Different loads can produce similar bone density distributions

Finite element models of a generic long bone and the proximal femur were used to identify important load characteristics and to determine whether small changes in load affect bone adaptation simulations. We also examined the effect of implants on the sensitivity of bone adaptation simulations to changes in loads. For each model, a primary load set was selected and incorporated in a bone adaptation simulation to generate a primary density distribution. A density-based load estimation method was used to determine a secondary set of loading conditions for each model. Each secondary load set was incorporated in a bone adaptation simulation and the resulting density distribution was compared to the corresponding primary density distribution. Nearly identical density distributions were produced for the natural generic long bone model (average nodal density difference 0.02 g/cm3). For the natural proximal femur model, the density distributions were very similar, but differences were apparent (average nodal density difference 0.07 g/cm3). The same primary and secondary load sets were used for bone adaptation simulations with implant models. For the proximal femur model, density distribution differences with the implant were very slightly less than those of the natural model. For the generic long bone model, the implant amplified differences between density distributions (average nodal density difference 0.14 g/cm3). Thus, variations in loading conditions may partially explain variations in long-term total joint outcome. The total equivalent stimulus load magnitudes for the two load sets for the generic long bone model were within 1%, and the stimulus-weighted average load directions were within 1 degree. The similarity of these parameters and the natural generic long bone density distributions indicate that the overall magnitude and average load direction are key factors affecting bone adaptation.