Polyethylene wear debris has been reported as the major cause of tibial osteolysis associated with total knee arthroplasty. 8,15 Its presence is associated with a reactive synovitis and formation of granulomatous tissue, which can be responsible for osteolysis and implant failure. 20,23 The articulating surface of the polyethylene insert is thought to be the main source of debris generated in total knee arthroplasty. Various damage modes of the articulating surface have been described. 11,17 Furthermore, according to some authors, wear of polyethylene inserts and osteolysis can be related to the insert design and the method of fixation of the metal tray. 5,10
The interface between the nonarticulating surface (backsurface) of the polyethylene insert and the metal tray has been considered a possible contributing source of polyethylene wear debris. 8,23 Because of its more direct access to the tibial metaphyseal bone-implant interface in the presence of screw holes in the tray, the polyethylene debris derived from backsurface wear is thought to play a potential role in the formation of tibial metaphyseal osteolytic lesions. 15,22,23
The purpose of the current study was to test the hypothesis that backsurface damage of polyethylene inserts may compromise longevity of total knee arthroplasty. To address the issue, backsurface wear and deformation on polyethylene tibial components retrieved postmortem from well-functioning total knee arthroplasty inserts were examined. The relationship between the degree of damage, and the histologic findings of particle-induced granulomas at the bone-implant interface were evaluated.
MATERIALS AND METHODS
Twenty-five total knee arthroplasty inserts were retrieved postmortem from 20 subjects (Table 1). The specimens consisted of 11 Miller-Galante and 14 Miller-Galante II (both made by Zimmer, Warsaw, IN) total knee arthroplasty polyethylene inserts. Both implant systems were posterior-cruciate retaining total knee prostheses with nonconforming articular surfaces in the sagittal plane and conforming surfaces in the coronal plane. The polyethylene insert was held in place by a central locking mechanism and was constrained around the periphery. The Miller-Galante inserts were net-shape molded from Himont 1900 resin conforming to the polyethylene standard American Society for Testing and Materials F 648 Type 3. 1 This material was consolidated by uniaxial compression molding without the addition of calcium stearate. The final implant geometry was determined by the molding process. The Miller-Galante II inserts were machined from ram extruded bar stock which was produced from GUR 4150 resin conforming to the polyethylene standard American Society for Testing and Materials F 648 Type 2. 1 This material was consolidated by ram extrusion with the addition of approximately 0.05% calcium stearate. The nominal thickness of the inserts ranged from 8.5 to 21 mm, the anteroposterior (AP) dimension ranged from 40 to 51 mm, and the mediolateral dimension ranged from 56 to 83 mm. Six new tibial inserts obtained from the manufacturer were measured to define the baseline geometry.
In comparison with the Miller-Galante design, the Miller-Galante II prosthesis was designed with a modified sagittal contour, placing the trochlea in a more anatomic position. In addition, the Miller-Galante II design increased patellofemoral congruence and replaced the metal-backed patellar component with an all-polyethylene patellar component. With the well-functioning postmortem retrievals reported here (no patellar component failures), it is unlikely that these geometric differences would substantially influence articular surface or backsurface tibial component wear and deformation.
The tibial trays were made of Ti-6A1–4V alloy and contained a peripheral rim and a central dovetail. Fifteen metal trays did not have screw holes and the bone opposing surface was precoated with polymethylmethacrylate for cement fixation. The remaining 10 trays had commercially pure, Ti fiber metal porous coating sintered to the bone-opposing surface for cementless implantation and contained four screw holes in eight cases and two posterior screw holes with a central post in two cases. All metal trays had a satin finish (maximum Ra, 30 microinches) on the polyethylene opposing surfaces.
The mean time of implantation was 64.1 months, ranging from 4 to 156 months. There were seven males and 13 females. The inserts were retrieved from the right knee in 11 of the cadavers and from the left knee in the remaining 14 cadavers. The average age of the patients at the time of surgery was 72.9 years (range, 60–90 years) with an average Hospital for Special Surgery knee score at the last followup of 93.3 points (range, 79–100 points). 12 The last clinical followup was, on average, 13.2 months (range, 1–29 months) before the postmortem retrieval.
