Cadaver Experiment
Six fresh-frozen human ankle specimens (five from female donors and one
from a male donor, with a mean donor age of eighty years [range, fifty-four to
ninety-eight years]) were obtained at autopsy. No deformities, contractures,
articular degeneration, or ligament injuries were evidenced by visual or
manual inspection. Each specimen was thawed at room temperature before testing
and was dissected free of soft tissue at the ankle except for all major
supporting ligaments, which were kept intact. For mounting in the testing
fixture, the midparts of the tibial and fibular shafts, the calcaneus, and the
distal phalanges were secured in three separate blocks of
polymethylmethacrylate.
Experimental loads were applied with a custom fixture
(Fig. 2) mounted in a
servohydraulic materials testing machine (Bionix, model 858.20; MTS Systems,
Eden Prairie, Minnesota). In this device, a specimen held at a predetermined
amount of ankle flexion was subjected to a primary axial load (600 N,
approximately one body weight), after which one of three modalities of
superimposed secondary load (anterior/posterior drawer force,
inversion/eversion torque, or internal/external rotation torque) was
separately applied to the talus, while motions associated with the two
remaining secondary load modalities were unconstrained. The magnitudes of
anterior/posterior drawer force were 40 and 80 N, similar to those occurring
during gait13.
Similarly, 150 and 300 N-cm of torque were utilized for the magnitudes of
inversion/eversion and internal/external rotation torques. The predetermined
ankle positions were 15° of dorsiflexion, 0° of flexion, 15° of
plantar flexion, and 30° of plantar flexion, values spanning the motion
range occurring during walking on level ground.
Contact stress between the superior-inferior tibiotalar surfaces of each
specimen was measured with a custom-designed real-time contact-stress sensor
(Tekscan, model 5033; Tekscan, Boston, Massachusetts), which was described in
detail elsewhere12.
The inserts are very thin (0.1 mm) and flexible, allowing conformation to
curved joint surfaces. The active area measures 27 mm (medial-lateral
direction) by 39 mm (anterior-posterior direction) and incorporates a
uniformly distributed 32-by-46 array of sensing elements (sensels), yielding a
spatial resolution of 0.694 mm2 per sensel. The size was chosen to
fully cover a typical tibiotalar articulation.
To account for differing joint curvatures and sensor variability, sensors
were calibrated in situ for each specimen. The sensor was inserted into the
joint through an anterior arthrotomy, and the specimen was mounted in the test
fixture. The servohydraulic testing machine was then used to apply a series of
known increasing axial loads. At each load level, observation of a real-time
graphic display of sensel array output verified that the load was transmitted
only across the active area of the sensor. Tekscan software performed an
integration of the sensel output over the total engaged area at each of the
known loads and derived a stress versus sensel output curve for that sensor.
This calibration curve was then used for the subsequent tests on that
specimen. During calibration and successive testing, sensor and joint surfaces
were kept well lubricated with petroleum jelly to eliminate any possible shear
artifact.
The tensile conditions of the major peri-ankle ligaments also were
monitored. Engineering strain (? = l/L0, where
l is the extension of the ligament and L0 is its
initial length determined for an incipiently taut condition) was measured for
each ligament, in a previously established
manner14,15,
with the use of miniature differential variable reluctance transducers
(Microminiature DVRT; MicroStrain, Williston, Vermont). The anterior
talofibular, calcaneofibular, and posterior talofibular ligaments as well as
the anterior, middle, and posterior bundles of the superficial deltoid
ligament complex were instrumented.
In the loading experiment, each specimen was subjected to all three
secondary load modalities (anterior/posterior force, inversion/eversion
torque, and internal/external rotation torque) in a prerandomized order. In
each test, prior to the secondary load application, a contact-stress
distribution under only the primary axial force (axial-force-only condition)
was recorded. The secondary load was then applied, and data registration was
repeated for each load magnitude, followed by a reregistration under an
axial-force-only condition. This sequence was repeated for all ankle
positions. The orderings of both load magnitude and ankle position were also
randomized. Immediately after each test load was applied, the contact-stress
data were collected for one second at a rate of 10 Hz and were temporally
averaged. (During data acquisition, the measured contact stresses were very
stable; typically, deviation of total force was within 1% and migration of the
contact center was <0.1 mm.)
During testing, contact stress and ligament strain were monitored. To avoid
the possibility of specimen damage, the limits of allowable local contact
stress and ligament strain were set at 15 MPa and 10%, respectively; both
values were based on previous studies in the
literature8,16.
Data Processing
For each of the three modalities of secondary load applied at each of the
four predetermined ankle positions, the data set for a specimen consisted of
six contact-stress maps: one baseline distribution map under the
axial-force-only condition, four maps under the corresponding
axial-plus-secondary load condition (one for each secondary load magnitude),
and a repeated baseline map. As a result, each specimen had a total of
seventy-two maps. To describe the locations of contact-stress changes
associated with secondary loading, an x-y coordinate system was established
for each specimen, for each loading modality, and for each ankle position. The
origin of each coordinate system was at the center of contact stresses under
only primary axial loading; the contact-stress-center position was determined,
with consideration of both location and magnitude, from the data averaging the
initial and repeated axial-force-only conditions. The absolute differences in
the contact-stress-center positions between the two axial-force-only
conditions were on average smaller than the spatial resolution of the sensor
(0.64 mm), confirming that any potential sensor migration was minimal. The
contact-stress sensor also was closely observed for motion relative to the
distal part of the tibia during the loading sequences; no motion was visually
detectable.
