Whether to resurface the patella in a total knee arthroplasty remains a
controversial
topic1-8.
Some authors have reported that patellar resurfacing improved pain relief and
the ability to climb
stairs2,3,7,8.
However, since patellar resurfacing is a source of early and late
complications2,9,10,
there has been a renewed interest in not resurfacing the patella. There is a
tendency for surgeons to resurface the patella in patients with inflammatory
arthritis7,8,11.
However, recently published prospective, randomized trials have revealed
conflicting results of patellar resurfacing in patients with
osteoarthritis4,6,7.
Previous studies of patellofemoral problems after total knee arthroplasty
have focused on the prevalence of prosthetic revision, patellar complications,
and functional
results1-10.
Long-term changes of the nonresurfaced patella and their relationship with
clinical outcomes have seldom been studied. Laughlin et
al.12 reported that
59% of eighty-nine resurfaced patellae were malaligned after total knee
arthroplasty in patients with osteoarthritis, but the relationship between
patellar maltracking and clinical outcome was not studied. Smith et
al.13 found that
patellofemoral symptoms, mostly due to patellar malalignment, occurred in 8%
of 112 knees treated with total knee arthroplasty without patellar
resurfacing. However, most of the patients in their study had rheumatoid
arthritis. Kawakubo et
al.11 reported few
patellar changes and little peripatellar pain after seventeen total knee
arthroplasties performed in patients with osteoarthritis.
Initial enthusiasm for patellar resurfacing in our institute was tempered
in the early 1990s by a high complication rate associated with metal-backed
patellar components. The attitude shifted toward not resurfacing the patella
except in patients with a severely degenerated patellofemoral joint or
inflammatory arthritis. However, we have occasionally observed gradual
deterioration of nonresurfaced patellae. The purposes of this study were to
examine the long-term radiographic changes of the nonresurfaced patella after
total knee arthroplasty, to correlate these changes with functional results,
and to identify risk factors predisposing to the development of these
changes.
Patients were eligible for inclusion in the study if they had had a primary
total knee arthroplasty without patellar resurfacing performed for the
treatment of osteoarthritis between 1992 and 1993 and had been followed
regularly for more than five years. Only patients who had had a total knee
arthroplasty with the Porous Coated Anatomic knee system (PCA; Howmedica,
Rutherford, New Jersey) were included. The basic design of the PCA femoral
component is characterized by an asymmetrical laterally raised flange with an
anatomically angled patellofemoral groove. One hundred and eighty-seven
patients with 235 total knee arthroplasties met the inclusion criteria. There
were twenty-one men and 166 women. The mean age of the patients at the time of
surgery was sixty-five years (range, fifty-six to seventy-eight years), the
mean body height was 152 cm (range, 136 to 176 cm), and the mean body weight
was 66 kg (range, 48 to 96 kg). The mean follow-up period was 8.5 years
(range, five to ten years).
The surgical technique was the same in all patients. A conventional medial
parapatellar approach was
used14. The femoral
component was externally rotated 3° and inserted without cement. The
tibial component was centered on the tibial tubercle and inserted with cement.
Treatment of the patella included removal of peripheral osteophytes,
chondroplasty by shaving the fibrillated cartilage with a power shaver, and
drilling of the eburnated bone. A lateral release was performed if the patella
did not track centrally as demonstrated by the so-called no-thumb
test15.
Functional Evaluation
Following the operation, patients were evaluated at six weeks, three
months, six months, one year, and annually thereafter. The same two nurses
performed all evaluations, without involvement of a physician. The functional
results were assessed with use of the Knee Society Clinical Rating
System16. In
addition, patients were carefully questioned regarding the location of any
knee pain and the ability to climb stairs, and they were observed while rising
from a chair.
Radiographic Evaluation
Postoperative radiographs, including standing hip-to-ankle anteroposterior,
lateral, and 45° Merchant
views17, were made
at three days, six weeks, three months, six months, and one year after the
operation and then yearly. Radiographic evaluation was performed by an
orthopaedic surgeon (L.-Y.S.) and a radiologist (Y.-C.W.) who had no knowledge
of the clinical data, treatment courses, or outcomes. Eight knees (six
patients) were excluded because of incongruency between the observers'
evaluations, leaving 227 knees (181 patients) for further analysis.