The backsurface of the inserts was inspected using a stereomicroscope at 18X magnification for abrasive wear, polishing, pitting, scratching, embedded particles of bone cement or metal, cracks, and delamination. 11,17,24 A wear score based on the ranking system used by Hood et al 11 and Wright et al 24 was given to the backsurface. Each inspected backsurface was divided into four quadrants (Fig 1): each quadrant was rated from zero to three points for the extent of the considered wear mode. 22 Zero points were assigned when no damage was observed, one point was given to those quadrants which had less than 10% surface involvement, two points were assigned to the quadrants whose surface was affected between 10% and 50%, and three points were assigned when more than 50% of the quadrant was damaged. In addition, to test for within-component variability of backsurface damage, the anteromedial and posteromedial quadrants were combined to calculate a medial wear score; the anterolateral and posterolateral quadrants were combined to calculate a lateral wear score; the anterolateral and anteromedial quadrants were combined to calculate an anterior wear score; and the posteromedial and posterolateral quadrants were combined to calculate a posterior wear score.
Extrusions of polyethylene into the screw holes of the tibial metal trays, occurring as a result of cold flow, were recorded according to their anatomic position: anteromedial, posteromedial, anterolateral, and posterolateral. 8,22,23 The height of these extrusions was measured using the Z-axis evaluation of a high resolution, digital, motorized zoom, optical system (SmartScope, Optical Gaging Products, Rochester, NY). The specified accuracy of the system was 4 μm at 350× magnification. The measurements were taken at the anterior, posterior, medial, and lateral edges of each extrusion and reported as average height.
The SmartScope Z-axis evaluation also was used to record profiles of the backsurface at a 20° angle from the AP axis (Fig 2) for each of the components. The points for the profile were recorded every 1 mm, starting at the posterior edge of the component to the anterior edge and the system was zeroed for each profile at the first point recorded. In addition to the backsurface of new and retrieved tibial inserts, the opposing surface of six new metal trays and three retrieved metal trays were measured for their profiles.
To test the backsurface profiles in situ, five (Cases 21–25) freshly retrieved total knee prostheses were investigated. After postmortem retrieval the tibial inserts were exposed, and the specimens were immersed in a 10% buffered formalin solution. An ultrasonic thickness gauge (25 DL, Panametrics, Waltham, MA) was used to detect the thickness of the layer of fluid between the polyethylene insert and the metal tray at six sites along the AP dimension of the insert. This method was validated by measuring fluid-filled void spaces between polyethylene and metal with known thickness. The technique was accurate in detecting layers of fluid of 150 μm or more. The same five implants then were disassembled and the backsurface profiles of the polyethylene inserts were immediately recorded with the SmartScope. The measurements were repeated after 2 weeks to allow the specimen to dry.
To assess the relation between the type of wear damage and histologic changes, 13 proximal tibias were processed and examined under the light microscope for the presence of granulomas. 16,21 Two coronal histologic sections through pegs of the implant, bone, soft tissues, and cement (when present) were obtained for seven cementless tibial components (Cases 1–7) and six cemented (Cases 11–14, 19, and 20) tibial components and were studied by light microscopy. The extent of penetration in millimeters of particle-induced granuloma was measured on the mediolateral axis from medial and lateral edges of the bone-implant interface for each of the sections. In addition, granuloma penetration in the proximodistal direction around the fixation screws of the cementless implants was recorded.
The data were analyzed for the effects of time of implantation, size and thickness of the component, gender and age of the patient, knee score, and granuloma penetration. A multivariate general linear model analysis of variance was used to test abrasive wear score, polishing wear score, and pitting wear score for the effect of manufacturing process (machined versus molded), design of tibial tray (for cemented versus cementless fixation), time of implantation, age of the patient at surgery, and knee score. The emerging effects then were investigated by means of the Mann-Whitney U tests and Spearman’s correlations. Spearman’s correlations also were used to test the relationship between the wear scores and the extent of granuloma penetration. To test for the effect of location of damage, anterior, posterior, medial, and lateral scores were compared by means of Sign tests. Linear regression analysis was used to investigate correlations between continuous parametric variables.
