Periprosthetic osteolysis is an increasingly prevalent complication of
total knee arthroplasty. Particulate debris displaced from the
articular1-3
and
backside4-11
surfaces of tibial polyethylene inserts is considered to be a principal cause.
Osteolysis tends to undermine fixation interfaces and can contribute to
implant loosening or predispose the periprosthetic bone to fracture.
From analyses of radiographs made between five and ten years after primary
total knee arthroplasty, we determined the prevalence of osteolysis associated
with a single knee-replacement system in which the proximal surface finish of
the tibial baseplate and the method of sterilization of the tibial
polyethylene insert had changed over the years. We suspected that polishing of
the proximal surface of the tibial baseplate counterface and improvement in
the polyethylene quality as a result of sterilization with gamma radiation in
an inert environment or with nonirradiation methods had decreased the
prevalence of osteolysis. In particular, we sought to identify whether these
design modifications and other factors (related to the patient, the
polyethylene insert, and the alignment of the knee and components) influenced
the risk of osteolysis.
Between 1987 and 1998, five surgeons performed a total of 795 total knee
arthroplasties with a posterior cruciate ligament-retaining Anatomic Modular
Knee implant (DePuy, Warsaw, Indiana) as a primary reconstructive procedure
for the treatment of osteoarthritis at our institution. The implant featured a
metal tibial baseplate and a minimally constrained ultra-high molecular weight
polyethylene insert. From this population, we studied 365 knees (300 patients)
for which a standing full-limb anteroposterior radiograph had been made during
the first postoperative year (at a mean of four months after the arthroplasty)
and follow-up radiographs had been made between five and ten years
postoperatively and that had traceable implant identification numbers. Four
hundred and thirty knees were not examined for this study, and the most common
reason for the exclusions was that no radiographs had been made in the five to
ten-year postoperative interval (285 knees). The 430 unexamined knees were
similar to the 365 study knees in terms of patient and implant factors.
The mean patient age and weight (and standard deviation) at the time of the
index arthroplasty were 68 ± 8 years and 84 ± 16 kg. Fifty-nine
percent (214) of the 365 total knee replacements were implanted in women. The
preoperative diagnosis was osteoarthritis with a varus deformity (a
radiographic tibiofemoral angle ranging from 5° of valgus to 22° of
varus) in 78% (285) of the 365 knees and osteoarthritis with a valgus
deformity (a tibiofemoral angle ranging from 6° to 29° of valgus) in
22% (eighty) of the 365 knees.
A titanium baseplate with a grit-blasted proximal surface finish (average
roughness [Ra] = 1.0
µm)12 was used
in 246 knees, with 239 of the 246 baseplates fixed with cement, whereas a
cobalt-chromium baseplate with a polished proximal surface finish
(Ra < 0.1 µm) was fixed with cement in 119
knees12. The
geometry of the proximal surface of the baseplate was the same in the two
groups (Fig. 1). All femoral
components had a polished cobalt-chromium articular surface, and 45% (166)
were placed with cement. All but eight patellae were resurfaced, and 346 of
the 357 resurfaced patellae had a cemented all-polyethylene button.
The initial minimum thickness of the polyethylene insert averaged 8
± 2 mm. The manufacturer furnished other data regarding the
polyethylene insert including the type of
resin13 (116 were
made with 412 resin; twenty-six, with 1900 resin; 219, with 415 resin; and
four, with 1050 resin), the forming method (all were machined to their final
shape, with 202 formed from a compression molded sheet and 163 formed from a
ram-extruded bar), the sterilization
method14 (four were
sterilized with ethylene oxide; 263, with gamma radiation in air; fifty-four,
with gamma radiation in nitrogen; and forty-four, with gas plasma), and the
sterilization date. The mean shelf age of the insert—i.e., the number of
years that had elapsed between sterilization and implantation—was 0.9
± 1.0 year, and the maximum was 7.1 years. Gas-plasma-sterilized
inserts differed in that the dovetailed slots in the distal surface were
sloped 0.5° relative to the distal surface to create interference during
assembly with the baseplate rails.
