Limitations of global morphometry in predicting trabecular bone failure.

Efforts in finding independent measures for accurate and reliable prediction of trabecular bone failure have led to the development of a number of morphometric indices characterizing trabecular bone microstructure. Generally, these indices assume a high homogeneity within the bone specimen. However, in the present study we found that the variance in bone volume fraction (BV/TV) in a single bone specimen can be relatively large (CV = 9.07% to 28.23%). To assess the limitations of morphometric indices in the prediction of bone failure for specimens in which the assumption of homogeneity is not met, we harvested 13 cadaveric samples from a single human spine. We tested these cylindrical samples using image-guided failure assessment (IGFA), a technique combining stepwise microcompression and time-lapsed micro-computed tomography (µCT). Additionally, we computed morphometric indices for the entire sample as well as for 10 equal subregions along the anatomical axis. We found that ultimate strength was equally well predicted by BV/TV of the entire sample (R(2)  = 0.55) and BV/TV of the weakest subregion (R(2)  = 0.57). Investigating three-dimensional animations of structural bone failure, we showed that two main failure mechanisms determine the competence of trabecular bone samples; in homogeneous, isotropic trabecular bone samples, competence is determined by a whole set of trabecular elements, whereas in inhomogeneous, anisotropic bone samples a single or a missing trabeculae may induce catastrophic failure. The latter failure mechanism cannot be described by conventional morphometry, indicating the need for novel morphometric indices also applicable to the prediction of failure in inhomogeneous bone samples.