The Effect of Contact Area on Wear in Relation to Fixed Bearing

and Mobile Bearing Knee Replacements

Shivani Sathasivam,1 Peter S. Walker,1 Patricia A. Campbell,2 Keith Rayner1
1
Biomedical Engineering Program, Cooper Union Research Foundation, 51 Astor Place, New York, New York 10003

2
Joint Replacement Institute, 2400 S. Flower Street, Los Angeles, California 90007

Received 3 April 2000; revised 9 October 2000; accepted 13 October 2000

Published online 00 Month 2001

Abstract: It is generally assumed that the wear rates in knee replacements are reduced as the
contact area is increased. Hence, fixed bearing or mobile bearing designs with large contact
areas throughout the full range of flexion wear less than partially conforming fixed-bearing
designs. This hypothesis was investigated in an experimental model, where flat-ended ultra
high molecular weight polyethylene pins of varying diameters were reciprocated and rotated
on polished metal plates under a constant load with serum lubrication. The pin diameters
ranged from 8 –23 mm, giving nominal contact pressures from 23.9 –2.8 MPa, covering the
range associated with a wide spectrum of total knees including mobile-bearings. For pin
diameters of 8 –12 mm, the mean wear rates were in the range of 5.0 –16.0 E-10 g/cycle. For
diameters of 17 and 23 mm, the mean wear rates were approximately 1.0 E-10 g/cycle. The
latter wear rates were significantly less than the former. Scanning electron microscopy
indicated milder wear processes with the larger diameters, while the smaller diameters
exhibited transverse ripples and cracks and detachment of thin layers from the surface. The
percentages of granules (mostly submicron), fibrils and flakes, and the sizes of these particle
types were similar among all pin diameters, except that, for the 23 mm pin diameter, the
percentage of fibrils increased and of flakes decreased. This work supports the hypothesis that
larger contact areas, up to the maximum area tested in our study, produce lower wear rates,
and suggests that there is no disadvantage regarding particle type or size associated with the
larger areas of contact. © 2001 John Wiley & Sons, Inc. J Biomed Mater Res (Appl Biomater) 58: 282–290,
2001

Keywords: wear debris; polyethylene wear; knee simulators; mobile bearing knees; total
knee replacement

INTRODUCTION increasing the contact area, while minimizing the constraint

and encouraging natural knee motion, by allowing the poly-
In retrieval studies, the polyethylene inserts of fixed bearing ethylene bearings to move freely on polished plates on the
knees have shown deformation, surface wear, and in many upper tibia. Comparison of a mobile bearing design with
cases, delamination.1–3 The most severe wear has generally fixed bearing designs in a knee simulator suggested that a
been in designs of low to moderate conformity. However, considerable reduction in gravimetric wear rate would be
even with higher conformity, severe wear has still been obtained with the former.5 However, another simulator study
observed, often due to torsional or shear movements not showed similar wear rates for a fixed bearing and a mobile
accommodated by the constraint of the surfaces. In addition, bearing design, although, in the latter, the motion between the
pitting has generally been more common with higher than polyethylene bearing and the plate included sliding as well as
with lower conformity designs.2,4 rotation.6,7 In a retrieval study of the Oxford compartmental
One of the major goals of mobile bearing knees (Fig. 1), or meniscal knee, which has full conformity throughout flexion,
meniscal knees, was to reduce the overall wear damage by the average wear rate of the upper and lower surfaces com-
bined was only 0.036 mm/year, with a maximum of 0.08
mm/year.8 The highest wear rates were found in those com-
Correspondence to: Peter S. Walker, Biomedical Engineering Program, Cooper ponents showing impingement; in the other components, the
Union Research Foundation, 51 Astor Place, New York, NY 10003 (e-mail: mean wear rate was only 0.01 mm/year, about one tenth that
[email protected])
Contract grant sponsor: Engineering and Physical Sciences Research Council in total hips. In a recent retrieval study, a knee where the
(EPSRC), UK polyethylene bearings slid in metal tracks showed scratching
© 2001 John Wiley & Sons, Inc. in the tracks.9 In a rotating platform design, there was cir-
282
 EFFECT OF CONTACT AREA ON WEAR IN KNEE REPLACEMENTS 283

Figure 1. (Left) A fixed bearing contact of low conformity with an average contact pressure in the
region of 20 MPa. (Right) A mobile bearing contact with complete conformity at the upper and lower
surfaces with an average contact pressure in the region of 3 MPa. (Note that the study is applicable
to all types of condylar TKRs.)

