Introduction: The first clinical applications of FE-strength estimates to classify osteoporotic fractures showed inhomogeneous results. We developed a FE model that correlated well with femur strength in-vitro (R2=0.9, 14 femurs). This work aims to verify if our model can classify osteoporotic fractures in three case-control studies: a retrospective and a prospective study on proximal femur fracture, and a retrospective study on prevalent osteoporotic fractures.
Methods: Femur retrospective study: 22 proximal femur fractures, 33 controls, all osteopenic or osteoporotic. Femur prospective study: 21 incident proximal femur fractures, 45 age-matched controls. Prevalent fractures study: 35 women with prevalent osteoporotic fracture (e.g. radius and vertebrae), 40 aBMD-matched controls. CT-based FE-strength estimates: FE-strength was defined as the load inducing εmax>εlim on the femoral neck surface (0.73% tensile and 1.04% compressive limit). For each patient, FE-strength was evaluated for a range of loading directions mimicking the in-vivo variability of hip reactions. The minimum FE-strength among all directions was retained for patient classification. Fracture classification: Odds- or hazard-ratios and area under roc curve (AUC) were derived for FE-strength and aBMD.
Results: Femur retrospective study: FE-strength was 33% lower in cases (vs 12% for aBMD). FE-strength classified fractures better than aBMD (AUC=0.88 vs 0.71) and remained associated with fracture in age- and aBMD-adjusted models. Femur prospective study: FE-strength was 19% lower in cases (vs 15% for aBMD). FE-strength showed higher fracture classification than aBMD (AUC=0.78 vs 0.72) and remained associated with fracture in aBMD-adjusted models. Prevalent fractures study: fractures and controls were aBMD-matched by design. FE-strength was 5% lower in fracture cases (non-significant difference).
Discussion: In postmenopausal women, i.e. the population at the highest risk of bone fracture, our linear FE model was highly associated with femur fracture. We confirmed that site-specificity is important since femur models can predict femur fractures in addition to aBMD, but not all prevalent osteoporotic fractures.
17 May 2014 - 20 May 2014