No abstract

The key to the development of a successful implant is an understanding of the effect of bone remodelling on its long-term fixation. In this study, clinically observed patterns of bone remodelling have been compared with computer-based predictions for one particular design of prosthesis, the Thrust Plate Prosthesis (Centerpulse Orthopedics, Winterthur, Switzerland). Three-dimensional finite-element models were created using geometrical and bone density data obtained from CT scanning. Results from the bone remodelling simulation indicated that varying the relative rate of bone deposition/resorption and the interfacial conditions between the bone and the implant could produce the trend towards the two clinically observed patterns of remodelling.

In experimental cartilage indentation studies, the indenter is typically a plane-ended or hemispherically ended cylinder and can be either porous or non-porous. Joints such as the hip and knee, however, have much higher radii of curvature than those used in experimental indentation testing. In order to interpret the results from such testing in a physiological context it is therefore useful to explore the effect of contact geometry on the pore pressure and strain distribution generated in the cartilage layer. Articular cartilage can be described as a saturated porous medium, and can be considered a biphasic material consisting of a porous, permeable solid matrix, and an interstitial fluid phase. This behaviour has been predicted in this study using the ABAQUS soils consolidation procedure. Four contact geometries were modelled: two typical experimental configurations (5 mm radii cylindrical indenter and hemispherical indenters) and two effective radii representative of the hip and knee (20 and 100 mm). A 10 per cent deformation, or a load of 0.9 kN, was applied over a ramp time of 2 s, which was then maintained for a further 100 s. The porous indenter generated less pore pressure compared with the equivalent non-porous indenter and produced higher values of compressive strain in the solid matrix. The predictions made using the cylindrical indenters, porous and non-porous, were dominated by the concentrations at the edge of the indenter and overestimated the peak compressive strain in the tissue by a factor of 21 and a factor of 14 respectively when compared with the hip model. The hemispherical indenter predicted peak strains in similar positions to those predicted using physiological radii, however, the magnitude was overestimated by a factor of 2.3 when compared with the knee and by 5.7 when compared with the hip. The pore pressure throughout the cartilage layer reduced significantly as the radius of the indenter was increased.

The resorption of bone in the human femur following total hip arthroplasty is recognized to be related to the loading in the bone surrounding the prosthesis. However, the precise nature of the mechanical signal that influences the biological remodelling activity of the bone is not completely understood. In this study, a validated finite element modelling methodology was combined with a numerical algorithm to simulate the biological changes over time. This was used to produce bone remodelling predictions for an implanted thrust plate prosthesis (Centerpulse Orthopedics Limited) in a patient specific bone model. The analysis was then repeated using different mechanical signals to drive the remodelling algorithm. The results of these simulations were then compared to the patient-specific clinical data, to distinguish which of the candidate signals produced predictions consistent with the clinical evidence. Good agreement was found for a range of strain energy based signals and also deviatoric remodelling signals. The results, however, did not support the use of compressive dilatational strain as a candidate remodelling signal.

Bone remodelling is the adaptation of bone mass in response to localized changes in loading conditions. The nature of the mechanical signal governing remodelling, however, remains the subject of continued investigation. The aims of this study were to use an iterative finite element (FE) bone remodelling technique to explore the effect of different remodelling signals in the prediction of bone remodelling behaviour. A finite element model of the turkey ulna, following that of Brown et al., was analysed using the ABAQUS package. The model was validated against the static predictions of the Brown et al. study. A bone remodelling technique, based on swelling algorithms given by Taylor and Clift, was then applied to predict the dramatic change in loading conditions imposed. The resulting changes in FE mid-shaft bone geometry were compared with the remodelling observed experimentally and showed good agreement. The tensile principal stress was found to be the best remodelling signal under the imposed conditions. Localized sensitivities in the remodelling patterns were found, however, and the definition of the dead zone was modified as a result. Remodelling with the new dead-zone definition showed that both the tensile principal stress and the tensile principal strain produced the remodelling patterns that agreed most closely with experiment.