Patient Demographics
From December 1989 to May 1994, all patients who were scheduled
for primary total knee arthroplasty by two of us (T. U. and T. N.)
and who were considered eligible for treatment with the Bisurface
knee prosthesis were enrolled in a prospective study. Patients were
not considered eligible for this type of arthroplasty if the knee
had varus or valgus deformity of more than 30 degrees or flexion
deformity of more than 60 degrees or if it had medial or lateral
instability of more than 30 degrees. Five patients were excluded
from the consecutive series on this basis, and they underwent total
knee arthroplasty with a prosthesis with more constraint or with
a hinge-type prosthesis. One hundred and sixty-six patients managed
with a total of 223 consecutive primary total knee arthroplasties
at our two institutions were enrolled in the prospective study of
the Bisurface knee prosthesis.
Seventeen patients (twenty-five knees) who had died less than
four years after the procedure were excluded from the study. Four
patients (four knees) were lost to follow-up. Four patients (seven knees)
could not be evaluated because of comorbidities unrelated to the
total knee arthroplasty. In addition, five patients (five knees)
had the prosthesis removed because of postoperative complications.
The remaining 182 knees in 136 patients formed the basis of the
clinical and radiographic evaluation in the present study.
The preoperative diagnoses included primary osteoarthritis in
117 knees, rheumatoid arthritis in sixty-one, posttraumatic osteoarthritis
in two, avascular necrosis in one, and postseptic osteoarthritis
in one. The mean age at the time of the index procedure was sixty-eight
years (range, thirty-nine to eighty-four years), and the mean weight
was 53.6 kilograms (range, 31.0 to 80.5 kilograms). Forty-six patients
had the procedure bilaterally. Of the 136 patients, twenty-two (thirty-one
knees) were male and 114 (151 knees) were female. The patients were
followed for a mean of 5.8 years (range, 3.9 to 9.0 years). Lateral
retinacular release was performed in eighty-five (47 percent) of
the knees (forty-six patients) for appropriate tracking of the patella.
Type-1 and type-2 prostheses were implanted in seventy-nine and
103 knees, respectively.
The Implant
The Bisurface knee (Fig. 1-A, Fig. 1-B, and Fig. 1-C) is a posterior-cruciate-substituting
prosthesis that has a unique ball-and-socket joint in the midposterior
portion of the femoral and tibial components. This joint functions
as a posterior stabilizing cam mechanism and as a load-bearing surface
in flexion, and it causes femoral rollback to avoid posterior impingement
of the femoral and tibial components. The femoral ball comes into
contact with the tibial socket at around 80 degrees of flexion.
Therefore, when there is more than 80 degrees of flexion, the prosthesis
has three contact areas on the tibial articular surface: the medial
and lateral condylar surfaces and the ball-and-socket joint. The
contact areas on the condylar surfaces translate posteriorly, while
the ball-and-socket joint rotates in situ (Fig. 2-A and Fig. 2-B). The posterior
condylar part of the tibial articular surface is flattened in the
anterior-posterior direction to provide axial-rotational freedom
around the ball-and-socket joint.
The femoral component is made of alumina ceramic and has a central
strut in the distal aspect to strengthen the component and to improve
fixation to the femur with bone cement. The tibial component is
made of titanium alloy with an ultra-high molecular weight polyethylene
articular insert. The thinnest part of the polyethylene articular
insert is in the deepest point of the socket, the thickness of which
ranges from four to ten millimeters depending on the thickness of
the tibial insert, which ranges from nine to fifteen millimeters.
The patellar component consists of a spherical dome made of ultra-high
molecular weight polyethylene and a four-cornered alumina-ceramic
peg.
The femorotibial articulation of the condyles was completely
flat in the medial-lateral direction in the original design of the
implant (type 1). Because the intercondylar eminence of the prosthesis is
short and anteriorly placed, owing to the midposterior location
of the socket on the tibial plateau, the medial-lateral stability
in midflexion provided by the articular geometry was considered to
be insufficient in the type-1 prosthesis. Two knees that had been
replaced with the type-1 prosthesis had recurrent medial-lateral
subluxations, which prompted us to modify the design to provide better
medial-lateral stability. In the type-2 implant, the medial and
lateral edges of the tibial insert were elevated so that the femorotibial articulation
had congruity and better stability in the medial-lateral direction.
Operative Technique and Postoperative Rehabilitation
All implants were fixed with bone cement. The technique of alignment
and soft-tissue balance was performed in a uniform manner by the
two surgeons. A lateral retinacular release was performed in knees
in which the patellar tracking was judged to be inadequate. The
incision in the knee was closed with the knee in 100 degrees of
flexion. A bulky dressing was applied with the knee in extension
and was worn for the first forty-eight hours, until removal of the
suction drainage. On the third day after the operation, knee flexion
and extension exercises were begun in a continuous-passive-motion
machine. Manipulation of the knee was performed in patients with
a poor range of motion (less than 70 degrees at two weeks after
the operation).