There were five permutations of the proximal surface finish of the
baseplate and the method of sterilization of the polyethylene insert. Four
inserts that had been sterilized with ethylene oxide and 242 that had been
gamma-irradiated in air were affixed to a grit-blasted baseplate surface.
Twenty-one inserts that had been gamma-irradiated in air, fifty-four that had
been gamma-irradiated in nitrogen, and forty-four that had been sterilized
with gas plasma were mated to a polished surface.
Knee alignment in the coronal plane and intercomponent hyperextension (the
angle of hyperextension between the femoral and tibial components) were
estimated from the standing full-limb radiographs (35 × 129 cm). Knee
alignment in the coronal plane—i.e., the distance by which the
hip-to-ankle mechanical axis was medial to the center of the tibial component
at the joint line—averaged 1 ± 10 mm medial to the knee center.
Intercomponent hyperextension, determined by comparing the tilts of the
femoral and tibial components on anteroposterior radiographs, averaged 4°
± 8°.
One pair of anteroposterior and lateral radiographs of the knee (35 ×
43 cm) made between five and ten years after the arthroplasty (i.e., the most
recent pair made within that interval) was evaluated for evidence of
osteolysis. The anteroposterior radiograph was made with the patient standing
on the limb under evaluation. The lateral radiograph was made with the patient
lying on the side with the knee in maximum flexion. The five to ten-year
postoperative interval was selected because the proximal surface of the
baseplate was not polished until 1992 and sterilization with gamma radiation
in air was not abandoned until 1993. We considered it inappropriate to compare
the most recent radiographs of total knee replacements with a
rough-proximal-surface baseplate and an insert that had been gamma-irradiated
in air, which were used as many as eighteen years before the time of the
study, with the most recent radiographs of knees with more contemporary
versions of the tibial component, which had been in use for only the previous
ten years.
Two experienced orthopaedists who specialized in hip and knee arthroplasty
(C.A.E. Jr. and J.P.McA.), and who had performed fourteen of the 365 total
knee replacements, reviewed the radiographs, independently deciding whether
the periprosthetic bone was positive, questionable, or negative for
osteolysis. Osteolysis was defined as any nonlinear region of periprosthetic
cancellous bone loss with delineable margins. The knee was deemed positive for
osteolysis if one observer rated it as positive for osteolysis and the other
rated the finding as positive or questionable. The knee was considered
negative for osteolysis if one observer deemed the radiographic finding to be
negative and the other rated it as negative or questionable. The knee was
considered to have a questionable finding of osteolysis if both observers
rated the finding as questionable or if one deemed it to be positive and the
other deemed it to be negative.
Statistical Methods
The Fisher exact test was used to compare prevalences of osteolysis.
Probability values of 0.05 or less (p = 0.05) were considered to indicate
significance. One-way analysis of variance with the Tukey post hoc test was
used to compare the durations of follow-up from the surgery to the date on
which the reviewed radiographs were made.
Cox proportional hazards regression analysis (SPSS, Chicago, Illinois) was
used to identify factors that influenced the risk of a knee having osteolysis
at five to ten years. The time variable of the regression was the number of
years from the arthroplasty to the date on which the follow-up radiographs
were made. The status variable was a positive finding of knee osteolysis (yes
or no). The twelve independent variables (gender, patient age, patient weight,
type of osteoarthritis, proximal surface finish of the baseplate, method of
forming the insert, method of sterilizing the insert, shelf age of the insert,
thickness of the insert, method of fixation of the femoral component, distance
of the postoperative mechanical axis from the center of the tibial component,
and intercomponent hyperextension angle) were entered into the regression
simultaneously, and then stepwise elimination (Wald backward method [SPSS]
with an exclusion threshold of p > 0.05) was used to isolate a regression
that included only variables significantly associated with osteolysis (p =
0.05). For each significant factor, we also report the odds ratio increase
calculated from the regression analysis to illustrate how the odds of a knee
having osteolysis would change on the basis of the presence of the factor or a
change in its magnitude.
