Abstract
INTRODUCTION
Knee contact force during activities after total knee arthroplasty (TKA) is very important, since it directly affects component wear and implant loosening. While several computational models have predicted knee contact force, the reports vary widely based on the type of modeling approach and the assumptions made in the model. The knee is a complex joint, with three compartments of which stability is governed primarily by soft tissues. Multiple muscles control knee motion with antagonistic co-contraction and redundant actions, which adds to the difficulty of accurate dynamic modeling. For accurate clinically relevant predictions a subject-specific approach is necessary to account for inter-patient variability.
METHODS
Data were collected from 3 patients who received custom TKA tibial prostheses instrumented with force transducers and a telemetry system. Knee contact forces were measured during squatting, which was performed up to a knee flexion angle that was possible without discomfort (range, 80–120°). Skin marker-based video motion analysis was used to record knee kinematics. Preoperative CT scans were reconstructed to extract tibiofemoral bone geometry using MIMICS (Materialise, Belgium). Subject-specific musculoskeletal models of dynamic squatting were generated in a commercial software program (LifeMOD, LifeModeler, USA). Contact was modeled between tibiofemoral and patellofemoral articular surfaces and between the quadriceps and trochlear groove to simulate tendon wrapping. Knee ligaments were modeled with nonlinear springs: the attachments of these ligaments were adjusted to subject-specific anatomic landmarks and material properties were assigned from published reports.
RESULTS
Total measured peak ground reaction force was 0.9–1.1 xBW (times of bodyweight) and measured peak knee contact force was 2.2 (±0.2) xBW during squatting. Model predicted peak tibiofemoral contact forces were within the cycle-to-cycle variations for each subject. Model predicted peak patellofemoral contact forces were 0.9–1.1 xBW and peak quadriceps forces were 1.3–1.6 xBW. Mean peak ligament tensions were 55.5 ± 8.8 N for the MCL and 47.1 ± 10.4 N for the LCL.
DISCUSSION
Small differences between predicted and measured forces were likely due to the complexity of the squatting activity, the inherent error in skin marker-based motion capture, and the fact that muscle force was computed from muscle shortening history. Trunk flexion significantly affected the contact force, especially at higher knee flexion angles. Trunk flexion reduced the external flexion moment at the knee leading to reduced quadriceps force and therefore reduced tibiofemoral contact force.
Peak patellofemoral contact forces and quadriceps muscle forces were also lower than previously reported. Although others have reported on hamstring muscle activity during the squat, hamstring forces were low in our models in qualitative agreement with the EMG data that we recorded during squatting. The lack of significant hamstring activity may explain the lower net tibiofemoral contact forces. This model would be very useful tool to predict the effect of surgical techniques on contact forces. Such a model could be used for implant design development to enhance knee function and to predict forces generated during other activities. Finally a subject-specific model could be useful for predicting clinical outcomes.