Femorotibial alignment was measured preoperatively and postoperatively on
standing hip-to-ankle anteroposterior radiographs. Patellar tracking was
assessed by measuring patellar displacement and tilt as described by Grelsamer
et al. (Fig.
1)18.
More than 5 mm of displacement or 5° of tilt was considered
abnormal11-13,18-21.
A congruence
angle17 of greater
than +16° was defined as patellar subluxation. The presence of
osteophytes, a decrease in cartilage thickness, and subchondral sclerosis of
the patella were also recorded.
Five patterns of temporal changes of the patellofemoral joints were
observed, including (1) persistent central tracking with preservation of the
cartilage thickness, (2) early-onset lateral tilt and/or displacement (seen on
the six-week or three-month postoperative radiographs)
(Fig. 2), (3) progressive loss
of lateral cartilage thickness with lateral displacement and/or tilt
(Fig. 3), (4) slowly
progressive lateral subluxation, and (5) persistent medial tilt. Every
radiographic series was graded as showing normal findings (neutral tracking
with preserved cartilage thickness) or abnormal findings (lateral tilt,
displacement, or subluxation) for further analysis.
Statistical Analysis
Statistical analysis was performed with the SPSS for Windows statistical
package (version 10.0; SPSS, Chicago, Illinois, 1999). The occurrence of each
clinical and radiographic variable was compared between the normal and
abnormal patellar groups with use of univariate analysis and then was examined
with use of a multiple logistic regression model to determine the independent
factors leading to the development of an abnormal patella postoperatively. The
association between functional variables (anterior knee pain, ability to climb
stairs, and ability to rise from a chair) and the postoperative status of the
patellofemoral joint was determined with use of the chi-square test or the
Fisher exact test when applicable. A two-tailed p value of <0.05 was used
to reflect findings not attributable to chance.
Radiographic Results
Preoperatively, 186 (82%) of the patellae tracked centrally, thirty-nine
(17%) tilted or displaced laterally, and two (1%) subluxated laterally. On the
immediate postoperative radiographs, 215 (95%) of the patellae tracked
centrally, five (2%) tilted or displaced laterally, three (1%) subluxated
laterally, and four (2%) tilted medially. Fifteen (7%) of the patellae that
had tracked centrally on the immediate postoperative radiographs showed
lateral tilt and/or displacement on radiographs made at six weeks or three
months after the operation. The quality of patellar tracking deteriorated very
rapidly in these fifteen patients. At the time of the latest follow-up, 133
(59%) of the patellae still tracked centrally with preservation of the
thickness of the patellofemoral cartilage, fifteen (7%) showed early lateral
tilt, sixty-eight (30%) showed progressive loss of the lateral cartilage
thickness with lateral tilt and/or displacement, seven (3%) had progressed to
lateral subluxation, and four (2%) tilted medially.
Analysis of Risk Factors for Postoperative Abnormality of the
Patellofemoral Joint
The preoperative femorotibial alignment was >15° of varus in six
knees, =15° of varus in 185 knees, neutral in eighteen knees,
=15° of valgus in sixteen knees, and >15° of valgus in two
knees. The average postoperative femorotibial alignment (and standard
deviation) was 5° ± 3° of valgus. With the numbers available,
neither the preoperative nor the postoperative femorotibial alignment differed
significantly between the normal and abnormal patellar groups (p = 0.89 and
0.91, respectively). The preoperative flexion arc ranged from 60° to
150° (mean, 115° ± 18°) with an average extension lag of
5.5°. The postoperative flexion arc ranged from 85° to 130° (mean,
105° ± 17°). Neither the preoperative nor the postoperative
flexion arc differed significantly between the two patellar groups (p = 0.74
and 0.76, respectively). Age, gender, body height, and body weight also did
not significantly affect the patellar changes (p = 0.20, 0.16, 0.65, and 0.50,
respectively).
