Normative Three-Dimensional Patellofemoral and Tibiofemoral Kinematics: A Dynamic, <emphasis emphasistype=\"italic\">in Vivo</emphasis> Study

(
Normative 3D Patellofemoral and Tibiofemoral Kinematics: a Dynamic, in vivo
Study

Andrea R. Seisler, MBE, Frances T. Sheehan, Ph.D.

Abstract—In order to advance biomechanical modeling, knee joint implant
design and clinical treatment of knee joint pathology, accurate in vivo
kinematic data of the combined patellofemoral and tibiofemoral joint during
volitional activity are critical. For example, one cause of the increased
prevalence of anterior knee pain in the female population is hypothesized
to be altered tibiofemoral kinematics, resulting in pathological
patellofemoral kinematics. Thus, the objectives of this study were to test
the hypothesis that knee joint kinematics vary based on gender and to
explore the correlation between the 3D kinematics of the patellofemoral and
tibiofemoral joints. In order to accomplish these goals, a large (n=34)
normative database of combined six degree of freedom patellofemoral and
tibiofemoral kinematics, acquired non-invasively during volitional knee
extension-flexion using fast-PC (dynamic) MRI, was established. In this
normative database, few correlations between tibiofemoral and
patellofemoral kinematics were found. In general, significant differences
could not be found based on gender. Specifically, tibial external rotation
did not predict lateral patellar tilt, as has been stated in previous
studies. Further investigation into these relationships in the presence of
pathology is warranted.

Index Terms—healthy, kinematics, knee, patellofemoral, tibiofemoral


INTRODUCTION

T
HE quality of biomechanical models, accuracy of clinical diagnosis, and
fidelity of joint implants are all dependent upon the quality of the in
vivo experimental data used in their creation. For this reason there have
been a host of experimental and modeling studies focused on knee joint
dynamics. Yet, current knowledge of this joint is limited by the fact that
complete six degree of freedom kinematics of this joint, inclusive of both
the patellofemoral (PF) and tibiofemoral (TF) joints, have been presented
for only 5 knee joints [1]-[2]. Since alterations in TF kinematics have
been hypothesized to result in pathological PF kinematics [3], a combined
PF and TF joint study is key to understanding pathologies such as anterior
knee pain and patellar maltracking.
For the TF joint, past research has focused on the finite helical axis
direction and location [4]-[5], the existence of the screw home mechanism
[6]-[7], the effects of ligament loss [5]-[6], total knee arthroplasty [8]-
[9] and cartilage contact patterns [10]. Gait analysis is commonly used to
collect in vivo data, but is susceptible to errors based on well documented
skin motion artifacts [11]. Biplane radiography [12] has proven to be an
accurate tool for studying TF kinematics. However, the required tracking of
beads, implanted within the bone, narrows the populations that can be
studied. Further, accurate in vivo patellar tracking is difficult with this
technique. Single-plane fluoroscopy has provided strong results for
sagittal-plane kinematics of total TF knee replacement [8], but tracking
motion in the other cardinal planes is not as accurate [13]. Recent studies
registering fluoroscopic images to CT bone models were able to eliminate
the need for directly tracking metallic beads or implants, but out-of-plane
tibiofemoral accuracies (translational: 8.4 mm to 22.3 mm; rotational: 1.3(
to 3.1(; assuming all data fell within 2 standard deviations of the
reported mean error ) are poor [14].
Numerous PF disorders are thought to arise from abnormal PF kinematics,
yet defining these kinematics has been quite difficult. An excellent review
[15] summarized 15 in vitro and 13 in vivo (primarily 2D) PF kinematic
studies with population sizes ranging from 2-32 specimens and 1-20 knees,
respectively. Katchburian et al. [15] noted that the inter-study
variability stemmed from various study design differences,


