Abstract
Background: The results of recent studies documenting
the backside wear of polyethylene inserts retrieved from total knee
implants call into question the stability of the locking mechanisms
of modular tibial components. Wear of the metal tibial baseplate
suggests that the capture mechanisms of some modular fixed-bearing
tibial components do not adequately restrict in vivo motion
of the insert. The purposes of this study were (1) to present a
method for evaluating locking-mechanism stability and (2) to investigate
the stability of modular tibial components after an interval in
vivo.
Methods: We measured the anteroposterior and mediolateral
motion between the polyethylene insert and the tibial tray in a
variety of modular total knee tibial components. A uniaxial mechanical
testing machine was used to evaluate the stability of ten unimplanted
components (control group), fifteen implants obtained from patients
who were undergoing revision total knee arthroplasty (revision group),
and fifteen devices retrieved post mortem (autopsy group). We applied
loads along the anteroposterior and mediolateral axes of the tibial component
and recorded the maximum insert displacement that occurred. From
this value, we calculated an insert-motion index, the magnitude
of a two-dimensional vector that represented the total motion in
the transverse plane.
Results: For the control group, the mean insert-motion
index was 64 ± 13 m (range, 6 to 157 m); for the
revision group, it was 341 ± 51 m (range, 104 to
718 m); and for the autopsy group, it was 380 ±
45 m (range, 122 to 657 m). The insert-motion index for the control
group was significantly lower than that for the revision group (p = 0.001) or
autopsy group (p < 0.001).
Conclusions: Motion between the polyethylene insert
and the metal baseplate in contemporary modular tibial designs increases
after a period of in vivo loading.
Clinical Relevance: Although there are several advantages
to the use of modular tibial components, these ad-vantages
must be weighed against the disadvantage of backside wear debris
secondary to motion of the modular insert. Debris from backside
wear combined with wear from the articular side might account for
the increasing prevalence of osteolysis since modular components have
become widely used.
Polyethylene wear debris from the bearing surfaces of contemporary
modular total knee components is widely recognized as the primary
cause of osteolysis after total knee arthroplasty. To our knowledge,
osteolysis was not reported as a clinical problem with the first
generation of one-piece tibial components. Only after the introduction
of modular polyethylene inserts in the mid-1980s was failure of
knee replacements due to osteolysis recognized as a major clinical problem1,2. In recent years, researchers also
have observed wear on the backside of retrieved polyethylene inserts3-5 (Figs. 1-A and 1-B). Perhaps the combination of articulating
surface wear and backside wear has produced a greater volume of
debris, which has caused the increased occurrence of osteolysis
observed with the use of modular implants.
The presence of backside polyethylene wear on retrieved components
indicates that the locking mechanism of fixed-bearing tibial trays
may not adequately eliminate motion between the modular elements2,3. Although motion of the insert
has been documented in a laboratory setting4,
to our knowledge the in vivo increase or reduction
of the insert motion of contemporary modular components has yet
to be reported. In the present study, we hypothesized that the motion
that is inherent in implants that have a locking mechanism would
increase over time with in vivo service and could
contribute to the generation of wear debris from the metal baseplate
and the backside of the polyethylene insert. Our purposes were (1)
to present a method for evaluating the stability of the locking
mechanism of new and retrieved modular tibial components, and (2)
to test the stability of modular components that were retrieved
post mortem and at revision of total knee replacements to determine whether
the stability of the locking mecha-nism deteriorated after in
vivo loading.
We tested forty tibial components from a variety of modular, fixed--bearing
total knee designs. The components were categorized into three groups:
unimplanted components, implants obtained during revision procedures,
and devices -retrieved post mortem (Table I). Thirty-two
components were cruciate-retaining, and eight were cruciate-substituting.
Of the unimplanted components, only the Insall-Burstein-II -im-plant
was cruciate-substituting. Seven of the retrieved components were
cruciate-substituting, including three Insall-Burstein II implants
(Zimmer, Warsaw, Indiana), two Genesis implants (Richards, Memphis,
Tennessee), one Or-tholoc implant (Wright Medical Technology, Arlington,
Tennessee), and one Press Fit Condylar implant (Johnson and Johnson,
Raynham, Massachusetts).
