Wear, delamination, and fatigue resistance of melt-annealed highly crosslinked UHMWPE cruciate-retaining knee inserts under activities of daily living



The wear, delamination, and fatigue resistance of artificially aged gamma irradiation-sterilized conventional polyethylene (CPE) and gas-plasma-sterilized melt-annealed highly crosslinked polyethylene tibial inserts (HXPE) were compared. Six CPE and 12 HXPE (six irradiated at 58 kGy and six at 72 kGy) left knee inserts were wear tested for 5.5 million cycles (Mc) under loads and motions that mimic activities of daily living, such as walking, chair rise, stair ascent, and deep squatting. Another six HXPE (72 kGy) and six CPE inserts were also tested under conditions that could produce severe delamination for 8 Mc. Ten other knees (five 72 kGy HXPE and five CPE) were subjected to posterior edge loading fatigue testing for 5 Mc. The HXPE inserts had an average wear rate reduction of about 80% relative to their CPE counterparts during all activities. All of the CPE inserts delaminated and fractured during high cycle deep squat (152° flexion) motions, while all the HXPE remained intact. None of the HXPE inserts delaminated after 8 Mc, while all of the CPE inserts developed delamination damage within 1.5–5.8 Mc of delamination testing. All CPE inserts developed subsurface cracks and delamination within 2.8 Mc during posterior edge loading fatigue studies, while none of the HXPE inserts showed cracking or delamination after 5 Mc. These results show that aged HXPE has higher wear and fatigue resistance than aged CPE, and offers potential long-term advantages for young active patients with sustained activities of daily living. © 2010 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 28:1120–1126, 2010

Total knee arthroplasty (TKA) has been successful in restoring function and relieving pain that results from osteoarthritis. A consensus statement developed by the National Institutes of Health reports that success of TKA is supported by >20 years of follow-up data with about 90% of patients experiencing an improvement in pain and functional status.1 Many authors report 10–15-year implant survival rates of >90%.2 However, wear, delamination, material degradation, and the potential for subsurface cracking of ultra high molecular weight polyethylene (UHMWPE) in combination with patient weight, activity level, and surgical technique issues are factors leading to surgical revision.3, 4 Major improvements have included sterilization of polyethylene components in oxygen-free environments, net-shape compression molding, and, most recently, the introduction of highly crosslinked materials to improve wear and delamination resistance. Despite these improvements, concerns still persist because radiation sterilization generates macroradicals that are precursors of oxidation.5 Polyethylene acetabular liners and tibial inserts that had been radiation-sterilized in a low-oxygen environment underwent similar in vivo oxidative mechanisms as polyethylene components that had been radiation-sterilized in air.5, 6 Wear accounted for several surgical revisions of oxygen-free sterilized tibial bearings.7 Electron beam irradiation crosslinking and melt annealing of UHMWPE has been shown to reduce the wear of total hip arthroplasty (THA) in both in vitro simulator tests8–10 and in vivo clinical studies.11, 12 This reduction was attributed to the fact that crosslinking reduces the strain softening phenomenon that arises from multidirectional shear stresses acting on UHMWPE during articulation.13 UHMWPE tibial knee inserts are also subjected to multidirectional shear stresses and exhibit strain softening-dependent wear behavior.14, 15 In vitro studies showed that crosslinking UHMWPE significantly decreases wear of tibial inserts.16

The knee joint experiences significantly higher contact stresses than the hip, and fatigue of tibial inserts has been reported as a prevalent damage mechanism17, 18 that may be exacerbated by the oxidation of UHMWPE. While crosslinking increases wear resistance, postirradiation annealing significantly reduces the amount of residual free-radicals and subsequent long-term oxidation, which assure long-term stability of the mechanical properties, including improved delamination resistance.19 However, the crosslinking and annealing process results in a reduction of some mechanical properties20 as measured in laboratory testing. With no ability to obtain a direct clinical measure of how those property reductions will impact the clinical performance, the clinical use of highly crosslinked polyethylene (HXPE) in the knee has met with resistance, particularly for young, active patients. Therefore, a study that investigates the response of HXPE tibial inserts to frequent high load and motion excursions, as can be expected in young active patients, is necessary. The potential for additional improvements in osteolysis prevention and subsequent enhanced longevity of these devices exists if the decreased wear and delamination rates of HXPE can be realized while maintaining other adequate mechanical performance. Cyclic fatigue induced delamination resistance, and crack initiation and propagation resistance are also important performance properties that need to be further evaluated.

