Polyethylene Wear Increases in Liners Articulating With Scratched Oxidized Zirconium Femoral Heads

Abstract

Background

Oxidized zirconium (OxZr) femoral heads were introduced in hip arthroplasty to reduce wear of the polyethylene compared with metallic heads and to reduce fracture risk compared with ceramic heads. Severe scratches have been reported on OxZr heads in patients undergoing revision for instability, but whether these scratches contribute to increased acetabular polyethylene wear remains unclear.

Questions/purposes

(1) How is the polyethylene of the acetabular liner affected by damage on the opposing OxZr head? (2) How does damage to the head affect the degree of polyethylene wear?

Methods

We assessed damage and deformation on all retrieved highly crosslinked liners that had articulated against OxZr heads collected at one institution between 2006 and 2013. Two observers used a visual subjective scoring system to assess polyethylene damage on the surface of the 42 retrieved liners. Polyethylene components were also laser scanned to measure dimensional changes to the liner. These outcomes were compared with the severity of scratching on the surface of the articulating OxZr head. We also used a 12-station hip simulator to measure wear over five million cycles (MCs) of pristine liners articulating against nine retrieved OxZr heads with varying degrees of scratching representing a spectrum of little to severe damage and three retrieved ceramic heads with severe metal transfer.

Results

Seventeen of the OxZr heads showed severe damage, of which 14 heads had been revised for dislocation. The retrieved liners that had articulated with these heads had greater damage scores for abrasion (mean score 0.4 versus 1.6; p = 0.008) and embedded debris (mean score 0.4 versus 1.4; p = 0.006) compared with liners that had articulated with less damaged heads. Four severely damaged OxZr heads wore at a higher rate than the others in the study with weight loss of 37.7 mg, 30.0 mg, 14.4 mg, and 2.6 mg after the first MC and a steady increase through testing to 5 MCs. Conversely, neither OxZr heads with less damage nor ceramic heads with severe metal transfer produced appreciable wear.

Conclusions

Surface scratching of OxZr heads from recurrent dislocation and reduction maneuvers leads to increased wear of the crosslinked polyethylene used as a bearing surface.

Clinical Relevance

Patients in whom such instability occurs in the presence of an OxZr head should be followed closely for the possibility of more severe wear.

Introduction

Long-term success of THA is dependent on minimizing polyethylene wear to avoid periprosthetic osteolysis. The combination of highly crosslinked acetabular bearings, as a result of their outstanding wear resistance [14, 17], and ceramic femoral heads with hard, scratch-resistant surfaces represent one approach in common use. Compared with metallic femoral heads, ceramic heads provide a smoother, harder bearing surface, which reduces abrasive and adhesive wear against polyethylene [3] and reduces the severity of mechanically assisted crevice corrosion at the head-taper junction [10, 21]. Despite these advantages, concern over fracture based on the brittle nature of ceramics remains, because early generations of ceramic heads had a worrisome risk of fracture [22]. With improved fabrication and manufacturing techniques, the fracture rate of modern alumina-zirconia composite ceramic heads (BIOLOX® delta; CeramTec AG, Plochingen, Germany) is reported to be 0.0010% (28 fractures in 28,000 heads over the past 16 years) [11].

To provide the superior wear resistance of a ceramic bearing surface with the toughness and fracture resistance of a metallic head, oxidized zirconium (OxZr) was commercialized in 2003 (Oxinium; Smith & Nephew, Memphis, TN, USA) for clinical use in THA. OxZr femoral heads are composed of a zirconium metallic alloy, which, when heated in the presence of oxygen, results in transformation of the zirconium near the surface into a stable zirconia oxide ceramic. The zirconia region extends only a short distance into the alloy (approximately 4 μm), whereas the remainder of the head remains the zirconium alloy. Thus, the bulk of the head maintains the strength and ductility of a metallic head, whereas the ceramicized outer surface provides the hardness and abrasion resistance inherent to ceramics.