The stress changes associated with secondary loading were calculated from
an axial-plus-secondary load map and the corresponding axial-force-only map,
by subtraction on a sensel-by-sensel basis
(Fig. 3). The resulting maps
displayed the distribution of contact-stress changes, including both elevation
(positive changes) and reduction (negative changes). To assess spatial
patterns of contact-stress changes, the centers of positive changes (again,
with consideration of both location and magnitude) were determined. This
analysis was applied for each secondary load direction/magnitude at all ankle
positions.
Computer-Model Analysis
The geometry of the interface between the superior talar dome and
corresponding tibial plafond surfaces was modeled as two adjacent spherical
sectors (Fig. 4). This was
implemented in MATLAB (version 7.0; MathWorks, Natick, Massachusetts). The
radius of each sphere was 25 mm, and separation of their centers was 20 mm.
The ankle motion axis was identified as the line connecting the two sphere
centers, and the midpoint of this line was defined as the center of the model.
The stress-change maps were positioned on this model surface such that the
origin was aligned with a vertical axis passing through the model center, and
with the medial-lateral axis parallel to the ankle axis. Contact stresses on
the model surface were assumed to act locally perpendicular to the sensel
surface, and the corresponding incremental force acting on each sensel was
resolved into three components: axial, anterior/posterior, and
medial/lateral.
For each stress-change map, the surface resistance associated with the
corresponding secondary load was calculated from the appropriate force
components. The sum of the anterior/posterior components was assumed to
represent the surface resistance to the applied secondary anterior/posterior
load. Version torque was the sum of the moments about the anterior/posterior
axis of the axial and medial/lateral force components. Internal/external
rotation torque about the vertical axis was calculated from the sum of the
moments of anterior/posterior and medial/lateral force components. As a
result, the effects of all measured contact-stress changes were counted in
these surface resistance calculations. However, (coupled) resistances to
motions in other degrees of freedom were not extracted, as they would be
expected to be inconsistent and of small magnitude. Friction was assumed to be
negligible.
For each of the three secondary loading modalities, and at each ankle joint
position, the resisting force (or torque) that was calculated was linearly
regressed against the magnitude of the secondarily applied force (or torque)
(Fig. 5). The regression
coefficient (the slope of the trend line) was considered to represent the
relative contribution of the superior-inferior tibiotalar surfaces to ankle
stability (i.e., a slope of 1.0 would correspond to surface resistance
accounting for 100% of the applied secondary load and would indicate that
ankle stability was completely dependent on articular surface restraint). This
analysis was applied individually to each specimen.
In the cadaver experiment, all tests at 15° of dorsiflexion, 0° of
flexion, and 15° of plantar flexion were completed for every specimen. At
30° of plantar flexion, in one specimen, peak contact stress beyond the
limit was observed under even the axial-force-only condition, so all tests at
that position for that specimen were omitted. In another specimen, excessive
strain of the anterior talofibular ligament occurred with 80 N of anterior
force at 30° of plantar flexion; therefore, contact-stress measurement
under this condition was also omitted.
In the anterior/posterior test at 0° of ankle flexion, anterior drawer
forces (applied to the talus) elevated contact stresses on the anterior region
of the superior-inferior tibiotalar interface, and posterior forces caused a
similar result on the posterior region
(Fig. 6). This trend was
consistent in all ankle positions, regardless of force magnitude. In the model
analysis, a nearly linear relationship between secondary load and surface
resistance was found in every specimen at every ankle position (R2
> 0.92), suggesting that the superior-inferior tibiotalar articulation
consistently contributed very substantially to ankle stabilization. These
stress changes accounted for approximately 70% of the total contribution to
passive ankle stability, at every ankle position
(Table I).
Inversion torques elevated contact stresses on the medial region, while
eversion torques caused a similar result on the lateral region, for all load
magnitudes and ankle positions (Fig.
7). At 15° of dorsiflexion, 0° of flexion, and 15° of
plantar flexion, the R2 values were >0.92 and the level of
resistive contribution ranged from 45% to 60%
(Table II). However, at 30°
of plantar flexion, two specimens had low R2 values (0.11 and 0.79)
and low contribution values (2% and 9%), suggesting that the superior-inferior
tibiotalar interface was no longer engaged in ankle stabilization. In the
remaining three specimens tested at 30° of plantar flexion, the
R2 values were >0.95 and the contributions to stability were
approximately 50%.
With internal rotation torques, positive changes in contact stresses
occurred at two locations; one was on the anteromedial region, and the other
was on the posterolateral region (Fig.