Examination of the backsurface for damage revealed that polishing and abrasive wear were the predominant wear modes of the 25 retrieved components. The damage was observed mainly at the periphery of the component and around the area corresponding to the site of the screw holes in the tibial plate. Polishing was recorded in four (36%) of the molded components and in each (100%) of the machined inserts. Abrasive wear as detected at the backsurface on five (45%) of the molded components and on none of the machined inserts. Pitting was present in 11 (100%) molded components and in 10 (71%) machined inserts. Pitting involved less than 1% of the area in 20 inserts (80%), between 1% and 10% in one component (4%), and was absent in four components (16%). Scratching, possibly attributable to disassembly procedures, was found around the front edge in five components (20%). None of the inserts showed bone cement or metal particles embedded in the backsurface using the light microscopic technique described above. Neither delamination nor cracks was present on the backsurface of any component. In general, the magnitude of wear on the backsurface of these components was limited and far less than the magnitude of damage that has been reported on the articular surface. 11,22,24,25
The manufacturing process had a significant effect on the pitting score (p = 0.018). Pitting had a higher incidence among the molded components than among the machined components (p = 0.015). The pitting score was inversely correlated with granuloma penetration along the posteromedial screws (r = 0.913; p = 0.004) in the cementless components. Significant interaction between the type of manufacturing process and the type of fixation of the tibial tray were found for abrasive wear (p = 0.004) and for polishing wear (p = 0.004). Therefore, abrasive wear was higher in molded components (p = 0.0002) and polishing wear was higher in machined components (p = 0.0043), but only in the trays designed for cemented fixation. There was no correlation between the wear scores as the dependent variable and the time of implantation, the age and gender of patients, or their knee scores as the independent variables. Neither were there any associations with the size or the thickness of components.
The posterior polishing score was significantly higher than the anterior score (p < 0.0005), but no other effects of damage location were found for any of the wear modes.
Extrusions were present on all 10 of the components that had screw holes in the tibial tray. Extrusions were not observed in the other 15 components.
The height of the extrusions of polyethylene on the backsurface of the inserts at the site of the screw holes ranged from an impression on the surface to as much as 317 μm. The biggest extrusions were observed in the posteromedial quadrant (mean height, 122 μm) being significantly larger than those on the anteromedial quadrant (mean height, 11 μm) (p = 0.020) and those on the posterolateral quadrant (mean height, 67 μm; p = 0.038). No measurable extrusions were found in the anterolateral quadrant.
Penetration of granulomatous membrane into the bone-implant interface at the medial and lateral edges of the tibial trays occurred in 10 of the 13 specimens evaluated histologically. It was not present in the two shortest-term specimens (Cases 5 and 6) and in a longer-term specimen (Case 1). The granulomatous tissue contained numerous histiocytes and multinucleated giant cells containing primarily polyethylene particles. The mean extent of penetration at the edges of the trays was 4.2 mm (range, 1–19 mm). Granuloma penetration depth into the interface was correlated to the type of the interface. According to the results of multiple regression analysis with granuloma penetration depth as the dependant variable and the type of the interface (cementless or cemented), thickness of the component, type of manufacturing (machined versus molded), time in situ, age, and knee score as the independent variables, the type of interface was the only variable which had a significant correlation with granuloma penetration depth. There was more penetration associated with the cemented interface in the anteromedial quadrant (cemented 7.6 ± 7.0 mm, cementless 0.71 ± 1.1 mm; p = 0.037, r2 = 0.44) and in the posteromedial quadrant (cemented 3.8 ± 0.9 mm, cementless 0.6 ± 1.1 mm; p = 0.004, r2 = 0.66).
Similar granulomatous membranes were present along two or more of the fixation screws of the five longer-term specimens (Cases 1–4, and 7) and absent from the two specimens retrieved at 4 months (Cases 5 and 6). The mean penetration of granuloma along the screws was 18.4 mm (range, 1–31 mm). A significant correlation was found between the distal penetration of the granuloma along the posteromedial screw into the tibial bone and the height of its corresponding extrusion (r2 = 0.91; p = 0.001;Fig 3). In addition, the height of the posterior extrusions was correlated to the penetration of granulomatous membrane from medial and lateral edges of the bone-implant interface (r2 = 0.623; p = 0.035 medially; and r2 = 0.72; p = 0.016 laterally).
