A Validated Two-Dimensional Model to Predict TKR Functional Kinematics

Abstract

INTRODUCTION Posterior Stabilized (PS) Total Knee Replacement (TKR) design utilises a cam and post mechanism to generate femoral rollback in higher flexion. The construct is designed to replicate the function of the posterior cruciate ligament (PCL). There is much interest in obtaining optimum post cam engagement to achieve normal femoral rollback and improved outcome after TKR. Previous studies [1] have observed that the PS knee may be ineffective in restoring normal kinematics in high flexion. However, such patient based fluoroscopic studies are demanding in terms of resources and time. Ideally, computer based modelling methodologies can offer a more suitable adjunct. The aims of this study were to develop a validated model capable of predicting the functional kinematics of PS TKR implanted knees, and compare results obtained using this model to those measured in vivo as validation. METHODS Functional kinematics were assessed using the Patella Tendon Angle (PTA) (the angle subtended between the patella tendon and the long axis of the tibia). PTA provides a valid assessment of relative tibial-femoral position [1] and as such is taken as the output of the model. The input to the model is the relative tibial femoral position. This allows the use of a patella-femoral modelling approach similar to that developed by Gill & O’Connor [2]. The model was executed using motion simulation software which uses the constraint force algorithm [3] to achieve a solution, the output parameter being the PTA. A group of ten patients implanted with Scorpio PS implants, at least one year previously for an underlying diagnosis of osteoarthritis, were recruited. The patients were asked to perform a step-up exercise and a deep knee bend whilst undergoing fluoroscopic imaging. The fluoroscopic images were subsequently corrected for distortion and the relative implant orientation was achieved using a 3D reconstruction method. The determined relative tibial femoral orientations were then input to the model. The mathematically obtained PTAs were then compared to those measured from in vivo fluoroscopy images using an established protocol [1]. The mean PTA values for each each data set were compared using paired t tests at 10 degree flexion intervals. Variation between model and in vivo data was further explored using root mean square analysis RESULTS The mean PTA measured in vivo ranged from 11 in extension to -1 at 100 flexion (fig.1). The mean PTA obtained using the model for the same patients was similar ranging from 10 in extension to 0 at 100 flexion. There was no significant difference between the PTAs for the data sets at any point in the range. A large standard deviation is present for both mathematically derived and in vivo data sets. The average root mean square error (RMSE) ranged between 1.17 and 2.10 over the flexion range. Table 1 shows the average RMSE and its standard deviation as well as the significance at 10 intervals.