From January 1984 through August 1989, seventy-two consecutive
posterior cruciate-retaining total knee arthroplasties were performed
in fifty-one patients with a primary diagnosis of rheumatoid arthritis.
There were thirty-nine women and twelve men. The mean age at the
time of the arthroplasty was fifty-seven years (range,
thirty to seventy-eight years). Twenty-one patients underwent
bilateral total knee arthroplasty; thirteen of these patients had
simultaneous procedures and eight had staged procedures. Eighteen
patients (twenty-four knees) died before a minimum of eight years
of clinical follow-up. Although these twenty-four
knees are included in the survivorship analysis, they are not included
in the analysis of the clinical and radiographic results after a
minimum follow-up of eight years. Two knees in one patient were
lost to follow-up, leaving forty-six knees in
thirty-two patients available for evaluation at a mean
of 10.5 years (range, eight to fourteen years).
The relatively unconstrained Miller-Galante I prosthesis (Zimmer,
Warsaw, Indiana) was used in all patients (Figs. 1-A and 1-B). The tibial component
featured a Ti6Al4V-alloy tray with four small fixation pegs and
a posterior recess to accommodate the retained posterior cruciate
ligament. The tibial component was inserted either with cement (in
which case it was precoated with polymethylmethacrylate) or without
cement (in which case it was fixed with screws that were placed through
holes in the fixation pegs). The polyethylene tibial insert had
a relatively flat articular surface. The femoral component also
was made of Ti6Al4V and could be inserted either with or without
cement. Overall, forty-eight arthroplasties were performed with
cement (Simplex-P; Howmedica, Rutherford, New Jersey), twenty-one
were performed without cement, and three were hybrid procedures
(two involved a cemented tibial and a cementless femoral component,
and one involved a cemented femoral and a cementless tibial component).
All patellar components were metal-backed (Ti6Al4V); fifty were
inserted with cement and twenty-two were inserted without cement.
All procedures were performed through a medial parapatellar arthrotomy.
A bone island was preserved around the tibial insertion of the posterior
cruciate ligament. The posterior cruciate ligament was not recessed
or otherwise balanced. The tibial cut was made with 10° of posterior
slope with use of an extramedullary guide. The femoral component
was slightly undersized, when possible, to avoid placing excessive
tension on the posterior cruciate ligament in flexion. A synovectomy was
not performed routinely.
Physical therapy for range of motion and walking was initiated
on the first postoperative day. Continuous passive-motion machines
were not used. The postoperative weight-bearing protocol prescribed
full weight-bearing for knees with cemented and hybrid components
and protected weight-bearing for six weeks for those with cementless
prostheses. Perioperative antibiotics were administered to all patients
for forty-eight hours, and warfarin was administered for four weeks
for thromboembolic prophylaxis.
Clinical and Radiographic Analysis
The clinical evaluation, which included an examination of the knee
and completion of the Hospital for Special Surgery knee-evaluation
form15, was conducted preoperatively
and at six weeks, three months, six months, one year, and annually
thereafter. Mediolateral and posterior ligamentous stability were assessed
by means of a manual examination, and the posterior drawer test
and the posterior sag test were used to assess posterior cruciate
ligament integrity.
Weight-bearing anteroposterior, non-weight-bearing
lateral, and Merchant radiographs of the involved knee were made preoperatively
and at each follow-up visit (at six weeks, three months,
six months, one year, and annually thereafter). The radiographs
were analyzed for component position and sizing, radiolucent lines,
and osteolysis. The alignment of the femoral component was determined
by measuring the angle between the distal condylar surfaces and
the shaft of the femur on anteroposterior and lateral radiographs.