The profiles recorded from the backsurface of the new polyethylene inserts resembled a flat surface with deviations less than 20 μm. The opposing surface of the new metal trays had the same variability, but in one of the six cases, a 50-μm step was present along the perimeter. The backsurface of the retrieved inserts showed a concave deformation in the AP direction in 48 of the 50 recorded profiles (Fig 4). The maximum deformation ranged from 50 to 380 μm on the medial side and from 40 to 276 μm on the lateral side. The components which had ultrasonic in situ measurement showed a detectable fluid layer between the polyethylene insert and the metal tray in three of the five cases. For these three cases, the profiles plotted using the in situ void space data matched the profiles recorded after disassembling the implants. In the other two cases, the maximum deformation was below the detection limits by ultrasound.
There were no statistically significant differences between the magnitude of the backsurface deformation measured immediately after disassembly and after 15 days. None of the variables related to the subject or the implant showed a significant correlation with this deformation.
Backsurface damage in the type of polyethylene components examined in this study was limited. The predominant wear modes affecting the backsurface of these postmortem retrieved components were polishing and abrasive wear. Pitting also was present, although to a lesser extent.
Polishing may result from cyclic loading and micromotion which leads to progressive obliteration of the original factory surface finish. This changed the original matte surface into a shiny appearance. Therefore, polishing was easier to detect than abrasive wear on the original matte surface of machined components. However, abrasive wear leads to a shredded appearance of an originally smooth surface. 11,17 This could have been why abrasive wear was more evident on the originally shiny surface of the molded components. Because of the small numbers of specimens and the limited damage on the backsurface of these components, the current study was not able to directly address the issue of whether molded or machined polyethylene was superior.
Limited pitting was observed on the backsurface of the inserts. Because contactinduced stresses near the backsurface of the polyethylene component are much smaller than those observed on the articulating surface, pitting on the backsurface may occur by a different mechanism than that on the articulating surface. One possible explanation for the formation of pits is the presence of third body particles and micromotion between the backsurface of the polyethylene insert and the metal backing. Although embedded third bodies were not seen on the backsurface of these devices using the light microscopic technique described above, this does not preclude the presence of finer embedded particles requiring specialized electron microscopic imaging for detection or the presence of particles within a reservoir of synovial fluid between the tray and the backsurface (see below) which abrade the polyethylene without becoming embedded in the surface.
Neither delamination nor cracks, which often are associated with production of large amounts of debris and gross component failure, were present on the backsurface of the polyethylene inserts. 2–4,25 Unlike the results reported from the studies based on retrievals for cause (obtained at revision surgery), the total wear score from the current backsurface analysis did not correlate with the clinical variables analyzed. 22,23 These postmortem retrievals were collected from well-functioning implants and were less influenced by the failure mechanisms which might have led other implants to need surgical revision. Considering the relatively low wear scores assigned to the examined components and the absence of wear modes producing large quantities of polyethylene debris, it seems that the backsurface of this particular tibial component design was not a major source of wear debris. This may be related to the stable central locking mechanism with peripheral circumferential constraint which may have reduced micromotion between the backsurface of the insert and tibial tray.
All the polyethylene tibial components implanted with a cementless metal tray showed extrusions of polyethylene into the screw holes. Cold flow of ultrahigh molecular weight polyethylene under cyclic load conditions from the tibial insert into cavities, such as those represented by the screw holes of cementless metal-backed implants, has been reproduced in an experimental computer-aided analysis showing that its amount increases with a decreasing thickness of the sample. 6 This correlation was not apparent in the current study, most likely because of the relatively small number of specimens and the multivariate nature of this process. The presence of extrusions of polyethylene has been reported previously and was a consistent finding in the current specimens when the polyethylene insert had been implanted with tibial trays that contained holes designed for cementless implantation. 8,22,23 The fact that extrusions were found on two inserts after only 4 months of implantation is consistent with one study, in which the amount of cold flow strain is high initially, and is followed by a reduced rate of cold flow that soon reaches a steady state. 14 The degree of progression of granuloma along fixation screws and the periphery of the tray-bone interface was correlated with the height of these extrusions. Because cold flow is caused by compressive forces across the knee, the height and location of extrusions may reflect the magnitude and duration of such forces at a particular location in the joint. Therefore, the presence of more extensive extrusions and granuloma in the same location may have resulted from a more intensive level of localized loading by virtue of greater activity levels or surgical technical factors (alignment, ligament balancing). This may result in higher wear rates and the generation of more particulate debris at the articulating and nonarticulating surfaces.