Multiple logistic regression analysis identified preoperative patellar
maltracking as the only independent factor associated with an abnormal
patellofemoral joint postoperatively (relative risk, 2.7; 95% confidence
interval, 2.21 to 3.30; p = 0.003). All 186 knees with a normal patella
preoperatively retained central tracking after surgery, and 133 (72%) retained
it at the time of the latest follow-up. All but one (98%) of the forty-one
knees with an abnormal patellofemoral joint preoperatively demonstrated
progressive degenerative changes on the lateral facet and had progression to
lateral displacement, tilt, or even subluxation of the patella.
Correlation of Functional Results and Radiographic Findings
Preoperatively, the forty-one knees with an abnormal patellofemoral joint
had a mean knee score of 67 ± 7 points and a mean functional score of
58 ± 7 points, whereas the 186 knees with a normal patellofemoral joint
had a mean knee score of 68 ± 10 points and a mean functional score of
61 ± 9 points. An abnormal patellofemoral joint postoperatively did not
significantly affect the knee and functional scores of the total knee
arthroplasty (p = 0.90 and 0.89, respectively). Both groups of patients showed
satisfactory improvement in function. In the group with a normal patella
postoperatively, the mean knee score was 91 ± 8 points and the mean
functional score was 89 ± 7 points at the time of the latest follow-up.
In the group with an abnormal patella, these scores were 90 ± 8 and 87
± 8 points, respectively.
At the time of the latest follow-up, twenty-two (10%) of the 227 knees were
reported to be mildly to moderately painful anteriorly, requiring either no
medication or occasional use of nonsteroidal anti-inflammatory drugs. The
patients who had anterior knee pain did not differ significantly from those
who did not have such pain with regard to age, gender, or weight, with the
numbers available. Eleven (6%) of the 186 knees with a normal patella
preoperatively, in contrast to eleven (27%) of the forty-one with an abnormal
patella preoperatively, were painful anteriorly. Four (3%) of the 133 knees
with a normal patella postoperatively, in contrast to eighteen (20%) of the
ninety with an abnormal patella postoperatively (excluding the four patients
with medial tilt), were painful anteriorly. These differences were significant
(p < 0.001 for both).
Fifty-six (25%) of the 227 knees caused difficulties with rising from a
chair. The patients who had such difficulties did not differ from those
without such difficulties with regard to age, gender, or weight, with the
numbers available. Thirty (16%) of the 186 knees with a normal patella
preoperatively, in contrast to twenty-six (63%) of the forty-one with an
abnormal patella preoperatively, caused difficulties with rising from a chair.
Eight (6%) of the 133 knees with a normal patella at the time of the latest
follow-up, in contrast to forty-eight (53%) of the ninety knees with an
abnormal patella at the time of the latest follow-up, caused difficulties with
rising from a chair. The differences were significant (p < 0.001 for
both).
Sixty-two (27%) of the 227 knees caused difficulties with climbing stairs.
The patients who had this problem did not differ significantly from those who
did not have it with regard to age, gender, or weight, with the numbers
available. Thirty-four (18%) of the 186 knees with a normal patella
preoperatively, in contrast to twenty-eight (68%) of the forty-one with an
abnormal patella preoperatively, caused this problem. Ten (8%) of the 133
knees with a normal patella at the time of the latest follow-up, in contrast
to fifty-two (58%) of the ninety with an abnormal patella at the time of the
latest followup, caused difficulties with climbing stairs. The differences
were significant (p < 0.001 for both).
The question of whether patellar tracking or the patellofemoral joint
deteriorated with time in patients treated with a total knee arthroplasty
without resurfacing of the patella has been debated. Smith et
al.13 reported
that, at an average of 6.5 years, patellar tracking was maintained with only a
small shift of patellar position in patients with rheumatoid arthritis and the
shift tended to stabilize with time. Kawakubo et
al.11 reported that
the patella became flattened and was associated with peripatellar pain in
patients with rheumatoid arthritis. However, patellar flattening was seldom
observed in patients with osteoarthritis and was not associated with
peripatellar pain. Laughlin et
al.12 reported that
laterally tilted patellar components became more laterally tilted with time.
