Understanding the in vivo motions of human joints
has become increasingly important. Researchers have used in
vitro (cadavers), noninvasive (gait laboratories), and in
vivo (fluoroscopy) approaches to assess human knee motion.
Unfortunately, previous attempts have been unable to track the in
vivo bearing-surface motion of the medial and lateral condyles
of the normal knee in three dimensions. The objective of this study
was to use fluoroscopy and computed tomography to accurately determine the
three-dimensional, in vivo, weight-bearing motion
of the normal knee.
Methods
Five normal knees, clinically determined to have no pain or ligamentous
laxity, were analyzed. With use of computed tomography scanning,
slices of the femur beginning 6 in (15.2 cm) proximal to the joint
line and slices of 6 in of the proximal part of the tibia (1.0-mm
slices near the bearing surfaces and 3.0-mm slices elsewhere) were
obtained (Fig. 1).
Three-dimensional bone models (Fig. 2-A) of each subject’s femur,
tibia, patella, and fibula were recreated from the three-dimensional
volume data (Fig. 2-B). Each subject was then asked to
perform five weight-bearing activities—deep knee bend,
normal gait, rising from a chair, sitting down in a chair, and descending
stairs—while under fluoroscopic surveillance. The fluoroscopic
images of each knee during each activity were downloaded to a computer
workstation. The computer-generated three-dimensional models of
each subject’s femur and tibia were overlaid onto the two-dimensional
fluoroscopic images and subsequently were analyzed at various knee-flexion
angles. For each activity, femorotibial contact paths were determined
for both the medial and the lateral condyle and were plotted with
respect to knee-flexion angle. Femorotibial contact anterior to
the tibial midline in the sagittal plane was denoted as positive,
and posterior contact was denoted as negative.
Results
During all five activities, there was significantly more motion
of the lateral condyle than of the medial condyle (p < 0.001).
During gait, the lateral condyle moved an average of 7.3 mm, while
the medial condyle moved an average of only 0.8 mm. During three
of the four deep flexion activities, the medial condyle moved minimally;
the exception was rising from a chair, during which there was an average
of 3.8 mm of motion. In contrast, the lateral condyle moved substantially
during the deep flexion activities, with translation of up to 25.2
mm during chair-sitting. The medial condyle moved substantially
in certain subjects, with one subject experiencing more medial motion
than lateral motion.
Discussion
The present study is the first in which medial and lateral femorotibial
contact pathways (three-dimensional kinematics) of the normal knee
were accurately determined under in vivo, weight-bearing
conditions with use of fluoroscopy and a computer model-fitting
process. The results from this study demonstrated that the three-dimensional
kinematic patterns of normal knees are significantly different from
the kinematic patterns in previously published studies pertaining
to posterior-cruciate-retaining and posterior stabilized total knee
replacements (p < 0.05). During all five activities, the
lateral condyle had significantly more anterior-posterior translation,
leading to normal axial rotation of the tibia relative to the femur.
Although on the average a medial pivot pattern was predominant,
there was variability among subjects with regard to the presence
of substantial translation of the medial condyle.
Controversy remains regarding the effectiveness of unicompartmental
knee arthroplasty. Some research studies have shown unicompartmental
knee arthroplasty to be unsuccessful, while others have demonstrated
it to be a viable option in specific cases. Other studies have demonstrated
a high prevalence of radiolucent lines associated with unicompartmental
knee arthroplasty. The main failure mode of unicompartmental knee
replacements has been polyethylene wear. However, although rates
of failure due to polyethylene wear have been high in some studies,
other investigators have reported excellent results and have highly
recommended unicompartmental knee arthroplasty. Recently, fluoroscopy
has been used to evaluate in vivo motions of total
knee replacements. The previous fluoroscopic analyses focused on
the kinematics of posterior-cruciate-retaining and posterior stabilized
total knee replacements. During these studies, it was determined
that a paradoxical anterior slide occurs during gait and deep flexion
activities in subjects with a posterior-cruciate-retaining total knee
replacement and during gait in subjects with a posterior stabilized
total knee replacement. From these studies, it has been hypothesized
that the anterior cruciate ligament plays an important role in knee
kinematics. Therefore, the purpose of the present study was to analyze
the kinematics of unicompartmental knee replacements in knees in which
the anterior cruciate ligament was intact at the time of the operative
procedure.