The group of unimplanted components served as our controls. The -control
group consisted of ten midsized components, each with a different -locking-mechanism
design (Table I).
For each component, we selected a polyethylene-insert thick-ness
that most closely approximated a tibial component thickness of 10
mm when combined with the tibial baseplate.
The revision group consisted of fifteen modular tibial components
that had been retrieved from fifteen patients during revision total
knee arthroplasty. We used our database to identify the modular
tibial components that had the same locking mechanism as the controls
and that had the longest in vivo duration for each
design and then selected those revision implants from our retrieval
laboratory. One or two implants of each locking-mechanism design
were selected. All fractured inserts and those with apparent damage
to the locking mechanism were excluded from the study.
Eleven of the revision components were from women, and four were
from men. At the time of implant removal, the average patient age
was 64.3 years (range, thirty-six to eighty years) and the average
patient weight was 90.3 kg (range, 52.2 to 145.2 kg). The mean time
that the implants were in situ was forty-four months
(range, two to eighty-eight months). The preoperative diagnosis
that necessitated the primary total knee arthroplasty was osteoarthritis
for thirteen patients and posttraumatic arthritis for two. The revision
total knee arthroplasty was performed for a variety of reasons,
including osteolysis (four patients); a deep infection (three);
a loose cemented tibial component (two); and knee instability, loose cementless
tibial and femoral components, a fracture of a femoral component,
oversized tibial and femoral components, arthrofibrosis, and patellar
subluxation (one patient each).
The autopsy group consisted of fifteen modular tibial components
harvested post mortem along with the supporting soft tissue and
bone. The specimens included components of four different tibial
designs, including three of the designs that were tested in the
control group. Seven implants were from five women, and eight were
from five men. The average age of the patients at the time of death
was seventy-eight years (range, sixty-nine to eighty-four years),
and the average weight was 80.4 kg (range, 55.8 to 131.6 kg). The
average time that the implants were in situ was
eighty-seven months (range, one to 158 months). All of the total
knee arthroplasties in the autopsy group had been performed because
of osteoarthritis, with the exception of two that had been performed
bilaterally in one patient because of rheumatoid arthritis. While
the patients were living, the average Hospital for Special Surgery
knee-rating score6 was 91.5 points
(range, 85 to 98 points) for four knees, and the average Knee Society
Score7 was 89.7 points (range,
73 to 99 points) for nine knees. All of the patients had reported
little or no pain at the final follow-up evaluation.
All but one insert retrieved at revision had been separated from
the tibial baseplate during the surgical procedure. We tested this
specimen and three autopsy implants both before and after disassembling
the components to determine whether disassembly affected motion
of the insert relative to the baseplate. For these four components,
only the test results after the reassembly were considered when
we compiled or compared the experimental data.
The anteroposterior and mediolateral motion between the polyethylene
insert and the baseplate of the tibial tray of each control and
retrieved implant was measured with use of the following protocol.
The control implants were assembled according to the manufacturer’s
instructions. The specimens in each group were preconditioned for
two weeks in a bath of saline solution at 37°C to simulate in
vivo physiological conditions, which caused the polyethylene
to swell and the locking mechanism to tighten8.
Each tibial baseplate, or proximal part of the tibia for the autopsy
specimens, was secured inside a rectangular solid mold with the
use of set-screws so that the anteroposterior, mediolateral, and
superior-inferior axes of the tray were aligned with the dimensions
of the mold. The mold was then filled with a quick-setting acrylic
(Bondo body filler; Bondo/Mar-Hyde, Atlanta, Georgia),
which, after hardening, yielded a fixture geometry that allowed
for the consistent attachment of the various tibial baseplate designs
to the testing machine. The baseplate and the polyethylene insert
were mounted separately on two metal frames (Fig. 2), which were
then mounted vertically on a materials testing machine (MTS Systems,
Eden Prairie, Minnesota) with a dovetail system. Metal brackets
were affixed to the polyethylene insert and the acrylic mold that
housed the baseplate with the use of hot glue. The brackets provided
reference points for an extensometer to record the anteroposterior
and the mediolateral displacement between the insert and the baseplate.
