A finite element model for investigating the influence of keel design and position on unicompartmental knee replacement cementless tibial component fixation

The cementless Oxford Unicompartmental Knee Replacement (OUKR) tibial component relies on an interference fit to achieve initial fixation. The behaviour at the implant-bone interface is not fully understood and hence modelling of implants using Finite Element (FE) software is challenging. With a goal of exploring alternative implant designs with lower fracture risk and adequate fixation, this study aims to investigate whether optimisation of FE model parameters could accurately reproduce experimental results of a pull-out test, which assesses fixation. Finite element models of implants with three methods of fixation (standard keel, small keel, and peg) in a bone analogue foam block were created, in which implants were modelled using an analytical rigid definition and the foam block was modelled as a homogenous linear isotropic material. The total interference and elastic slip were varied in these models and optimised by comparing simulated and experimental results of pull-out tests for two (standard and peg) implant geometries. The optimised interference and elastic slip were validated by comparing simulated and experimental data of a third (small keel) implant geometry. The optimisation of parameters established an interference of 0.16mm and an elastic slip of 0.20mm as most suitable for modelling the experimental force-displacement plots during pull-out. This combination of parameters accurately reproduced the experimental results of the small keel geometry. The maximum pull-out forces from the FE models were consistent with experimental data for each implant design. This study shows that experimental pull-out tests can be accurately modelled using adjusted interference values and non-linear friction and outlines a method for determining these parameters. This study demonstrates that complex problems in modelling implant behaviour can be addressed with relatively simple models. This can potentially lead to the development of implants with reduced risk of failure.