Abstract
Currently, preclinical mechanical wear testing of total knee replacements (TKRs) is done using ideally aligned components using standardized TKR level walking under either force or displacement-control regimes. To understand the influence of implant alignment and testing control regime, we studied the effect of nine component alignment parameters on TKR volumetric wear in silico. We used a computational framework combining Latin Hypercube sampling design of experiments, finite element analysis, and a numerical model of polyethylene wear, to create a predictive model of how component alignment affects wear rate for each control regime. Nine component alignment parameters were investigated, five femoral variables and four tibial variables. To investigate perturbations of the nine implant alignment variables, two separate 300-point designs were executed, one for each control regime. The results were then used to generate surrogate statistical models using stepwise multiple linear regression. Wear at the neutral position was 4.5mm(3)/million cycle and 8.6mm(3)/million cycle for displacement and force-control, respectively. Stepwise multiple linear regression surrogate models were highly significant for each control regime, but force-control generated a stronger predictive model, with a higher R(2), more included terms, and a lower RMSE. Both models predicted transverse plane rotational mismatch can lead to large changes in predicted wear; a transverse plane alignment mismatch of 15° can elicit a change in wear of up to 5mm(3)/million cycle, almost double that of neutral alignment. Therefore, transverse plane alignment is particularly important when considering failure of the implant due to wear.