Polyethylene particulate wear debris continues to be implicated in the aetiology of aseptic loosening following knee arthroplasty. The Oxford unicompartmental knee arthroplasty employs a spherical femoral component and a fully congruous meniscal bearing to increase contact area and theoretically reduce the potential for polyethylene wear. This study measures the in vivo ten-year linear wear of the device, using a roentgenstereophotogrammetric technique.
In this in vivo study, seven medial Oxford unicompartmental prostheses, which had been implanted ten years previously were studied. Stereo pairs of radiographs were acquired for each patient and the films were analysed using a roentgen stereophotogrammetric analysis calibration and a computer-aided design model silhouette-fitting technique. Penetration of the femoral component into the original volume of the bearing was our estimate of linear wear. In addition, eight control patients were examined less than three weeks post-insertion of an Oxford prosthesis, where no wear would be expected. The control group showed no measured wear and suggested a system accuracy of 0.1 mm. At ten years, the mean linear wear rate was 0.02 mm/year.
The results from this in vivo study confirm that the device has low ten-year linear wear in clinical practice. This may offer the device a survival advantage in the long term.
Polyethylene wear remains a major factor in the failure of knee replacements.1 In the shorter term, catastrophic failure can result in early revision.2 In the medium- to long-term wear particles are implicated in the pathogenesis of osteolysis and the production of aseptic loosening.3
Theoretically reducing contact stress at the bearing surface can decrease wear.4 The use of a fully congruent mobile bearing increases contact area, decreases contact stress and therefore reduces the potential for formation of wear particles.5 The Oxford unicompartmental knee arthroplasty (UKA; Biomet, Swindon, UK) employs a meniscal bearing and was designed in an attempt to reduce wear.6
In vitro measurement of polyethylene wear using laboratory simulators and retrieval studies have suggested that the device can produce low wear rates, when functioning normally.7,8 However, these findings have not been confirmed in vivo. This study combines roentgenstereophotogrammetric analysis (RSA) and computer-aided design models to produce a system for calculating in vivo polyethylene wear as previously applied to assess a total knee replacement.9,10 Our aim was to measure linear wear occurring in meniscal-bearing uni-compartmental knee replacements at a minimum follow-up of ten years.
Patients and Methods
A stereo pair of radiographs was simultaneously taken of each knee studied. The technique required the subject to stand on both legs, with the study knee placed within a calibration frame, which is a rigid, triangular open-ended structure. The three planes of the frame contained tantalum marker balls (0.8 mm in diameter) in accurately-known positions relative to a coordinate system embedded in the calibration frame. The marker balls gave rise to image points on the stereo radiographs. The relationship between the three-dimensional coordinate system of the frame and the two-dimensional coordinate system of each radiograph was then determined. The process of calculating this relationship is called calibration. Once both images were calibrated, the three-dimensional coordinates of any object visible in both images could be calculated with a high degree of accuracy.
The images were then processed to detect the silhouettes of the metal components of the implant. These were separated into the femoral and tibial component silhouettes. For each component, a virtual silhouette could be generated using the photogrammetric relationship determined by calibration and the appropriate computer-aided design (CAD) model of the component. The model was considered to be placed within a virtual representation of the calibration frame, the projected image giving rise to the virtual silhouettes on each radiograph. The position of the CAD model was systematically adjusted to minimise the mismatch between the actual and virtual silhouettes. This allowed accurate calculation of the three-dimensional location of each component. Validation of this method has been described previously.10
The Oxford UKA has a flat tibial component and a spherical femoral component. Determination of the three-dimensional positions of each allows the smallest perpendicular distance between the articulating surfaces of these components to be calculated. This represents the minimum thickness of the bearing, which separates the two metal components. Minor differences in extra-articular bearing shape do not affect the method. By comparing the RSA-calculated bearing thickness with the known thickness of the bearing at time of insertion, the linear wear can be estimated.
Estimating bearing thickness at insertion.
The nominal size of each bearing studied was known, but manufacturing tolerance introduces some variation to this value. To assess the degree of variation the minimum thickness of 20 unused bearings was measured with a dial gauge. The mean difference in measured thickness to nominal thickness was + 0.05 mm (SD 0.07). Therefore, for all other calculations the thickness of the polyethylene bearing before implantation was taken to be the nominal thickness + 0.05 mm.
As a control, eight patients who were undergoing a medial Oxford UKA were recruited and each had RSA radiographs taken after their operation. The implants were all phase 3 Oxford UKA. This prosthesis is manufactured using a CAD model, which had been made available for the experiment by the manufacturer. Using the system described above, the linear penetration of the femoral component into the bearing was measured, within three weeks of insertion, when it could be assumed no wear had occurred. The measured thickness of polyethylene was compared with the thickness of the bearing at insertion. The details of these patients are given in Table I⇓.
Patients were recruited from a series of medial Oxford UKAs performed by John Goodfellow, one of the designers of the prosthesis.4,11 An accurate database of the 144 knees in 114 patients was available. All patients had undergone a medial Oxford UKA for anteromedial osteoarthritis, with a functioning anterior cruciate ligament and correctable varus deformity of the knee at the time of surgery. The recruitment of suitable patients from this database involved the following criteria: 1) the patients had to be alive, 2) have received the phase 2 prosthesis to enable the use of a single CAD model, 3) be living in the Oxford-shire region, and therefore not precluded by distance from reattending and 4) have received their prosthesis approximately ten years earlier. From the original 114 patients, seven were identified of whom five (seven knees) agreed to take part in the study. All patients completed an Oxford Knee Score questionnaire12 and their knees were evaluated using the American Knee Society (knee and function) score.13 The demographic and clinical details of the patients are given in Table II⇓.
