Total knee arthroplasty is an effective treatment for severe arthritis of
the knee. The technical aspects of performing the arthroplasty are known to
affect both the short and the long-term outcome. One of the technical details
considered to be important is to reproduce the anteroposterior patellofemoral
size.
As the patellar retinaculum is relatively fixed in length, some believe
that increasing the anteroposterior size of the patella, the femur, or both
(and thus increasing the tension on the lateral retinaculum) adversely affects
the outcome of total knee
arthroplasty1-6.
This situation, in which the anteroposterior size of the patella or femur (or
both) is increased, is referred to as stuffing of the patellofemoral
joint1,2.
Some studies have shown that stuffing of the joint causes reduced range of
motion1,3,4,7-10.
It also has been suggested that stuffing might result in pain and inferior
function of the
joint2,3,5,6.
Although frequently speculated upon, stuffing has been infrequently
evaluated clinically. The purpose of the present study was to determine the
effect of stuffing on the outcome of total knee arthroplasty.
We conducted a retrospective review of 1100 consecutive primary total knee
arthroplasties that had been performed in 742 patients in 1997 and 1998.
Institutional review board approval was obtained from both participating
centers. The exclusion of fifty-three knees (4.8%) that had been lost to
follow-up, 208 knees (18.9%) that had fewer than two years of follow-up, and
nine knees (0.8%) in patients with rheumatoid arthritis rather than
osteoarthritis left 830 knees with a minimum two-year follow-up, which are the
focus of the present report. These 830 knees represented 75.5% of all
arthroplasties and 74.9% (556) of all patients included in the initial review.
The mean duration of follow-up (and standard deviation) for these 348 women
(62.6%) and 208 men (37.4%) was 4.4 ± 1.1 years (range, two to seven
years; mode, five years). The mean age of the study patients at the time of
the index arthroplasty was 68.9 ± 8.7 years) (range, thirty-five to
eighty-eight years).
All procedures had been performed by one of the six surgeon authors
(J.L.P., M.A.R., E.M.K., P.M.F., J.B.M., or M.E.B.). A posterior
cruciate-retaining prosthesis (Anatomic Graduated Component AGC Knee; Biomet,
Warsaw, Indiana) was used in 715 knees (86.1%), and a posterior
cruciate-substituting prosthesis (Legacy Posterior Stabilized/LPS Knee;
Zimmer, Warsaw, Indiana) was used in 115 knees (13.9%). Four of the surgeons
routinely used posterior cruciate-retaining prostheses, and two routinely used
posterior cruciate-substituting prostheses. No surgeon used both methods. All
prostheses were cemented. The patella was resurfaced in all of the knees with
an all-polyethylene component. The femoral component was rotationally aligned
along the epicondylar axis.
Preoperative and postoperative radiographic measurements of anterior
patellar displacement, anteroposterior femoral size, combined anteroposterior
patellofemoral size, anterior femoral offset, and posterior femoral offset
were made. One-sixth of all radiographic evaluations were done by each
surgeon. Radiographs were randomly assigned for review, and the reviewers were
blinded with regard to the clinical outcome. Measured anteroposterior femoral
size was compared with the absolute anteroposterior femoral size by entering
the known size of the prosthetic implants.
Radiographic techniques were standardized for each radiographic view. The
radiographs were made by two radiography technicians, each of whom had twenty
years of experience. The latest available postoperative radiographs were used
for analysis unless they were determined to be inadequate for accurate
radiographic measurements, in which case the best available postoperative
radiographs were used.
The skyline view was made with the knee in 45° of flexion. Preoperative
anterior patellar displacement was determined on the skyline view and was
defined as the anteroposterior distance from the anterior cortex of the
patella to the femoral trochlear groove
(Fig. 1; Measurement A).
Postoperative anterior patellar displacement was determined on the skyline
view and was defined as the anteroposterior distance from the anterior margin
of the resurfaced patella to the prosthetic femoral trochlear groove
(Fig. 2; Measurement A').
Preoperative anteroposterior femoral size was determined on the lateral
knee radiograph (made with the knee positioned in 45° of flexion on the
radiography table and with a tube-to-tabletop distance of 101.6 cm) and was
defined as the distance from the most anterior aspect of the distal part of
the femur to the posterior cortex of the femoral condyles
(Fig. 3; Measurement B). In the
event that the lateral radiograph was not a true-lateral (oblique) radiograph,
the posterior reference point was the midpoint between the posterior margins
of the medial and lateral femoral condyles. Postoperative anteroposterior
femoral size was determined on the lateral radiograph and was defined as the
distance from the anterior margin of the prosthetic femur to the posterior
margins of the prosthetic femoral condyles
(Fig. 4; Measurement B'). In
the event that the lateral radiograph was not a truelateral (oblique)
radiograph, the posterior reference point was the midpoint between the
posterior margins of the medial and lateral femoral condyles.
