PRESIDENTIAL GUEST LECTURE: Tribology of Alternative Bearings

Abstract

The tribological performance and biological activity of the wear debris produced has been compared for highly cross-linked polyethylene, ceramic-on-ceramic, metal-on-metal, and modified metal bearings in a series of in vitro studies from a single laboratory. The functional lifetime demand of young and active patients is 10-fold greater than the estimated functional lifetime of traditional polyethylene. There is considerable interest in using larger diameter heads in these high demand patients. Highly cross-linked polyethylene show a four-fold reduction in functional biological activity. Ceramic-on-ceramic bearings have the lowest wear rates and least reactive wear debris. The functional biological activity is 20-fold lower than with highly cross-linked polyethylene. Hence, ceramic-on-ceramic bearings address the tribological lifetime demand of highly active patients. Metal-on-metal bearings have substantially lower wear rates than highly cross-linked polyethylene and wear decreases with head diameter. Bedding in wear is also lower with reduced radial clearance. Differential hardness ceramic-on-metal bearings and the application of ceramic-like coatings reduce metal wear and ion levels.

Polyethylene-on-metal bearings in hip prostheses have suffered long-term failure because of polyethylene wear debris-induced osteolysis.⁵⁷ This is a consequence of complex biomechanical and biological interactions that are dependent on the volume of wear debris, the size and shape of the debris, the resulting biological activity, and individual patient-specific reactions.^31,32 Increased volumetric wear and osteolysis with historical polyethylene acetabular cups has been attributed to third body damage to the metallic femoral head,² oxidative degradation of gamma in air sterilized material,¹⁸ and increased femoral head size.³⁷ Smaller submicron size polyethylene wear particles produce a more intense inflammatory response in macrophages.²⁴ The incidence and location of the periprosthetic osteolysis is also dependent on the access of the wear particles, prosthesis design, and the effective joint space.⁴⁹ With historical gamma-irradiated in air polyethylene, osteolysis and failure occur after 5 to 15 years.^31,37 Assuming 1.5 million steps per year, this corresponds to a total lifetime of 10 to 20 million cycles. A wide range of wear rates have been observed in failed prostheses and there is an association between high wear rate and early failure. One retrieval study of Charnley prostheses, with a mean lifetime at failure of 12 years, recorded a mean wear rate of 60 mm³ per year (linear penetration 0.17 mm per year) and an overall mean wear volume of 785 mm³ of polyethylene at failure.⁵² The total volume of wear at failure was similar to values reported in another study.¹⁰ Assuming 1.5 million steps per year, this wear rate was similar to wear rates found in hip simulators of 40 mm³ per million cycles for the same polyethylene material.^1,2

However, younger and more active patients who have longer life expectancies in the range of 20 to 40 years and higher levels of activity, with up to a million steps per year,⁵⁰ the lifetime tribological demand may increase 100 to 200 million steps, up to a 10-fold increase on the tribological demand, causing failure with historical polyethylene. In addition, there is an increased interest in the use of larger size femoral heads to provide a greater range of motion, reduce impingement, and improve stability.⁵ However, with historical polyethylene, clinical studies³⁷ and tribological studies¹⁶ have shown a higher volumetric wear rate with increased head size because of the greater sliding distance of a large head for the same angular motion of flexion. Ideally, an alternative bearing should have a 10-fold improvement in performance compared with historical polyethylene. This substantial increase in tribological performance cannot be met by conventional polyethylene, which has been the primary reason developing alternative bearing materials such as highly cross-linked polyethylene, ceramic-on-ceramic, and metal-on-metal over the last 10 years.²⁸ It will be at least another 10 years before the comparative long-term clinical performance of these new alternative bearings becomes available, and current long-term clinical performance can only be extrapolated from preclinical in vitro simulation studies and early clinical results.

