QCT-based finite element models predict human vertebral strength in vitro significantly better than simulated DEXA

Abstract

While dual energy X-ray absorptiometry (DXA) is considered the gold standard to evaluate fracture risk in vivo, in the present study, the quantitative computed tomography (QCT)-based finite element modeling has been found to provide a quantitative and significantly improved prediction of vertebral strength in vitro. This technique might be used in vivo considering however the much larger doses of radiation needed for QCT. Vertebral fracture is a common medical problem in osteoporotic individuals. Bone mineral density (BMD) is the gold standard measure to evaluate fracture risk in vivo. QCT-based finite element (FE) modeling is an engineering method to predict vertebral strength. The aim of this study was to compare the ability of FE and clinical diagnostic tools to predict vertebral strength in vitro using an improved testing protocol. Thirty-seven vertebral sections were scanned with QCT and high resolution peripheral QCT (HR-pQCT). Bone mineral content (BMC), total BMD (tBMD), areal BMD from lateral (aBMD-lat), and anterior-posterior (aBMD-ap) projections were evaluated for both resolutions. Wedge-shaped fractures were then induced in each specimen with a novel testing setup. Nonlinear homogenized FE models (hFE) and linear micro-FE (μFE) were generated from QCT and HR-pQCT images, respectively. For experiments and models, both structural properties (stiffness, ultimate load) and material properties (apparent modulus and strength) were computed and compared. Both hFE and μFE models predicted material properties better than structural ones and predicted strength significantly better than aBMD computed from QCT and HR-pQCT (hFE: R² = 0.79, μFE: R² = 0.88, aBMD-ap: R² = 0.48-0.47, aBMD-lat: R² = 0.41-0.43). Moreover, the hFE provided reasonable quantitative estimations of the experimental mechanical properties without fitting the model parameters. The QCT-based hFE method provides a quantitative and significantly improved prediction of vertebral strength in vitro when compared to simulated DXA. This superior predictive power needs to be verified for loading conditions that simulate even more the in vivo case for human vertebrae.