To better understand how the examined risk factors influenced the
prevalence of osteolysis before surface polishing of the baseplate was
initiated and after sterilization with gamma radiation in air was
discontinued, the Cox proportional hazards regression analysis was repeated
with two study subgroups. In an analysis of the 242 knees in which the
baseplate counterface had been grit-blasted and the insert had been
gamma-irradiated in air, two of the twelve independent variables of the
regression were not entered. (The baseplate surface finish and the
sterilization method did not vary in this subgroup.) In an analysis of the
ninety-eight knees with a polished baseplate counterface and an insert that
had been gamma-irradiated in nitrogen or had been sterilized with gas plasma,
three of the twelve independent variables of the regression were not entered.
(The baseplate surface finish did not vary in this subgroup, and the methods
for forming the insert and for fixing the femoral component varied
negligibly.)
Prevalence of Osteolysis
The overall prevalence of osteolysis was 27% (ninety-nine of 365 knees).
The prevalence of osteolysis decreased when the baseplates with a grit-blasted
proximal surface and the inserts sterilized with gamma radiation in air were
changed to baseplates with a polished proximal surface and inserts sterilized
by alternative methods (Table I
and Appendix). The highest prevalence (34%; eighty-two of 242 knees) was found
when an insert that had been gamma-irradiated in air had been mated with a
grit-blasted baseplate surface. The lowest prevalence (6%; three of fifty-four
knees) was observed when an insert that had been gamma-irradiated in nitrogen
had been mated with a polished baseplate surface. Knees in which the insert
had been sterilized with gas plasma were followed for a significantly shorter
period than were knees in which the insert had been gamma-irradiated in air
and those in which it had been gamma-irradiated in nitrogen (p < 0.01 for
each, one-way analysis of variance with Tukey post hoc test). The prevalence
of osteolysis differed significantly (p < 0.01) between the knees treated
with a baseplate with a grit-blasted proximal surface and an insert that had
been gamma-irradiated in air (34%; eighty-two of 242) and the knees with a
baseplate with a polished proximal surface and an insert sterilized with gamma
radiation in nitrogen or with gas plasma (9%; nine of ninety-eight).
Ninety-nine knees were rated positive for osteolysis; the two observers
each chose the positive rating more often when the proximal surface of the
baseplate had been grit-blasted (fifty-seven of the eighty-three positive
results in the grit-blasted group were considered positive by both observers)
than when it had been polished (four of the sixteen positive results in the
polished group were considered positive by both observers), as osteolysis
tended to be more severe in the former group. Men with a varus deformity, a
baseplate with a grit-blasted proximal surface, and an insert that had been
gamma-irradiated in air had a particularly high prevalence of osteolysis
(forty-four [53%] of the eighty-three knees were rated as positive). Women
with a varus deformity, a baseplate with a grit-blasted proximal surface, and
an insert that had been gamma-irradiated in air had a 30% prevalence of
osteolysis (thirty-one of 103 knees). Valgus osteoarthritic knees (seventy-one
in women and nine in men) had a relatively low prevalence of osteolysis (nine
[11%] of the eighty knees).
Analysis of Risk Factors for Osteolysis
Six of the twelve examined factors were associated with osteolysis at five
to ten years postoperatively (p = 0.05, Cox regression). These factors
included male gender, proximal surface finish of the baseplate, method of
forming the insert (the type of polyethylene from which the insert was
machined), method of sterilizing the insert, shelf age of the insert, and
hyperextension angle of the femoral component relative to the tibial component
(Table II). Men were 3.6 times
more likely to have osteolysis than were women. Knees in which the baseplate
had a grit-blasted proximal surface were 2.6 times more likely to be affected
by osteolysis than were knees treated with a polished-surface baseplate. Knees
with an insert that had been gamma-irradiated in air were 4.0 times more
likely to have osteolysis than were knees with an insert that had been
gamma-irradiated in nitrogen. The risk of osteolysis increased by a factor of
1.5 with any 1.0-year increase in the shelf age of the insert (for example,
the risk of osteolysis was 3.4, or 1.53.0-0.0, times greater in a
knee with a 3.0-year-shelf-age insert than it was in a knee with a
0.0-year-shelf-age insert).
