gave informed consent for participation in the study, which was approved by the local Ethics Committee.
Patients were selected from our database of those receiving a medial UKR for antero-medial osteoarthritis of the knee. All had an intact anterior cruciate ligament and correctable varus deformity at the time of surgery, with operations performed through a midline incision using a medial parapatellar approach. We identified 141 knees (121 patients) which had retained their implants for more than 17 years. Of these, 98 were in patients who had died, in whom there were five revisions, three for lateral progression, one for pain of unknown cause and one for bearing dislocation, and in 12 who were lost to follow-up. Of the 31 surviving knees (25 patients), four had been revised, two for lateral progression, one for loosening of the femoral component and one for deep infection. A total of 12 patients (12 knees) were too frail or lived too far away to attend and two (two knees) declined the invitation. Therefore nine patients (13 knees) attended for stereoradiography and clinical assessment. There were eight women and one man with a mean age at surgery of 64.3 years (55.1 to 73.0) and a mean follow-up of 20.9 years (17.2 to 25.9). Of the identified patients, five (six knees) had phase-1 and four (seven knees) had phase-2 implants Oxford UKRs (Biomet, Swindon, United Kingdom).
The phase-1 bearing with the longest follow-up was made from machined polyethylene while all the other bearings, both phase 1 and phase 2, were compression moulded. All were made from Hostulen RCH 1000 (Hoescht, Oberhausen, Germany) and were gamma-irradiated in air. The anterior lip of the phase-2 bearing was lower than that of the phase-1 to decrease the risk of impingement. The nominal bearing thickness at insertion was noted for all patients. All the operations were performed by one of two surgeons. The Oxford knee score17 (OKS, 0 to 48) and American Knee Society scores18 (AKS, objective and functional, 0 to 200) were obtained when follow-up radiographs were taken.
For RSA, paired radiographs were obtained in a standard fashion with the patient fully weight-bearing.19,20 Two sets of images were obtained for each patient, with the knee in full extension and in 30° of flexion. Each patient stood within a calibration frame which had been manufactured to provide an accurate set of known three-dimensional calibration points defined by tantalum marker balls 0.8 mm in diameter. The radiographs were taken obliquely angled 30° off a standard anteroposterior (AP) radiograph in each direction. Model-based RSA software version 3.21 (Medis Specials, Leiden, The Netherlands) was then used to estimate the position of both the femoral and tibial component having first identified the outline of the implant on each radiograph using a Canny edge detection algorithm.21 The RSA software has been validated and described in detail elsewhere.15,22
The shortest linear distance between the estimated positions of the two components was calculated and taken as the thickness of the bearing. This was then implemented in Matlab version 18.104.22.1687 software (The Math Works Inc., Natick, Massachusetts) using a customised routine which has been described by Simpson et al.16
The maximum linear penetration was calculated by subtracting this distance from an estimated bearing thickness at insertion. The estimated bearing thickness was taken as the nominal bearing thickness at implantation (provided by the manufacturer) plus an offset of 0.05 mm, which was deduced from a previous study in which the thickness of 20 unused bearings had been measured.6
The linear penetration for each bearing was calculated for the flexed and extended RSA examinations and the mean of the two estimates was determined. The linear penetration for each joint was then divided by the number of years in situ to give the mean yearly rate of linear penetration.
Each radiograph was independently assessed for the presence of osteolysis by two authors (BJLK, DJS), without discussion or considering cases together.
The precision of the linear wear measurement has previously been published,16 with sd of the difference between the estimated and measured bearing thickness of 0.163 mm. The mean difference bias was 0.03 mm.
All data (linear penetration rate, bearing thickness, OKS and AKS scores) were not normally distributed and therefore non-parametric tests were used. The Mann-Whitney U test was used for comparison of the linear penetration rate and the OKS and AKS. Spearman’s rank correlation was used for correlation. SPSS version 15 software (SPSS Inc., Chicago, Illinois) was used for all the statistical analysis. A p-value ≤ 0.05 was considered to be significant.