Hip simulation systems have been used extensively over the last 10 years to support the research and development of new bearing systems for hip prostheses.^4,6,35,55 However, different methods used in various laboratories make it difficult and inadvisable to compare data from different groups. Whereas considerable numbers of studies are performed in commercial laboratories, limited or selected publication means it is difficult for users to gain a full understanding of the overall landscape and options. We have within our laboratory accumulated a portfolio of independent studies over time and are able to make within laboratory comparisons of different bearing types.¹⁶

Concerns have also been raised regarding the convention of only studying wear under a single set of conditions-the walking cycle. Within our laboratory a wider range of conditions have been considered. These include the influence of kinematics,^1,13 varying levels of peak load,⁴⁴ swing phase load,⁵⁹ and joint microseparation^51,58 with validation against clinical retrievals.⁴³ It has been common practice only to consider volumetric wear rate when comparing the performance of alternative bearings and yet it is clear the intensity of osteolysis depends on the shape and size of the particles and resultant biological and chemical activity.³³ To compare alternative bearings it is desirable to also isolate and characterize the wear particles^52,53 and determine the relative biological activity of the real wear debris as demonstrated for polyethylene,^21,33 ceramic,^27,54 and metal.^15,24 By combining the volumetric wear rate, particle characteristics, and the macrophage response to the particles, a model of functional biological activity was developed which allows direct comparison of the overall tribological and biological performance of different bearing combinations.^11,17

Our aim was to compare the tribological performance, wear rate, and functional biological activity of alternative bearings for hip prostheses for active patients by analyzing the results of a portfolio of tribological, wear particle, and cell culture studies from a single laboratory. The wear rate and functional biological activity of highly cross-linked polyethylene was compared with conventional polyethylene for different head types. The wear rate and functional biological activity of alumina ceramic-on-ceramic bearings were compared with polyethylene under standard and microseparation conditions. We considered the influence of head size and radial clearance on the wear rate of cobalt-chrome metal-on-metal hips. We also compared the wear rate and biological activity of modified metal-on-metal bearings with differential hardness (ceramic-on-metal) and ceramic-like coatings with standard metal-on-metal bearings.

MATERIALS AND METHODS

The wear of highly cross-linked polyethylene (GUR1050; molecular weight, 4-6 million irradiated with 10 MRad irradiation and remelted) was compared with conventional polyethylene (GUR1020; molecular weight 2-4 million GVF, gamma irradiated with 4 MRad in a vacuum and foil packed) acetabular cups in the Leeds hip joint simulator articulating against 28 mm cobalt-chrome femoral heads.¹¹ Four specimens of each type of material were studied and a further station was loaded without articulation to determine creep. Each articulating station was subjected to twin peak physiological loading (max 3 kN), with flexion/extension and internal/external rotation generating an open elliptical wear track.^1,2,11 We conducted tests to 5 million cycles in 25% (v/v) new born calf serum (15 mg/L protein concentration) and determined the volume change of the cups every 1 million cycles using a coordinate measuring machine. The volume change in the first 1 million cycles included creep plus wear followed by a steady state volume change which was predominantly wear from 1 to 5 million cycles. An additional set of tests with conventional polyethylene on alumina ceramic heads was also performed under the same conditions, along with a study of highly cross-linked polyethylene sliding against size 36 mm cobalt-chrome and alumina ceramic femoral heads. Wear debris was isolated during the steady state wear regions from three independent stations and analyzed.^11,21,53 We determined the number and volume of wear particles as a function of size. As a measure of biological activity, we used the levels of TNF-alpha released per unit volume of debris from primary human macrophages as a function of particle size;³⁹ the specific biological activity of the debris (SBA; biological activity per unit volume) was determined.^17,57 The specific biological activity was combined with the volumetric wear rate to predict the comparative functional biological activity (FBA),¹⁷ which is a measure of functional osteolytic potential.

