Novel Method for Standardizing Contact Stresses on Tibial Plateaus during
Simulated Gait: A Cadaveric Model
Hongsheng Wang, PhD, Tony Chen, PhD, Matthew F. Koff, Ph.D., Hollis Potter, M.D., Russell Warren, M.D., Suzanne A. Maher,
PhD.
Hospital for Special Surgery, New York, NY, USA.
Disclosures:
H. Wang: None. T. Chen: None. M.F. Koff: None. H. Potter: None. R. Warren: None. S.A. Maher: None.
Introduction: Surprisingly little is known about the contact mechanics of the human knee during activities of daily living. This is
in part caused by our inability to directly measure joint contact forces in vivo, and the inadequacies of cadaveric tests that
cannot appropriately mimic the complex multidirectional and dynamic forces of daily activities. Even in cadaveric tests that can
apply physiological loads[1,2,3], knee-to-knee variability in osseous and soft tissue geometry, and ligament constraints hinder
our ability to identify common contact patterns that might exist[2]. The objective of this study was to determine if common
contact stress distribution and center of contact trajectories exist across human knees during the activity of walking. To
accomplish this objective, a geometry-based normalization method was developed and applied to assess the contact mechanics
of a set of cadaveric knees subjected to loading conditions that mimic gait.
Methods: 12 human cadaveric knees were stripped of musculature and augmented with a thin electronic sensor (4010N,
Tekscan Inc., MA) which was placed across the tibial plateau. The sensor was sutured to the tibial attachment of ACL and the
posterior capsule. Each knee was mounted on a load-controlled Stanmore KC Knee Joint Simulator (University College of
London)[3]. Axial force, anterior-posterior force, internal/external torque, and flexion angle were programmed to mimic the
activity of gait (ISO 14243-1). Contact stress data was recorded at a frequency of 100 Hz, and exported to a custom Matlab
program (MathWorks Inc., MA) for analysis. To align and scale the contact stress maps based on individual knee geometry, the
stress maps at peak axial force for each cadaveric knee were selected. Using a least square method, circles were fit to the
periphery of each meniscus on the contact stress map (Fig. 1A). To ensure that this method accurately identified the meniscus
size, five of the twelve knees were scanned using MRI. A three dimensional (3D) CUBE scan[4] was acquired on a 3T clinical
scanner (GE Healthcare, Waukesha, WI) with an 8 channel phased array knee coil (Invivo, Gainesville, FL) to generate a
volumetric dataset for meniscal segmentation: echo time = 31 ms, repetition time = 2500 ms, receiver bandwidth = ± 41.7 kHz ,
with voxel size: 0.3 x 0.3 x 0.6 mm3. Images were manually segmented using custom software, and 3D models of menisci were
created to evaluate meniscal geometry. Points on the peripheral edges of each meniscus were identified and circumscribed
circles were computed (Fig. 1A). Finally, the radii of the menisci measured from the 3D model were compared to those
calculated from the 2D contact stress map. Pearson’s correlation coefficient was calculated for this comparison.
The centers of the best-fit circles of all knees were ‘normalized’ to match the fit circle of a template knee which has an average
size and geometry (Fig. 2B). After normalization, the tibial plateau was divided into four quadrants (Fig. 1A): posterior peripheral
(PP), anterior peripheral (AP), anterior central (AC), posterior central (PC). The peak contact stress was quantified and compared
(paired t-test) within each quadrant at two points of the cycle: early (14%) and late (45%) in the stance phase. To assess the
location of the center of contact, the weighted center of contact Stress (WCoCS) was calculated (WCoCS = sum(Si*Pi)/sum(Si),
where Si denotes the contact stress at sensing element-i, Pi is the coordinate).
Results: The approach of estimating the meniscal radius using the contact stress maps yielded significant linear correlations with
that measured from MRI models (medial R2 = 0.68, lateral R2 = 0.72, p < 0.01). On the medial plateau, the peak stresses in the
anterior quadrants were significantly increased (p < 0.05) during late stance as compared to early stance (Fig. 2). The posterior
quadrants had similar peak contact stress during early and late stance phases. On the lateral plateau, all quadrants had similar
peak contact stresses at early and late stance (Fig. 2), but the AC quadrant had the overall highest contact stress on both the
medial (6.03 MPa) and lateral plateaus (6.02 MPa). There was a greater anterior-posterior excursion of WCoCS on the medial
plateau than on the lateral (Fig. 3). The rotation of the axis drawn through the medial and lateral WCoCSs demonstrated a lateral
pivot, indicating that the center of rotation is mainly on the lateral plateau. The WCoCS axis first internally rotates during loading
response, and then external rotates during midstance and early terminal stance phases. During the pre-swing phase, the femur
rotates internally.
Discussion: By way of a novel geometry-based normalization method, we identified common contact stress distributions and
common center of contact trajectories across twelve human knees during gait. During late stance, the anterior aspect of the
medial tibial plateau experienced significantly higher peak contact stresses than the posterior aspects. This may be due to the
fact that when the contact center moves anteriorly as the knee approaches full extension during late stance, contact moves over
the anterior horn of medial meniscus which is thinner and less able to distribute joint forces. Given that the anterior region of
the medial plateau has the thickest articular cartilage (Andriacchi et al.[5] and Li et al.[6]) the higher contact stresses
experienced by this region suggests that the morphology of knee cartilage is conditioned to meet the higher loads. The anteriorposterior excursion of WCoCS on the medial tibial plateau was larger than that on the lateral; moreover the motion of the axis
connecting the WCoCS on both plateaus throughout gait suggests that lateral pivoting is occuring. While controversial, these
findings are similar to results previously described by Kozanek et al.[7] and Koo et al.[8]. The greater translational magnitude on
the medial plateau implies an internal tibial rotation for the majority of stance phase. The differences in contact mechanics
between the medial and lateral knee compartments may have contribution to the clinically observed faster cartilage
degeneration following lateral meniscectomy compared to that following medial meniscectomy [9].
Significance: Despite variability in knee geometry and soft tissue constraints across human knees, common stress patterns exist
during the activity of walking. Identifying these commonalities will help us better understand the conditions to which scaffolds
or implants will be subjected to in vivo.
Acknowledgments: Research reported was supported by National Institutes of Health (R01 AR057343) and NIH funded Clinical
and Translational Science Center at Weill Cornell Medical College (KL2RR000458). We also thank The Clark and Kirby Foundation,
The Leo Rosner Foundation, and The Russell Warren Chair in Tissue Engineering.
References: 1. Bedi et al. Arthroscopy 2012; 2. Bedi et al. JBJS Am. 2010; 3. Bedi et al, Proc Inst Mech Eng H 2013; 4. Gold et al.,
AJR Am J Roentgenol 2007; 5. Andriacchi et al, JBJS Am. 2005; 6. Li et al, Clin Biomech 2005; 7. Kozanek M et al. J Biomech. 2009;
8. Koo S and Andriacchi T et al. J Biomech. 2008; 9. Rodeo SA, Am J Sports Med.
2001
ORS 2014 Annual Meeting
Poster No: 1686