When the regression analysis was limited to knees with a grit-blasted
baseplate counterface and an insert that had been gamma-irradiated in air,
which had a 34% prevalence of osteolysis (eighty-two of 242 knees), male
gender, the method of forming the insert, and the shelf age of the insert were
significantly associated with osteolysis (p < 0.01 for each). When the
regression analysis was limited to knees with a polished baseplate counterface
and an insert that had been gamma-irradiated in nitrogen or had not been
irradiated, which had a 9% prevalence of osteolysis (nine of ninety-eight
knees), only the intercomponent hyperextension angle was significantly
associated with osteolysis (p = 0.01). This suggests that hyperextension of
the femoral component relative to the tibial component is one variable that
may explain why some knees were affected by osteolysis despite the changes in
the baseplate surface and the method of sterilizing the insert
(Fig. 2).
Over its history, the modular tibial component of the Anatomic Modular Knee
system evolved from one featuring an insert that had been gamma-irradiated in
air and joined to a baseplate with a grit-blasted surface to one featuring an
insert sterilized with another method (gamma radiation in an inert environment
or a method not involving radiation) joined to a baseplate with a polished
surface. The geometries of the standard tibial articular surface, proximal
surface of the baseplate, and articular surface of the femoral component were
maintained. Thus, surgeons who used this implant had a unique opportunity to
examine how polishing of the baseplate counterface and theoretical improvement
of the polyethylene quality influenced the prevalence of osteolysis.
The aforementioned shifts to a polished baseplate surface (in 1992) and
away from polyethylene that had been gamma-irradiated in air (in 1993)
markedly reduced the rate of osteolysis associated with this prosthetic
design. At five to ten years postoperatively, osteolysis was identified in 34%
(eighty-two) of 242 knees in which the baseplate had a grit-blasted proximal
surface finish and the polyethylene insert had been gamma-irradiated in air,
whereas it was found in 9% (nine) of ninety-eight knees in which the baseplate
had a polished proximal surface and the insert had been gamma-irradiated in
nitrogen or sterilized with gas plasma. Polishing the baseplate counterface is
believed to have reduced loss of polyethylene from the backside of the insert
(as compared with that measured in inserts retrieved from knees with a
grit-blasted baseplate
counterface6),
despite probable increases in interface motion related to the reduced
friction12. Use of
gamma radiation in an inert gas environment and non-radiation sterilization
methods may yield a more fatigue-resistant bearing surface. As osteolysis was
found in relatively few knees with a polished baseplate counterface and an
insert that had been gamma-irradiated in an inert gas or sterilized with gas
plasma, longer-term study may be required to more clearly elucidate the
importance of the sterilization method (and other potential risk factors)
within this cohort.
It is no surprise that the reported prevalences of osteolysis following
total knee arthroplasty have varied
widely2,15-27.
Each series has featured unique definitions of osteolysis, different
observers, observer bias, different patient populations, dissimilarities among
the implants, and variable follow-up. We defined osteolysis as any nonlinear
area of cancellous bone loss with distinct margins, and we elected to consider
the findings to be positive when one observer rated the radiograph as positive
for osteolysis and the second observer considered the finding to be
questionable. Detection can be difficult. Anteroposterior radiographs permit
limited inspection of the femoral epicondylar bone. Posterior femoral condyle
overlap28 and the
central box of posterior-stabilized
components29 limit
evaluation of the condyles on lateral radiographs. Oblique radiographs afford
better scrutiny of femoral condylar
bone29,30.