cumferential scratching of the metal plate and considerable mobile bearing, suffer less wear and damage than those with
wear on the lower surface of the polyethylene bearing.2 There smaller contact areas. Even the laboratory studies differ in
was also delamination wear on the upper surface in some their findings on this subject. We, therefore, carried out a
samples. It is noted that, in this design, there was full con- laboratory study of wear as a function of contact area, to
formity in early flexion but only partial conformity in high determine whether a larger area metal-polyethylene contact
flexion. would produce a lower rate of gravimetric wear compared to
Concerning the wear particles themselves, the size distri- a small area contact under the same loading conditions. In
butions10 and tissue response have been compared for tissues addition, we asked whether the polishing mechanisms ex-
from periprosthetic tissue recovered at revision surgery of pected for a low-pressure interface would produce particles of
total hips and total knees.11 For hips, most of the particles a smaller size than for a high-pressure interface, where more
were submicron, and there were numerous plump macro- severe wear mechanisms might be expected.
phages containing these particles, with only occasional giant
cells. For the knees, most of the particles were in the 2–20
micron range, and these elicited a giant cell reaction with few MATERIALS AND METHODS
macrophages. When the particle sizes from different types of
total knee were measured, it was found that the less conform-
Design and Materials of the Test Components
ing the design, the higher was the surface damage, but the
larger the particle size.9 Mobile bearing implants actually had To maintain a constant area of contact, it was decided to use
a higher rate of production of particles than the fixed bearing cylindrical specimens of ultra high molecular weight poly-
designs. ethylene (UHMWPE) sliding on flat plates made from cobalt-
The earliest laboratory study of the effect of contact pres- chrome alloy. The diameter of the pins ranged from 8 –23
sure on wear rate, in the pressure range relevant to joint mm, producing average contact pressures from 23.9 –2.8 MPa
replacements, showed an exponential increase in wear rate under a load of 1200 N (Fig. 2). This configuration was
with pressure.12 A similar result was found in a later study intended to represent the contact area of one condyle of a total
using a different configuration of the test specimens.13 How- knee, ranging from a low conformity fixed bearing design, to
ever, the opposite effect, a decrease in wear factor with an a mobile bearing design with complete conformity, under a
increase in contact pressure, was found in a reciprocating total joint force of approximately 3 body weight. The pins
pin-on-plate study.14 were held in metal housings with half the diameter protrud-
It is evident that there is no consistent finding that total ing. A semi-circular lug at the bottom of the hole in the metal
knees with larger contact areas, whether fixed bearing or housing prevented the plastic pins from rotating. To evaluate
284 SATHASIVAM ET AL.

Figure 2. The five diameters of UHMWPE pins that were tested,

producing average contact pressures from 2.8 –23.9 MPa. Half the
diameters of the pins were protruding from the metal holders.

the most appropriate shape for the plastic specimens to pro-

duce a uniform pressure distribution, a finite element analysis
was carried out using Abaqus software (Sorensen Inc., Paw-
tucket, RI), the variables being the length of the pin, the Figure 4. Each test station consisted of a tibial tray from a mobile
length protruding from the metal housing, and chamfer vs. no bearing knee, on which two identical plastic pins were loaded.
chamfer (Fig. 3).
The polyethylene was made from 4150 HP powder from
tended to simulate condylar replacement knees and those
Hoechst Celanese. The powder was ram extruded by Poly Hi
types of mobile bearing knees where both displacement and
Solidur in the same way as for standard production of UH-
rotation occur in function. A computer simulation was pro-
MWPE for the orthopedic industry. The tensile modulus was
duced of the superimposed motions, where the paths of an
916 MPa, the tensile yield was 24 MPa, the density was
array of supposed asperities on the metal surface were traced
0.932, and the crystallinity was 59%. The material was not
over the plastic surface for one complete cycle of motion,
sterilized prior to testing to avoid issues concerning oxida-
showing cross-over of the individual wear tracks (Fig. 5).
tion. The metal trays were implantable components of the
Six test samples and two control samples of each size of
Mobile Bearing Knee (MBK) (Zimmer, Warsaw, ID) made
UHMWPE pin were manufactured. Two additional 10 mm
from cast cobalt-chrome alloy, the upper surface polished to
samples and 4 additional 12 mm samples were made, due to
0.01– 0.03 ␮m Ra value.
inconsistent data from the first 6 samples of each. The control
samples were inserted into the metal housings, immersed in
Wear Test Specifications
lubricant, and left beside the test machine. The purpose of the
Each tibial tray was housed in a chamber containing the controls was to monitor the changes in weight of the different
lubricant which was 30% calf serum (Fig. 4). Two plastic pin diameters due to intake of lubricant. Immediately before
pins were loaded against the tray, as shown. The tibial tray the test, samples were loaded in the machine for 96 h to allow
was reciprocated anterior-posterior with a total travel of 10 plastic deformation to occur, which would tend to produce a
mm. The plastic pins were rotated by ⫾5° about their own uniform pressure distribution. As a check, Fujifilm was used
axes, in phase with the displacement. This motion was in- to measure the contact area of the test samples after this initial