Clinical Evaluation
Yearly follow-up evaluation was carried out, independent of the
treating surgeons, by four of us (M. A., K. N., E. O., and Y. M.).
Preoperative and postoperative knee function was evaluated and rated according
to The Hospital for Special Surgery knee-rating system13,22. In addition, one of us (M. A.)
conducted an evaluation based on patient self-administered questionnaires.
Excluding the patients who had died, each patient was sent by mail
a questionnaire that specifically addressed the severity, location,
and frequency of pain; subjective knee instability; the ability
to perform activities of daily living; and the level of satisfaction
with the operative result. If there was a discrepancy between the
data on the chart and those on the questionnaire, a telephone interview
was performed to clarify the matter. With regard to subjective instability
of the knee, the questionnaire asked if the patient had felt looseness
during his or her activities of daily living. Knee instability was
also evaluated objectively according to The Hospital for Special
Surgery knee-rating system, in which knee instability is graded
as none, mild, moderate, or severe. Postoperative satisfaction was
graded on the questionnaire as very satisfied, satisfied, partly
dissatisfied, or very dissatisfied.
Radiographic Evaluation
Radiographs were made preoperatively, early after the operation,
at six months, and at one-year intervals thereafter. Radiographs
were analyzed by three of us (M. A., K. N., and Y. M.), independent of
the surgeons who had performed the operative procedures. Preoperative
and postoperative radiographs, which included anteroposterior and
lateral radiographs and a skyline patellar radiograph, were assessed
for alignment of the limb and the presence and location of any radiolucent
lines at the bone-cement interface, according to the recommendation
of the Knee Society7. Radiolucent
lines in each zone were evaluated according to the method of Ranawat
et al.19.
Statistical Analysis
Correlations between the preoperative and postoperative ranges
of flexion were investigated with regression analysis. Kaplan-Meier
survivorship analysis was used to determine the cumulative rate of
survival of the implant during the period of the study. The relationship
between objective and subjective instability was investigated with
use of the chi-square test.
Clinical Results
The mean preoperative Hospital for Special Surgery knee score
was 44.5 points (range, 7 to 66 points). At the time of latest follow-up,
the mean knee score was 86.3 points (range, 46 to 100 points). The
clinical result was excellent for 114 knees (63 percent), good for
fifty-seven (31 percent), fair for nine (5 percent), and poor for
two (1 percent). One poor result was in a patient who had severe
rheumatoid arthritis, and the other was in a patient who had Parkinson
disease.
Range of Motion
The mean preoperative range of flexion was 119 degrees (range,
35 to 155 degrees). The mean postoperative range of flexion was
124 degrees (range, 60 to 155 degrees). Although the number of knees with
flexion exceeding 145 degrees was reduced from thirty preoperatively
to nineteen postoperatively, 70 percent (127) of the knees had postoperative
flexion of more than 120 degrees. Regression analysis showed good
linear dependency (coefficient of correlation, r = 0.65; p < 0.0001)
between the preoperative and postoperative ranges of flexion (Fig. 3). The mean preoperative
range of extension was -9.8 degrees (range, 25 to -40 degrees),
and the mean postoperative range of extension was -0.3 degree (range,
10 to -25 degrees). Preoperatively, 132 knees (73 percent) had a
flexion contracture (mean, 14 degrees; range, 5 to 40 degrees);
postoperatively, thirty-nine knees (21 percent) had a flexion contracture
(mean, 8 degrees; range, 5 to 25 degrees).
Pain
Seventy patients (ninety-seven knees; 53 percent) had no pain
in the knee at the latest follow-up evaluation. Thirty-five patients
(forty-eight knees; 26 percent) reported infrequent discomfort in
the knee. Twenty-eight patients (thirty-four knees; 19 percent)
reported mild pain in the knee at the beginning of motion or when
walking for a long distance. Of these twenty-eight patients, thirteen
took non-narcotic analgesics infrequently to control the symptom
and two needed medication every day. Three patients (three knees;
2 percent) reported moderate pain in the knee during walking and needed
non-narcotic analgesics every day to control the symptom. In nine
patients (eleven knees; 6 percent), the pain was in the anterior
aspect of the knee and was presumed to be attributable to the patellofemoral
joint. There was no association between anterior knee pain and the
presence of a flexion contracture. Fourteen patients (fourteen knees;
8 percent) had infrequent mild pain associated with looseness of
the knee while performing daily living activities.
Ability to Walk
Seventy-four patients (101 knees; 55 percent) were able to walk
an essentially unlimited distance (more than 1000 meters) without
pain, forty-six patients (sixty knees; 33 percent) were able to walk
five to ten blocks, thirteen patients (eighteen knees; 10 percent)
were able to walk only in the house, and three patients (three knees;
2 percent) were unable to walk even in the house without assistance.