Polyethylene quality also may have an impact on the nature of the polyethylene damage observed. The broad term polyethylene quality encompasses numerous characteristics including the presence of inclusions, fusion defects, inhomogeneous molecular weight distributions, and the degree of oxidation. In this study, the destructive tests necessary for a full characterization of polyethylene quality were not done. Although it is plausible that variations in polyethylene quality could account for some of the observed variability in the degree of damage, the relatively small number of retrieved postmortem devices available for analysis makes it difficult to show this effect.
An interesting finding in this study was the presence of a concave deformation of the backsurface in all retrieved inserts and in none of the unimplanted inserts. A simulation of stresses with nonlinear finite element analysis in polyethylene components for total knee arthroplasty subjected to cyclic moving loads showed that stress and deformation reached a steady state after the first five cycles, and that plastic deformation was related to the presence of horizontal tensile residual stresses at the articulating surface and to horizontal compressive residual stresses at the backsurface. 9 The current study detected the presence of deformation on the two components retrieved after only 4 months in situ. However, the direction of this deformation, contrary to the finite element study, suggests the presence of tensile rather than compressive residual stresses at the backsurface. This supports the theory that this plastic deformation results from stresses imparted at the time of the implantation by virtue of the geometry of the tibial component locking mechanism and peripheral constraining feature. This deformation is likely a function of the dimensional mismatch between the tibial insert and the peripheral rim of the tibial tray. The possibility that deformation occurs after disassembly of the component was excluded by the in situ ultrasonic evaluation of the freshly retrieved knee components which clearly showed the presence of a void space between the polyethylene insert and the metal tray. Similarly, dehydration did not influence the status of the deformation: repeated measurements taken immediately after disassembly of the components and 2 weeks later were comparable.
The presence of a concave deformation at the backsurface of the tibial component can create a reservoir of synovial fluid and particulate debris, generated at the backsurface or at the articulating surface. Assuming that joint loads during alternate gait cycles could cyclically compress the potential space, the contents of this reservoir could be forced into the metaphyseal bone through the screw holes in the metal tray, in a manner similar to that described in a study on hip simulation. 13 This then represents a possible preferential pathway for particulate migration to the underlying bone which may promote the development and progression of granuloma at the screw-bone interface. The preferential nature of this pathway was supported by two findings. First, there was more extensive penetration of granuloma along the screw-bone interface than along the tray-bone interface in the cementless components. Second, the degree of granuloma penetration at the tray-bone interface was significantly greater in the cemented devices in comparison with the cementless devices, suggesting that particulate migration and granuloma penetration was diverted away from the tray-bone interface by the presence of screw holes in cementless devices. When considering the potential biologic long-term effects, metal trays without screw holes would be preferable to limit the distal extent of access to the tibial metaphyseal bone. These considerations must be balanced, however, with the enhancement in short-term stability offered by screw fixation. 19
The authors thank Aaron Rosenberg, MD; Steven Gitelis, MD; Mitchell B. Sheinkop, MD; Wayne Paprosky, MD; and Victor Goldberg, MD, for providing patient information for this study.
1. American Society for Testing and Materials: Standard Specification for Ultra-High-Molecular Weight Polyethylene Powder and Fabricated Form for Surgical Implants. Designation: F 648–96. 1997 Annual Book of ASTM Standards. West Conshohocken, PA, American Society for Testing and Materials 139–144, 1997.
2. Bloebaum RD, Nelson K, Dorr LD, Hofmann AA, Lyman DJ: Investigation of early surface delamination observed in heat-pressed tibial inserts. Clin Orthop 269:120–127, 1991.
3. Blunn GW, Joshi AB, Lilley PA, et al: Polyethylene wear in unicondylar knee prostheses. Acta Orthop Scand 63: 247–255, 1992.