Although midterm wear rates of polyethylene acetabular liners bearing against OxZr femoral heads are low [6–8], scratching and surface damage of OxZr heads have been reported in recent case series. This damage accompanied joint dislocation and reduction, because the head rubbed across the edge of the metallic backing of the acetabular component [18] exposing the underlying, softer metallic alloy, increasing the local surface roughness around the damaged area and potentially generating third-body debris [13]. We previously reported that among 59 retrieved OxZr femoral heads, 75% of the 24 heads retrieved from patients who had experienced recurrent dislocation exhibited severe surface damage and increased surface roughness [16]. These findings raised our concern that acetabular liners articulating with damaged OxZr heads could lead to increased polyethylene wear.

The true clinical importance of these findings and the potential for increased polyethylene wear as a result of articulation against a damaged OxZr head remain unclear. Therefore, we sought to determine how the severity of damage to an OxZr head affects the wear on the opposing polyethylene acetabular liner. We asked: (1) How is the polyethylene of the acetabular liner affected by damage on the opposing OxZr head? (2) How does damage to the head affect the degree of polyethylene wear?

Materials and Methods

Along with the 59 OxZr femoral heads included in our previous publication [16], we had a cohort of 42 heads for which the articulating crosslinked polyethylene (XLPE) liners were also collected during revision surgery. All of the polyethylene liners were neutral in design (did not have an elevated rim) and had been retrieved as part of the institutional review board-approved protocol followed as part of our ongoing implant retrieval program. The mean time to revision surgery for the 42 retrievals was 17 months (range, 0.3-84 months). Reasons for revision surgery were recurrent dislocation (21 cases), infection (nine), aseptic loosening of the acetabular or femoral component (five), femoral stem subsidence (two), heterotopic ossification (two), leg length discrepancy (two), and acetabular malposition (one). The mean age of the patients (22 men, 20 women) at index surgery was 60 years (range, 43-88 years). Because our previous study showed greater varying degrees of damage on the surface of OxZr femoral heads [16], we divided the 42 retrieved polyethylene liners into three groups: (1) liners that articulated against heads with a pristine surface (Grade 1 head damage); (2) liners that articulated against heads with one or two scratches (> 0.5 cm in length; Grade 2 head damage); and (3) liners that articulated against heads with severe damage with multiple scratches, surface effacement, or metal transfer (Grade 3 head damage).

Analyses of Retrieved Polyethylene Liners

For visual assessment of polyethylene damage to the bearing surfaces of the retrieved liners, each liner was divided into quadrants and graded for damage to the articulating surface using a modified version of the Hood subjective grading scale [4]. Grading was done under low-power (x 10) light stereomicroscopy (Type 355110; Wild Heerbrugg, Heerbrugg, Switzerland) by two independent observers (AC, CNK) blinded to the clinical information. When an individual score (eg, the score for burnishing in quadrant 1) differed by > 2 between the two graders, a third independent investigator (CIE) provided a grade to resolve the discrepancy. Each liner was evaluated for seven damage modes: surface deformation, embedded debris, scratching, burnishing, delamination, pitting, and abrasion. The damage from each mode was assigned a score from 0 to 3, where 0 indicated no damage, 1 indicated damage to < 10% of the zone, 2 indicated damage to 10% to 50% of the zone, and 3 indicated damage to > 50% of the zone.

After visual scoring, the bearing surfaces from all 42 liners were digitized using a commercially available desktop three-dimensional (3-D) scanner (Range 7; Konica Minolta, Tokyo, Japan; accuracy = 40 µm). Each insert was coated with aerosol talc and scanned twice every 36° of revolution to ensure that the entire bearing surface geometry was captured. The resulting point clouds from all 20 scans were then aligned and merged using Geomagic Studio software (Version 12; 3D Systems, Rock Hill, SC, USA) to create a single 3-D polygon model. Design- and sized-matched pristine liners were purchased and scanned in the same manner to produce pristine polygon models for comparison. Using Geomagic Qualify software (Version 2015; 3D Systems), each retrieval was then aligned and compared with its design- and size-matched pristine counterpart using the “Best Fit Alignment” tool and a previously described iterative closest point algorithm [20]. Deviations in dimensions between the retrieved and pristine inserts were then depicted colorimetrically to depict the combination of wear (removal of material from the surface) and deformation (yielding or creep of the polyethylene) that had occurred while the liners had been implanted. We have used this method previously to evaluate dimensional changes in total knee tibial inserts [12, 20]. To ensure there was proper registration of the retrieved liner to the pristine liner, we visually confirmed that the areas of greatest deviation on the maps corresponded to obvious wear on the actual insert. For each liner, the mean deviation and greatest maximal deviation were recorded.