Similarly, the position of the tibial component was determined by
measuring the angle between the surface of the plateau and the shaft
of the tibia on anteroposterior and lateral radiographs. The position
of the patellar component was assessed by measuring the angle between
the cut surface of the patella and a line drawn along the medial
and lateral anterior extensions of the trochlear groove on the skyline
radiograph. Medial or lateral displacement of the patellar component
from the trochlear groove was recorded as a measure of patellar
tracking. Sizing of the tibial component was assessed on the anteroposterior
radiographs and on the basis of whether the edges of the component
lined up with the edges of the cut tibial surface, overhung the
edges of the bone (indicating an oversized component), or did not
reach the edges of the bone (indicating an undersized component).
Radiolucent lines were characterized by width (<1 mm, 1
to 2 mm, or >2 mm), and they were classified by location
with use of a nine-quadrant zoning system for the femur, tibia,
and patella. Osteolysis was evaluated on all radiographs, with the
findings on the preoperative or early postoperative radiographs
used as a baseline.
An attempt to determine posterior cruciate ligament integrity was
made by examining plain radiographs. The tibiofemoral contact point,
defined as the point at which the femoral component contacted the
articular surface of the tibial polyethylene, was identified on
non-weight-bearing lateral radiographs made with the knee in 45°
of flexion. The distance from the posterior edge of the tibial component
to the tibiofemoral contact point was measured and expressed as
a percentage of the total width of the tibial component (from the
anterior to the posterior edge) (Fig. 2). The resulting contact-point ratio
was calculated on the initial postoperative and latest follow-up
radiographs.
Survivorship analysis was performed with use of the Kaplan-Meier
method. Chi-square tests with Yates and Bonferroni corrections were
used for comparisons of partial radiolucencies and contact-point
ratios.
The average Hospital for Special Surgery score increased from
42 points (range, 13 to 75 points) preoperatively to 87 points (range,
46 to 100 points) at the final follow-up examination. Of
the forty-six knees that were followed for at least eight years,
thirty (65%) had an excellent result; fourteen (30%),
a good result; one (2%), a fair result; and one, a poor result.
The one poor result was for a patient who required a contralateral
below-the-knee amputation. At the final follow-up examination, twenty-nine
knees (63%) were pain-free, twelve (26%)
were slightly painful during long walks, three (7%) were
moderately painful during stair-climbing, two (4%) were
painful during walking on a level surface, and none were painful
at rest. Thirty-five knees (76%) caused no limp, eight
(17%) caused a slight limp, two (4%) caused a
moderate limp, and one (2%) caused a severe limp. Twenty-eight
knees (61%) were in patients who did not require a cane,
eight (17%) were in patients who required a cane only for
long walks, six (13%) were in patients who required a cane full-time,
and four (9%) were in patients who required a walker. No
patient required two canes. All but two knees demonstrated anterior-posterior
stability in flexion, with negative results on the posterior drawer
and posterior sag tests. Two knees in one patient had posterior
instability on clinical examination.
Total knee flexion averaged 98° (range, 5° to 135°) preoperatively
and 104° (range, 45° to 130°) at the final follow-up examination.
Flexion contractures decreased from a mean of 7° (range, 0° to 35°)
preoperatively to a mean of 1° (range, 0° to 10°) at the final follow-up.
Nine (13%) of the original seventy-two knees
had revision of the implant. Six knees had revision of the patellar
component because of delamination of the metal backing. The six
patellar failures occurred at a mean of fifty-nine months
(range, twenty-three to 116 months) postoperatively. One of these
six knees also had revision of a well-fixed femoral component because
of surface abrasions resulting from metal-on-metal contact
with the patellar component. Three additional knees had revision
of a well-fixed component. Specifically, one knee had revision of
the tibial component because of pain of unknown cause at seventeen
months, one knee had an arthrodesis because of severe wound complications
at three months, and one knee had removal of the femoral component
and insertion of an allograft-prosthesis composite because of a periprosthetic
femoral fracture at 123 months. There were no revisions for aseptic
loosening.
There were three additional repeat operations that did not involve
component revision. Two separate patellar-realignment procedures
were performed in one knee because of patellar subluxation, and
one below-the-knee amputation was performed one week postoperatively
because of vascular insufficiency in a patient who had preexisting
peripheral vascular disease.