Fig. 1: Subject position within the imager.
including TF orientation, coordinate system definitions, muscular loading
conditions, study type (static, quasi-static or dynamic), and motion
direction (flexion versus extension). The potential variability resulting
from the gender of the subjects studied was not discussed.
Potential gender-based kinematics differences [3]-[16]-[17] may be key to
understanding the higher prevalence of anterior knee pain in the female
population. Structural differences, muscular strength differences,
sociologic factors, and hormonal factors have all been hypothesized to lead
to increased anterior knee pain among women [3]. Yet, a direct link between
these factors and anterior knee pain or altered PF kinematics has not been
shown and few studies have investigated gender-based differences in PF
kinematics. A single study did directly investigate gender differences at
the PF joint and found increased static PF contact pressures in female
cadaver specimens [10]. TF kinematic and kinetic gender-based differences
have been shown using gait analysis during various tasks [18-21]. Such TF
kinematic variations have been hypothesized to lead to PF instability and
anterior knee pain [3]. Unfortunately, few studies have quantified the
kinematic relationships between these two joints. Li et al. [22] found that
hamstrings co-contraction under static loading conditions increased tibial
posterior translation and external rotation relative to the femur with an
accompanying increase in PF contact. In addition, anterior cruciate
ligament loss has been shown to alter both TF and PF kinematics [23].
Thus, the purpose of this study was to explore whether excessive tibial
external rotation results in PF maltracking by correlating PF and TF
kinematics and to test the hypothesis that knee joint kinematics vary with
gender. In order to accomplish these goals, a large (n=34) normative
database of combined six degree of freedom patellofemoral and tibiofemoral
kinematics, acquired non-invasively during volitional knee extension-
flexion using fast-PC (dynamic) MRI, was established.



Methods

Twenty-five healthy subjects (14 female, 11 male; age = 26.7 ± 8.8 years;
weight = 67.5 ± 12.7 kg, height = 172.3 ± 7.5 cm) participated in this IRB
approved study and gave informed consent upon entering the study. If time
allowed, both knees were imaged resulting in data from 34 knees (9 knee
pairs, 18 Left, 16 Right; 20 female, 14 male). Subjects were excluded from
this study if they had a history of knee problems or pain, were diagnosed
with any knee pathology, had previous knee joint surgery or had any
contraindications to having an MRI scan.
Subjects were placed supine in a 1.5-Tesla MR imager (CV-9.1M4 or LX-
9.1M4; GE Medical Systems, Milwaukee, WI, USA – Fig. 1). A cushioned wedge
was placed under the thigh without contacting the posterior knee such that
full extension of the leg was attainable. Cushioning supported the head,
neck and lower back. A custom designed coil holder [24] was used to
stabilize two phased array torso coils medial and lateral to the knee. An
optical trigger, placed on the imaging bed

Fig. 2: Anatomically based coordinate systems. All axes were defined in the
image representing the fully extended position for that subject. For all
three bones, the x-axis was defined first. A) Tibial image at full
extension (20 mm below the tibial patellar tendon insertion). The anterior
edge of the tibia defined Tx. The average angle between this line and the
femoral x-axis (Fx) in the normative population was 41°. To ensure a medial-
lateral x-axis, the tibial x-axis (Tx) was defined as a unit vector rotated
41° away from Tx in the axial plane for all subjects. B) Femoral image (at
the level of the epicondyles). Note, this image was intentionally selected
not to be in full extension so that the mid-patella could be visualized.
The posterior edge of the femoral condyles in the femoral image defined Fx.
The patellar lateral posterior edge in the patellar image defined Px. The
dashed line represents the plane used for the fast-PC data collection. C)
Sagittal-oblique fast-PC anatomical images at full extension. A temporary y-
axis (Tytemp , Pytemp, Fytemp ) was established for each bone in the
sagittal image. Tytemp was defined as a unit vector directed along the
tibial anterior edge (TiTs), Pytemp was defined as a unit vector directed
along the patellar posterior edge (PiPs), and Fytemp was defined as a unit
vector that bisected the angle subtended by the femoral anterior and
posterior edges (FiFs). The z-axis for each bone was then defined as the
unit cross-product between that bone's x-axis and its temporary y-axis.
Finally, the y-axis was defined as the unit cross-product between that
bone's z-axis and x-axis.

beneath the ankle, was used to synchronize data collection to the motion
cycle.
Dynamic MRI sequences acquired the data from which the 3D kinematics were
determined. During dynamic imaging,
subjects were asked to extend and flex their knee from maximum attainable
flexion to full extension and back at 35 cycles per minute to the beat of
an auditory metronome. Prior to data collection, subjects practiced the
task until they could comfortably repeat the motion. A dynamic exam
involved three movement trials. During the first trial, anatomic images at
the mid-patellar level were collected using an axial fast cine gradient
echo sequence (Fastcard-1, Table I). This image set was used to select the
proper sagittal-oblique plane for the next dynamic trial. A full fast-PC
(Fast-PC, Table I) data set was collected using a single sagittal-oblique
plane that was perpendicular to the posterior femoral condyles, bisected
the
Table I: Key imaging parameters for the dynamic sequence.