All bearings assessed were made using the same manufacturing process and tolerance. There were minor differences in the extra-articular shape of the phase 2 and 3 bearings used in this study, but importantly the intra-articular shape remained the same.
The patients who agreed to enter the study had RSA radiographs taken and linear wear was calculated. For six knees, two sets of radiographs were taken, one in full extension and the other at 30° of flexion. One patient (one knee) declined the second set of radiographs. The full congruency of the articulation should not affect the measurement of bearing thickness regardless of the degree of knee flexion. Therefore, comparison of the two measurements provided a test of reproducibility.
Ethical approval for our radiographic study was given by the local ethics committee.
The penetration measurements of the control bearings are displayed in Figure 1⇓. The mean difference was 0.05 mm (95% confidence interval (CI) 0.01).
The details of all linear wear measurements made for the study group are shown in Table III⇓. From the six knees within the study group where two different pairs of radiographs had been taken, the mean difference between penetration measurements was 0.01 mm (95% CI 0.07).
In one bearing, the mean penetration measurement was negative, but positive values were recorded for the others. Individual linear wear rates for each bearing have been calculated by dividing measured penetration (mm) by the time since insertion (years). The highest linear wear rate was 0.05 mm/year, but four bearings were wearing at approximately 0.02 mm/year. The mean penetration rate was 0.02 mm/year (95% CI 0.01).
The results are summarised in Figure 2⇓, which displays the linear wear measurements for each of the study and control bearings plotted against time. The error bars around the study bearings represent the error of the system (SD 0.01), estimated from the confidence interval around the control and repeated measurements. The mean linear wear rate of the study bearings is represented by the gradient of the line.
A combination of RSA techniques and CAD technology provided the methodological basis for measuring wear. The reported control experiments describe a system error of approximately 0.10 mm. The repeated measurements from the study group confirm that the method was repeatable. The radiation exposure from a stereo pair of RSA radiographs represents a very small relative risk to the patient, with the total dose equivalent to approximately three weeks of background radiation. It is concluded that the in vivo method for measuring polyethylene penetration is accurate, reliable and presents minimal risk to patients. The value for penetration represents a combination of creep and linear wear. Delineating the relative contribution of the two processes is a limitation of this method and the measurements represent an overestimation of wear. However the majority of creep is believed to occur within three months of implantation.14,15 Therefore at ten years post-operation the proportion of true wear to creep will increase, reducing the contribution of creep to the penetration value.
The mean in vivo wear rate for the group of ten-year Oxford UKAs was 0.02 mm/year. This figure represents a low wear rate and is similar in magnitude to both the 0.01 mm/year measured with in vitro wear testing, and the value estimated from retrieved bearings with normal macroscopic wear patterns.7,8 The patients within the study group had good clinical outcomes and the results provide evidence that low wear rates are maintained for up to ten years in patients in whom the device works well. A low wear rate in itself would not be expected to produce any functional outcome benefit for the patient in the short-term, but should offer the prosthesis a survival advantage, enhancing its longevity. The range of wear rates was −0.004 to 0.046 mm/yr (Table III⇑). The apparent increase in bearing thickness suggested by the negative value probably reflects the error introduced in estimating the original bearing thickness.
The mean wear rate of 0.02 mm/year measured in the in vivo study compares favourably with the published results of polyethylene penetration for other forms of arthroplasty which use a metal-on-polyethylene bearing. The value is approximately ten times less than the penetration rates of 0.1 to 0.2 mm/year reported for total hip arthroplasty.16–18 Despite many published retrieval studies there are few reports of the linear penetration rate in total knee arthroplasty. Plante-Bordeneuve and Freeman19 reported a mean wear rate of 0.025 mm/year for a conforming total knee arthroplasty (TKA), with a large contact area. However, this device did not afford axial rotation. Contemporary, unconstrained TKAs have penetration rates of around 0.2 mm/year, reflecting their lower contact area.9,20 Ashraf et al21 reported a penetration rate of 0.15 mm/year for the St. Georg Sledge fixed-bearing UKA prosthesis. There are no other published penetration rates for fixed-bearing devices, but early catastrophic polyethylene wear has been a problem with many designs and it is safe to assume that linear penetration rates were high in these cases.2,22,23 Despite high linear wear rates measured in some retrieved specimens, catastrophic wear was not a feature with the Oxford device. The use of in-air gamma-irradiated polyethylene may have been a factor in these findings, but low conformity and high contact stress may predispose fixed-bearing devices to more aggressive wear patterns.
The RSA system reported in this paper has a number of possible applications. It can be used as an outcome tool for monitoring the wear of established prostheses in the long term and provides a complimentary method to the other forms of assessing polyethylene wear. Additionally, it could be used for studying wear in new designs of prosthesis as they are introduced. The current RSA system can measure wear of approximately 0.1 mm. Wear rates in TKA are approximately 0.1 mm/year, therefore, the system could potentially be used to measure wear at two to three years post-implantation. The system could be employed to identify accelerated wear, which may predispose to early failure.
The Oxford meniscal-bearing arthroplasty was designed to reduce wear, a factor that has adversely affected the results of fixed-bearing unicompartmental replacements. The results from this in vivo study confirm that low wear rates can be achieved and maintained in the long term.
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 non-profit organisation with which one or more of the authors are associated.
- Received January 24, 2005.
- Accepted June 14, 2005.
- © 2005 British Editorial Society of Bone and Joint Surgery