Figure 3. The contact pressures on the surfaces of two alternate pin shapes, determined by finite
element analysis. The plain cylindrical shape has a more uniform distribution.
 EFFECT OF CONTACT AREA ON WEAR IN KNEE REPLACEMENTS 285

a cylindrical pin, and a pin with a chamfer (Fig. 3). It was

thought that the pressure with the chamfer might be more
uniform. However, there was a concentration of stress around
the periphery, in contrast to the plain cylindrical pin, which
showed a much more uniform distribution. Hence, the cylin-
drical shape was chosen for the tests. The Fujifilm pressure
patterns measured just before the start of the testing showed
a uniform pressure distribution for all pin diameters. This
indicated that, even if there was some lack of uniformity on
immediate loading as was indicated by the FEA model, creep
during the 96 h pretest loading produced uniformity.
For all the diameters except 8 mm, the profilometer traces
showed the highest wear to occur in arcs in line with the
sliding direction [Fig. 6(A)]. This corresponded to the great-
est areas of metal asperity overlap determined in the com-
Figure 5. A computer simulation of the paths taken on the plastic pin puter model (Fig. 5).
surfaces by an array of asperities on the metal surface. The crossing However, for the 8 mm diameter pins, the nominal pres-
tracks can be seen when displacement and rotation were combined. sure of 23.9 MPa was excessive and resulted in deformation,
which expanded the diameter to approximately 9 mm. In one
loading, and at the intervals during the test when wear mea-
of the samples, a crack developed across the contact area. The
surements were made. This also checked the alignment of the
profilometry at the end of the 5 million cycle test showed an
components in the machine. The wear test was run at 1 Hz for
uneven surface, indicating nonuniform wear over the surface
5 million cycles at 20°C on a 6-station hydraulic machine,
[Fig. 6(B)]. The center of the specimen wore and deformed to
testing a total of 12 specimens in one run (Fig. 3). The
a depth of 0.2 mm more than the surrounding areas.
lubricant was topped up with distilled water during each run,
The wear of all the specimens plotted against the number
and was changed every 48 h.
of sliding cycles is shown in Figure 7. The wear of the 17 and
Quantification of Wear 23 mm pins was almost undetectable. The curves for the 8
After every one million cycles, the test was stopped and the mm pins were clustered just above the highest 23 mm curve.
serum lubricant was agitated to generate a suspension of wear The one 8 mm sample that cracked showed a high early wear
particles in the lubricant. Around each pin, 30 mL of the rate, but the subsequent wear rate was only slightly higher
lubricant was collected and frozen. The plastic pins were than for the other 8 mm pins. This one sample was not
removed from their holders, washed, dried, and then left in an included in the averaged data. The next cluster of curves was
oven for 24 h before their weight losses were measured. for the 10 mm pins, while the highest wear rates occurred for
When the test ended after 5 million cycles, a pooled the 12 mm diameter pins. However, there were 2 curves for
solution was made for each pair of pins tested containing 4 ⫻ the 12 mm pins with approximately half the wear rate. The
30 mL of lubricant taken from the test after 2, 3, 4, and 5 differences were attributed to the presence of fine scratches
million cycles. 10 mL of each of these pooled solutions was on the metal plates of the samples with the higher wear rates.
processed in a similar manner to that for extracting particles Due to this, the 4 additional samples were run. However,
from tissue around joints.15,16 The particles were examined these also showed the higher wear rates, along with fine
by scanning electron microscopy, with a computerized image scratching of the metal plates. The 2 additional 10 mm
analysis program being used to characterize the particles. The
wear particles were classified into three groups: granules
(submicron, rounded and smooth), fibrils (elongated shapes),
and flakes (submicron or micron sized, irregular in shape and
texture), the numbers of each type being quantified and their
sizes measured.
At the end of the test, the shapes of the ends of the pins
were determined by soft-probe profilometry. The diameter of
the probe was 1 mm, while the resolution in a vertical
direction was 10 ␮m. The worn surfaces of the pins were also
viewed by scanning electron microscopy to investigate the
different wear mechanisms occurring.