patella and did not pass through the popliteal artery (Fig. 2). The last
dynamic trial was a fastcard acquisition (Fastcard-3, Table I) at three
axial slice locations (patellar image: through the mid-patella, femoral
image: through the femoral epicondyles, and tibial image: through the
tibia, approximately
2 cm below the tibial insertion of the patellar tendon). For each subject,
these slice locations were selected from the fast-PC anatomic image
representing full extension (the most extended position).
As previously described [24], 3D rigid body rotations and translations
(jointly referred to as attitudes) of the femur, tibia and patella were
quantified through integration of the fast-PC velocity data. It is
important to note that although the fast-PC acquisition was based on a
single imaging plane (due to time constraints), the 3D velocity data
allowed the attitude of all three bones to be accurately tracked three-
dimensionally throughout the movement. To account for differences in
translation, due skeletal size variations across subjects, the translations
were scaled by the ratio of the average epicondylar width for all 34 knees
(77.3 mm) to the epicondylar width for that individual knee. Unlike earlier
cine MRI experiments [25], these kinematics were based on an anatomical
coordinate system (Fig. 2), which was identified in a single time frame
only. The fast-PC data were then used to track the bones' changes in
attitude through all time frames. To reduce variability within the
kinematics due to the sensitivity of bone shape to the bone's attitude
within the imaging plane [26], clear rules were created for identifying
imaging planes and establishing coordinate systems. As part of this study,
improvements to the original coordinate systems [24] were made. The
original tibial anatomical landmarks were selected in the sagittal plane
only. As this may cause imprecision in quantifying out-of-plane motion,
this coordinate system is now selected in two orthogonal planes (Fig. 2).
In addition, the patellar superior-inferior was previously defined by its
most superior and inferior points. The posterior flat edge of the patella
is more consistent to define and is now used for the patellar superior-
inferior line.
The interdependence of the 12 parameters defining the PF and TF kinematics
(6 translations and 6 rotations) was evaluated using a Pearson's linear
correlation
To create population averages and analyze differences between groups, each
kinematic variable was interpolated to single degree knee angle increments.
Certain subjects achieved greater or lesser than the full 40° of range of
motion, depending on their individual leg length. Thus, not all subjects
are represented in the average kinematics at all knee angles. Data
representing 3 or fewer subjects were eliminated from the total group
average. In order to explore the relationship between knee joint kinematics
and gender, the kinematic database was divided into two subpopulations
(male and female) and the value of each kinematic parameter for both joints
was compared. Statistical differences were investigated at each knee angle
using a 2-tailed student t-test, assuming unequal population sizes and
unequal variances. Prior to any statistical analysis, the data at each knee
angle were checked for normality. All data are presented for the extension
portion of the movement only.
As a follow-up to previous investigations into the accuracy of fast-PC MRI
[27]-[28], the PF attitude in the axial plane was measured based on the
visual identification of landmarks in the fastcard images [29]. This was
then compared to the results derived from the fast-PC data for two separate
time points. Two subjects, K#13 and K#24, were selected for this analysis
based on the fact that they had the most lateral and medial, respectively,
absolute value of patellar tilt (at full extension) and slope of patellar
tilt versus knee angle. At full extension the patellar and femoral images
were used to define the patellar and femoral 2D attitudes, respectively.
These were then combined to define the PF attitude. The second time point
analyzed was early extension, selected such that the view of the patella in
the femoral image visually matched the patellar image at full extension as
closely as possible. The PF attitude was then defined.

Table II: Tabulated ranges of motion for the average PF and TF kinematics
during TF extension for the current study and the swing phase from the work
of Lafortune [1]




Results

The kinematics of the PF and TF joints (Fig. 3 and 4, Appendix A) were
similar to those previously measured using cine-PC MRI [24], but patellar
rotation was more varus and the translation of the patellar origin was more
superior and posterior. In comparing the average range of kinematic values
to the range of values obtained during the swing phase of Lafortune's [1]
study (one of only two studies that have reported combined 3D PF and TF
kinematics) the rotational data were similar. The translational data had
different ranges (Table II).
All data were determined to be normally distributed (P