Our study showed that, after a mean duration of follow-up of 8.5 years,
patellar tracking and the patellofemoral joint remained normal in about 60% of
knees with a nonresurfaced patella. Gradual loss of cartilage thickness in the
patellofemoral joint, osteophyte formation, and deterioration of patellar
tracking were the most common abnormal radiographic changes (found in
sixty-eight [30%] of the patellae). Our observations support those of Laughlin
et al. We believe that the temporal changes are comparable with the
degenerative changes of the natural patellofemoral joint due to abnormal
stress distribution; once those changes are initiated, the patella will
deteriorate slowly but relentlessly.
The early onset of lateral tilt and/or displacement of the patella was
found in fifteen patients. This change was attributed to gradual loosening of
the capsular sutures that compromised the healing of the conventional medial
parapatellar arthrotomy wound. A classic medial parapatellar
approach14 includes
incision of part of the quadriceps tendon at its insertion to the patella,
which may destabilize patellar tracking. The repair site may stretch or relax
with
time22,23.
The natural kinematics of patellar tracking favor a lateral traction force
because of the strong vastus lateralis muscle, which exaggerates the tracking
abnormality. We later modified our approach to incise the extensor mechanism
along the most medial aspect of the quadriceps tendon, and we carefully
repaired the arthrotomy wound with nonabsorbable suture materials to avoid
this pattern of abnormal patellar tracking.
In this study, abnormal patellar tracking preoperatively was identified as
the only independent factor predisposing to the development of patellar
maltracking postoperatively. This finding is consistent with those of
Bindelglass et
al.19 and Chan and
Gill20. However,
Kawano et al.24
reported that preoperative patellar tilt had no effect on patellar tracking
when the patella had been resurfaced. They found that medialization of the
patellar component on the patella and an oblique bone resection of the patella
were effective in achieving proper tracking following total knee
arthroplasties with patellar resurfacing. However, these techniques involving
the patella (oblique bone resection and medialization of the patellar
component) are not applicable to total knee arthroplasties without resurfacing
of the patella. In our series of total knee arthroplasties without patellar
resurfacing, forty (98%) of forty-one knees with preoperative maltracking of
the patella demonstrated progressive degenerative changes of the
patellofemoral joint and/or patellar maltracking postoperatively. On the basis
of this observation, we suggest resurfacing of the patella for patients with
preoperative maltracking.
Degenerative changes of the patellofemoral joint and/or patellar
maltracking remain a major source of disability following total knee
arthroplasty. The correlation of clinical results and patellar tracking had
been discussed
extensively4,6,13,19.
Bindelglass et
al.19 reported no
relationship between patellar tilt and pain, flexion, or fixation. In our
study, the Knee Society knee and functional scores were not adversely affected
by the radiographic abnormalities. However, some patients did have symptoms
and signs related to the patellofemoral joint. Mild anterior knee pain was the
most frequently mentioned symptom, which was attributed to gradual
degenerative changes of the patellofemoral joint with erosion of cartilage and
surface incongruity. The true number of painful knees may be greater since
many patients adjusted to these symptoms.
We relied heavily on the Merchant view to assess the changes of the
patellofemoral joint and patellar
tracking17.
Radiographic evaluation of this joint is difficult, as it is in the normal
knee21. The
information yielded depends greatly on the technique of obtaining this view.
Despite the use of a standard frame and experienced x-ray technicians, some
Merchant radiographs were made with the knee in different positions. The
effect of radiographic variation was minimized by observing the pattern of
temporal changes in each patellofemoral joint.
The results of this study are specific to the PCA total knee system.
Recently developed prostheses designed to be more compatible with the native
patella may have different results.
Note: The authors gratefully acknowledge the valuable
contribution of Yen-Chiu Liang, RN, and Hui-Ling Chu, RN, who performed the
assessments of the functional results and kept the records.