Our objectives were to determine the long-term wear, delamination, and fatigue resistance of aged HXPE cruciate-retaining tibial inserts under various activities of daily living (ADL) and compare them to those of aged conventional γ-irradiation-sterilized GUR 1050 UHMWPE (CPE). Wear debris generated from CPE and HXPE tibial inserts during various ADL were also characterized. We hypothesized that aged HXPE tibial inserts will have a superior long-term wear performance, and equivalent or better delamination and fatigue resistance, when subjected to strenuous ADL conditions than that of aged CPE tibial inserts.


All CPE knee inserts were machined from compression molded slabs of GUR 1050, packaged in nitrogen, and γ-irradiated to a maximum dose of 37 kGy. The HXPE inserts were machined from compression molded GUR 1050 slabs that had been e-beam irradiated at doses of 58 and 72 kGy, melt-annealed at 150° C in air, and gas plasma sterilized. All inserts (left knees) were aged using a standard protocol21 for 2 weeks following ASTM Standard F2003 Method B 22 (i.e., 70°C in pure oxygen at 0.55 MPa). All tests were performed in undiluted bovine calf serum lubricant that was maintained at 37 ± 3°C and changed every 0.5 million cycles (Mc).

Wear Testing

Wear tests were performed using 6-station displacement control knee simulators (AMTI, Watertown, MA). Gravimetric wear measurements were made every 0.5 Mc. Load-soak controls were run during each test to correct for fluid absorption by the polymer inserts. Wear differences were analyzed using the Student's t-test with a confidence interval set at 95%.

Walking Gait

Eighteen cruciate-retaining NexGen® CR (Zimmer Inc., Warsaw, IN) tibial inserts (six e-beam irradiated at 58 kGy, six at 72 kGy, six CPE inserts) were wear tested to 5 Mc walking gait cycles, after which eight of the HXPE and four of the CPE inserts were further tested to 20 Mc under a modified ISO 1424323 waveform24 with femoral extension of 0 to −58°, external tibial rotation from −1.9 to 5.7°, and an anterior-posterior (AP) femoral translation range of 5.2 mm. The peak load was 3,188 N. The ISO 14243-3 waveform calls for a constant 168 N during the swing phase while the load waveform used in our study had two load peaks during swing at 378 N at 74% and 1,195 N at 95% (see Fig. 1a)

Figure 1.

Load and motion waveforms used to simulate various ADLs: (a) walk, (b) stair ascent, c) chair rise, and (d) deep squat.

Activities of Daily Living

The ADL frequencies and proportions were determined from work of Morlock et al.25 The load waveforms for chair rise, stair ascent, and deep squat activities were adopted from Ellis et al.,26 Morrison,27 and Dahlkvist et al.,28 respectively. The kinematics data were from the recent work of Dyrby.29 Twelve NexGen® CR-Flex (six CPE and six HXPE) high flexion knee implants were tested for 5.5 Mc in three phases. Phase I consisted of 3 Mc of walking gait. In Phase II, implants were subjected to a multigait sequence of walking, chair rise, stair climb, and 125° flexion for 2 Mc. Phase III consisted of 2,500 loops of 150 cycles of walking, followed by 50 cycles of 152° flexion, and was intended to test the implant's ability to withstand severe conditions with a high frequency of occurrence. The femoral component fixture was cut back short of the posterior edge of the femoral condyle to avoid fixture/insert impingement during high flexion activities. At 152° flexion (24.5 mm AP displacement and −28° external rotation under 2,200 N), the edge of the lateral condyle of the femoral component was located within 2 mm of the posterior edge of the insert. No rollover of the femoral component occurred during the entire high flexion activity simulation. The ADL distributions, loads, and motions waveforms and peak ranges are shown in Figure 1, and Tables 1 and 2.