To test our hypothesis that polyethylene damage and deformation would be greatest in XLPE liners from patients who experienced dislocation, we compared the damage scores and deviation measurements from the OxZr dislocation group with the OxZr nondislocation group. These heads had been scored in our previous study [15]. Briefly, 15 of the 21 (71%) dislocated OxZr heads had severe (Grade 3) damage to their bearing surfaces with evidence of multiple scratches and surface effacement compared with only three of 21 heads with severe damage in the nondislocated group.

Although we compared patients who have experienced a dislocation with those who had not had a dislocation for the impact of dislocation on polyethylene surface damage, we did not correlate head quadrant damage to liner quadrant damage because we did not know the relative orientation of the two components when implanted. We did not mark the components to maintain the orientation of the head relative to the liner. Furthermore, no elevated liners were included, which could have helped define the in vivo orientation of the components, and the wear patterns were not a reliable method to determine orientation. Nonetheless, we could still consider total head damage and total liner damage.

Hip Simulator Analysis

To assess the direct impact of damage to the OxZr bearing surface on wear of an opposing polyethylene acetabular bearing surface, three groups of retrieved femoral heads were chosen. The first group included six OxZr heads randomly selected from the 21 heads retrieved from patients who had been revised for dislocation. Three of these heads were 32 mm in diameter and three were 36 mm in diameter; all had Grade 3 damage. This group was compared with a second group of three heads randomly selected from the 21 other OxZr heads retrieved from patients revised for reasons other than dislocation. One of these was a 32-mm head revised for femoral component subsidence, another a 32-mm head revised for periprosthetic fracture, and the third a 36-mm head revised for infection; all three had Grade 2 damage. In addition, a third group of three alumina-zirconia composite (BIOLOX delta; two 32 mm in diameter and one 36 mm in diameter) ceramic femoral heads were tested. These three heads had been retrieved from patients with a history of recurrent dislocation and with visual evidence of severe (Grade 3) surface damage primarily in the form of large metal transfer scars [2]. In total, therefore, 12 femoral heads were analyzed: six OxZr heads from patients who had dislocated, three OxZr heads from patients without dislocation, and three ceramic heads from patients who had dislocated (Fig. 1). The mean surface roughness of the damaged areas in the first group (patients who had OxZr and had dislocated) ranged from 1.4 to 8.6 μm, in the second group (patients who had OxZr and did not dislocate) from 3.7 to 6.9 μm, and in the third group (patients who had BIOLOX delta and had dislocated) from 0.62 to 2.5 μm.

Fig. 1.

Photographs are shown of the 12 retrieved femoral heads used for the hip simulator study, including the six OxZr heads that had severe Grade 3 damage and had been retrieved at revision surgery for recurrent dislocation; the three OxZr heads that had moderate Grade 2 damage and had been retrieved at revision surgery for reasons other than recurrent dislocation; and the three BIOLOX delta heads that had severe damage in the form of metal transfer and had been retrieved at revision surgery for recurrent dislocation.

Seven pristine XLPE liners (R3^TM; Smith & Nephew) with an inner diameter of 32 mm and five XLPE liners with an inner diameter of 36 mm were used in the wear test. Six additional liners (three 32 mm and three 36 mm) were used for load, soaked controls. All had an outer diameter of 52 mm; their inner diameters were chosen to match the sizes of the retrieved heads. All liners were sterilized with plasma gas and, to recreate in vivo conditions, they were artificially aged through heating to 80° C in air for 21 days [19]. All liners were presoaked in distilled water and run in the simulator at the same time. The 12 pristine XLPE liners described were then enclosed in polyurethane molds in the inverted position with the cups mounted at 23° to the load axis to undergo 46° of biaxial rocking relative to the load axis in a 12-station orbital bearing-type hip simulator (Shore Western Manufacturing, Monrovia, CA, USA). This simulator was previously used in published investigations on polyethylene acetabular liner wear [15, 19]. OxZr and ceramic heads were mounted in self-aligning connection fixtures above each liner. Each of the 12 articulating stations was filled with lubricant composed of filter-sterilized bovine serum diluted with 0.2% sodium azide and 20 mmol/L EDTA. Heads were worn against the liners under a double-peak Paul-type load cycle that is representative of normal gait with a 2000-N maximum with motion and loading synchronized at one cycle per second. Six pristine liners were used as loaded, soak controls. These liners were axially loaded in a separate station of the simulator using the same loading cycle as those in the wear stations, but without any motion, all while being continuously soaked in the same serum lubricant as was used in the wear stations.