We studied size 28 mm alumina ceramic-on-ceramic bearings wear in the Leeds hip joint simulator under the standard walking conditions described above (n = 3) for 5 million cycles. Hips were tested in 25% (v/v) bovine serum and the wear volume was determined by gravimetric analysis and by surface profilometry. Hips were also studied under microseparation conditions in which the head was allowed to contact the superior rim of the cup,⁵¹ replicating the stripe wear found in clinical retrievals for 28 mm and 36 mm bearings.⁴³ Sufficient wear debris could not be isolated from the standard simulator tests, but debris from simulator studies run under microseparation conditions was cultured with primary human macrophages to determine the specific biological activity.²⁶ The functional biological activity index was determined as above and compared with polyethylene.¹⁷ The results from the simulator studies, the volumetric wear rate, and debris characteristics were compared with retrievals studies.^43,45

The wear of metal-on-metal bearings depends on a wider range of variables. We therefore performed a range of parametric studies to independently identify the role of these variables. The influence of the carbon content of the wrought alloy was investigated in simulator studies to 5 million cycles in 25% (v/v) bovine serum with gravimetric measurements of wear volume every 1 million cycles. Simulator studies were performed under different kinematics conditions^13,15 and with different swing phase loads⁵⁹ (n = 3 specimens for each condition) with size 28 mm wrought cobalt-chrome alloy bearings with a radial clearance of 30 μm.

The influence of head size on wear was studied in the simulator with using (n = 5) size 39 mm and size 54 mm diameter cast cobalt-chrome alloy bearings to 5 million cycles. In all studies, wear was defined to 1 million cycles as bedding in wear and from 1 to 5 million cycles as steady state wear. We investigated the influence of radial clearance on initial bedding in wear using size 28 mm diameter bearings (n = 10) with radial clearances in the range of 15 to 150 μm,¹² and in larger diameter surface replacement bearings. Fluid lubrication conditions were analyzed based on elastohydrodynamic mechanism,³¹ and the influence of radial clearance on initial bedding in wear was analyzed by three-dimensional geometry.³⁰

Wear particles were analyzed using high resolution transmission electron microscopy (TEM) and field emission gun scanning electron microscopy (FEGSEM).²² Biocompatibility was evaluated using culture with L929 fibroblasts to assess cytotoxicity, and primary human macrophages to determine the inflammatory response.²²

The potential for reduction in the volume metallic wear debris was investigated by using differential hardness ceramic-on-metal bearings¹⁴ and ceramic-like coatings, including CrN and CrCN, which were greater than 10 μm thick.^19,20,60 We performed simulator studies to 5 million cycles. Wear was determined every 1 million cycles and cytotoxicity tests were performed on the wear debris.

RESULTS

The highly cross-linked polyethylene showed a reduction in wear rate and functional biological activity compared with conventional polyethylene. Polyethylene wear was reduced with ceramic femoral heads for both polyethylene materials, and an increase in head size increased polyethylene wear with highly cross-linked and conventional material. There was an eight-fold reduction in the volumetric wear rate of the highly cross-linked polyethylene compared with conventional polyethylene (Fig 1A).^11,21 Ceramic femoral heads reduced the wear of conventional polyethylene by 35% and highly cross-linked polyethylene by 40% (Fig 2). However, the highly cross-linked (higher molecular weight) polyethylene produced smaller wear debris with a larger proportion of its volume in the submicron size range, making it more biologically active than the conventional polyethylene wear debris.^11,21 This was consistent with direct cell culture studies that showed an increase in molecular weight (GUR1050) and cross-linking independently increased the biological reactivity of the wear debris.³³ The specific biological activity of the highly cross-linked polyethylene was close to twice the conventional polyethylene (Fig 1B), which resulted in a functional biological activity (FBA) of highly cross-linked polyethylene (Fig 1C) four-fold lower than conventional polyethylene.^11,21 Ongoing studies in our laboratory with highly cross-linked polyethylene (10 MRad E beam) with 36 mm diameter heads have shown a higher wear rate of 10.6 ± 1.4 mm³/million cycles, more than twice the value found for size 28 mm diameter heads.

Fig 1A: C. (A) A graph shows the volumetric wear rate for highly cross linked and conventional polyethylene. Data are presented as the mean (n = 4) ± 95% confidence limits. (B) A graph shows the specific biological activity (SBA) of highly cross linked and conventional polyethylene. (C) A graph shows the functional biological activity (FBA) of highly cross linked and conventional polyethylene.

Fig 2: A graph shows a reduction in the wear rate of conventional polyethylene with an alumina ceramic femoral head compared with a metallic femoral head. Data are presented as the mean (n = 4) ± 95% confidence limits.

