Prediction of strength and fracture location of the proximal femur by a CT-based nonlinear finite element method - Effect of load direction on hip fracture load and fracture site -

Introduction: The occurrence rate of hip fractures due to osteoporosis is rapidly increasing, representing one of the most serious and urgent social problems. We focused on a computed tomography-based finite element method (CT/FEM) to quantify structural strength, thereby developing a nonlinear CT/FEM to achieve accurate assessment of strength of the proximal femur [1]. The aim of this study was to investigate the effect of load direction on fracture risk of the proximal femur. For this purpose, we evaluated changes in magnitude of strength for the proximal femur with changes in load direction by analyzing the contralateral femur in patients with hip fracture using the nonlinear CT/FEM. We also verified changes in fracture risk by site. From these analyses, we identified load and boundary conditions that could increase risk of hip fracture and clarified that this could possibly cause the fracture types commonly seen in clinical situations. Materials and Methods: Twenty eight femora in female patients with contralateral hip fracture (age: 80 91, average: 85.2)(femoral neck fracture: 13 patients, trochanteric fracture: 15 patients). The study protocol was approved by our ethics committee and the patients were enrolled after informed consent was given. Within 7 days after admission, axial CT images of the proximal femur were obtained (slice thickness: 3 mm, Aquilion Super 4, Toshiba Medical Systems Co., Tokyo, Japan) as well as scans of a calibration phantom. The CT data were transferred to a workstation and 3D finite element models were constructed from the CT data using Mechanical Finder (Research Center of Computational Mechanics Inc., Tokyo, Japan). Trabecular bone and the inner portion of cortical bone were modeled using 3 mm linear tetrahedral elements, while the outer cortex was modeled using 3 mm triangular plates (0.4 mm thick)[1]. On average, there were 75,212 tetrahedral elements and 4,103 triangular plates. Force was applied to the femoral head at an angle γ to the shaft in the frontal plane and at an angle δ to the neck axis in the transverse plane (Fig. 1). For stance configuration (SC), γ and δ were set at 160° and 0°. For fall configuration (FC), γ and δ were set at 120° and 0° (FC1), 60° and 0° (FC2), 60° and 15° (FC3) or 60° and 45° (FC4), respectively [4, 5]. Materially nonlinear finite element analysis was performed by the NewtonRaphson method. Fracture was defined as occurring when at least one shell element failed. Fracture loads were predicted and sites at fracture risk were identified [1]. Correlations between predicted fracture load and load direction were investigated. Predicted fracture type was compared with contralateral actual fracture type. Pearson’s correlation analysis, Friedman test, Scheffe’s post hoc test and Fisher’s exact test were used for statistical analyses and the results were considered significant when p values were less than 0.05. Results: The average predicted fracture loads for SC was 3080 N (standard deviation (SD): 551 N), 2210 N (SD: 606 N) for FC1, 1047 N (SD: 236 N) for FC2, 970 N (SD: 199 N) for FC3 and 700 N (SD: 167 N) for FC4, respectively. The predicted fracture loads for SC were significantly higher than those for all fall configurations except for FC1 (p < 0.001). In comparisons of predicted fracture loads for all fall configurations, loads were significantly higher for FC1 than for FC2, FC3 or FC4 (p=0.02, p<0.001, p<0.001, respectively). The predicted fracture loads for FC2 were significantly higher than those for FC4 (p < 0.001). The predicted fracture loads for FC3 were significantly higher than those for FC4 (p < 0.01).The correlations of the predicted fracture loads for all configurations were shown on Table 1. The predicted fracture sites located at sub-capital region in all patients for SC. The predicted sites located at trochanteric region in all patients for all fall loading configurations except for FC1. For FC1, the predicted sites located at sub-capital region in13 patients, but in 15 patients, they located at trochanteric region. For 20 patients, contralateral actual fracture type corresponded to predicted fracture type. Predicted fracture type corresponded significantly to contralateral actual fracture type (p<0.01). Discussion: As δ increases, the fall tends to be directed more posteriorly. Falls in a posterolateral direction were thus indicated to increase fracture risk more than falls to the side. Each of the predicted fracture loads from various loading conditions displayed poor correlation with each other, even though most correlations were significant. Strength of the proximal femur should thus be evaluated under multiple loading conditions. Intertrochanteric fractures were predicted to occur under all fall loading conditions except FC1. Hirsch et al. reported that compressive force along the long axis of the femoral neck is necessary for femoral neck fractures to occur [6]. Mean neck-shaft angle of the femur is known to be 120°-130°, so FC1 was considered as the condition that could cause neck fractures. If we assume that no morphological differences exist between right and left femora in each patient [7], in all fall loading conditions except FC1, the loading condition would possibly be the only decisive factor for fracture type, irrespective of the morphological characteristics of each patient. Conversely, in FC1, fracture type might differ depending on morphological characteristics of each patient. Keyak et al. reported relationships between loading direction and magnitude of predicted fracture load [4]. However, they reported results from only 4 patients and statistical analyses were not conducted. In addition, they lacked information on predicted fracture sites. The present study could contribute to providing us with useful information for the establishment of effective measures to prevent hip fractures. References: [1] Bessho et al. J Biomech 40: 1745-53, 2007 [2] Keyak et al., J Biomed Mater Res 28: 1329-36, 1994 [3] Keller et al., J Biomech 27: 1159-68, 1994 [4] Keyak et al., J Orthop Res 19: 539-44, 2001 [5] Fujii et al., Nippon Seikeigeka Gakkai Zasshi 61: 531-41, 1987 [6] Hirsch et al., J Bone Joint Surg Br 42: 633-40, 1960 [7] Boston et al., Injury 14: 207-10, 1982