Every 500,000 cycles, the 12 liners being wear-tested and the three loaded, soak control liners were removed from the simulator. The wear test liners and soak test controls were cleaned, dried, weighed, and then inspected for visible evidence of cracking or failure. The wear volume was determined by increasing the measured weight loss of the worn liners by the mean weight gain of the soak control liners and dividing by the density of XLPE (0.93 g/cm³). A decrease in liner weight was interpreted as resulting from polyethylene wear. Photographs were analyzed to reveal any focal areas of liner wear and any evidence of metal transfer or third-body debris. After weighing, the liners were placed back on the simulator with fresh serum.

Liners were worn for a total of five million cycles (MCs) to represent 2 to 5 years of ambulation in a patient undergoing standard THA [23], comparable to the length of other hip simulator wear studies [14, 15, 19]. At the completion of the test, the 12 XLPE liners were laser scanned using the same protocol as was used for the retrieved liners to assess the deviations of the articulating surfaces consistent with any wear and deformation that had occurred from the simulator test.

Statistical Analyses

Patient demographics were compared between the OxZr dislocator and nondislocator groups using Student’s t-test. Polyethylene visual damage scores and the laser scanning surface deviations were compared against OxZr head roughness using a Kruskal-Wallis one-way analysis of variance. A Pearson correlational coefficient was calculated to examine the relationship between XLPE liner wear and the time from index dislocation to revision surgery. For the hip simulator wear data, XLPE liner weight change was compared among the OxZr heads from patients with dislocation, OxZr heads from patients without dislocation, and ceramic heads from patients without dislocation using repeated-measures one-way analysis of variance. All continuous variables were tested for normality using the Kolmogorov-Smirnov test. Significance was set at p < 0.05. For multivariable comparisons, post hoc Bonferroni correction was performed. All statistical analyses were performed using SigmaPlot 12 (Systat Software, Inc, San Jose, CA, USA).

Results

Retrieval Analysis of XLPE Acetabular Liners

There was no difference in polyethylene damage of the acetabular liners from patients revised for instability compared with those revised for other reasons. However, greater damage of the OxZr heads was associated with more damage to the retrieved liners. The 18 XLPE retrieved liners that had been retrieved with Grade 3 damaged OxZr heads had greater abrasive damage (mean score 1.6 versus 0.4; p = 0.01), embedded debris (mean score 1.4 versus 0.4; p < 0.01), and total damage scores (mean score 18.5 versus 15.1; p = 0.02) compared with the 24 liners with Grade 1 and 2 damaged heads (Table 1; Fig. 2). The embedded debris was black in color. Damage scores for scratching and pitting, the two most common types of surface damage observed on all of the retrieved liners, were not higher for the liners matched with Grade 3 heads (p = 0.75 and p = 0.65, respectively; Table 1). No liners showed visible surface delamination.

Table 1.

Polyethylene damage scores and the severity of damage were compared based on the level of damage (Grades 1 and 2 versus Grade 3) observed on the opposing OxZr femoral heads

Fig. 2 A-C.

(A) Mean damage scores (± SD) are shown for each damage mode (scratching [SCR], pitting [PIT], burnishing [BURN], abrasion [ABR], delamination [DELAM], surface deformation [SURF DEF], and embedded debris (EMB DEB]) as a function of the damage grade of the OxZr head against which they had articulated while implanted. (B) An example is shown of an abraded area of polyethylene damage. (C) Third-body debris (arrows) was found embedded in the articulating surface of some of the liners.