The mean weight at the final follow-up was 67.8 kg (50.8 to 88.9) and the mean body mass index (BMI) was 25.2 (22.0 to 30.3) in phase-1 patients. The respective values were 68.6 kg (60.3 to 82.5) and BMI 25.2 (22.0 to 29.9) in phase-2 patients. There was no significant difference between these groups (t-test, p = 0.873) and no significant difference in age at the initial operation between them (t-test, p = 0.774). As expected, there was a difference between the two phases for mean time since surgery (phase 1, 22.5 years (21.6 to 25.9); phase 2, 19.5 years (17.2 to 20.5) (Table I⇓). The mean OKS was 40.2 (35 to 46) and the mean total AKS 164.2 (149 to 194). There was no statistically significant difference for the OKS or the AKS score and functional scores between those patients with phase-1 and those with phase-2 implants (Table I⇓).
The mean linear penetration for the entire group was 1.041 mm (0.307 to 2.151). There was a significant difference (p = 0.007) between the two phases (Table II⇓). The range of penetration rates in phase 2 was narrow and was between 0.016 mm/year and 0.03 mm/year (mean 0.022 mm/year). By contrast the phase-1 penetration rates were spread within a wider range of between 0.023 mm/year and 0.099 mm/year (mean 0.07 mm/year) (Fig. 1⇓). There was a significant difference between the mean wear rates (p = 0.01).
Our patients had received varying thicknesses of bearing, ranging from size 0 (3.5 mm) to size 5 (8.5 mm) (Table II⇑). There was no statistically significant difference when linear penetration rates were compared for those bearings thicker (n = 6) and those thinner (n = 7) than 6 mm (p = 0.775). The mean linear penetration rate for thin bearings was 0.040 mm/year and for thick bearings 0.050 mm/year. Direct correlation analysis between nominal thickness and wear rate showed a Spearman’s rho correlation coefficient at −0.035 (p = 0.910) showing no correlation.
None of the radiographs showed osteolysis beneath the tibial component.
In our study we measured the distance between the articular surfaces of the femoral and tibial components, subtracting this measurement from the known thickness of the bearing at implantation. Since both the upper and lower surfaces of the bearing have fully congruent contact with the articular surfaces of the femoral and tibial components the resulting thickness is a manifestation of creep and wear in the polyethylene. Creep is important early after implantation, but after 20 years its effect is probably negligible and can be ignored.23 If there was appreciable creep the measurement of penetration would be an overestimate of wear. Wear occurs on both the upper and lower surfaces of the bearing and therefore the measurement of penetration made in our study can be considered to be the combined linear wear of both of these surfaces. We were not able to differentiate between wear on the upper and lower surfaces.
Our study has shown that low rates of linear wear (mean 0.045 mm/year, maximum 0.099 mm/year) can be maintained in the Oxford UKR to the end of the second decade after implantation. Although our study does not provide direct evidence of low wear in young patients, we can infer that a well-functioning implant will have a low rate of wear regardless of the age of the patient. This is important because the number of replacements performed in younger patients is increasing.24 Our study has also shown that the rate of linear wear at 20 years in phase-2 bearings (0.022 mm/year) is the same as that measured at ten years.6 We are not aware of any long-term studies in vivo on wear in fixed-bearing UKRs. However, at a mean of 4.6 years, Witvoet, Peyrache and Nizard25 showed radiologically detectable wear of up to 7 mm in a fixed-bearing UKR. In a retrieval study on fixed-bearing UKR9 the mean rate of wear was 0.15 mm/year which is substantially higher than that of the fastest wearing mobile bearing in our study.
The variability in the wear of bearings retrieved at revision of the Oxford UKR has been described by Psychoyios et al12 and more recently by Kendrick et al.13 The rate of wear on the articular surfaces was related to the amount of damage caused to the bearing by impingement against bone or cement. Bearings which were functioning normally, with no evidence of impingement, all had low rates of wear < 0.03 mm/year. The mean rate of wear in these bearings was 0.01 mm/year in both studies. However, as the degree of extra- or intra-articular impingement increased so did the wear rate, presumably because the debris released by impingement caused third-body wear. The highest rate of wear was found in those bearings which showed evidence of loss of congruency between the femoral component and the mobile bearing. In our study, as shown in Figure 1⇑, about half the bearings had very similar low rates of wear (< 0.03 mm/year). The remainder of the bearings had more variable and higher wear rates. Therefore, we suggest that those bearings in our study with low penetration rates, ≤ 0.03 mm/year, are likely to be functioning well with no impingement. All the phase-2 bearings had wear rates of < 0.03 mm/year which suggested that they were not impinging. By contrast, the phase-1 bearings had rates of penetration which ranged between 0.023 mm/year and 0.099 mm/year, suggesting that all but one of the phase-1 bearings were probably impinging.