RESULTS
Figure 6. The surface height patterns of a 12 mm diameter pin and an
The finite element analysis (FEA) model was used to calcu- 8 mm diameter pin after the test, determined by soft-probe profilom-
late the pressure distribution on the end of the plastic pin for etry. The dark color represents a lower surface level. (Scale: mm.)
286 SATHASIVAM ET AL.

Figure 7. Weight loss plotted against number of cycles for all the pins
tested. The data is corrected for fluid uptake based on the nonworn
soak controls.

samples were run due to the scatter of the data for the initial
6 samples.
The mean wear rates for each diameter pin were calculated
(Fig. 8). ANOVA was carried out on the data (SPSS software,
Chicago, IL), and using the Bonferroni test, there were sig-
nificant differences between the wear rates of all pairs of pin
diameters at p ⬍ 0.05, except for the following pairs: 8 and
10 mm, 8 and 17 mm, 8 and 23 mm, and 17 and 23 mm.
Typical scanning electron micrographs at 2000 magnifi-
cation are shown in Figure 9. For the 17 and 23 mm diameter
pins, there were scratches 1–3 ␮m wide and deep, local areas
of ripples at 2– 4 ␮m spacing perpendicular to the main
sliding direction, and raised patches a few microns in size that
could indicate incipient flake generation. For the 12 mm
diameter pins, the ripples were more extensive, often had Figure 9. Scanning electron micrographs at 2000 ⫻ of typical areas
of the worn surfaces: (a)–(b) 17 mm dia; (c)–(d) 12 mm dia; (e)–(f) 8 mm
cracks in between them, were larger at 4 –10 ␮m spacing, and
dia. Milder wear processes are evident for the larger diameters.

had fibrils projecting from the crests oriented in the sliding

direction.
The 10 mm pins were similar to the 12 mm pins, but had
distinct areas where a thin surface layer was breaking up to
form flakes several microns in size. For the 8 mm pins, all the
above phenomena were represented, but there was evidence
of more generalized surface disruption, particularly surface
cracking and the partial detachment of sheets and flakes from
the surface.
A typical appearance of the granules, fibrils, and flakes is
shown in Figure 10. Visually there were no apparent differ-
ences in the morphologies of the 3 particle types among the
pin diameters. For the diameters from 8 –17 mm, the percent-
ages of the three types of particles were similar. However, for
Figure 8. Mean and standard deviation data for the wear rate plotted
the 23 mm diameter pin, there was a higher percentage of
against pin diameter. The wear rates for the two larger diameters are fibrils and a lower percentage of flakes (Fig. 11). The per-
significantly less than for the three smaller diameters. centage of granules remained similar for all pin diameters,
 EFFECT OF CONTACT AREA ON WEAR IN KNEE REPLACEMENTS 287

while there was also a remarkable similarity in the sizes of the

granules (Table I). On the other hand, the length of the fibrils
and the diameters of the flakes tended to reduce for the higher
pin sizes, especially notable for the 23 mm diameter pin.