Table 1. Distribution of ADL motions (W = walk)
 Phase I 1.0–3.0 McPhase II 3.0–5.0 McPhase III 5.0–5.5 Mc
  1. ADL, activities of daily living; W, walk.

ActivityWWChairWStairW125° flexionW152° flexion
ADL (%)10043.74.543.
Table 2. Simulator load and motion inputs for various ADL
ActivityPeak Load (N)F/E Range (°)I/E Range (°)A/P Range (mm)Frequency (Hz)
  1. ADL, activities of daily living; A/P, anterior-posterior, F/E; flexion-extension, I/E, internal–external rotation.


Wear Debris Analysis

Wear debris particles from spent lubricant were isolated by the method of Scott et al.30 using 0.05 µm pore size polycarbonate filters. Photomicrographs of representative areas of the debris-laden filters were obtained using a Cambridge S360 scanning electron microscope operated at 15 kV and magnification of 10,000 (LEO Electron Microscopy, Thornwood, NY). One thousand particles from each material and activity type were characterized using Image Pro™ image analysis code (Media Cybernetics Inc., Silver Spring, MD). Each debris particle was defined using four parameters: equivalent circle diameter (ECD), roundness (R), aspect ratio (AR), and elongation (E) as outlined in ASTM F1877-989.31

Oxidation and Delamination

The extent of oxidation of the tibial inserts was estimated by calculating the surface oxidation index (SOI) from Fourier Transform Infrared Spectroscopy (FTIR) absorption spectra according to ASTM Standard F2102-01.32 To evaluate and compare their delamination resistances, six HXPE (72 kGy) and six CPE inserts were articulated in a reciprocating pattern against 16-mm radius spherical CoCrMo balls under 1,779 N load (contact stress of about 28 MPa) in a 12-station custom-built delamination tester for 8 Mc. The stroke length was 20 mm, and the sliding waveform was sinusoidal with a frequency of 1 Hz (maximum sliding speed of 63 mm/s), which was within the range of the physiological sliding speed in the tibio-femoral joint during the stance phase of gait.33 The inserts were frequently visually inspected to detect the onset of delamination, at which point the test was stopped and the cycle count was recorded.

Posterior Edge Fatigue

This severe test replicated posterior femoral sliding and positioning that occurs subsequent to heel strike in an ACL-deficient TKA. A 2,224 N constant load was applied vertically on the medial plateau of each insert through the medial femoral condyle (Fig. 2). The lateral condyle was machined away to avoid loading of the lateral tibio-femoral compartment. The measured contact area of the medial condyle was 139.5 mm2, giving an average contact stress of 16 MPa and maximum contact pressure of 24 MPa. This is at the high end of contact stresses expected for the tibio-femoral contact at full extension.34

Figure 2.

Schematic of the posterior edge fatigue test set up.

Five 72 kGy HXPE and five CPE inserts were articulated against CoCrMo femoral condyles (Fig. 2) at a frequency of 2 Hz. The sinusoidal sliding motion was controlled such that the apex (the most distal point at full extension) of each femoral condyle was 2 mm from the posterior edge of the opposing tibial insert when the femur was at its most posterior position. All implants were set at full extension with the tibial component at a posterior slope of 7°. The stroke of the reciprocating motion was 10 mm, resulting in a peak sliding speed of 63 mm/s. Tibial inserts were visually examined periodically for signs of fatigue cracking.


Walking Gait

The wear rates (mean ± standard deviation) were 14.4 ± 2.8 mg/Mc, 3.7 ± 1.1 mg/Mc, and 1.7 ± 0.6 mg/Mc for CPE, 58 kGy HXPE, and 72 kGy HXPE, respectively (Fig. 3). The HXPE inserts wore significantly less (72%–85% wear reduction) than their CPE counterparts (p < 0.05). The HXPE inserts had a total thickness reduction, at the lowest point on the articulating surface, of 0.10 ± 0.04 mm versus 0.23 ± 0.06 mm for the CPE inserts after 10 Mc. The CPE and HXPE load soaks had a thickness reduction of 0.06 ± 0.002 mm over the first 10 Mc of testing, indicating that both materials had similar creep behaviors.

Figure 3.

Weight loss of CPE and HXPE and tibial inserts during 20 Mc in vitro walking gait test.