Of these three patients, one had severe polyarticular rheumatoid
arthritis, one had delayed union of a supracondylar fracture of
the contralateral femur at the time of latest follow-up, and one had
Parkinson disease. Sixty patients (eighty knees; 44 percent) did
not need support for walking, sixty-four patients (eighty-seven
knees; 48 percent) used a cane, one patient (two knees; 1 percent)
used a crutch, and eight patients (ten knees; 5 percent) used a
walker. The three patients (three knees; 2 percent) who were unable
to walk used a wheelchair outdoors.
Patellofemoral Articulation
No patient had a postoperative patellar fracture or loosening
of the patellar component. On examination, one knee had patellofemoral
noise through the arc of the motion, and this was attributed to friction
between the femoral component and the peg of the patellar component.
A revision procedure was recommended for this patient at the time of
latest follow-up, and the original knee replacement was recorded
as a failure in the survivorship analysis. Eleven knees (6 percent)
had a symptomatic patellofemoral joint.
Instability
In addition to the two patients who underwent a revision operation
because of instability, two patients had medial-lateral subluxations
of the femorotibial articulation. One patient was managed successfully
with a hinged knee brace and had no recurrence during 7.5 years
of follow-up. The other patient had recurrent medial subluxations
but had little disability when using a hinged knee brace. These
subluxations developed only with the type-1 prosthesis. According
to The Hospital for Special Surgery knee-rating system, thirty-six
(20 percent) of the 182 knees had no objective instability, ninety-two
knees (51 percent) had mild instability, fifty-three knees (29 percent)
had moderate instability, and one knee (less than 1 percent) had
severe instability. According to patients' reports, 146 knees (80
percent) had good subjective stability; however, thirty-six knees
(20 percent) felt loose in activities of daily living. There was
a significant relationship between objective and subjective knee
instability (chi-square test, p < 0.0001) (Table I). A feeling
of looseness was reported in sixteen (20 percent) of the seventy-nine
knees with a type-1 prosthesis and in twenty (19 percent) of the
103 knees with a type-2 prosthesis.
Satisfaction with Outcome
One hundred and twenty-two patients (166 knees; 91 percent) were
very satisfied with the outcome of the operation, seven patients
(nine knees; 5 percent) were satisfied, six patients (six knees;
3 percent) were partly dissatisfied, and one patient (one knee;
less than 1 percent) was very dissatisfied. Of the six patients
who were partly dissatisfied, three had instability of the knee
with or without pain, one had recurrent subluxations of the knee
joint and required a knee brace for knee stability, one had not
regained the ability to assume a fully squatting posture, and one
was confined to a wheelchair because of delayed union of a supracondylar
fracture about the contralateral knee. The patient who was very
dissatisfied had not regained an appreciable ability to walk after
the operation, owing to Parkinson disease.
Radiographic Results
The mean postoperative tibiofemoral angle (the angle formed by
the long axis of the tibia and the long axis of the femur) was 6
degrees of valgus (range, 5 degrees of varus to 15 degrees of valgus). No
radiolucent lines were observed around the femoral strut, the tibial
keel, or the patellar peg (Table II). None of the knees had a radiolucent
line in more than two contiguous zones. No knee had grade-4 radiolucency
at the bone-cement interface of any component. Therefore, no component
in any knee that had not been revised was suspected of being loose
on the basis of radiographic appearance.
Revision Operations
A revision operation was performed in eight knees. Five of the
revisions were undertaken because of infection of the involved joint,
and two of these five revisions involved replacement or removal
of the prosthesis. Two additional revisions were performed because
of instability; one of the two was done because of genu recurvatum
that developed four years after the operation in a patient with
rheumatoid arthritis, and the other was undertaken because of attenuation
of the medial collateral ligament that developed 3.5 years after
the operation in a patient who had osteoarthritis with severe valgus
deformity. The eighth revision was performed because the peg of
the patellar component broke when the patient fell on the knee.
Only the patellar component was retrieved, and a new patellar component
was implanted 2.8 years postoperatively.
Survivorship Analysis
Kaplan-Meier survivorship analysis was performed with revision
of the knee or recommendation for revision as the end point. The
projected rate of survival of the knee prostheses at six years was
94 percent. Forty-seven knees remained at risk, and the 95 percent
confidence interval was 90 to 98 percent six years postoperatively.
The Bisurface knee prosthesis has a unique design to improve
the range of flexion: a ball-and-socket joint. This unique joint
was developed by inversion of the post-and-cam mechanism of the
posterior stabilized prosthesis and can provide a weight-bearing
surface with the cam mechanism during flexion of more than 80 degrees.
