Vertebral fractures are the hallmark of osteoporosis, yet the failure mechanisms involved in these fractures are not well understood. Current approaches to predicting fracture risk rely on average measures of bone mineral density in the vertebra, which are imperfect predictors of vertebral strength and poor predictors of fracture risk. Prior research has established that substantial regional variations in density exist throughout the vertebra and has suggested several biomechanical consequences of these variations. The overall goal of this dissertation was to characterize failure mechanisms in human vertebrae, with specific emphasis on the role of intra-vertebral heterogeneity in density and microstructure and on identifying clinically feasible techniques for predicting fracture risk.

Using images obtained from micro-computed tomography (IJCT) and quantitative computed tomography (QCT), the intra-vertebral heterogeneity in bone density was quantified in cadaveric specimens. Quantitative measures of this heterogeneity improved predictions of vertebral strength as compared to predictions based only on mean density. Subsequently, the intra-vertebral heterogeneity in density was measured via QCT in a cohort of post-menopausal women and was found to be lower in those who had sustained a vertebral fracture vs. in age-matched individuals without fracture.

The next set of studies focused on assessing the accuracy of finite element (FE) models for predicting vertebral failure. Digital volume correlation (DVC) was used to measure the deformations sustained throughout the vertebra during compression tests. These results were compared against deformation patterns predicted using FE models created from QCT images of the vertebrae. Good agreement was found between predicted and measured deformations when the boundary conditions were accurately defined, despite simplifications made in representing material properties.

The outcomes from this dissertation demonstrate that the intra-vertebral heterogeneity in density contributes to bone strength and has promise as a clinically feasible indicator of fracture risk. OCT-based FE models, which by definition account for this heterogeneity, are another promising technique, yet will likely require non-invasive techniques for estimating vertebral loading to provide the requisite accuracy in failure predictions. These two engineering approaches that account for the spatial heterogeneity in density within the vertebra may lead to more sensitive and specific indicators of fracture risk.