The combined anterior patellar displacement and anteroposterior femoral
size was calculated by adding the values of these measurements (Measurement A
+ Measurement B = preoperative combined anterior patellar displacement and
anteroposterior femoral size; Measurement A' + Measurement B' = postoperative
combined anterior patellar displacement and anteroposterior femoral size).
Anterior femoral offset, defined as the distance between the anterior
femoral cortical margin and the anterior margins of the femoral condyles, also
was measured preoperatively (Fig.
3, Measurement C) and postoperatively
(Fig. 4, Measurement C').
Similarly, posterior femoral offset, defined as the distance between the
posterior femoral cortical margin and the posterior margins of the femoral
condyles, was measured both preoperatively
(Fig. 3, Measurement D) and
postoperatively (Figure 4,
Measurement D').
Postoperative data on range of motion, the Knee Society Knee Score, the
Knee Society Function Score, the Knee Society Pain Score, and lateral
retinacular release were retrieved electronically from our patient database,
which reflects data collected by the operating surgeon at the time of office
follow-up visits. The technique for measuring range of motion did not vary by
surgeon or over time. Passive flexion was measured with use of a standard
goniometer while the patient was supine. Because of the large number of
patients, we were able to examine Knee Society Function and Pain Scores
separately to identify whether differences in the Knee Society Knee Score were
due to function, pain, or both. Consistent with the Knee Society clinical
rating scale, pain was defined as no pain or as mild through severe pain. The
indication for lateral retinacular release was the same for all surgeons. A
lateral release was performed when patellar tilt or lateral patellar
subluxation were present when the extensor mechanism was approximated with a
towel clip or suture at the superior pole of the patella with the knee flexed
to 90°.
Linear regression was used to analyze the effect of the measured
radiographic variables on Knee Society Knee Scores. Log-linear regression was
used to analyze the effect of the measured radiographic variables on range of
motion and Knee Society Function Scores. Logistic regression was used to
analyze the effect of the measured radiographic variables on Knee Society Pain
Scores and the need for a lateral retinacular release. Statistical analysis
was performed with use of SAS Version 8.2 statistical software (SAS Institute,
Cary, North Carolina).
All regression models were analyzed for power, with the level of
significance set at 0.05 and power equal to 0.80 (probability of type-II-beta
error = 0.20). These models were adequately powered (n = 830) to detect any
variable explaining >0.9% of the total variance in postoperative range of
motion, the Knee Society Knee Score, the Knee Society Function Score, the Knee
Society Pain Score, and the need for lateral retinacular release. Power
analysis was performed with nQuery Advisor 5.0 (Statistical Solutions, Cork,
Ireland).
Changes in anterior patellar displacement had a significant effect on the
range of motion (p = 0.0079) and the Knee Society Function Score (p <
0.0001) but had no association with the Knee Society Knee Score (p = 0.1824)
or the Knee Society Pain Score (p = 0.4838)
(Table I). Changes in
measured anteroposterior femoral size had a significant effect on
range of motion (p = 0.0182) and the Knee Society Function Score (p = 0.0094)
but had no correlation with the Knee Society Knee Score (p = 0.8030) or the
Knee Society Pain Score (p = 0.2688). Absolute anteroposterior
femoral size was less strongly associated with outcomes than was measured
anteroposterior femoral size, most likely because of the tendency for any
nonsystematic magnification and measurement error to average out with the use
of a consistent preoperative and postoperative measure and to approach zero
with the large number of study subjects. Changes in combined anterior patellar
displacement and anteroposterior femoral size had no significant effect on
range of motion, the Knee Society Knee Score, the Knee Society Function Score,
or the Knee Society Pain Score (p = 0.8869, p = 0.3127, p = 0.3191, and p =
0.5372, respectively). Both anterior femoral offset (p = 0.0092) and posterior
femoral offset (p = 0.0385) had small negative effects on function. The Knee
Society Function Score decreased by 0.27 and 0.20 points for every millimeter
of increase in anterior and posterior femoral offset, respectively. Posterior
offset alone was related to range of motion, with a small positive correlation
being observed (0.12° of flexion obtained for every millimeter of increase
in posterior offset, p = 0.0492).
Simple correlations of stuffing measurements and
preoperative-to-postoperative changes in outcomes were consistent with the
findings of regression analysis (Table
II). The data in Table
II largely mirror the findings of regression analysis shown in
Table I, indicating that
preoperative outcome scores did not selectively favor "stuffed" or
"unstuffed" knees. Discrepancies between the results in Tables
I and
II are accounted for by the
difference between the regression and correlation procedures. The former
controlled for the effects of covariates such as gender, age, body mass index,
and preoperative range of motion, whereas the latter did not.