At a fixed rate of 10 N/sec, we applied a compressive
load of 100 N to the frame that held the polyethylene insert, which induced
downward motion of the polyethylene insert relative to the stationary
tibial baseplate. After attaining 100 N of compression, the system
was loaded to 100 N of tension at the same loading rate, displacing
the insert upward relative to the baseplate. This loading profile
was applied for at least one additional cycle, and the resulting
plot was inspected to ensure reproducibility of the output. The
fixture was then rotated 90° about the superior-inferior axis of
the component to align the alternate axis with the direction of
loading. The above test was repeated to determine the permissible
displacement in the plane of the tibial baseplate. The recorded
parameters during each test of anteroposterior or mediolateral insert displacement
were time (in seconds), load (in newtons), and exten-someter
displace-ment (in micrometers).
This protocol ensured that the total slack of the locking mechanism
was recorded, regardless of the initial position of the insert relative
to the baseplate. We defined insert motion as the displacement that
occurred between the insert and the baseplate with no mechanical
resistance (Fig. 3).
In order to characterize the stability of the locking mechanism,
we then calculated the -insert-motion index, the magnitude
of a two-dimensional vector that represented the total motion in
the transverse plane, for each specimen with use of the equation:
Insert-Motion Index = AP2+ML2, where AP is
the insert motion in the anteroposterior direction and ML is the
insert motion in the mediolateral direction.
We used analysis of variance with a Tukey post hoc test
to analyze the differences in the insert-motion index among the
three groups. A detectable difference of 200 m between groups was
considered clinically relevant. A sample power analysis revealed
that, for the standard deviation levels for the -anteroposterior
and mediolateral motions, a sample size of eight would lead to a
power of 0.95 (alpha = 0.05). Statistical relationships
between the insert-motion- index and other variables were
assessed with the use of the Mann-Whitney U- test or the
Spearman rho test.
The mean insert-motion index (and standard error) was 64 ±
13 m (range, 6 to 157 m) for the ten tibial components in the control
group, 341 ± 51 m (range, 104 to 718 m) for the
fifteen components in the revision group, and 380 ±
45 m (range, 122 to 657 m) for the fifteen autopsy specimens. The
insert-motion index of the control group was significantly lower
than that of the revision group (p = 0.001, analysis of
variance, Tukey post hoc test) and that of the
autopsy group (p < 0.001, analysis of variance,
Tukey post hoc test). There was no significant
difference between the insert-motion indices of the revision and
autopsy groups (p = 0.8, analysis of variance, Tukey post
hoc test).
Included among the two retrieval groups were twenty-seven specimens
of the ten designs tested in the control group. The insert-motion
index for each of these retrieved specimens exceeded that of its
control counterpart by an -average of 273 m (range, 43
to 696 m). The insert-motion indices were similar in the left and
right components from three patients in the autopsy group who had
received the same type of total knee implant bilaterally and in
whom the in situ durations of the two implants
had been equivalent. The values for the two implants in each of
these three patients were 122 m compared with 231 m, 465 m compared
with 488 m, and 561 m compared with 625 m. In contrast, the test results
varied among components of the same design that had been implanted
in different patients.
In the revision and autopsy groups, no correlations were found
between the insert-motion index and the in situ duration
(p = 0.8, Spearman rho test), the patient’s weight
(p = 0.9, Spearman rho test), or the patient’s
gender (p = 0.6, Mann-Whitney U test). In situ duration
was significantly longer for the autopsy group than for the revision
group (p < 0.003, Mann-Whitney U test). In addition, no
association was found between the mechanism of failure that had
necessitated revision and the insert-motion index. The four specimens
revised because of osteolysis had highly variable insert-motion
indices (121, 178, 276, and 718 m) relative to the general population
of retrieval specimens.