DISCUSSION

In comparing and evaluating the relation between wear rate

and contact pressure in different studies, the test conditions
need to be considered. In the study of Rostoker and Ga-
lante,12 distilled water was used as the lubricant and stainless
steel as the counterface material. Rather than varying the
contact area, they varied the applied load in order to vary the
contact pressure. Wear was measured by change in linear
height over relatively small time periods, which could have
Figure 11. The percentages of the three types of particles for all the
introduced creep effects. Nevertheless, their finding of higher pin diameters.
wear rates for higher pressures agreed with ours in general.
Rose et al.13 found a similar relation, although their sliding
speed was higher than physiological and they applied a PV bends. Morra et al.18 calculated contact stresses of up to 26
(pressure ⫻ velocity) criterion to explain the increased wear MPa at 1950 Newtons for contemporary total knees, using
at high pressures. The findings of Barbour et al.14 were the finite element analysis. The Freeman–Samuelson design,
opposite, lower wear rates for higher contact pressures. They which has a roller-in-trough geometry, has an average contact
used truncated UHMWPE cones of only 3.4 –5.3 mm diam- stress of 3.1 MPa,19 which is similar to that of the upper
eter sliding without rotation along a track 30 mm in length. bearing surfaces of the Rotaglide Mobile Bearing Knee de-
The contact stresses of 3.4 –28.3 MPa were achieved with signed by Polyzoides et al.20 Mobile Bearing designs with
small loads. Their data had a large scatter, especially at the complete conformity on the upper surface, such as the Ox-
lower pressures. Hence, in various respects, particularly the ford8 and the MBK,6 have average contact pressures in the
relation between the areas of contact and the sliding distances region of 2 MPa. This value is applicable to the lower bearing
and the absence of rotation, the tests did not closely resemble surfaces of mobile bearing knees, where the flat plastic sur-
the contact situation in total knees. faces slide on the polished surface of the metal baseplate. The
In relating our test to total knees, one consideration is the range of pressures used in our test of 2.8 –23.9 MPa covers
range of average contact pressures. For different fixed bear- the range in total knees, except for the even lower contact
ing condylar knees under 2000 Newtons load, Szivek et al.17 stresses seen in some mobile bearing knees.
measured a range from 13–25 MPa at flexion angles of 0° and In our tests, an attempt was made to reproduce the contact
60° flexion. Pappas et al.5 calculated values as high as 60 conditions as closely as possible across the spectrum from
MPa for low conformity designs in stairs and deep knee condylar knees to mobile bearing knees, using a simple test
configuration. A metal pin sliding on a plastic surface was
considered, but for a cylindrical pin, there would be serious
edge effects, while for a spherically ended pin, the contact
areas would change shape and the pressures would not be
uniform or constant over time.21 For the plastic pin on the flat
metal surface, a uniform pressure could be achieved, as
determined by the Fujifilm. By setting the linear displacement
and the rotation to typical physiological values, the relative
plastic-metal sliding and the excursions relative to the dimen-
sions of the contact area itself would be reasonably repro-
duced. Varying the contact areas of the plastic specimens
while maintaining the same loading conditions was also seen
as important for reproducing the actual situation in the knee,
rather than changing the load for specimens of constant area.
However, the test does not fully simulate all the phe-
nomena that occur in total knees. For example, partially
conforming fixed bearing total knees are subject to tractive
rolling as well as linear sliding, rotation, and pure rolling,
Figure 10. The three particle types that were characterized: granules, which can significantly change the wear mechanisms on
fibrils, and flakes. the plastic.22 Also, the conditions in the mobile bearing
288 SATHASIVAM ET AL.

TABLE I. Sizes of Granules, Fibrils, and Flakes

Granules Fibrils Flakes

8mm max dia min dia length max dia min dia
ave 0.5201 0.3860 4.3293 3.4587 1.9357
s.dev 0.0851 0.0767 2.4618 2.1642 0.9412
median 0.5059 0.3626 3.6874 2.8100 1.8000
min 0.4098 0.2898 1.1305 0.7806 0.4837
max 0.7352 0.5724 12.6629 12.6079 4.6520
n 20 20 72 101 101
% 10.36 10.36 37.31 52.33 52.33
10mm max dia min dia length max dia min dia
ave 0.4344 0.3193 4.2464 2.9357 1.7647
s.dev 0.0836 0.0723 3.2641 2.6350 1.5017
median 0.4344 0.3159 3.2759 1.9792 1.2653
min 0.2520 0.1768 0.8969 0.4858 0.3563
max 0.6476 0.4898 16.0461 15.1302 10.8113
n 81 81 136 221 221
% 18.49 18.49 31.05 50.46 50.46
12mm max dia min dia length max dia min dia
ave 0.4544 0.3317 4.2083 4.2017 2.5996
s.dev 0.0768 0.0675 3.0039 5.1293 3.1926
median 0.4327 0.3175 3.3924 2.1892 1.3709
min 0.3182 0.2021 0.9482 0.6361 0.3137
max 0.6504 0.4878 14.8099 29.4765 19.2346
n 77 77 146 239 239
% 16.67 16.67 31.60 51.73 51.73
17mm max dia min dia length max dia min dia
ave 0.5413 0.4108 4.3011 2.2860 1.4741
s.dev 0.0672 0.0551 2.9730 1.8632 1.0353
median 0.5373 0.4082 3.2917 1.7250 1.1353
min 0.4106 0.3096 1.3170 0.5691 0.3902
max 0.6956 0.5423 15.5756 14.6378 7.3828
n 33 33 90 198 198
% 10.28 10.28 28.04 61.68 61.68
23mm max dia min dia length max dia min dia
ave 0.4706 0.3444 2.5314 1.9794 1.3047
s.dev 0.0897 0.0763 1.5197 1.5116 1.0123
median 0.4561 0.3415 1.9895 1.4555 0.9479
min 0.3467 0.2353 0.6341 0.6700 0.4206
max 0.7540 0.5882 9.3836 7.5854 5.5181
n 103 103 294 154 154
% 18.69 18.69 53.36 27.95 27.95