To further test and confirm the regression findings, we defined stuffing as
a 15% increase in anterior patellar displacement and compared mean outcome
scores among "stuffed" and "unstuffed" groups on the
basis of this definition (Table
III). The absence of significant differences between these groups
supports the results of our regression analyses as presented in
Table I. Mean outcome scores
also were examined with stuffing defined as a 15% increase in combined
anterior patellar displacement and anteroposterior femoral size
(Table IV). Contrary to the
regression findings that indicated no changes in knee and pain scores as this
measure of stuffing increases, the Knee Society Knee Score (mean, 97.5 for
nineteen stuffed knees as compared with 94.1 for 769 unstuffed knees; p <
0.0001) and the Knee Society Pain Score (mean, 49.7 for nineteen stuffed knees
as compared with 47.5 for 769 unstuffed knees; p < 0.0001) were
significantly higher for the stuffed knee group. Large differences in the
number of knees in these two study groups (nineteen compared with 769) and
differences in the sensitivity of descriptive and regression procedures may
account for this discrepancy.
An increase in anterior patellar displacement was associated with a lower
probability of the need for a lateral retinacular release (odds ratio =
0.50/cm; p = 0.0022) (Table V).
Furthermore, an increase in measured anteroposterior femoral size was
associated with a higher probability of the need for a lateral retinacular
release (odds ratio = 1.9/cm; p = 0.0010)
(Table V). These relationships,
even when combined, explained only 10.1% of the observed variance in the need
for a lateral release.
Increases in anterior femoral offset also increased the odds of lateral
release (odds ratio = 2.2/cm; p = 0.0006)
(Table V). Measurement B
(anteroposterior femoral size) overlaps and encompasses measurement C
(anterior femoral offset). The two measures were analyzed in separate
regressions to ensure that the effect of length C was uniquely represented.
The results indicated that anteroposterior femoral size (odds ratio = 1.9/cm;
p = 0.0010) and anterior femoral offset (odds ratio = 2.2/cm; p = 0.0006)
equivalently increased the likelihood of lateral release.
Table VI demonstrates
additional factors associated with the probability of lateral retinacular
release. The forward/stepwise selection procedure of the logistic regression
revealed that gender, large as opposed to medium patellar size, and absolute
femoral component size were more important than either anterior patellar
displacement or measured anteroposterior femoral size
(Table VI).
There were no differences in outcome between the AGC or LPS knee systems
with regard to the Knee Society Knee Score, the Knee Society Function Score,
the Knee Society Pain Score, or the need for lateral release (p = 0.9949, p =
0.7898, p = 0.0851, and p = 0.0612, respectively). There was a +4.4°
difference in the postoperative range of motion for the LPS prosthesis
compared with the AGC prosthesis (p < 0.0001), and this effect was
accounted for in the analysis. Interactions between stuffing measures that had
a significant effect on range of motion and the use of either prosthesis were
not significantly related to range of motion (anterior patellar displacement,
p = 0.5755; anteroposterior femoral size, p = 0.0924).
Increasing the anteroposterior size of the patella or the distal part of
the femur, or both, during total knee arthroplasty is referred to as stuffing
the patellofemoral
joint1,2.
Some authors have observed that stuffing of the patellofemoral joint adversely
affects the outcome of knee
arthroplasty1-7,9-11.
Shoji et al. reported that 17% of 141 knees with a =10% increase in
anteroposterior patellar thickness after total knee arthroplasty achieved
>120° of flexion postoperatively, compared with 50% of ninety knees
with a <10% increase in patellar
size4. Ryu et al.
observed that postoperative patellar thickness was significantly greater in
twenty-one knees with flexion of =90° after total knee arthroplasty
than in twenty-nine knees with flexion of
=120°10.
Daluga et al. reported an increased need for manipulation following total knee
arthroplasty in ten of ten patients with a postoperative increase of 12% in
combined patellofemoral anteroposterior
size3. Koh et al.
compared fifty-six patients with residual patellar bone thickness of =12 mm
with sixty-six patients with residual patellar bone thickness of >12
mm12. Postoperative
differences in the Knee Society Score, Knee Society Function Score, and range
of motion were not observed, but the authors noted that a higher rate of
patellar complications may occur when the patellofemoral articulation is
"excessively" increased. It is important to note that those
studies involved relatively small numbers of patients and
knees3,4,10,12.