The comparison of insert-motion indices before and after disassembly
of the modular tibial component demonstrated variable changes in
insert stability. After reassembly, one revision and one autopsy
implant had increased insert motion (+19 and +39
m, respectively), and two autopsy implants had decreased
insert motion (-24 and -69 m). All four inserts had greater motion
than their respective control specimens, both before and after modular
disassembly.
The results of our study demonstrate that the instability inherent
in contemporary modular tibial tray locking mechanisms increases
with in vivo physiological loading. The elimination
of all motion between modular parts that have markedly different
moduli of elasticity is not possible. We believe that after implantation
a competition phenomenon occurs between the stability of the modular
tibial tray locking mechanism and the stresses acting upon it. The
ability of the locking mechanism to counteract repetitively applied
stresses is affected by the initial insert motion, the in
situ duration, and the stresses transmitted at the modular
interface. We speculate that the instability of the locking mechanism
begins with deformation of the polyethylene and then gradually increases
as wear of the polyethylene insert further alters the fit within
its metal housing.
Parks et al.4 previously reported
the interface motion of nine types of unimplanted modular tibial
trays. Using the same implant designs, we confirmed their finding
that motion can occur in assembled tibial components with each capture-mechanism design.
The magnitudes of motion under 100-N and 400-N loads reported by
Parks et al. represented a combination of polyethylene deformation
and insert motion. We chose to isolate insert motion from polyethylene
deformation by identifying the displacement that occurred before
resistance from the locking mechanism was encountered (Fig. 3). Therefore,
the lower values in our study are inherent to the nature of our
loading protocol. The insert-motion index expresses the magnitude
of the motion vectors in the transverse plane, a Cartesian combination
of both anteroposterior and mediolateral motion. We believe that
this is a valid method for testing the locking mechanisms of new
and retrieved modular fixed-bearing inserts and is more representative
of motion that might occur under minimal in vivo loading
conditions.
The amount of motion that occurs with in vivo loading
is not necessarily represented by the amount measured in the autopsy
and retrieval specimens. We did not account for axial or torsional
loads or for other factors that interact with the stability of the
locking mechanism in vivo. Axial loads due to muscle
forces and weight-bearing would be expected to affect insert motion in
vivo, but we did not apply such loads in the mechanical
testing. For ease of measurement, we recorded insert motion that
was parallel to the anteroposterior and mediolateral axes of the
baseplate under loads directed parallel to these axes, although characteristic
wear patterns on explanted tibial tray baseplates confirm that rotational
motion also occurs at the interface in many designs (Fig. 1-A). We acknowledge
that rotation in the same plane also occurs, as does motion perpendicular
to the tibial component.
We speculated that removal of the insert might alter the stability
of the modular tibial components. However, we found that the insert-motion
indices before and after disassembly were comparable, suggesting
that insert removal did not appreciably alter insert stability.
Although only four different designs were available in the postmortem
group, the test results for these implants were similar to the results
for the wide variety of revision implants tested.
No difference in insert motion was found between the revision and
autopsy specimens. Perhaps the shorter length of time for revision
components to exhibit insert motion equivalent to that of the autopsy
specimens was associated with the conditions that had necessitated
revision surgery and the younger average age of the patients in
the revision group.
We had speculated that the four implants revised because of osteolysis
would have more insert motion than the other revision components.
However, we did not find a relationship between the reason for component
revision and insert stability. We also had thought that the demands
on the stability of the locking mechanism might be greater for the
more conforming fixed-bearing implants, such as the posterior-stabilized
and varus-valgus constrained implants. In theory, a higher degree
of articular constraint would transmit higher loads to the insert-baseplate
interface in these designs. However, we did not find more insert
motion in the six posteriorly stabilized components.
Tibial tray-locking mechanisms come in a variety of designs that
could lead to some differences in insert stability4. We found that tongue-in-groove components
blocked motion least effectively in the direction of the tongue
and groove. Therefore, Anatomic Modular Knee implants demonstrated greater
anteroposterior motion, and Insall-Burstein-II implants showed greater
mediolateral motion. In the control group, the snap-fit implants,
such as the Press Fit Condylar and Miller-Galante-II components,
were marginally more stable than the tongue-in-groove components.