Activities of Daily Living

During walking (0–3 Mc), the HXPE inserts showed a 72% wear improvement over CPE in agreement with earlier results (Table 3). Phase II produced a 30% increase in average wear rate of CPE inserts and no increase in that of HXPE inserts. The wear rate of CPE inserts increased from 14.9 ± 3.2 mg/Mc in Phase I to 19.3 ± 4.9 mg/Mc during Phase II. The CPE testing was stopped at 5.1 Mc due to delamination and fracture of the aged inserts (Fig. 4a). The corresponding Phase III wear rate was >40 mg/Mc. The HXPE wear rate increased from 3.4 ± 0.4 mg/Mc in Phase I to 8.6 ± 2.2 mg/Mc after 5.5 Mc in Phase III, with no visible delamination or fracture of the aged inserts (Fig. 4b).

Table 3. Wear rates of HXPE and CPE inserts during various ADL
 MaterialPhase IPhase IIPhase III
  1. ADL, activities of daily living; CPE, conventional polyethylene; HXPE, highly crosslinked polyethylene.

Wear rates (mg/Mc)CPE14.9 ± 3.219.3 ± 4.9> 40
 HXPE3.4 ± 0.43.8 ± 0.78.6 ± 2.2
Figure 4.

Representative photographs of tibial inserts: (a) HXPE after 5.5 Mc and (b) CPE after 5.1 Mc of ADL testing.

Wear Debris Analysis

The CPE inserts generated significantly more debris than their HXPE counterparts (Fig. 5). Over 95% of all the particles had an ECD <1 µm (Table 4). The HXPE and CPE debris populations had significantly overlapping size ranges, with proportionally more fine debris in the HXPE mix than in the CPE one. The debris morphologies were independent of material and activity type, with 5%–10% fibrillar, 30%–40% spheroidal, 10%–15% granular, and 30%–35% others in all cases.

Figure 5.

Scanning electron micrographs of UHMWPE wear debris (a) walking gait, CPE; (b) walking gait, HXPE; (c) high flexion, CPE; and (d) high flexion, HXPE.

Table 4. Shape parameters of HXPE and CPE wear debris from knee tibial inserts
  1. AR, aspect ratio; CPE, conventional polyethylene; ECD, equivalent circle diameter; HXPE, highly crosslinked polyethylene; R, roundness.

Phase I0.21 ± 0.080.53 ± 0.192.86 ± 2.970.38 ± 0.090.58 ± 0.222.39 ± 1.49
Phase III0.23 ± 0.130.60 ± 0.202.25 ± 1.500.29 ± 0.210.53 ± 0.202.98 ± 1.60

Oxidation and Delamination

The aged HXPE and CPE inserts had surface oxidation indices of 0.05 ± 0.04 and 0.89 ± 0.02, respectively. All of the CPE inserts showed delamination within 5.8 Mc (Fig. 6a), while none of the HXPE exhibited delamination after 8 Mc of testing. Delamination initiated as a subsurface white patch that progressively propagated to the surface and led to the detachment of loose fragments of material.

Figure 6.

Delamination (a) and fatigue (b) resistance of CPE and HXPE tibial inserts. Each bar/oval dot represents a sample.

Posterior Edge Loading Fatigue

All CPE inserts developed visible delamination and fatigue cracks within 2.8 Mc of testing (Fig. 6b). Cracks ran predominantly perpendicular to the sliding direction. No fatigue cracks were observed on any of the aged HXPE inserts after 5 Mc.


We showed that the wear of aged HXPE tibial inserts is significantly lower than that of aged CPE tibial inserts under various activities of daily living. Aged HXPE inserts had better delamination and fatigue resistance than aged CPE inserts. Clinical studies suggest that oxidation-induced degradation of mechanical properties leads to, or is a precursor to, delamination of tibial inserts.35, 36 Implanted inserts are exposed to oxygen in body fluids and thereby may undergo in vivo oxidation that may lead to time-dependent in vivo mechanical degradation of UHMWPE.36 Delamination of polyethylene tibial inserts is multifactorial and patient specific, but HXPE's ability to resist oxidation could delay the onset of failure by delamination. The fatigue data from severe posterior edge loading showed a clear performance improvement for aged HXPE when compared to the aged CPE. All aged CPE inserts showed significant fatigue damage below 3.5 Mc of testing, but none of the aged HXPE samples showed any damage after 5 Mc. Likewise, all aged CPE tibial inserts showed delamination between 1.5 Mc to 5.8 Mc, but none of the HXPE delaminated during 8 Mc. These data suggest that the long-term mechanical performance of HXPE may be better than that of gamma sterilized and inert gas/vacuum packaged inserts.