knee vary between the upper and lower bearing surfaces, distance combined with the rotation allowed for sufficient
and whether the lower surfaces allow only rotation or both replenishment of fluid in the contact region. As the diameter
sliding and rotation. It is clear that no simple test can increased to 12 mm, the fluid replenishment would be much
reproduce all conditions of all designs, but can neverthe- less, possibly promoting starved lubrication and even fluid
less give valuable indicators under a controlled set of degradation. This may have produced the worst combination
conditions. Ultimately, individual designs need to be of contact pressure and lubricating conditions, as evidenced
tested in an appropriate knee simulating machine, a more by the tendency for scratching of the metal plate. For the
extensive and costly procedure. larger diameter pins, however, the contact pressures would be
Explanations can be proposed for the relation between sufficiently low that fluid flow could still occur between the
wear rate and contact area in our test; namely, an increase in surface irregularities of the plastic and metal, because the
wear rate from 8 –12 mm diameter, and then a sharp decrease pressures would be insufficient to completely compress them.
in wear rate at 17 and 23 mm diameters. For the 8 mm It must be noted, however, that the constant load used in our
diameter, although the wear processes were severe, so long as tests would have imposed more severe conditions than phys-
there was no gross breakdown, such as delamination, the iological, where lubricant replenishment could occur during
wear rate was low. This may be because the 10 mm sliding swing phases of activity.
 EFFECT OF CONTACT AREA ON WEAR IN KNEE REPLACEMENTS 289

The scratching of the metal plates per se is not easy to For the areas of contact associated with mobile bearing
explain. Such scratching of metal has been reported by others knees, represented by the 17 and 23 mm diameter pins, the
on both retrievals and test specimens.1–3,6,23 Possible expla- wear rates were significantly lower than for the areas repre-
nations are hard contaminants such as calcium oxide in the senting fixed bearing designs, even if the wear rates are
polyethylene, carbide particles released from the surface of doubled to account for the upper and lower bearing surfaces.
the metal, particles in the lubricant, or even extraneous debris This would suggest that the wear rates of mobile bearing
from the wear test machine. knees are much less than that of fixed bearing knees. How-
The scanning electron microscopy data showed that the ever, the relative amounts of sliding in fixed vs. mobile
wear processes for the three smaller diameters appeared to be bearing knees needs to be considered. Evidence from knee
similar, predicting granules, fibrils, and flakes. Generation of simulator testing suggests that, in fact, the sliding and rotation
flakes is probably dependent on sufficient contact pressure to in a mobile bearing knee can be less than that of a fixed-
produce high shear stresses over sufficient areas of plastic bearing knee.27 This has also been borne out in fluoroscopy
surface, at least several microns in size. However, for the studies of patients ascending a step,28,29 where there was less
larger diameters, the individual contact patches were evi- motion than reported for a range of fixed-bearing designs.30
dently smaller in area, favoring the generation of granules Knee simulator testing showed that the wear in fixed-bearing
and fibrils from small ripples. The tendency for smaller sized designs was much greater than for the LCS Rotating Plat-
particles is consistent also. form.5 Bell et al.,6 however, showed similar wear rates in the
The size of the particles in our study can be compared with MBK Mobile Bearing design and the Insall–Burstein II de-
that found from in vivo sources. Schmalzreid et al.11 found sign, mainly due to the higher wear on the lower plastic
that most of the particles in the tissue around fixed-bearing surface compared with the upper plastic surface interfacing
total knees were in the size range of 2–20 microns, with with the femoral condyles. This may have been due to both