However, the retrieved snap-fit components demonstrated overall
motion equivalent to that of the tongue-in-groove components, but with
greater variability in the direction of motion. Components with
a partial capture wall, such as the Genesis and the Duracon implants,
tended to allow motion in the direction without a barrier to insert
motion.
Caution is needed when the femoral articulating surface and the
insert-tray interface are being compared. A few hundred micrometers
of motion at the tray-insert interface is much smaller than the
motion that occurs at the tibiofemoral articulation as the knee
flexes through an arc of >90°. However, the articular side
has a highly polished surface, whereas the baseplates of most modular
tibial components do not. Certainly, two relatively rough surfaces
moving against each other with loads exceeding three times body
weight will generate substantial amounts of debris. Our findings
warrant additional studies to examine the instability of locking
mechanisms and the correlations with the extent of wear of the counterfaces.
Modular tibial components have a number of advantages. Inventory
is reduced when variable thickness polyethylene inserts can be combined
with tibial baseplates of different sizes. Positioning and insertion
of the components is far easier with use of modular implants. Cementless
fixation of the tibial tray is enhanced because screw augmentation
is allowed. Revision surgery is easier because of improved access
to the tibial fixation interface with removal of the modular insert. Insert
exchange provides an alternative to complete revision if polyethylene
wear becomes a clinical problem. However, until problems with modularity
are resolved, these relative advantages of modular components need
to be considered against the consequences of more wear debris.
The results of this study indicate that locking-mechanism designs
for modular implants need to be improved. An acceptable amount of
motion of modular components has not been established. The mechanical
stability of contemporary modular components should be documented
by benchmark testing under repetitive physiological loads and by
additional testing of retrieved modular implants. If insert motion
cannot be reduced to an acceptable level with the use of modular
components, the orthopaedic community should consider -establishing
International Organization for Standardization standards for the
surface finish of modular elements. Alternatives to reduce wear
debris, such as nonmodular implants and polished tibial baseplates,
should be considered until the stability of modular components can
be ensured.
Peters PC, Engh GA, Dwyer KA,Vinh TN. Osteolysis after total knee arthroplasty without cement. J Bone Joint Surg Am,1992;74: 864-76. 74864
1992
[PubMed]
Engh GA, Dwyer KA,Hanes CK. Polyethylene wear of metal-backed tibial components in
total and unicompartmental knee prostheses. J Bone Joint Surg Br,1992;74: 9-17. 749
1992
[PubMed]
Wasielewski RC, Parks N, Williams I, Surprenant H, Collier JP,Engh G. Tibial insert undersurface as a contributing source of
polyethylene wear debris. Clin Orthop,1997;345: 53-9. 34553
1997
[PubMed]
Parks NL, Engh GA, Topoleski LD,Emperado J. The Coventry Award. Modular tibial insert micromotion.
A concern with contemporary knee implants. Clin Orthop,1998;356: 10-5. 35610
1998
[PubMed][CrossRef]
Gabriel SM, Dennis DA, Honey MJ,Scott RD. Polyethylene wear on the distal tibial insert surface
in total knee arthroplasty. Knee,1998;5: 221-8. 5221
1998
[CrossRef]
Insall JN, Ranawat CS, Aglietti P,Shine J. A comparison of four models
of total knee--replacement prostheses. J Bone Joint Surg Am,1976;58: 754-65. 58754
1976
[PubMed]
Insall JN, Dorr LD, Scott RD,Scott WN. Rationale of the Knee Society clinical rating system. Clin Orthop,1989;248: 13-4. 24813
1989
[PubMed]
Clarke IC, Starkebaum W, Hosseinian A, McGuire P, Okuda R, Salovey R,Young R. Fluid--sorption phenomena in sterilized polyethylene
acetabular prostheses. Biomaterials,1985;6: 184-8. 6184
1985
[PubMed][CrossRef]