Types of vertebral fractures could influence the selection of clinical bone mineral measures to predict biomechanical properties

Areal and volumetric BMD (aBMD and vBMD) measured by DXA and quantitative CT (QCT), respectively, are usually employed to predict vertebral fracture risks. In this study, we induced compression and wedge vertebral fractures to test if the types of fracture could influence the selection of bone mineral measures to predict biomechanical properties of vertebral bodies. DXA and QCT were employed to scan twenty-four male cadaveric vertebral bodies of humans for bone mineral content (BMC) and aBMD measures, and vBMD measures, respectively. We computed vBMD measures from three kinds of volumes of interest: intact structures (vertebral body, cortical compartment, and trabecular core), axially middle sections (1.250-1.875 cm height) of the intact structures, and clinically used elliptical regions of trabecular bone. We loaded vertebral bodies to failure for properties of strength (Pu), failure displacement (delta u), and stiffness (K). Thirteen vertebral bodies sustained compression fractures and the remaining sustained wedge fractures. Linear and power regression models were used to test bone mineral predictions for Pu, delta u, and K. We also did equality tests of correlation coefficients. Our results showed aBMD, BMC, and vBMD of the middle section of trabecular bone had the strongest correlations with Pu (R-2 = 0.6420, p < 0.001), delta u (R-2 = 0.4619, p < 0.001), and K (R-2 = 0.5992, p < 0.001) in power regression models, respectively when compression and wedge fractures were mixed. Considering compression fractures only, vBMD of the intact vertebral body displayed the strongest correlations with both Pu (R-2 = 0.6529, p < 0.001) and K (R-2 = 0.6354, p < 0.001) while BMC showed the strongest correlation with delta u (R-2 = 0.4376, p < 0.001) in linear regression models. When only wedge fractures were analyzed, vBMD of the elliptical regions of trabecular bone exhibited the strongest correlations with both Pu (R-2 = 0.5845, p < 0.001) and K (R-2 = 0.6420, p < 0.001) in power regression models, however, no bone mineral measure could significantly correlate with delta u. These results may suggest the type of fracture could influence the determination of bone mineral measures to predict biomechanical properties of vertebral bodies.