Cortical bone finite element models in the estimation of experimentally measured failure loads in the proximal femur

Highly accurate nonlinear finite element (FE) models have been presented to estimate bone fracture load. However, these complex models require high computational capacity, which restricts their clinical applicability. The objective of this experimental FE study was to assess the predictive value of a more simple cortical bone simulation model in the estimation of experimentally measured fracture load of the proximal femur. The prediction was compared with that of DXA, and with the prediction of our previous, more complex FE model including trabecular bone. Sixty-one formalin-fixed cadaver femora (from 41 women and 20 men, age 55-100 years) were scanned using a multi-detector CT and were mechanically tested for failure in a sideways fall loading configuration. Trabecular bone was completely removed from the FE models and only cortical bone was analyzed. The training set FE models (N = 21) was used to establish the stress and strain thresholds for the element failure criteria. Bi-linear elastoplastic FE analysis was performed based on the CT images. The validation set (N = 40) was used to estimate the fracture load. The estimated fracture load values were highly correlated with the experimental data (r(2) = 0.73; p<0.001). The slope was 1.128, with an intercept of -360 N, which was not significantly different from 1 and 0, respectively. DXA-based BMD and BMC correlated moderately with the fracture load (r(2) = 0.41 and r(2) = 0.40, respectively). The study shows that the proximal femoral failure load in a sideways fall configuration can be estimated with reasonable accuracy by using the CT-based bi-linear elastoplastic cortical bone FE model. This model was more predictive for fracture load than DXA and only slightly less accurate than a full bone FE model including trabecular bone. The accuracy and calculation time of the model give promises for clinical use. (C) 2012 Elsevier Inc. All rights reserved.