Methods
In vivo kinematics were determined for twenty
subjects who had been treated with a unicompartmental knee arthroplasty
by the same surgeon. Seventeen knees were implanted with a medial
unicompartmental knee replacement, and three were implanted with
a lateral unicompartmental knee replacement. All of the knee arthroplasties
were judged to be clinically successful (a Hospital for Special
Surgery score of >90 points), with no ligamentous laxity
or pain. Under fluoroscopic surveillance, each subject was asked
to perform successive weight-bearing deep knee bends to maximum
flexion. The fluoroscopic images were stored on videotape for subsequent
redigitization with use of a frame-grabber. The kinematics were
assessed at full knee extension and at 15°, 30°, 45°, 60°, 75°,
and 90° of knee flexion.
The contact position between the medial condyle (in the patients
with a medial unicompartmental knee arthroplasty) or the lateral
condyle (for those with a lateral unicompartmental knee arthroplasty) and
the tibia was determined with use of a three-dimensional model-fitting
technique. The fluoroscopic images were captured onto a workstation computer.
The three-dimensional computer-aided-design solid models of the
femoral and tibial components were overlaid onto the two-dimensional fluoroscopic
perspective images (Fig. 3). Once the three-dimensional components
were fit, the medial or lateral femorotibial contact position was
determined with respect to the midline of the tibia in the sagittal
plane (Figs. 4 and 5). A contact position
anterior to the midline was denoted as positive, and a position
posterior was denoted as negative. The coronal view was used to assess
axial rotation. The angle between the longitudinal axis that passes
through the femoral component of the unicompartmental knee replacement (posterior
to anterior) and the fixed axis through the tibial component was
measured (Figs. 6 and 7).
Results
Anteroposterior Translation
On the average, subjects with a medial unicompartmental knee
replacement had minimal motion, but the kinematic patterns varied
among the subjects (Fig. 8). The average contact position
in subjects with a medial unicompartmental knee replacement was 0.0
mm (range, 10.7 to -6.8 mm) at full extension, -1.4 mm (range, 10.5
to -9.8 mm) at 15° of knee flexion, -2.1 mm (range, 6.4 to -10.5
mm) at 30° of knee flexion, -3.1 mm (range, 4.9 to -12.7 mm) at 45°
of knee flexion, -2.1 mm (range, 2.7 to -9.9 mm) at 60° of knee
flexion, -1.2 mm (range, 2.5 to -5.4 mm) at 75° of knee flexion,
and -0.8 mm (range, 3.3 to -5.4 mm) at 90° of knee flexion. The variability
in the data decreased as the flexion angle increased. Seven of the
seventeen subjects had posterior femoral rollback of the medial
condyle from full extension to 90° of knee flexion. Five of the seventeen
subjects had minimal motion (<1.0 mm) of the medial condyle
and another five had an anterior slide of the medial condyle from
full extension to 90° of knee flexion. At full extension, eight
of the seventeen subjects had anterior contact of the medial condyle.
All eight of these subjects had either minimal motion of the medial
condyle or posterior femoral rollback of the medial condyle.
On the average, subjects with a lateral unicompartmental knee
replacement had posterior femoral rollback (Fig. 9). The average
contact position for subjects with a lateral unicompartmental knee
arthroplasty was -1.95 mm (range, 0.0 to -3.9 mm) at full extension,
-5.1 mm (range, -2.7 to -6.4 mm) at 15° of knee flexion, -5.1 mm
(range, -1.6 to -6.9 mm) at 30° of knee flexion, -5.8 mm (range,
-1.1 to -9.9 mm) at 45° of knee flexion, -6.7 mm (range, -1.0 to
-11.2 mm) at 60° of knee flexion, -6.0 mm (range, -1.2 to -11.8 mm)
at 75° of knee flexion, and -4.5 mm (range, -0.6 to -9.4 mm) at
90° of knee flexion. On the average, there was increased rollback
of the lateral condyle in the posterior direction from full extension
to 60° of knee flexion. From 60° to 75° of knee flexion, minimal
motion of the lateral condyle was detected on the average, and from
75° to 90° of knee flexion, an average of 1.5 mm of anterior motion
was detected. As stated earlier, with both medial and lateral unicompartmental
knee arthroplasty there were variable kinematic patterns, with an
anterior slide with increasing knee flexion (especially with medial
unicompartmental knee arthroplasty) occurring at either 30° or 60°
of knee flexion.