Under standard simulator conditions the size 28 mm diameter alumina ceramic-on-ceramic bearings had a wear rate of less than 0.1 mm³/million cycles, which was 50-fold lower than highly cross-linked polyethylene and 350 times lower than standard polyethylene. However, standard simulator conditions did not replicate the stripe wear found on the head of retrievals,^43,45 which is thought related to head and cup rim contact after microseparation. Simulation of microseparation in the hip simulator replicated the stripe wear seen on retrievals and gave a steady state wear rate of 1.4 ± 0.2 mm³/million cycles.⁵¹ When microseparation was simulated on every gait cycle the steady state wear of the alumina ceramic was four times lower than the highly cross-linked polyethylene and 25 times lower than the conventional polyethylene. Analysis of the wear debris from the microseparation simulations showed a bimodal size distribution, with nanometer size debris from the normal articulation surfaces, larger micron size granules from grain boundary failure, and pull out from the stripe wear area.⁵⁴ This was consistent with debris found from retrievals.²⁷ Cell culture studies showed the alumina debris produced from microseparation simulation studies was less reactive than conventional and cross-linked polyethylene debris, with a specific biological activity of 0.18 (compared with Fig 1B). When combined with the reduced wear volume, this resulted in a substantially lower functional biological activity, which was almost 20 times lower than predicted for highly cross-linked polyethylene (Fig 3) and 80 times lower than conventional polyethylene. This low propensity for osteolysis is reflected in retrieval studies in which low wear rates have been identified and with a corresponding low incidence of osteolysis.^25,45 These retrieval studies also suggest that, in contrast with polyethylene, since wear was predominantly associated with head and cup rim contact an increase in wear with increased bearing size was not found.⁴⁴

Fig 3: A graph shows the functional biological activity of alumina ceramic and highly cross linked polyethylene bearings.

The overall wear rate of the low carbon content alloy was six-fold greater (p < 0.01) than the high carbon content alloy (Fig 4), indicating low carbon content alloys should not be used in metal-on-metal bearings. An increase in the applied swing phase load produced a higher wear (p < 0.05) for both cup positions (Table 1).^13,15,59 The elevated wear with elevated swing phase load was also associated with an elevated coefficient of friction. These changes in friction and wear with increased swing phase load were indicative of a reduction in fluid lubrication.

Fig 4: A graph shows the overall wear rate of low carbon content and high carbon content cobalt chrome alloy bearings. Data are presented as the mean (n = 3) ± 95% confidence limits.

TABLE 1: Overall Wear Rate of 28 mm Metal-on-Metal Hips with Different Swing Phase Loads

Wear of metal-on-metal hips reduced as the head diameter increased. For similar radial clearances the wear of the size 55 mm diameter bearing was lower than the size 39 mm bearing (Fig 5). Both bearings showed the characteristic initial high wear rate with metal-on-metal bearings during the bedding in phase up to 1 million cycles and then a lower steady state wear rate. The initial bedding in wear volumes at 1 million cycles, 2.58 mm³ for size 39 mm compared with 1.15 mm³ for 55 mm were different (p <0.01). Three-dimensional geometric analysis showed for the same radial clearance, the bedding in wear volume reduced with head diameter.³⁰ The steady state wear rate of 0.3 mm³/million cycles for 39 mm bearings was higher (p = 0.06) than the 0.13 mm³/million cycles for 55 mm bearings. Lubrication analysis³¹ showed as the head size increased the fluid film thickness increased, hence reducing wear (Fig 6).

Fig 5: A graph shows the wear volume as a function of number of cycles for 39 mm top and 55 mm bottom metal on metal surface replacements. Data are presented as the mean (n = 5) ± 95% confidence limits.

Fig 6: A graph shows the lubricating film thickness as a function of diameter and radial clearance for metal-on-metal bearings.