Large femoral rollback provided by the ball-and-socket joint permits
maximum flexion of 140 degrees (Fig. 2-A and Fig. 2-B). Whereas the need for extensive
bone loss in the intercondylar region is one of the defects of the posterior
stabilized prosthesis13,25,31,
the Bisurface knee requires relatively little bone loss for the
femoral central strut. Posterior dislocation of the cam mechanism4,8,9,14,15 and the patellar clunk
syndrome2,12 are not possible
with this prosthesis.
The mean range of motion of 124 degrees in the present group
was greater than the mean range of motion of 101 to 112 degrees
reported for the cruciate-retaining design21,24,28,32 and
the mean range of motion of 107 to 115 degrees reported for the
posterior-cruciate-substituting design13,16,19,25,27.
Also, 70 percent (127) of the knees in the present study had more
than 120 degrees of postoperative flexion.
Although previous studies have shown that the preoperative range
of flexion is the most important determinant of the postoperative
range of flexion11,18, Wright
et al.32 found that the coefficient
of correlation between preoperative and postoperative flexion was
only 0.37 for patients with the posterior-cruciate-retaining kinematic
prosthesis. However, in the present study, regression analysis of
the preoperative and postoperative ranges of flexion showed good
linear dependency (coefficient of correlation, r = 0.65; p < 0.0001).
This result implies that arthroplasty with the Bisurface prosthesis
rarely resulted in a large loss of flexion postoperatively and that the
restriction of motion caused by the articular geometry of the femorotibial
joint might be reduced or almost eliminated.
Another unique feature of the Bisurface knee is that the femoral
component is made of alumina ceramic. The surface roughness of the
alumina component (Ra = 0.02 micrometer) is substantially less than
that of chromium-cobalt components (usually Ra = 0.05 micrometer).
Furthermore, the ceramic bearing surfaces have been reported to
have a good resistance to third-body wear damage and to produce
the least amount of ultra-high molecular weight polyethylene wear
debris5,6,17,33. The wear property
of the alumina ceramic femoral component is promising. However,
the longevity of the ultra-high molecular weight polyethylene insert
must be investigated in long-term follow-up studies and in additional
studies of retrieved components.
Recently, cost-effectiveness of medical treatments has been an
important concern. In our country, the Ministry of Public Welfare
has fixed the cost of orthopaedic implants, including total knee
prostheses. The official price of the alumina ceramic femoral component
is only 4.7 percent higher than that of the chromium-cobalt femoral
component used widely in Japan. Thus, the cost of the knee prosthesis
with an alumina ceramic femoral component is almost comparable with
that of the standard chromium-cobalt implant.
There are some concerns about breakage of the alumina femoral
component during and after the operation, owing to the brittleness
of the ceramic material. Before clinical application, mechanical testing
was carried out according to "The Guidance Document for the Preparation
of Premarket Notifications for Ceramic Ball Hip Systems"10 devised by the United States Food
and Drug Administration, and clinical application appeared to be
safe on the basis of the results of that testing. In fact, none
of the alumina femoral components broke in the present study. However,
as the study population was limited to Japanese patients, the concern
that problems with implant breakage might arise in North America
or Northern Europe cannot be completely eliminated; the patient
population in Japan clearly differs from that in North America or
Northern Europe, especially with respect to weight.
Excessive rollback may have several disadvantages with regard
to the durability of the prosthesis. Increased translation of the
femorotibial contact point can result in higher shear stresses and
compressive forces on the polyethylene, possibly leading to premature
fatigue failure of the polyethylene3,26,29.
Although only four prostheses were retrieved by the time of writing,
there was no indication of severe fatigue failure of the polyethylene
articular surface. Excessive rollback also can increase rocking
of the tibial component, possibly leading to early loosening of
the component20. Radiographic
observation in the present study demonstrated no problems with regard
to component fixation with the bone cement that was used. None of
the knees had grade-4 radiolucency. These results show that fixation
of the tibial component enabled it to withstand the extensive rollback
provided by the cam mechanism.
The questionnaire revealed that 20 percent (thirty-six) of the
knees felt somewhat loose during various daily activities. The prevalence
of this mild sensation of knee instability was not reduced after we
switched to the type-2 Bisurface prosthesis. These results prompted
us to enhance the intrinsic stability of the prosthesis by improving
the congruity of the ball-and-socket joint. This new prosthesis
is called type 3-Plus and has been used since July 1998.
In the present series, total knee arthroplasty with the Bisurface
prosthesis resulted in an excellent range of motion and a high level
of satisfaction with the outcome, and the durability of the prosthesis
was promising. We expect that the introduction, after this study,
of a modification that improved the congruity of the joint and the
more correct ligamentous balance provided by a new instrumentation
system will reduce the prevalence of patients who feel looseness
in the knee.