Simulation of the Fatigue Performance of Spinal Vertebrae

Over 25% of the population are expected to suffer a vertebral fracture over the course of their lifetime (Palastanga and Soames, 2011). This can lead to severe pain and a dramatically reduced quality of life for the patient. Vertebroplasty is a surgical intervention for the treatment of osteoporotic vertebral compression fractures, and there is contradictory evidence as to the efficacy of the procedure. Experimental and finite element (FE) investigations have been undertaken to evaluate the mechanical behaviour of damaged vertebrae, and the effects of vertebroplasty immediately after cement injection. However further work is required to investigate the longer-term behaviour of damaged and treated vertebrae. The aim of this research was to develop experimental and FE fatigue simulation techniques suitable for investigating the longer-term mechanical behaviour of fractured vertebrae, both when left untreated and following vertebroplasty. A combined experimental and FE approach was adopted for this study. Experimental fatigue methods were established by first developing a damage model and vertebroplasty repair techniques in bovine tail vertebrae. Fatigue testing was then carried out on both cement augmented and untreated specimens, and quantified for different levels of loading. Subsequently specimen-specific FE models were created and used to optimise a density to Young’s modulus conversion parameter, where density was found from micro CT images, allowing for variation in bone stiffness to be captured in the models. Yield properties were then determined, also using optimisation, to capture varying yield behaviour across the bone in an elastic perfectly-plastic FE model. Models were compared against experimental data and shown to predict stiffness well and adequately predict yield. A fatigue simulation method was then developed by creating an automated script to implement material property changes in the models on an iterative basis. These models were then directly compared to experimental fatigue displacement data, and microCT images of fatigue failure, in the un-treated vertebrae. The experimental fatigue testing showed no significant difference in the number of cycles withstood before failure occurred for the un-treated and cement augmented groups. Differences were difficult to identify between groups due to large variations in fatigue response between specimens. However, there was some evidence to suggest that augmented vertebrae retain mechanical stiffness through fatigue testing to a greater degree than un-treated vertebrae. It was found that the fatigue simulation methods showed a good correlation between predicted displacements after large numbers of cycles and experimental displacements at failure, in cases where the plastic strain response in the FE model was not affected by the assumed boundary conditions. However, in some cases, the boundary conditions resulted in a poor distribution of plastic strain, and poor correlation. Additionally, the models showed the potential to give a reasonable indication of fracture locations in some cases. Further work is required to improve the representation of experimental boundary conditions in the models. Although further work is needed to simulate the vertebroplasty procedure in the FE models, the methods developed have the potential to be applied to examine the fatigue behaviour of human vertebrae and a range of different treatment scenarios.