Axial Tibiofemoral Rotation
On the average, subjects with a medial unicompartmental knee
replacement had, on the average, normal axial rotation of 3.3° from
full extension to 90° of knee flexion (Fig. 10). Subjects with a lateral unicompartmental
knee replacement also had, on the average, normal axial rotation
of 11.2° (Fig. 11).
Twelve of the seventeen subjects with a medial unicompartmental
knee replacement had normal axial rotation from full extension to
90° of knee flexion. Two of the seventeen subjects had minimal axial
rotation (<1.0°), and three of the seventeen had an opposite
axial rotation. Two of the three subjects with a lateral unicompartmental
knee arthroplasty had normal axial rotation from full extension to
90° of knee flexion. One of the three subjects had an opposite axial
rotation. Axial rotation patterns, similar to anteroposterior translation,
were variable, with normal and opposite rotation patterns occurring
at varying flexion angles.
Discussion
The results from this study were surprising and suggest that
progressive laxity of the anterior cruciate ligament may occur over
time. Previously, other researchers reported that the anterior cruciate
ligament plays a substantial role in knee kinematics. In our study,
it appeared that eight subjects had an anterior contact position
at full extension, which may lead to the hypothesis that the anterior
cruciate ligament was not functioning properly in twelve subjects,
since the anterior cruciate ligament was unable to provide an anterior
constraint force with the necessary magnitude to thrust the femur
in the anterior direction at full extension. This suggests inconsistent
function of the anterior cruciate ligament following unicompartmental
knee arthroplasty and may account, at least in part, for the premature
polyethylene wear occasionally seen after unicompartmental knee
arthroplasty. Therefore, our results support the findings of other
studies that the anterior cruciate ligament plays an important role
in maintaining satisfactory knee kinematics, which also may, in
part, contribute to the longevity of unicompartmental knee replacement.
The objective of this study was to determine the presence of
condylar lift-off in a group of patients who had a conforming posterior
stabilized total knee prosthesis, which was implanted with attention
to positioning of the femoral component along the epicondylar axis.
The hypothesis was that condylar lift-off is related to the rotational
position of the femoral component.
Methods
Twenty-five patients received a Legacy Posterior Stabilized knee
prosthesis (Zimmer, Warsaw, Indiana), with the implantation performed
by two surgeons (J.I. and G.S.) with a similar technique. With use
of the classic method of bone resection and the appropriate soft-tissue
releases, the femoral component rotation was positioned along the
epicondylar axis. The femoral instrumentation was posterior, referenced
so that the variable resection was anterior, and the goal was to
create a reproducibly tight and symmetrical flexion space. All of
the total knee arthroplasties were judged to be clinically successful,
with Hospital for Special Surgery and Knee Society scores of >90
points.
Condylar lift-off was assessed for all patients, and a second
analysis was conducted on the five subjects with the maximum amounts
of condylar lift-off. Each subject was asked to perform three successive deep
knee bends to maximum flexion under fluoroscopic surveillance in
the sagittal plane (Fig. 12). Each subject placed the foot
of the leg to be studied on a designated marker. The subjects were
initially examined fluoroscopically with the knee at full extension
and then throughout the flexion cycle.