The maximum and mean deviations measured from the laser scans of the bearing surfaces of the retrieved inserts were not affected by the grade of damage to the retrieved OxZr head (Table 2). There was no relationship between length of implantation and maximum deviations from laser scanning (Pearson correlational coefficient = -0.124, p = 0.48), mean deviations from laser scanning (Pearson correlational coefficient = 0.0593, p = 0.739), or total damage scores of the retrieved liners (Pearson correlational coefficient = -0.1, p = 0.57).

Table 2.

Deviation measurements for the retrieved XLPE liners as a function of the damage score of the retrieved OxZr heads

Hip Simulator Study

OxZr heads with Grade 3 surface damage and a history of dislocation produced higher amounts of wear when loaded against the pristine XLPE acetabular liners in the hip simulator (Fig. 3). Four of the six Grade 3 damaged OxZr heads wore at a higher rate than the others in the study with weight loss of 38 mg, 30 mg, 14 mg, and 3 mg after the first MC and a steady increase through testing to 5 MCs. The highest amount of wear was noted in the first 500,000 cycles (14 + 11 mm³), the second highest occurred from 500,000 to 1 MC (10 + 6 mm³), and from then on until the end of the test, a smaller but persistent rate of wear (7-14 mm³/MC) was observed (Fig. 3). At every 500,000 cycles point of the test, the six Grade 3 damaged OxZr heads created more wear in the opposing XLPE liners than did the other six heads (the three Grade 2 damaged OxZr heads and the three Grade 3 damaged BIOLOX delta heads) with p values ranging from 0.023 to 0.035.

Fig. 3.

Wear in cubic millimeters of polyethylene is plotted against the number of cycles for the hip simulator test of 12 pristine XLPE liners against the 12 heads shown in Figure 1. The photographs of the heads along with the laser scan colorimetric maps of the articulating surfaces of the XLPE liners at the end of the test are shown next to the curves for the six Grade 3 OxZr heads, four of which showed considerable wear.

In contrast, neither the OxZr heads with Grade 2 damage and no history of dislocation nor the BIOLOX delta heads with severe surface damage and a history of dislocation produced any appreciable wear in the pristine XLPE liners throughout the 5 MC simulator study (Fig. 3). This finding was reflected by the gradual gaining of weight of the XLPE liners over time that matched the weight gains of the loaded, soaked controls.

The scans of the XLPE acetabular liners taken at the end of the test reflected the relative weight losses that had occurred over the 5 MC simulator test. The wear patterns for the four liners that showed appreciable wear were evident from the colorimetric deviation maps obtained at the end of the simulator test (Fig. 3). The matching four Grade 3 damaged OxZr femoral heads that had produced the wear in these liners had scratches located in the equatorial and apex regions of the head. When seated fully in the XLPE liners during the simulator test, these scratches passed over the area that was experiencing the peak load applied by the hip simulator. These locations of large deviations on the maps (Fig. 3) were consistent with visual evidence of heavy scratches on the XLPE surface.

Conversely, the two Grade 3 OxZr femoral heads that did not produce appreciable wear in the simulator test had scratches that were located in the subequatorial region of the head. When seated fully in the XLPE liners, these scratches did not pass over the area that was experiencing the peak load applied by the simulator; hence, the colorimetric maps showed little change in deviation as a result of the wear test (Fig. 3) and far less visual surface damage.

Discussion

Although scratches have been demonstrated on OxZr heads in retrieval studies [9, 16], it is unclear to what extent the damage observed might contribute to increased wear of the articulating polyethylene liner. We combined observations made on retrieved implants with laboratory wear simulator tests on some of these same retrievals to test the hypothesis that greater damage to an OxZr femoral head causes more wear to the opposing highly XLPE acetabular liner. We demonstrated that not only was the surface damage to the retrieved liners more severe when the OxZr heads were themselves more damaged, but also that these same heads were capable of creating greater amounts of wear when worn against pristine XLPE liners in a hip simulator test.