particles less than 1 micron being less common. In contrast, translation and rotation occurring at the lower bearing sur-
most of the particles in tissue around total hips was less than face, a facility intended to provide improved knee kinematics
1 micron in size. Hirakawa et al.9 studied the particles around and a greater freedom of position. Collier et al.2 found that
three types of total knee, with low-medium conformity, me- the lower bearing surfaces on mobile bearing knees were
dium-high conformity, and mobile bearing knees. The tissue usually scratched, which could account for accelerated wear
was obtained at revision for loosening. The mobile bearing on these surfaces, especially considering the accelerated wear
knees (LCS Rotating Platform) had a larger number of par- that could be produced even for modest scratching.31 The
ticles per month at implantation time greater than 10 months, wear rates of retrieved Oxford knees, a completely conform-
compared with the other two groups. The fixed-bearing de- ing unicompartmental design, was very small in 1–13 year
sign of low-medium conformity had a higher percentage of follow-up.8 The average wear depth was 0.036 mm per year
larger particles than the other two groups. Beaule et al.24 with a maximum of 0.08 mm. However, when the cases with
measured the mean granule diameter, fibril length, and flake bone-plastic impingement were discounted, this figure was
diameter of particles around medium-high conformity fixed- reduced to 0.01 mm per year. In light of the above, caution
bearing knees at 0.39, 3.37, and 1.14 microns. This can be must be given to directly extrapolating the results of simple
compared with our data, averaging all pin diameters, of 0.42, tests to specific designs of mobile bearing (or fixed bearing)
3.92, and 2.39 microns. The percentages of the different knees. For example, factors such as the actual area of contact,
particle sizes in our study agreed more with the data of which can be very large on the lower bearing surface, and the
Schmalzreid et al.11 and Hirakawa et al.9 than with Beaule et flatness of the polished tibial plate may be important factors.
al.,24 but that might to some extent reflect the degree of In summary, considering all the above evidence, it appears
conformity of the different knee designs studied. that the wear rates of total knees with large contact areas,
Our data as a whole supports the use of large contact areas such as mobile bearing knees, can be much less than that of
in total knees, in order to reduce the volumetric wear rate. fixed-bearing designs, for the same plastic material. How-
However, paradoxically, for fixed bearing designs, it ap- ever, each specific design would need to be tested individu-
peared that an 8 mm diameter contact area with a pressure of ally in a suitable simulating machine in order to account for
23.9 MPa would be better than a larger area, for example 12 the design, kinematics, manufacturing, and materials factors
mm diameter. However, for the small area, the severe wear that come into play.
processes and the likelihood of delamination with some types The authors acknowledge the collaboration of Dr. Gordon Blunn
of polyethylene would suggest caution. Nevertheless, with and Dr. Carol Bell at the Centre of Biomedical Engineering, UK.
molded polyethylene, such contact areas in a low conformity SEM analysis was provided by AEA Technology, Didcot, Oxford-
modular knee have indeed produced low wear without del- shire, UK. The tibial trays were provided by Zimmer, Swindon, UK.
This project was funded by the Engineering and Physical Sciences
amination in long-term follow-up.23 The absence of severe Research Council (EPSRC), UK.
wear at 8 –12 year follow-up has also been noted for the M-G
unicompartmental knee, and for the AGC knee, also made REFERENCES
from molded polyethylene.2,25 Molded polyethylene has been 1. Blunn GW, Joshi AB, Minns RJ, Lidgren L, Lilley P, Ryd L,
found to be much less susceptible to oxidation, which may Engelbrecht E, Walker PS. Wear in retrieved condylar knee
explain the lack of delamination.26 arthroplasties. J Arthropl 1997;12:281–290.
290 SATHASIVAM ET AL.