An increased radial clearance increased the wear in metal-on-metal hips. The initial bedding in wear increased with radial clearance (Fig 7) for size 28 mm hips.¹² As the radial clearance increased, the bedding in wear increased from 0.5 to 4 mm³ over a range of radial clearances from 20 to 150 μm.¹² For larger size surface replacement hips, the bedding in wear was 1.1 ± 0.2 mm³ for a radial clearance of 51 μm (Fig 5) and 1.8 mm³ for a radial clearance of 150 μm. This increase in bedding in wear volume with increase in radial clearance was predicted by geometrical analysis.³⁰ Lubrication analysis also predicted a reduction in lubricating film thickness and increased wear with increased radial clearance (Fig 6).

Fig 7: A graph shows the bedding in wear as a function of radial clearance for 28 mm bearings. Data are presented as the mean (n = 10).

Metallic cobalt-chrome wear particles were in the nanometer size range (mean size, 28 nm), and were cytotoxic to cells at concentrations of 5 μm³ per cell and greater.²² Because of the effect of the metal ions on the macrophages, the metallic particles did not stimulate an inflammatory response in the same way as polyethylene particles. The cytotoxic effect of the metallic debris remains a concern and has been a substantial factor in seeking a reduction in the volume of metal wear and reduction in ion levels.

The wear of ceramic-on-metal bearings was 100-fold lower than with metal-on-metal bearings (28 mm) (Fig 8). The 100-fold reduction in wear rate with ceramic-on-metal bearings was attributed to the differential hardness of the bearing surfaces, smoother surfaces, improved lubrication, and a reduction in corrosive wear.¹⁴

Fig 8: A graph shows the wear rate of ceramic-on-metal compared with metal-on-metal bearings. Data are presented as the mean (n = 3) ± 95% confidence limits.

The wear of thick CrCN bearings showed a 100-fold reduction in wear (Fig 9) compared with metal-on-metal bearings.^19,20 The wear debris from CrCN was more bio-compatible and less cytotoxic than the metallic debris.⁶⁰

Fig 9: A graph shows the wear rate of CrCN coated bearings compared with metal-on-metal. Data are presented as the mean (n = 3) ± 95% confidence limits.

DISCUSSION

Concern regarding polyethylene wear debris-induced osteolysis and the desire to use larger femoral head sizes for more active patients has led to the development and clinical use of a range of alternative bearings during the last 10 years.²⁸ However, it will be many years before the long-term clinical performance and relative in vivo tribological performance can be determined, therefore we rely on in vitro simulator studies and isolated retrievals to compare the performance of different material bearings. We used simulation methods, wear particle analysis, and particle/cell culture studies from a single laboratory to compare the tribological performance, wear particles, and functional biological activity of a range of alternative bearings including highly cross-linked polyethylene, ceramic-on-ceramic, metal-on-metal, and modified metal-on-metal.

The recognized limitations of in vitro laboratory studies include variation in the methods used by different laboratories. This makes it difficult to compare results but was overcome by comparing data from a single laboratory. Also, only a standard gait walking cycle is used in hip simulations. This may not replicate the wear occurring in other activities or under other loading conditions, and so may lack clinical relevance. We developed a range of simulator conditions with different kinematics and loading conditions and validated wear rates, mechanisms, and debris generated against retrievals of historical bearings. However, it must be noted a limited envelope of conditions were selectively used. Many researchers consider only the wear volumes, while different alternative bearings have different types of debris and biological reactions. We characterized wear particles from all materials and determined the biological activity of real wear particles directly from in vitro cell culture studies. It should be recognized many biological markers could be used to determine biological activity, and in practical terms, cell viability was used as a measure of cytotoxicity and macrophage release of TNF-alpha used as a measure of inflammatory response.²³ It also must be recognized as different types of debris illicit different types of biological reactions, it can be difficult to make a direct quantitative comparison. Additionally, it is necessary to consider the combination of the volume of wear and the biological response. Using TNF-alpha as a marker of osteolysis, we developed an indicator of osteolysis, the functional biological activity. We only used a single cytokine, involved in vitro assays, and adopted an integration of simple linear functions. Additionally, the in vitro model does not take account of in vivo transport of particles, which may be size and material-dependent. It does, however, attempt to combine both tribology and biology for the first time, and can provide additional insight beyond comparing simple mechanical wear.