No revisions were performed because of tibiofemoral instability.
Two knees in one patient were found to have posterior instability
secondary to posterior cruciate ligament deficiency on both clinical
examination and lateral radiographs (Figs. 3-A and 3-B). At the time of the primary knee
arthroplasties, the posterior cruciate ligament had been intact
bilaterally. The patient had progressive evidence of anterior femoral
translation on flexion radiographs, beginning at the forty-eight-month
follow-up examination. She first demonstrated clinically
symptomatic posterior instability and a positive posterior drawer
sign at the ten-year follow-up examination. Although the patient’s weight
of 340 lb (154 kg) may have contributed to the symptoms, there was
no antecedent trauma. The patient’s knee score was 74 points
bilaterally at the last evaluation (at ten years), compared with
25 points preoperatively. The patient required a walker for household
ambulation (walking inside the home). Although a revision procedure
was offered, it was declined.
The ten-year survival rate was 81% ± 5% with
any reoperation as the end point, 93% ± 4% with
femoral or tibial revision for any reason as the end point, 88% ± 5% with patellar revision for any reason
as the end point, and 100% with femoral or tibial revision
for aseptic loosening as the end point.
Eighteen patients (twenty-four knees) died before a minimum follow-up
of eight years, and one patient (two knees) was also lost to follow-up.
The patients in this group were followed for a mean of thirty-five
months (range, three to eighty-four months). At the time of the
last follow-up examination, ten knees (38%) had an excellent
result; ten (38%), a good result; three (12%),
a fair result; and three (12%), a poor result. The mean
Hospital for Special Surgery score increased from 35 points (range,
11 to 59 points) preoperatively to 74 points (range, 31 to 97 points)
at the final examination. There were no cases of instability, no
revisions, and no repeat operations in this group at the time of
the last follow-up.
Radiographic Results
A complete series of radiographs was available for thirty-six of
the forty-six knees that were followed for at least eight years.
The mean duration of radiographic follow-up in this group
was 10.5 years (range, eight to fourteen years). At the time of
the latest follow-up, the femoral component was in a mean
of 5° of valgus (range, 1° to 8° of valgus) and 2° of flexion (range,
4° of extension to 8° of flexion). The tibial tray was in a mean
of 1 of varus (range, 6° of varus to 3° of valgus) and had a mean
of 8° of posterior slope (range, 2° to 13° of posterior slope).
The tibial tray was oversized in four knees and undersized in none.
The patellar component had a mean of 1 mm of lateral displacement
(range, 5 mm of medial displacement to 8 mm of lateral displacement)
and 4° of lateral tilt (range, 0 to 18 of lateral tilt).
There were no complete radiolucent lines around any component
in any knee, all radiolucent lines measured less than 2 mm in width,
and no partial radiolucent lines were noted to be progressive. No
component was radiographically loose, and there was no migration
or subsidence of any component in any knee. None of the patellar
fixation pegs had failed.
The mean tibiofemoral contact point, expressed as a percentage
of the width of the tibial tray as measured from the posterior to
the anterior margin, was 47% (range, 22% to 61%)
on the initial postoperative lateral radiographs and 45% (range, 28% to
98%) on the most recent radiographs. This difference was
not significant (p = 0.61). Only two knees (one patient) demonstrated
a change of more than 20%. One of these knees progressed
from 41% to 79%, and the contralateral knee progressed
from 54% to 98%. These radiographic findings were seen
in the one patient who had clinical evidence of bilateral posterior
instability secondary to deficiency of the posterior cruciate ligament.
A complete series of postoperative radiographs was available for
nineteen of the twenty-six knees in the nineteen patients who
died or were lost to follow-up before the eight-year examination.
No complete radiolucent lines or osteolytic lesions were seen on
the radiographs of any of these knees. Two patients who had a cementless
prosthesis had an incomplete radiolucent line around the tibial
component; these lines were nonprogressive and measured <2
mm in width. No femoral or patellar radiolucencies were identified.
