Vertebral fractures affect at least 12–20% of men and women over the age of 50, and the risk of fracture increases exponentially with age. Despite their high prevalence, the failure mechanisms leading to these fractures are not well understood. For example, clinical observations of fractured vertebra often note that one or both vertebral endplates have collapsed, but the precise involvement of the endplates in the initiation and progression of failure has not yet been defined. The mechanisms of failure may also relate to spatial variations in the density and microstructure of the porous trabecular bone within the vertebra as well as to the health of the adjacent intervertebral discs (IVDs) which transfer loads directly to the vertebral endplates. Delineating the contributions of these factors would shed light on the etiology of vertebral fractures and would aid in development of clinically feasible, patient-specific finite element (FE) models of the vertebra. These models are built from a patient’s quantitative computed tomography (QCT) scan and have shown tremendous promise for accurate, patient-specific estimates of bone strength and fracture risk. Further validation studies are required to assess the impact of the choices of material properties and boundary conditions, as a prerequisite for broad implementation of these FE models in clinical care.

The overall goal of this work was to define the failure processes involved in vertebral fractures and to evaluate the accuracy of patient-specific FE models in simulating these processes. Mechanical testing of human spine segments, in conjunction with micro-computed tomography, enabled the assessment of deformation at the vertebral endplate and deformation throughout the entire bone, as the vertebra was loaded to failure under both axial compression and anterior flexion. These data were compared against predictions of vertebral deformation obtained from QCT-based FE models. The impact of the choice of boundary conditions was specifically examined by comparing the accuracy of the FE predictions between models that simulated applied loads based on measured distributions of pressure within IVDs and models that used highly idealized boundary conditions.

The results of these studies demonstrated that sudden and non-recoverable endplate deflection is a defining feature of biomechanical failure of the vertebra, for both compression and flexion loading. The locations of endplate collapse as vertebral failure progressed were associated with the porosity of the endplate and the microstructure of the underlying trabecular bone. FE analyses incorporating the experimentally observed endplate deflections as boundary conditions provided more accurate predictions of displacements throughout the rest of the vertebra when compared to FE models with highly idealized boundary conditions. Under anterior flexion, the use of boundary conditions informed by measurements of IVD pressure mitigated, but did not eliminate, the inaccuracy of the idealized boundary conditions. No further improvement in accuracy was found when using boundary conditions based on pressure measurements corresponding only to IVDs whose level of degeneration matched that observed in the IVDs adjacent to the vertebra being modeled. Overall, the accuracy of the FE predictions of vertebral deformation was only moderate, particularly near the locations of endplate collapse. The outcomes of this work indicate that the vertebral endplate is principally involved in vertebral fractures and that current methods for QCT-based FE models do not adequately capture this failure mechanism. These outcomes provide a biomechanical rationale for clinical diagnoses of vertebral fracture based on endplate collapse. These outcomes also emphasize that future studies of patient-specific FE models should incorporate physiologically relevant loading conditions and also material properties that more accurately represent the vertebral endplate in order to obtain higher fidelity predictions of vertebral failure.