2. Collier JP, Mayor MB, McNamara JL, Surprenant VA, Jensen weight polyethylene tibial inserts. A Finite element study. At-
RE. Analysis of the failure of 122 polyethylene inserts from lanta: Proc AAOS Conf; 1996.
uncemented tibial knee components. Clin Orthop 1991;273: 19. Plante–Bordeneuve P, Freeman MAR. Tibial high-density poly-
232–242. ethylene wear in conforming tibiofemoral prostheses. J Bone
3. Landy MM, Walker PS. Wear of ultra high molecular weight Joint Surg B 1993;75:630 – 636.
polyethylene components of 90 retrieved knee prostheses. J 20. Polyzoides AJ, Dendrinos GK, Tsakonas H. The Rotaglide total
Arthropl 1988;3S:S73–S85. knee arthroplasty. J Arthropl 1996;11:453– 459.
4. Wright TM, Rimnac CM, Faris PM, Bansal M. Analysis of 21. Estupinan JA, Bartel DL, Wright TM. Residual stresses in
surface damage in retrieved carbon-fibre reinforced and plain ultra-high molecular weight polyethylene loaded cyclically by a
polyethylene tibial components from posterior stabilized total rigid moving indenter in nonconforming geometries. J Orthop
knee replacements. J Bone Joint Surg A 1988;70:1312–1319. Res 1998;16:80 – 85.
5. Pappas MJ, Makris G, Buechel FF. Wear in knee prosthetic 22. Wimmer MA, Andriacchi TP, Natarajan RN, Loos J, Karlhuber
knee joints. Washington DC: Proc AAOS Conf; 1992.
M, Petermann J, Schneider E, Rosenberg AG. Astriated pattern
6. Bell CJ, Walker PS, Sathasivam S, Campbell PA, Blunn GW.
of wear in ultra high-molecular-weight polyethylene compo-
Differences in wear between fixed bearing and mobile bearing
knees. Trans Orthop Res Soc 1999;24:962. nents of miller-galante total knee arthroplasty. J Arthropl 1998;
7. Menchetti PPM, Walker PS. Mechanical evaluation of mobile 13:8 –16.
bearing knees. Am J Knee Surg 1997;10:73– 82. 23. Blunn GW, Joshi AB, Lilley PA, Engelbrecht E, Ryd L, Lidgren
8. Psychoyios V, Crawford RW, O’Connor JJ, Murray DW. Wear L, Hardinge K, Neider E, Walker PS. Polyethylene wear in
of congruent meniscal bearings in unicompartmental knee ar- unicondylar knee prostheses: 106 retrieved Marmor, PCA and
throplasty. J Bone Joint Surg B 1998;80:976 –982. St. Georg tibial components compared. Acta Orthop Scand
9. Hirakawa K, Bauer TW, Stulberg BN, Wilde AH, Borden LS. 1992;63:247–255.
Characterization of debris adjacent to failed knee implants of 3 24. Beaule PE, Campbell PA, Walker PS, Schmalzreid TP, Blunn
different designs. Clin Orthop 1996;331:151–158. GW, Bell C, Dorey FJ, Amstutz HC. Characterization of poly-
10. Landry ME, Blanchard CR, Mabrey JD, Wang X, Agrawal CM. ethylene wear particles retrieved from total knee replacements
Morphology of in vivo generated ultra high molecular weight worn in vivo and in a knee simulator. Orlando: Trans Orthop
polyethylene wear particles as a function of contact conditions Res Soc; 2000.
and material parameters. J Biomed Mater Res (Appl Biomater) 25. Ritter MA, Worland R, Saliski J, Helphenstine JV, Edmondson
1999;48:61– 69. KL, Keating EM, Faris PM, Meding JB. Flat-on-flat, noncon-
11. Schmalzreid TP, Jasty M, Rosenberg A, Harris WH. Polyeth- strained, compression molded polyethylene total knee replace-
ylene wear debris and tissue reactions in knee as compared to ment. Clin Orthop 1995;321:79 – 85.
hip replacement prostheses. J Appl Biomater 1994;5:185–190. 26. Furman BD, Awad JN, Chastain KE, Li S. Material and per-
12. Rostoker W, Galante JO. Contact pressure dependence on wear formance differences between retrieved machine and moulded
rates of ultra high molecular weight polyethylene. J Biomed I-B type total knee arthroplasties. Trans Orthop Res Soc 1997;
Mater Res 1979;13:957–964. 22:643.
13. Rose RM, Goldfarb HV, Ellis E, Crugnola AM. On the pressure 27. Des Jardins JD, Walker PS, Haider H, Perry J. A comparison
dependence of wear of UHMWPE. Wear 1983;92:99 –111. between the uni-axial and dynamic walking cycle laxities of
14. Barbour PSM, Barton DC, Fisher J. The influence of contact
multiple TKR designs. Trans Euro Orthop Res Soc 1999;32.
stress on the wear of UHMWPE for total replacement hip
28. Des Jardins JD, Walker PS, Haider H, Perry J. The use of a
prostheses. Wear 1995;181–183:250 –257.
15. Campbell PA, Ma S, Schmalzreid T, Amstutz HC. Tissue di- force-controlled dynamic knee simulator to quantify the me-
gestion for wear debris particle isolation. J Biomed Mater Res chanical performance of total knee replacement designs. J Bio-
1994;28:523–526. mech, to appear.
16. Campbell PA, Ma S, Yeom B, McKellop H, Schmalzreid TP, 29. Walker PS. Consequences regarding wear of providing rotation
Amstutz HC. Isolation of predominantly submicron-sized UH- in TKR. J Bone Joint Surg B 1999;81SII:126.
MWPE wear particles from periprosthetic tissues. J Biomed 30. Dennis DA, Komistek RD, Hoff WA, Gabriel SM. In vivo
Mater Res 1995;29:127–131. kinematics derived using an inverse perspective technique. Clin
17. Szivek JA, Cutignola L, Volz RG. Tibiofemoral contact stress Orthop 1996;331:107–117.
and stress distribution evaluation of total knee arthroplasties. J 31. Hailey JL, Ingham E, Stone M, Wroblewski BM, Fisher J.
Arthropl 1995;10:480 – 491. Ultra-high molecular weight polyethylene wear debris gener-
18. Morra EA, Postak PD, Greenwald AS. Effects of plateau ge- ated in vivo and in laboratory tests; the influence of counterface
ometry on delamination wear failures in ultra high molecular roughness. Proc Inst Mech Eng 1996;210H1:3–10.