The most common site for impingement is on the upper anterior edge of the bearing which can hit the bone in front of the femoral component in full extension. In phase 1 the femur was prepared using three flat saw cuts to accommodate the femoral component (Fig. 2⇓). It was not appreciated that the front of the bearing could impinge on the femur in extension and therefore no attempt was made to remove bone to prevent it.26 The introduction of the femoral condyle mill when phase-2 implants were inserted, made it possible to position the femoral component accurately in order to balance the ligaments, but it left a rim of bone anterior to the femoral component which obviously impinged on the bearing in extension. Anterior impingement was then recognised and care was taken to remove bone in this area to prevent it. In addition, the bearing was modified to minimise the risk of impingement by lowering the height of the anterior edge (Fig. 2⇓). As a result of these changes the risk of anterior impingement, which was high in phase 1, was reduced in phase 2.
The evidence from this and other studies is that if the bearing is functioning well, with no impingement, the rate of wear is less than 0.03 mm/year, and does not change with time.12,13 Therefore, under these circumstances, demonstration of wear of 1 mm will require the bearing to have been in use for over 33 years. All the patients in our study were mobile with an exercise tolerance > 100 metres according to the American Knee Society functional questionnaire. Although not an entry criterion, a good level of mobility increases the likelihood of gaining an accurate measurement of wear. Assessment of patients who were less mobile would have given an artificially low penetration measurement. This supports the view that mobile UKR can be a definitive treatment rather than a procedure for postponing the need for eventual TKR.
In our study standard polyethylene has been used. Although cross-linked polyethylenes have been shown to have lower rates of wear than standard polyethylene they have lower fracture toughness. Since fracture of thin bearings has occasionally been described in large male patients we feel that until the fracture toughness of cross-linked polyethylene has been improved it remains safer to use the standard material.
It has been recommended for fixed-bearing UKR that thin polyethylene bearings should be avoided,2,3 with suggestions that a minimum thickness be used, such as 8 mm with the Miller-Galante UKR (Zimmer, Warsaw, Indiana)27 or 6 mm with the Robert-Brigham UKR (DePuy, Johnson & Johnson, Warsaw, Indiana).28 It is suggested that in a fully congruent mobile bearing the contact stress is so low that the wear rate is independent of the bearing thickness. Since the wear rates are very low thin polyethylene can be used, with the thinnest available for the Oxford UKR being 3.5 mm. Our results have shown that the bearing thickness did not affect the linear penetration rate, suggesting that the continued use of thin bearings is justified.
The OKS for our patients showed that good function can be maintained into the third decade, comparing well with scores from patients of a similar age range with no knee pathology.29 However, we acknowledge that our patients could have produced an artificially high range of scores since those unable to attend possibly had poorer function and therefore would presumably have poorer scores.
The main weakness of our study is the possibility of selection bias. However, we have tried to avoid this by inviting all surviving patients to attend, but, inevitably, there will be loss to follow-up. There were also four patients with bilateral implants. We feel that the small numbers do not allow reliable subgroup analysis. However, with few patients able to attend we felt that it was preferable to include all the available joints. It is also recognised that our study used a limited series of patients from a single specialist unit, which may not have been representative of all patients with a medial Oxford UKR at 20 years.
In conclusion, at 20 years the linear wear of a fully congruent mobile-bearing UKR is low (mean 1.041 mm, maximum 2.151 mm). However, the wear at 20 years for the phase-2 implant, which was inserted using a femoral condyle mill that reduced the risk of impingement, is lower (mean 0.437 mm, maximum 0.567 mm).
Financial support has been received from the NIHR Biomedical Research Unit into Musculoskeletal Disease, Nuffield Orthopaedic Centre and University of Oxford and Arthritis Research UK. The authors are grateful to Mr J. W. Goodfellow and Mr C. A. F. Dodd for performing the operations.
The author or one or more of the authors have received or will receive benefits for personal or professional use from a commercial party related directly or indirectly to the subject of this article. In addition, benefits have been or will be directed to a research fund, foundation, educational institution, or other nonprofit organisation with which one or more of the authors are associated
- Received July 23, 2010.
- Accepted November 29, 2010.
- © 2011 British Editorial Society of Bone and Joint Surgery