When we evaluated whether a difference existed between patients who dislocated and those who did not, we found no difference in the severity of damage on the surface of retrieved liners with the numbers available (Table 3). Patients who experienced a dislocation did not differ from those who had not experienced a dislocation in terms of key clinical factors (age, sex, and length of implantation), although the patients who did not experience a dislocation had a higher body mass index (29 versus 25 kg/m²). Three explanations exist for why we found no difference in the severity of damage. First, we may be underpowered to find a difference between 21 patients with dislocation and 21 without; however, we were limited by the number of implants in our retrieval system. Second, we could not account for whether patients experienced recurrent dislocation events or a single dislocation event. This influences whether a surface had multiple scratches or just a single scratch. Finally, we could not control the length of time after the original procedure. Some patients who experienced dislocation(s) were likely revised early, so the degree to which the polyethylene could be damaged was limited.

Table 3.

Polyethylene damage scores and the reason for revision surgery of retrieved OxZr heads

Our study has other limitations as well. First, the OxZr heads and XLPE implants were retrieved from revision THAs, so they may not represent the state of these implants in well-functioning and clinically asymptomatic patients. However, the goal of our study was to investigate the relevance of a particular failure mechanism with OxZr heads, namely effacement of the bearing surface caused by dislocation (and reduction) of the head from the liner. Thus, the focus on damaged heads at retrieval was necessary to test our hypothesis. The preponderance of implants revised for dislocation also explains the lack of correlation between XLPE damage scores and deviation measurements with length of implantation. Liners were damaged and deformed by the dislocation process itself, so even liners with short periods of implantation had severe damage and wear. Second, we did not use scanning electron microscopy coupled with energy dispersive x-ray analysis to determine whether the black debris seen on the surface of the retrieved liners was in fact small pieces of the black ceramicized surface from the damaged OxZr heads, although it is likely, because previous studies showed damage on the femoral heads consisted of surface effacement of the OxZr with exposure of the underlying zirconium alloy [9, 16].

Another limitation of the unmarked orientation of retrieved heads and liners was that we could not orient the heads in the 12 stations of the hip simulator to recreate their in vivo orientations relative to the cyclic loads applied by the simulator. The impact of this limitation can be seen by comparing the colorimetric deviation maps of the retrieved liners with the liners that had been worn against the same retrieved OxZr heads in the hip simulator. The deviation patterns on the retrievals differed from those on the liners tested in the hip simulator (Fig. 4). This was likely the result of the fact that the heads had dislocated in vivo, leading to considerably more deformation at the rim, which contributed to the measured maximum deviations. Nonetheless, we were able to demonstrate a range of wear behaviors that corresponded to the location of the surface damage on the heads relative to the locations of maximum applied load during the wear test.

Fig. 4.

Two examples of the Grade 3 OxZr heads are shown along with the laser scan colorimetric maps of the articulating surfaces of the XLPE liners that were retrieved with these two compared with the corresponding maps of the pristine XLPE liners that were tested against the same heads for 5 MCs in the hip simulator.

A final limitation to our study is the use of simulator testing itself; simulator wear tests are limited to small sample sizes, and small variations in test conditions can affect the results. To limit this possibility, all testing was done coincidentally. Nonetheless, the diameters and overall surface shape of the pristine liners were likely not identical to those of the original retrieved liners that articulated against the retrieved heads while in vivo the liners in this study. This could have affected comparison between the wear seen on the retrieved liners with that from the pristine liners used in the simulator. However, we believe that the match between wear locations and amounts as reflected in the scans of the retrieved liners with the wear measurements on the pristine liners showed that they were behaving similarly when worn against the same femoral head.

Our study demonstrates that Grade 3 damage to an OxZr femoral head is associated with damage to the corresponding acetabular liner. The assessment of surface damage through subjective damage scoring [4] and measurements of the deviations of the articular surfaces of polyethylene implants with laser scanning [20] or CT [1] have long been used to study how design and material factors affect polyethylene wear. Adding a wear simulator test provides additional objective measures of the direct effect that surface damage to an OxZr head has on polyethylene wear.

