Could initial guess of the ligament parameters during estimation procedures affect post-operative predictions of knee laxity following total knee arthroplasty?

Musculoskeletal modelling can assist in subject-specific surgical planning of operations like Total Knee Arthroplasty (TKA) [1]. For this purpose, subject-specific models are developed based on medical images, while the needed ligament properties are obtained from literature values or via optimization procedures [2]. These optimization procedures adopt literature ligament values based on cadaveric experiments as initial guess and vary them until simulated joint biomechanical behavior matches respective experimental measurements, often laxity tests. Although optimization procedures are, in general, depended in their initial guesses, not enough evidence exist on how such initial guesses influence procedures for obtaining subject-specific ligament parameters and what effects they might have on clinical decisions. Does the choice of initial guess of ligament parameters, during the optimization-based estimation procedure, influence TKA preoperative planning? Pre and post-operative medical images and laxity profiles were obtained from one cadaveric specimen. After the post-operative laxity tests, tensile tests were performed on isolated ligamentous structures to obtain ligament force-elongation curves. The pre-operative medical images were used to develop a subject-specific model of the tibiofemoral joint and the anterior and posterior crucial, medial (MCL) and lateral collateral (LCL) and anterior lateral ligament (ALL). The laxity tests were used to optimize ligament parameters starting from two different initial guesses: set 1 had literature values and set 2 included subject-specific stiffness values obtained from the ligament tensile tests. The optimization procedures provided two different solutions with similar preoperative laxity behavior. The two different sets of optimized ligament parameters were implemented into post-operative musculoskeletal models that were used to predict post-operative laxity. The predictions of the models were computed by static equilibrium among the externally applied load, the contact forces of the articulate cartilage and the ligament forces [3]. The laxity predictions were compared against the ones obtained experimentally by tracking the tibia and femoral bones with bone pin trackers and motion analysis. Similar post-operative laxity profiles were obtained for the two different sets of ligament parameters (Fig. 1) however, significant differences were observed for the ligament forces. Although good predictions of the post-operative laxity were obtained with both sets of optimized ligaments, the observed differences in the ligament loading suggest that different decisions would have been made pre-operatively. Possibly different sizes of the tibial inserts would have been used for the first set of parameters to distribute the ligament forces more equally to the surrounding structures. Alternatively, a different soft tissue balancing approach would have been adopted to release the high observed ligament loads. This case study highlights possible dangers of using computationally predicted ligament properties starting from generic values. As true ligament properties are a priori unknown in clinical settings, both experimental and computational methods are needed to ensure the best possible ligament parameters estimation.Download : Download high-res image (185KB)Download : Download full-size image