This study draws on results from published papers over a period of 8 years from one academic laboratory to allow comparisons between the tribological performances of different alternative bearings and compares these finding with other work. We did not attempt a comprehensive literature review, which can be found in other articles.^16,31,32

Like other studies, we found an eight-fold reduction in wear of highly cross-linked polyethylene in the laboratory simulator systems.^6,40-42 However, unlike others who have reported close to zero wear and the retention of machining marks on the polyethylene,^6,42 we found finite long-term steady state wear rates of 5 and 10 mm³ per million cycles, loss of machining marks, and clearly identified micron and submicron size polyethylene debris.²¹ It is interesting to contrast these findings and the different methods used by others. We used a physiologically relevant level of protein lubricant (15 mg/L); three times lower than the 90% serum used in other studies. High serum concentrations produced lower wear in a simulator study.⁵⁵ We used geometrical measurements to determine wear, which avoids the artifacts due to moisture uptake found in other studies used gravimetric measurements, which can mask low levels of wear. The geometrical measurements allowed direct comparison with clinical penetration rates showing early phase deformation and creep, followed by lower longer term penetration because of wear. The comparison of our simulator studies with these time-dependent volume changes found clinically^7,28,29,38 and our evidence of the wearing away of machining marks, which has been validated by clinical studies,³ shows clinically relevant finite wear rates for highly cross-linked polyethylene. Other studies⁶ have not shown an increase in wear with increased head size for cross linked polyethylene, while our most recent work has shown an increase in volumetric wear with size 36 mm heads to over 10 mm³/million cycles, consistent with the tribological laws for boundary lubricated systems and previous clinical studies of conventional polyethylene.³⁷ The use of a ceramic femoral head on highly cross-linked polyethylene showed a reduction in wear of 40% compared with metallic heads, which was consistent with studies of conventional polyethylene and clinical studies.

The analysis of polyethylene wear debris down to 10 nm in size using FEGSEM^21,33 and the ability to directly culture real wear debris with cells has allowed us to differentiate between the effect of both molecular weight and cross linking on biological activity. The levels of TNF-alpha produced by macrophages was dependent on the volumetric distribution of the particle sizes, with a greater proportion of the wear debris volume being less than 1 μm, resulting in a greater specific biological activity and level of response. Increasing the molecular weight of the polyethylene and/or cross-linking the polyethylene increased the proportion of smaller particles producing debris that was more reactive.^11,21,33 Overall this model predicts a three- to four-fold reduction in functional biological activity, an indicator of osteolytic potential for highly cross-linked polyethylene with 28 mm bearings. This leads to the question of whether this is sufficient to meet the demand to use larger 36 mm head sizes (with twice wear rate) with highly cross-linked polyethylene when patients' lifetime demands exceed 100 million cycles. The laboratory studies indicate the improvement in tribological performance of highly cross-linked polyethylene when coupled with elevated biological activity of the wear particles may not be sufficient to meet the lifetime functional demands of highly active patients.

Ceramic-on-ceramic bearings arguably have the best tribological performance, the lowest wear, and generate particles with the lowest biological reactivity of the three types of alternative bearings. The long-term clinical performance of first generation devices is supported by these simulator, particle, and biological activity studies^{26,27,43,51,54} which have been validated by retrieval studies.^43,45 The discovery of head-rim contact and stripe wear, which may be associated with microseparation, a steep cup angle, or stem/neck impingement has highlighted the importance of surgical technique. However, the level of tribological performance and biocompatibility of the debris, which is available in 28 and 36 mm, does provide a smaller and larger head solution to meet the functional requirements of 100 million cycles for active patients today. Of course, the risk of component fracture at less than one in 1000 and some limitations in design flexibility and insert cup thickness may be cited as less favorable features. The next technological challenge for ceramic-on-ceramic bearings is to find larger diameter head solutions which preserve bone.

