The application of finite element modelling based on clinical pQCT for classification of fracture status.

Abstract

Fracture risk assessment using dual-energy X-ray absorptiometry (DXA) frequently fails to diagnose osteoporosis amongst individuals who later experience fragility fractures. Hence, more reliable techniques that improve the prediction of fracture risk are needed. In this study, we evaluated a finite element (FE) modelling framework based on clinical peripheral quantitative computed tomography (pQCT) imaging of the tibial epiphysis and diaphysis to predict the stiffness at these locations in compression, shear, torsion and bending. The ability of these properties to identify a group of women who had recently sustained a low-trauma fracture from an age- and weight-matched control group was determined and compared to clinical pQCT and DXA properties and structural properties based on composite beam theory. The predicted stiffnesses derived from the FE models and composite beam theory were significantly different (p < 0.05) between the control and fracture groups, whereas no meaningful differences were observed using DXA and for the stress-strain indices (SSIs) derived using pQCT. The diagnostic performance of each property was assessed by the odds ratio (OR) and the area under the receiver operating curve (AUC), and both were greatest for the FE-predicted shear stiffness (OR 16.09, 95% CI 2.52-102.56, p = 0.003) (AUC: 0.80, 95% CI 0.67-0.93). The clinical pQCT variable total density (ρtot) and a number of structural and FE-predicted variables had a similar probability of correct classification between the control and fracture groups (i.e. ORs and AUCs with mean values greater than 5.00 and 0.80, respectively). In general, the diagnostic characteristics were lower for variables derived using DXA and for the SSIs (i.e. ORs and AUCs with mean values of 1.65-2.98 and 0.64-0.71, respectively). For all properties considered, the trabecular-dominant tibial epiphysis exhibited enhanced classification characteristics, as compared to the cortical-dominant tibial diaphysis. The results of this study demonstrate that bone properties may be derived using FE modelling that have the potential to enhance fracture risk assessment using conventional pQCT or DXA instruments in clinical settings.