4. Blunn GW, Joshi AB, Minns RJ, et al: Wear in retrieved condylar knee prostheses. J Arthroplasty 12: 281–290, 1997.
5. Cameron HU: Tibial component wear in total knee replacement. Clin Orthop 309:29–32, 1994.
6. Chang DC, Goh JC, Teoh SH, Bose K: Cold extrusion deformation of UHMWPE in total knee replacement prostheses. Biomaterials 16: 219–223, 1995.
Collier JP, Mayor MB, McNamara JL, Surprenant VA, Jensen RE: Analysis of the failure of 122 polyethylene inserts from uncemented tibial knee components. Clin Orthop 273: 232–242, 1991.
8. Engh GA, Dwyer KA, Hanes CK: Polyethylene wear of metal-backed tibial components in total and unicompartmental knee prostheses. J Bone Joint Surg 74B: 9–17, 1988.
9. Estupinan JA, Bartel DL, Wright TM: Residual stresses in ultra-high molecular weight polyethylene loaded cyclically by a rigid moving indenter in nonconforming geometries. J Orthop Res 16: 80–88, 1998.
10. Ezzet KA, Garcia R, Barrack RL: Effect of component fixation method on osteolysis in total knee arthroplasty. Clin Orthop 321: 86–91, 1995.
11. Hood RW, Wright TM, Burstein AH: Retrieval analysis of total knee prostheses: A method and its application to 48 condylar prostheses. J Biomed Mater Res 17: 829–842, 1983.
12. Insall JN, Ranawat CS, Aglietti P, Shine J: A comparison of four models of total knee-replacement prostheses. J Bone Joint Surg 58A: 754–765, 1976.
13. Khalily C, Tanner MG, Williams VG, Whiteside LA: Effect of locking mechanism on fluid and particle flow through modular acetabular components. J Arthroplasty 13: 254–258, 1998.
14. Lee KY, Pienkowsky D: Compressive creep characteristics of extruded ultrahigh molecular-weight polyethylene. J Biomed Mater Res 39:261–265, 1998.
15. Lewis PL, Rorabeck CH, Bourne RB: Screw osteolysis after cementless total knee replacement. Clin Orthop 321: 173–177, 1995.
16. Pidhorz LE, Urban RM, Jacobs JJ, Sumner DR, Galante JO: A quantitative study of bone and soft tissues in cementless porous-coated acetabular components retrieved at autopsy. J Arthroplasty 8: 213–225, 1993.
17. Rostoker W, Chao EYS, Galante JO: The appearances of wear on polyethylene: A comparison of in vivo and in vitro wear surfaces. J Biomed Mater Res 12: 317–335, 1978.
Schipplein OD, Andriacchi TP: Interaction between active and passive knee stabilizers during level walking. J Orthop Res 9: 113–119, 1991.
19. Sumner DR, Berzins A, Turner TM, Igloria R, Natarajan R: Initial in vitro stability of the tibial component in a canine model of cementless total knee replacement. J Biomech 27: 929–939, 1994.
20. Urban RM, Hall DJ, Sapienza CI, et al: Interface tissues in cemented and cementless total knee replacement tibial components retrieved at autopsy. Trans Orthop Res Soc 23: 246, 1998.
21. Urban RM, Jacobs JJ, Sumner DR, et al: The bone-implant interface in femoral stems with non-circumferential porous coating in specimens retrieved at autopsy. J Bone Joint Surg 78A: 1068–1081, 1996.
22. Wasielewski RC, Galante JO, Leighty RM, Natarajan RG, Rosenberg AG: Wear patterns on retrieved polyethylene tibial inserts and their relationship to technical considerations during total knee arthroplasty. Clin Orthop 299:31–43, 1994.
23. Wasielewski RC, Parks N, Williams I, et al: Tibial insert undersurface as a contributing source of polyethylene wear debris. Clin Orthop 345: 53–59, 1997.
24. Wright TM, Rimnac CM, Faris PM, Bansal M: Analysis of surface damage in retrieved carbon fiber-reinforced an plain polyethylene tibial components from posterior stabilized total knee replacements. J Bone Joint Surg 70A: 1312–1319, 1988.
25. Wright TM, Rimnac CM, Stulberg SD, et al: Wear of polyethylene in total joint replacements. Clin Orthop 276: 126–134, 1992.
Richard S. Laskin, MD—Guest Editor