Removal of the cortical endplates has little effect on ultimate load and damage distribution in QCT-based voxel models of human lumbar vertebrae under axial compression

Every year, 500,000 osteoporotic vertebral compression fractures occur in Europe. Quantitative computed tomography (QCT)-based finite element (FE) voxel models predict ultimate force whether they simulate vertebral bodies embedded in polymethylmethacrylate (PMMA) or vertebral sections with both endplates removed. To assess the effect of endplate removal in those predictions, non-linear FE analyses of QCT-based voxel models of human vertebral bodies were performed. High resolution pQCT images of 11 human lumbar vertebrae without posterior elements were coarsened to clinical resolution and bone volume fraction was used to determine the elastic, plastic and damage behavior of bone tissue. Three model boundary conditions (BCs) were chosen: the endplates were cropped (BC1, BC2) or voxel layers were added on the intact vertebrae to mimic embedding (BC3). For BC1 and BC3, the bottom nodes were fully constrained and the top nodes were constrained transversely while both node sets were freed transversely for BC2. Axial displacement was prescribed to the top nodes. In each model, we compared ultimate force and damage distribution during post-yield loading. The results showed that ultimate forces obtained with BC3 correlated perfectly with those computed with BC1 (R2=0.9988) and BC2 (R2=0.9987), but were in average 3.4% lower and 6% higher respectively. Moreover, good correlation of damage distribution calculated for BC3 was found with those of BC1 (R2=0.92) and BC2 (R2=0.73). This study demonstrated that voxel models of vertebral sections provide the same ultimate forces and damage distributions as embedded vertebral bodies, but with less preprocessing and computing time required.