The wear debris had ECDs in the lower range of those isolated from periprosthetic tissues from 36 revised TKAs by Schmalzried et al.37 and by Shanbhag et al.38 This may be due to in vivo/in vitro kinematic differences and the pore size of polycarbonate filters used. The referenced studies employed polycarbonate filters with 0.2 µm and 1 µm pore sizes, while we used 0.05 µm pore size filters. During gait, the wear debris from HXPE and CPE inserts had similar morphological characteristics and overlapping size distributions but with a proportionately higher population of finer particles in the HXPE mix. This finding agrees with that from an in vitro study by Endo et al.39 who showed smaller wear particles from HXPE acetabular cups than from CPE ones. Using peri-implant tissues from 18 revision surgeries from aseptically loose TKAs, Kobayashi et al.40 studied the influence of size, shape, and number of polyethylene particles on the pathogenesis of osteolysis and found no correlation between particle morphology and occurrence of osteolysis. However, they found a highly significant association between number of particles and the presence of osteolysis. Illgen et al.,41 studied the macrophage inflammatory response induced by hip simulator-generated HXPE and CPE wear particles (the HXPE particles are finer than CPE) by measuring the secreted levels of tumor necrosis factor-alpha (TNF-α), vascular endothelial growth factor (VEGF), and interleukin 1α (IL-1α) cytokines. They found concentration-dependent VEGF and TNF-α responses with increased reactivity of HXPE particles compared to CPE, but no difference in IL-1α response. Fisher et al.42 showed that the HXPE wear debris had higher specific TNF-α activity than conventional counterparts when co-cultured with human macrophages, and attributed the difference to the higher proportion of fine particles in the HXPE debris population. However, Illgen et al.43 concluded that the significant reduction in the total volume of debris generated from HXPE inserts when compared to CPE inserts could be expected to offset their increased inflammatory response, and thus reduce the potential for osteolysis.

Our study also showed that high flexion activity did not significantly modify the wear debris size and morphology distribution of HXPE and CPE particles from those produced during walking. All motions produce overlapping particle size ranges with proportionately finer particles in the HXPE population. This suggests that the use of HXPE in young and active patients with a higher frequency of ADL may not produce an increased risk of osteolysis.

There are several limitations to this study. Only aged inserts were tested, which favor HXPE over CPE in terms of mechanical properties. Though non-aged HXPE has lower mechanical properties (tensile fatigue, ductility, and impact strength) than its CPE counterpart, aging-induced oxidation can significantly reduce the impact strength and ductility of CPE and thereby lower its fatigue strength, while HXPE properties remain unchanged. The continuous application of 50 cycles of deep squat (during Phase III) did not allow intermittent relaxation of the polyethylene and represents an extreme worst case. Aged CPE components with compromised fatigue strength may be more prone to delaminate and fracture under the conditions used in this study. It is surprising that wear debris particles with sizes <100 nm were not found even though 50-nm filters were used. This may be due to the resolution limits of the microscope. Low-voltage field emission gun scanning electron microscopes with better image resolution could be used to further investigate the particle size distribution. Moreover, the particle morphologies were deduced from 2D measurements of shape parameters on SEM microphotographs that may obscure the depth information. However, the absence of shadows on these images, even when the filters were tilted relative to the electron beam, indicates that the third dimensions of the particles were of the same order of magnitude as the other two.

In conclusion, we showed that HXPE may have better long-term in vivo stability than CPE inserts. The long-term higher wear and delamination resistance coupled with adequate fatigue behavior of HXPE may offer significant advantages over CPE for younger patients with sustained levels of strenuous activities of daily living that expect prolonged implant longevity.