The tribological performance of metal-on-metal hips provides substantially lower wear than highly cross-linked polyethylene bearings and the widest range of head sizes of the three bearing surfaces currently available. However, they are the more tribologically and biologically complex, with a wider range of clinical wear rates,⁴⁷ and potentially adverse biological reactions^8,35,56 as a result of elevated ion levels, metallosis, or the development of hypersensitivity, albeit at very low levels of incidence. Tribologically, metal-on-metal bearings can benefit from enhanced fluid lubrication to reduce wear, but this does lead to a sensitivity of wear performance depending on load and kinematics conditions,^13,59 lubricant,⁵⁹ metallurgy,^15,47 design diameter,³⁵ and radial clearance.³⁰ The influence of material and design has been controversial with different commercial solutions being available and results with these have helped clarify the influence of key variables on wear performance. The influence of metallurgy on wear has been clarified and low carbon content alloys should not be used (Fig 4). The influence of cast or wrought processing and heat treatment has been more controversial. However, a review paper from three tribological centers showed these process variables have little effect on wear of metal-on-metal hips.⁴⁷

The initial higher bedding in wear followed by lower steady state wear has been reported in many studies.^13,15,47,59 This two-phase wear has to be considered when studying the influence of the key design variables, the head diameter, and radial clearance. The reduction in wear with increased head diameter is a feature of metal-on-metal bearings associated with fluid lubrication (Fig 5), and was first described in our theoretical paper.³⁵ This has been supported by a range of other studies.^9,47 In addition, our theoretical study of bedding in wear shows, for the same radial clearance, the larger diameter head has lower bedding in wear, a finding confirmed experimentally. Therefore, there is a body of sound experimental and theoretical tribological evidence to support a reduction in wear with larger diameter metal on metal bearings, which supports the use of surface replacement hips. There is less of a consensus regarding the influence of radial clearance on wear despite tribological science being based on the same premise as accepted for an increase in head size. An early study by Farrar¹² on the effect of radial clearance on bedding in wear (Fig 7) showed the clear relationship between increased wear with radial clearance in size 28 mm bearings, which was explained by the three-dimensional theoretical modelling.³⁰ This increase in wear has been supported by two independent experimental studies in other centers.^9,48 There is a minimum radial clearance for any bearing to accommodate manufacturing tolerances, deformation, and deflection to avoid equatorial bearing and jamming that is design manufacturer-specific. Early generation metal-on-metal hips required large radial clearances simply because the manufacturing tolerances and sphericity were not well controlled. This is one reason why it is difficult to interpret historical data. Theoretically,³⁵ a larger radial clearance increases steady state wear, but the experimental evidence to support this is less strong.^9,48 However, there is a clear dependency of the overall wear rate on radial clearance. Concerns have been expressed that conforming geometries and low radial clearances may lead to equatorial bearing and jamming in the long term. However, computational wear models run to 50 million cycles do not support these concerns.³⁶

Concerns about ion release from metal wear particles and hypersensitivity have led to interest in modified metal-on-metal bearings, including ceramic-on-metal and thick ceramic-like coatings. The former is suitable for sizes up to 36 mm and the 100-fold reduction in wear, wear debris, and ion release found in simulator studies has led to a multicenter international clinical trial. Ceramic-like coatings are attractive for larger heads and surface replacements and laboratory studies of both CrN and CrCN coatings are showing considerable potential for reducing wear, ion levels, and wear debris which is more biocompatible.

The analysis of data from 28 different studies carried out in one laboratory has allowed the comparison of the tribological and in vitro biological performance of three types of alternative bearings currently available for clinical use in high demand patients. Highly cross-linked polyethylene had improved performance over conventional polyethylene, is suitable for low and medium demand patients, and can be used with large diameter heads for low demand patients requiring additional stability. However, it does not appear to have sufficient improvement in tribological and biological performance to meet the life-time tribological demands of highly active patients. Ceramic-on-ceramic bearings showed the best overall tribological and biological performance, but this has to be set in the context of head size constraints (36 mm limit) and design constraints on neck length and acetabular component thickness. Metal-on-metal bearings provide good tribological performance and low wear, when designed correctly, and offer a wider range of head sizes, but there remain some concerns about their biological performance in a few patients. Future modified metal-on-metal bearings should address these concerns.

Acknowledgments

The authors thank T. D. Stewart, J. Nevelos, P. Firkins, S. Williams, A. Galvin, J. Ingram, A. Hatton, M. Germain, I. Leslie, and R. Farrar for their contributions to the results and publications referred to in this article.