Three-Dimensional Interactive Model-Fitting
Technique
Individual fluoroscopic frames at specified degrees of flexion
were digitized. The images were projected onto the image plane,
and the corresponding implant models were added to the scene. The
operator initially positioned the models in the fluoroscopic scene,
and then the iterative model-fitting algorithm accurately fit both
the femoral and the tibial component (Fig. 13). The correct fit was achieved
when the silhouettes of the femoral and tibial components perfectly matched
the corresponding components on the fluoroscopic image (Fig. 14). The pose
of each component was then recorded, and measurements of interest
were extracted with use of a computer-aided-design modeling program. The
process was performed at 0°, 30°, 60°, and 90° of knee flexion,
and the distances from the medial and lateral condyles to the tibial
plateau were measured. The difference between these two measurements
was denoted as condylar lift-off.
Computed Tomography Scanning
Each subject in the study was then reanalyzed with 1.0-mm computed
tomography slices to determine if the femoral component of the total
knee arthroplasty was malaligned relative to the transepicondylar
axis. Using the slice that allows for the determination of the transepicondylar
axis, an angle was measured from the transepicondylar axis to the femoral
condylar axis (Fig. 15). This angle was then correlated
with the lift-off values determined with the model-fitting process.
Results
Condylar Lift-Off
Ten (40%) of the twenty-five subjects in this study had
condylar lift-off at one of the four flexion angles (Table I). Of the twenty-five
subjects, five had condylar lift-off, which was medial in one and
lateral in four, at full extension. Six had condylar lift-off, which
was medial in two and lateral in four, at 30° of knee flexion. At
60° of knee flexion, four subjects had condylar lift-off, which
was medial in one and lateral in three, and at 90° of knee flexion,
five had condylar lift-off, which was medial in three and lateral
in two. Condylar lift-off occurred predominantly on the lateral
side. In the twenty-five patients, the average amount of condylar
lift-off was 0.6, 0.6, 0.5, and 0.5 mm at 0°, 30°, 60°, and 90°
of knee flexion, respectively. In the subjects with condylar lift-off,
the average amount was 1.3, 1.3, 1.6, and 1.1, respectively. The
average amount of condylar lift-off determined at any angle, in
the subjects who had lift-off, was 1.3 mm. The maximum amount of
condylar lift-off was 2.3 mm.
A second trial was conducted on the subjects in whom the largest
amount of lift-off had been identified in the first trial (Table II). Each subject
had very similar amounts of condylar lift-off during the two trials,
and lift-off of the same condyle occurred in each of the five subjects. The
maximum amounts of lift-off in each of these five subjects averaged
1.9 mm during the first trial and 1.7 mm during the second.
Axial Tibiofemoral Rotation
Axial tibiofemoral rotation was correlated with condylar lift-off
for each subject who had lift-off (Fig. 16). This correlation was conducted
to ensure that extreme internal/external rotation of the
femoral component on the tibia, which could lead to the presumption
of lift-off, was not occurring. If the lateral condyle was more
anterior than the medial condyle, the angle was denoted as negative,
and if the lateral condyle was more posterior than the medial condyle,
the angle was denoted as positive. The average amount of axial tibiofemoral
rotation was -1.5° (range, 4.7° to -9.4°), -0.3° (range, 5.0° to -9.9°),
0.1° (range, 5.2° to -4.7°), and 0.5° (range, 8.1° to -7.8°) at
0°, 30°, 60°, and 90° of knee flexion, respectively. During the
full deep knee bend, the average amount of axial tibiofemoral rotation was
2.0° (range, 14.3° to -3.8°). Nineteen of the twenty-five subjects
had normal axial tibiofemoral rotation from 0° to 90° of knee flexion,
whereas six subjects had an opposite axial tibiofemoral rotation. On
the basis of these results, it was determined that, at each occurrence
of condylar lift-off, minimal axial tibiofemoral rotation was present.
Anteroposterior Translation
We also analyzed the anteroposterior translation of both the
medial and the lateral condyle for the twenty-five subjects. A contact
position anterior to the midline of the tibia in the sagittal plane
was denoted as positive, and a position posterior was negative.