Eight patients had both early postoperative and follow-up radiographs
made with the knee in 45° of flexion. None of these patients had
a change of >20% in the contact-point ratio.
Although many studies of posterior cruciate-retaining total knee
arthroplasty in patients with rheumatoid arthritis have demonstrated
good results9-13, the issue of
late posterior instability remains a concern14,16,17.
The purpose of the present prospective study was to analyze the
results of posterior cruciate ligament-retaining total knee arthroplasty
in patients with rheumatoid arthritis. We found that the procedure
yielded excellent clinical and radiographic results, with a ten-year
survival rate of 93% ± 4% with
tibial or femoral revision or loosening used as the end point. The
ten-year survival rate with use of any reoperation as the end point
was 81% ± 5%. Only two
knees in one patient had posterior instability secondary to late
posterior cruciate ligament insufficiency. We therefore believe
that posterior cruciate-retaining total knee arthroplasty remains
an excellent treatment option for patients with rheumatoid arthritis.
In this series, the most common mode of failure (accounting for
six of the nine revision procedures) involved the patellar component.
Several factors affected the patellofemoral articulation. One was
the metal backing of the patellar component, with resultant decreased
polyethylene thickness and increased edge-loading, leading to delamination
and failure18. In addition, at
the time that these operations were performed, the standard surgical
technique involved rotationally aligning the femoral component parallel
to the posterior condyles. This technique led to internal rotation
of the femoral component, with resultant lateral maltracking of
the patellar component. The femoral component also had a more prominent
anterior patellar-flange profile, which has been shown to result
in increased quadriceps moment in late stance and thus greater patellar
loads19. Today, these are well-recognized
causes of patellofemoral complications and failure. The guides that
we currently use for the femoral component include the epicondylar
axis, 3° of external rotation from the posterior condylar line,
and the anteroposterior axis of Whiteside20.
Radiographs demonstrated that the tibiofemoral contact point was
anteriorly displaced by >20% of the width of the
tibial tray in only two knees (one patient), both of which were
also found to be posteriorly unstable on clinical examination. However,
no revision procedures were performed because of instability and
none of the other knees in the study had clinical or radiographic
signs of instability. Thus, in this group of patients with rheumatoid
arthritis who were followed for a mean of 10.5 years (or until death),
the rate of flexion stability secondary to posterior cruciate ligament
deficiency was 2.8% (two of seventy-two knees) or 2.0% (one
of fifty-one patients). In our study of patients without rheumatoid
arthritis who were followed for at least eight years (or until death),
the rate of revision for posterior instability was 2.0% (three
of 151 knees) or 2.1% (three of 140 patients)21. Therefore, we found that the prevalence
of late posterior instability in patients with rheumatoid arthritis
was very low and was similar to that in patients without rheumatoid
arthritis.
The reported prevalence of late posterior instability of the knee
in patients with rheumatoid arthritis who are treated with a posterior
cruciate ligament-retaining prosthesis varies considerably. Studies
ranging in size from fifty-five to 367 knees have demonstrated excellent
results, with a rate of posterior instability of 0% to
0.5% after four to eleven years of follow-up10,11,22,23. Other studies ranging
from twenty-five to 202 knees have identified a much higher prevalence
of 8% to 50% at intermediate-term follow-up14,16,17. The reasons for the differing
rates of late instability reported in the literature are not known;
however, they may be related to prosthetic design, polyethylene
wear, or surgical technique.
In summary, posterior cruciate-retaining total knee arthroplasty
in patients with rheumatoid arthritis demonstrated satisfactory
clinical and radiographic results, a minimal prevalence of late
posterior instability, and excellent prosthetic fixation at a mean
of 10.5 years of follow-up. In contrast to recent reports,
the present study demonstrated a low prevalence of posterior cruciate-ligament
insufficiency leading to posterior instability (found in approximately
2% of the patients). Therefore, we believe that posterior
cruciate-retaining total knee arthroplasty remains an excellent
treatment option for patients with rheumatoid arthritis.