Our simulator data showed that Grade 3 damaged OxZr heads from retrievals associated with dislocation were associated with higher amounts of wear in pristine XLPE liners. These data can be compared with other wear simulator measurements on highly XLPE. For example, in the classic study by McKellop and colleagues [14] of crosslinking’s effect on polyethylene wear, acetabular liners were exposed to different levels of radiation, imparting different amounts of crosslinking to the polyethylene material. When tested in the same type of hip simulator and with the same test protocol as was used in our study, increased crosslinking imparted ever greater wear resistance (Fig. 5). The XLPE liners used in our simulator study had been crosslinked with 10 MRad. Superimposing the results for the Grade 3 OxZr that produced the greatest wear rate in our test (the top curve in Fig. 3) shows a wear rate nearly as great as that of a polyethylene liner that had received only 2.5 MRad, that is, a historically conventional liner that would not be considered highly crosslinked. Even when the mean wear rate for all six of the Grade 3 OxZr heads tested in our simulator study is considered, the wear rate remains elevated over the 10-MRad results from McKellop et al. (Fig. 5). Interestingly, even after correction of the weight changes by using soak controls, some liners showed weight progressive loss (ie, “negative” wear). This suggests that either little or no wear occurred, that these pristine liners were gaining more as a result of fluid absorption than the soak controls, or that perhaps particles of debris were embedding into the liners’ bearing surfaces during the testing, thus increasing their weights.

Fig. 5.

The wear rate of the worst case of the Grade 3 severely damaged OxZr heads tested in the hip simulator (the top curve in Fig. 3) and the mean wear rate of all six Grade 3 heads tested in the simulator are plotted onto the results from a wear test by McKellop et al. [15] in which different amounts of radiation were imparted to polyethylene liners to affect different levels of crosslinking.

We are aware of one other study similar to ours. In that report, Jaffe et al. [5] obtained two severely damaged and two minimally damaged OxZr heads retrieved after failed, attempted closed reductions. They tested the heads in a hip wear simulator and included BIOLOX delta heads that had been artificially damaged in vitro before the test. The mean volumetric wear rate of the XLPE liners that articulated against the two severely damaged OxZr heads in the simulator was 28 mm³/MC, whereas that of the two minimally damaged OxZr heads was 1.8 mm³/MC. The three BIOLOX delta heads generated an XLPE wear rate of < 1 mm³/MC. These results corroborate our findings that severe damage to OxZr heads can lead to considerable polyethylene wear; however, the number of specimens used by Jaffe and colleagues was smaller, and the ceramic heads had been artificially damaged by creating only a single wear scar. More importantly, their simulator test was only conducted to 1 MC instead of 5 MC as in our study, so their results were likely dominated by the considerable deformation that occurs over the first million cycles of a simulator test as the head beds into the polyethylene liner.

Clinical wear rates reported for OxZr bearings against XLPE are quite low, in keeping with our simulator results that showed no appreciable wear of XLPE liners when articulating against moderately damaged OxZr heads. For example, Jassim et al. [6] determined linear wear rates radiographically over 5 years in a randomized clinical trial that included OxZr heads. Their linear wear rate for OxZr heads articulating against XLPE liners was 0.023 mm/year (SD = 0.010 mm/year). Similarly, Karidakis and Karachalios [8] showed a low linear wear rate of 0.07 mm/year (SD = 0.02 mm/year) at 10 years. Conversely, Jonsson et al. [7], using radiostereometric analysis, reported a higher linear wear rate of 0.10 mm/year (SD = 0.10 mm/year) at 5 years. However, patients in the Jassim et al. study who were revised for recurrent dislocations were removed from the study. The other two reports are unclear as to the inclusion of patients who had dislocated, so no clear clinical data exist on the wear rates of OxZr-XLPE couples in the presence of dislocation.

In conclusion, our results demonstrate that scratching of the bearing surfaces of OxZr heads such as can occur with recurrent dislocation and subsequent attempts at reduction may affect the wear performance of the XLPE liners used here. We have not followed patients after dislocation to demonstrate a direct link between increased damage to the head and increased wear (and subsequent osteolysis) as a cause of failure. Nonetheless, the wear rates measured in the simulator for our most damaged heads are of a magnitude commensurate with liners with no crosslinking tested earlier in the same wear simulator [14]. This leads us to conclude that increased wear (and subsequent osteolysis) might indeed be possible. We believe that patients with a history of instability and an OxZr-XLPE bearing couple should continue to be followed for the possibility of accelerated wear.