Our study evaluated the effect of stuffing the patellofemoral compartment
on the outcome of total knee arthroplasty in a large sample of 556 patients
and 830 knees. For every millimeter of increase in anterior patellar
displacement, we observed a corresponding 0.18° decrease in range of
motion (p = 0.0079) and a 0.52-point increase in the Knee Society Function
Score (p < 0.0001). It is possible that anterior displacement shifted the
extensor mechanism, limiting its full excursion and adversely affecting range
of motion, and improved the efficiency of the quadriceps mechanism by
increasing the moment arm of the quadriceps. Clinically, however, our findings
indicate that even a 1-cm change in anterior patellar displacement, which
rarely (if ever) occurs, could result in only a 1.8° change in the range
of motion and a 5.2-point change in the function score. For every millimeter
of increase in anteroposterior femoral size, the range of motion increased by
0.14° (p = 0.0182) and the Knee Society Function Score decreased by 0.23
points (p = 0.0094). Combined anterior patellar displacement and
anteroposterior femoral size had no effect on outcomes. Both anterior femoral
offset and posterior femoral offset had small negative effects on function
scores. Posterior offset alone was related to range of motion, with a small
positive correlation being observed. None of the measures were associated with
significant changes in the Knee Society Knee Score or the Knee Society Pain
Score.
The patellar components used in the present study ranged in thickness from
7.5 to 10 mm, and our attempts to resolve even a 5° loss in range of
motion due to anterior patellar displacement was only marginally successful (p
= 0.0512, F-test, H0: ß = 5). Similarly, an effect size of
=5 points in the Knee Society Function Score for a 1-cm change in anterior
patellar displacement or anteroposterior femoral size was equally unattainable
in this population (p < 0.0001, F-test, H0: ß = 5).
Thus, the significant effects that we observed do not appear to be clinically
relevant in terms of patient outcomes.
The probability of the need for lateral retinacular release decreased as
the anterior patellar displacement increased and as anteroposterior femoral
size increased. The latter finding is consistent with the concept that
increasing anteroposterior femoral size tightens the lateral retinaculum and
increases the likelihood of lateral patellar subluxation and tilt. However,
increasing anterior patellar displacement theoretically produces a similar
negative effect on patellar tracking, and yet we found that an increase in
anterior patellar displacement was associated with a lower probability of the
need for a lateral release. Moreover, the probability of needing a lateral
release was more significantly related to gender, large as opposed to medium
patellar size, and absolute femoral component size than it was to either
anterior patellar displacement or anteroposterior femoral size. This
observation suggests that the probability of requiring a lateral release most
likely involves a more complex set of factors than just stuffing of the
patellofemoral joint.
The present study was limited by its retrospective design, the potential
for measurement errors on the radiographs, the inability to correct for
patellar erosion and the effect of patellar tilt and/or subluxation on
measured anterior patellar displacement, and the absence of an outcome survey
specific to patellofemoral problems and function. Multiple surgeons performed
the operations but, with the exception of the two different implant designs,
all surgeons employed standardized procedures and approaches to patient care
based on shared physician practice protocols and evidence-based standards of
care. Although the present study reflects a retrospective review of historical
cases and not a prospective collection of data from new cases, the outcome
data were collected by the operating surgeons and are therefore subject to
observer bias.
Our study evaluated the outcomes associated with a posterior
cruciate-retaining prosthesis (AGC Knee) and a posterior cruciate-substituting
prosthesis (LPS Knee). With the numbers available, the analysis did not
indicate differences in outcomes with the exception of range of motion. Thus,
we believe that our findings are relevant to both cruciate-retaining and
cruciate-substituting total knee arthroplasty designs.
Our data do not support the concept that stuffing of the patellofemoral
joint is associated with adverse outcomes. As discussed earlier, it is not
likely that any of the small effect sizes that we observed would result in
meaningful changes in flexion, function, or pain postoperatively. Although we
still advocate the practice of attempting to reproduce anterior patellar
displacement, anteroposterior femoral size, and the combined patellofemoral
anteroposterior size during a total knee arthroplasty, our data show that
precision of reproduction of these sizes did not affect the outcome of the
procedure.
The findings of the present study have important clinical implications. In
the evaluation of a patient who has pain or stiffness following a total knee
arthroplasty, caution should be exercised in attributing these problems to
stuffing of the patellofemoral joint even if radiographs show evidence of an
increase in anterior patellar displacement, anteroposterior femoral size, or
the combination of anterior patellar displacement and anteroposterior femoral
size. Because our data do not demonstrate a relationship between stuffing and
pain, we do not recommend a revision to correct stuffing of the patellofemoral
joint in the absence of another identifiable cause of the pain, such as
prosthetic loosening, osteolysis, infection, or instability. ?