8). Similarly, positive changes with external rotation torques
occurred on the anterolateral and posteromedial regions. The R2
values averaged >0.90 at every ankle position, although the surface's
relative contribution to torque resistance was only 22% to 30%
(Table III).
In the cadaver experiment, with every secondary load modality tested,
redistribution of contact stresses in response to an altered external load
showed a reproducible and load-direction-dependent pattern. In the geometrical
model analysis, it was found that each redistribution pattern was effective in
changing articular surface resistance so as to restrain the ankle against the
altered external load. This resistance arose spontaneously, contributing to
maintenance of the physiologic apposition; the highly congruent
three-dimensional conformation of the superior-inferior tibiotalar
articulation is probably responsible for this effect. The hypothesized
reproducible patterns of ankle contact-stress changes were demonstrated.
The model analysis included calculation of the contribution of the
articular surface to passive ankle stability. For computational simplicity,
the generalized model geometry did not account for the effects of individual
topographic variability. For example, the spherical radii of the model (25 mm)
were larger than the actual sagittal radius of the talar dome in any specimen
(20 to 24 mm was the best fitting radius on lateral radiographs). To
appreciate the influence of computational geometrical simplifications, a set
of illustrative cases was run with use of series-average contact stresses, but
individual features of the computational model surface were perturbed. For a
20% reduction in overall computational model surface size, the contribution
level to anterior/posterior stabilization was calculated to be 17% larger,
suggesting possible underestimation of the contribution levels associated with
the larger model radii. Regarding version stability, the larger model scale
might have caused overestimation (the same scale-down model calculated a 12%
smaller contribution level). Similarly, the relatively deep central groove of
the model might have caused overestimation of the contribution to
inversion/eversion and internal/external rotation stability (for a 50%
shallower model groove, the contribution levels were calculated to be 10% and
5% smaller, respectively). The assumption that the entire axial load passed
across the sensor surface also might have caused some overestimation, since
physiologically the fibula transmits 3.7% to 13.0% of axial forces, depending
on ankle
positioning17. As a
result of these possible overestimations and underestimations, the reported
articular contribution levels may not fully represent the function of natural
ankle topography. However, the data appear to be meaningful for studying the
general roles of the joint geometry.
The relative contribution data suggest that articular surface restraint
plays the primary role in anterior/posterior ankle stabilization. A major role
of articular surface restraint in inversion/eversion and internal/external
rotation stability also is suggested. In two previous cadaver studies,
Stormont et al.10
and Watanabe et
al.11 utilized
ligament-sectioning paradigms to estimate the relative contribution of
articular surface restraint to passive stability of the ankle under axial
loading. The reported contribution levels accounted for 100% of
anterior/posterior
stability11, 100%
of version
stability10, and
30%10 or
60%11 of
internal/external rotation stability. Considering that the present study did
not include the effect of the talomalleolar articulations, which appear to
play important roles in version and internal/external rotation stability, the
overall trend found in all three studies is consistent, suggesting that the
methodology that we employed is valid.
The patterns of contact-stress redistribution in the present study suggest
that different parts of the superior-inferior tibiotalar articulation play
different specific roles in ankle stabilization. The talomalleolar
articulations, although not specifically addressed by this study design, are
postulated to function similarly. Presumably, passive ankle stability under
weight-bearing conditions is dictated by the integrity of the articular
surface geometry. This implies that, in the clinical setting, any
abnormalities that alter ankle geometry, either intra-articular (e.g., a
defect or surface incongruity) or extra-articular (e.g., periarticular
malalignment), may affect ankle stability under weight-bearing conditions,
possibly leading to abnormal joint kinematics during locomotive activities.
The effect of such pathological conditions on articular contact stress has
been previously assessed in quasistatic loading experiments with cadaver
models of a defect of the posterior
malleolus18,
angular deformity of the distal part of the
tibia19,20,
and lateral or proximal displacement of the lateral
malleolus21-24.
In these studies, although no injurious-level peak stresses were demonstrated,
every intervention reduced contact stress at the specific area of interest. In
a recent cadaver experiment that explored ankle contact stress during a
dynamic loading sequence, slight step-off of the anterolateral plafond surface
caused significant elevation of instantaneous temporal and spatial gradients
of contact stresses (p < 0.001 and p < 0.05,
respectively)25,26.
These biomechanical effects should be considered in the management of
intra-articular fractures or of osteoarthritis, especially when a surgical
procedure is being planned.
In conclusion, our experimental results documented that, in the ankle under
weight-bearing conditions, redistribution of contact stress in response to
alteration of external load involves reproducible patterns specific to the
direction of the altered external load. In the geometrical model analysis,
each of these patterns was found to be effective in changing articular surface
resistance to maintain ankle stability. To our knowledge, this is the first
study that experimentally supports the conceptual mechanism of articular
surface restraint in the human ankle, advancing our understanding of the
mechanisms of ankle motion control under weight-bearing conditions.
Geometrical model analysis focusing on specimen-specific anatomical
information would provide more detailed information, such as the effects of
individual variability or of traumatic alteration. ?