As our study was retrospective and we used standard anteroposterior and
lateral radiographs, we probably underestimated the prevalence of
osteolysis.
Few investigators have examined risk factors for the development of
osteolysis. Our findings parallel those of a study of a contemporaneous
implant design (PFC [Press-Fit Condylar] Modular; Johnson and Johnson
Orthopaedics, Raynham, Massachusetts) that featured a cruciate-retaining or
cruciate-substituting insert that had been gamma-irradiated in air and was
affixed to a baseplate with a grit-blasted
surface18. That
study showed that male gender, a young age, the type of polyethylene from
which the insert had been machined (the sheet supplier and the type of resin),
the insert tumbling method (a process used to remove rough edges from the
bearing), and the shelf age of the insert increased the risk of wear-related
failure or radiographic evidence of major osteolysis (an area of >100
mm2). Our Cox regression analysis of knees in which the insert had
been gamma-irradiated in air and the baseplate counterface had been
grit-blasted yielded similar findings: male gender, the type of polyethylene
from which the insert had been machined (the type of resin, and sheet molding
as opposed to bar extrusion), and shelf age were associated with an increased
risk of osteolysis. The documented importance of insert-processing factors and
shelf age in both studies attests to the role of polyethylene quality in the
development of osteolysis. Such studies allow surgeons to identify at-risk
joints and to appreciate risk factors associated with osteolysis. Partial
revision arthroplasties precipitated by wear or
osteolysis31-34
are less likely to provide an active patient with a durable revision if a
baseplate style associated with osteolysis is retained, the insert is
exchanged for an aged insert that had been gamma-irradiated in air, or knee or
intercomponent malalignment that had contributed to wear is not identified and
addressed.
Varus placement of the postoperative mechanical axis has been associated
with increased medial compartment wear of inserts that had been
gamma-irradiated in
air35,36.
However, we could not identify an association between the postoperative
mechanical axis and osteolysis.
Sagittal plane alignment may also affect the risk of osteolysis.
Hyperextension of the femoral component relative to the tibial component has
become a concern with
cruciate-substituting37
and
cruciate-retaining38
polyethylene inserts. When intercomponent hyperextension is present (as a
result of flexion of the femoral component, posterior slope of the tibial
component, knee hyperextension, or a combination of the three), the femoral
component sulcus can engage the central eminence or stabilizing post of the
insert, thereby displacing additional particulate debris and transferring
greater demands to the backside interface. We used a trigonometric method to
quantify the angle of intercomponent hyperextension in stance from the
full-limb anteroposterior radiograph and found that the result was associated
with osteolysis (see Appendix). With attention to how the vertical thickness
of the baseplate and the gap between the posterior condyles and the trochlea
of the femoral component change as each implant is rotated in the field of the
beam, one can better recognize intercomponent hyperextension on a standing
anteroposterior radiograph. We did not assess intercomponent rotation in the
transverse plane.
In this radiographic review performed five to ten years following primary
total knee arthroplasties done with one design of implant, periprosthetic
osteolysis was associated with multiple factors. These factors were related to
the patient (male gender), the design of the tibial baseplate (proximal
surface finish), the surgical technique (intercomponent hyperextension), and
the tibial insert (the type of polyethylene from which it was machined, the
sterilization method, and the shelf age). A notable reduction in the
prevalence of osteolysis occurred after the baseplate counterface was polished
to limit backside wear and the method for sterilization of the polyethylene
insert was refined to limit articular wear. In our view, the keys to
minimizing osteolysis are implantation of a polyethylene bearing of optimum
quality, continued refinements in the design of the modular tibial component
to reduce backside wear, and innovation in surgical instrumentation and
component design to limit postoperative malalignment.
Note: The authors acknowledge the efforts of multiple
individuals at DePuy, a Johnson and Johnson Company, who traced the processing
data of the polyethylene tibial components.