INTRODUCTION: Two major causes of failure of total knee replacements (TKRs) are component misalignment  and polyethylene wear . Studies on component alignment and its effect on wear is relevant not only clinically but in the mechanical testing field, as small variations in alignment may account for the large inter- and intra-lab variability seen in wear tests . In this study, the effect of 9 component alignment parameters on predicted volumetric wear in displacement control TKR simulations is investigated using a validated TKR finite element analysis (FEA) wear model combined with a Latin Hypercube Sampling (LHS) design of experiments (DOE) approach. A surrogate model generated using stepwise multiple linear regression is used to evaluate the effect on volumetric wear of each of the input parameters. METHODS: The effect of 9 component alignment parameters on TKR volumetric wear was investigated using a previously validated TKR FEA wear model  incorporated into a computational framework developed in the Isight Execution Engine (Dassault Systèmes). This framework was used to run a 180-point LHS DOE of 9 component alignment parameters: the internal/external (IE) rotation of the femoral component, the IE rotation of the tibial component, the femoral component varus/valgus (VV) angle, the tibial insert anterior/posterior (AP) position, the tibial component rotation in the flexion/extension (FE) axis (tibial slope), the location of the femoral center of rotation (CoR) in both the AP and superior/inferior (SI) directions, and the location of the tibial IE axis in the medial/lateral (ML) direction. Baseline value (ISO), and bounds for the 9 parameters used in this study are found in Table 1. ISO 14243-3:2014 displacement control kinematic and loading inputs were used. To evaluate the effect each parameter had on wear and generate a surrogate model for use in the prediction of wear, step-wise multiple linear regression up to second order terms was performed on the results from the LHS DOE using MATLAB® v2017a (The Mathworks Inc.). In the step-wise approach, model terms that do not contribute to the explained variance of the model are excluded. An effects plot of volumetric wear was generated using the remaining terms in the model. The effects plot visualizes the maximum change on the output that can be elicited as the result of a change in a particular input parameter using the regression model, within the original bounds of the data. The range within which this change occurs is automatically calculated. RESULTS: Volumetric wear varied from 2.7 to 17.2 mm3/million cycles (MC) within the simulations performed. The step-wise linear regression algorithm generated a model that included all first order terms except for femoral component VV angle, and several interactions and second order terms (Table 1). All terms were significant (p<0.05) except for the location of the femoral CoR in the AP direction and the ML location of the tibial IE axis. The regression model was highly significant to p << 0.001. Model R2 ¬ was 0.72, explaining most of the variance in the FEA prediction. The output for this study was volume change, with more negative values representing higher volumetric wear. According to the effects plot, the three most influential parameters were femoral component IE angle, tibial insert IE angle, and tibial slope (Figure 1). For femoral component IE angle, more wear occurs (-4.7mm3/MC) when going from -2.2° (internal rotation) to 15° (external rotation) indicating more wear for high external rotation of the femoral component. For tibial insert IE angle, more wear occurs (-3.6mm3/MC), when modifying its value from 3.9° (external rotation) to -15° (internal rotation) indicating more internal rotation of the tibial insert leads to high wear. DISCUSSION: In this study, the effect of TKR component alignment on wear was evaluated. Nine alignment parameters were investigated and eight were found to have significant influences on wear, with femoral and tibial component IE angles and tibial slope the most influential. The linear regression model generated was able to explain 72% of the variability in the output data. As a surrogate model, the regression model can enable predictions of the effect of component alignment in a fraction of the time it would take to run additional FEA simulations and help to better understand how component alignment affects wear in TKRs. The results are comparable to a similar study done by Pal et al. 2008, although the previous investigation was performed in force control. Despite the differences in study and model design, the comparable results lend credibility to both investigations, help to validate the results, and can reveal differences in force versus displacement control. Clinically, component alignment is considered primarily when discussing patellar tracking and not wear. In addition, component alignment is considered primarily in the frontal plane and not the transverse plane. This study suggests that TKR wear is highly sensitive to transverse plane alignment and care should be taken to properly align components. SIGNIFICANCE: This study demonstrates that component rotational alignment in the transverse plane can have a large influence on TKR wear and therefore care should be taken to properly align components both during surgery and during mechanical testing REFERENCES:  Schroer WC et al. J Arthroplasty 2013, 28:116–19.  Fraser et al., J Knee Surg, 2015, 28(2) 139-144.  Pal et al. Wear 2008, 264:701-7  Mell SP et al. Proc Inst Mech Eng H 2018, 232:545-52. ACKNOWLEDGEMENTS: Partial funding NIH R01 AR 059843. Thanks to Zimmer for providing CAD models of the NexGen Cruciate Retaining TKR.
No datasets are available for this submission.
No license information is available for this submission.