The average anteroposterior position of the medial condyle was -5.4
mm (range, -0.5 to -10.1 mm), -7.1 mm (range, -3.7 to -12.6 mm),
-6.7 mm (range, -3.1 to -10.5 mm), and -7.1 mm (range, -2.9 to -11.6
mm) at 0°, 30°, 60°, and 90° of knee flexion, respectively (Fig. 17). The average
anteroposterior position for the lateral condyle was -4.2 mm (range,
0.3 to -12.2 mm), -6.9 mm (range, -2.0 to -11.5 mm), -6.8 mm (range, -3.7
to -10.2 mm), and -7.5 mm (range, -1.9 to -13.4 mm) at 0°, 30°,
60°, and 90° of knee flexion, respectively. On the average, there
was posterior femoral rollback of the medial condyle from 0° to 30°
of knee flexion (1.4 mm), but the medial condyle remained relatively
stationary from 30° to 60° (-0.4 mm) and from 60° to 90° (0.4 mm)
of knee flexion. From 0° to 30° of knee flexion, the lateral condyle
rolled in the posterior direction, 2.3 mm on the average, but it
remained stationary from 30° to 60° (-0.1 mm) and it rolled more
posteriorly from 60° to 90° (0.7 mm) of knee flexion. During the
full deep knee bend, the medial condyle rolled an average of 1.7
mm (range, -3.1 to 8.2 mm) posteriorly and the lateral condyle rolled
an average of 3.3 mm (range, -2.9 to 13.5 mm) posteriorly.
Computed Tomography Scanning
A comparison of condylar lift-off and femoral component alignment
was performed for only twenty-six of the thirty trials. During eighteen
(69%) of the twenty-six trials, there was a correlation
between femoral condylar lift-off and malalignment of the femoral
component with the transepicondylar axis.
Discussion
Numerous analyses of the abduction and adduction moments at the
knee during normal gait have been conducted. It has been reported
that an abduction moment occurs at heel strike but quickly reverses
to an adduction moment throughout the remainder of stance phase.
This adduction moment has been shown to have an estimated magnitude
between 36 and 50 N, increasing if coexisting varus deformity is
present. During the midstance phase of gait, the medial compressive
loads increase to a range of 70% to 75% of the
load at the knee, secondary to the adduction moment occurring at
midstance. In order to resist this adduction moment, numerous compensatory
mechanisms are active at the knee joint. These mechanisms include:
(1) redistribution of the condylar loads, (2) co-contraction of
antagonist muscle groups, (3) increased tension in the lateral convex
soft tissues, (4) increased tension in the cruciate ligaments, (5)
increased swaying of the body in the lateral direction, (6) decreased
stride length, and (7) decreased inversion moment at the ankle accomplished
by out-toeing. Although, these compensatory mechanisms aid in knee
stabilization, they may lead to increased joint reaction forces
at the knee. If these compensatory mechanisms are inadequate, femoral
condylar lift-off may occur.
The findings in this study support our impression that both surgical
technique and implant design influence femoral condylar lift-off.
With use of the classic method of bone resection, alignment of the femoral
component along the epicondylar axis will facilitate creation of
a symmetrical and balanced flexion space. This technique, along
with consistently measured posterior resection, will create a tight
flexion space. Although some lift-off might occur, it will be less
frequent and of a lesser magnitude than that occurring with techniques
that do not position the femoral component in the appropriate degree
of rotation or include a variable resection of the posterior femoral
condyles. Even with adherence to these principles, it may be difficult
to be technically perfect. Correlation of the fluoroscopic images
and the computed tomography measurements demonstrates that the accuracy
of this alignment technique, within 2°, is acceptable.
If condylar lift-off does occur, it may be beneficial to have
a prosthesis that was designed with a conforming frontal articulation.
The present observations show that, when lift-off occurs with a
round-on-round articulation, the conformity of the articulation
eliminates the edge-loading that has been observed on flat-on-flat
and flat-on-round designs. A posterior-cruciate-substituting prosthesis
with a conforming round-on-round articulation eliminates the kinematic
conflict of articular conformity that is seen with cruciate-retaining
implant designs.
In conclusion, the results of this study suggest that there is
a correlation between condylar lift-off and rotational alignment
of the femoral component. Reducing lift-off with a conforming articulation reduces
edge-loading and in turn should reduce polyethylene damage.
Acknowledgments: National Science Foundation; Zimmer; Radiographic & Data
Solutions, Minneapolis, Minnesota.