Construction of the finite element analysis model
A three-dimensional (3D) model of composite femurs (4th generation, Sawbones Worldwide, WA) was constructed by computed tomography (CT) imaging (Mimics 16, Materialise, Software & Services for Biomedical Engineering, Leuven, Belgium) of the data obtained from CT (Eclos-4S, Hitachi, Otawara, Tochigi, Japan) [22]. The periprosthetic femoral fracture model was assembled in a 3D-computer aided design software (UG NX 5, SIEMENS, Plano, TX), and the stem position was determined based on radiographs and CT data of an experimental THA model. A transverse fracture was created 10 mm below the tip of the Exeter femoral stem (Stryker, Kalamazoo, MI) and the construct was fixed using two different fixation methods (Fig 1). The single-plate method was performed by fixing a 9-hole LCP-DF locking plate (Depuy Synthes, West Chester, PA) laterally with four proximal uni-cortical locking screws and three distal bi-cortical locking screws. The double-plate method was performed similarly to the single-plating method with an additional anterior 7-hole metaphyseal locking plate (Depuy Synthes) with two proximal uni-cortical locking screws and three distal bi-cortical locking screws. To provide additional fixation to the proximal fragments, two cerclage cables (Depuy Synthes) were used with a tension of 400 N.
Material properties
All sections were assigned isotropic material properties with an elastic modulus of 16.3 GPa for cortical bone [23], 0.15 GPa for cancellous bone [24], 2.8 GPa for polymethylmethacrylate (PMMA) cement [25], 195 GPa for Orthinox stainless steel [26], and 110 GPa for Titanium [27]. A Poisson’s ratio of 0.3 was used for all materials [26].
Finite element analysis modelling
A finite element pre-processor was generated using HyperMesh 13 (Altair Engineering, Troy, MI). Tetrahedral primary elements were used, whereas the number of elements and nodes were 1,023,382 and 224,630 in the single-plate fixation method, and 1,047,309 and 231,601 in the double-plate fixation method, respectively. To set up the boundary conditions, the cortical and trabecular bones were fixed by glue, with a coefficient of friction of 0.1, 0.1, 0.3, 0.1 and 0.1 used at the bone-stem, bone-plate, bone-screw, bone-cable, and cable-fastener interfaces, respectively [28,29]. The distal end of the femoral model was fixed with cement. These constructs were positioned at 20 degrees of frontal plane adduction and aligned vertically in the sagittal plane. This position was to simulate the anatomical one-legged stance. Thereafter, the constructs were tested under an axial load of 1,500 N (Fig. 2) as previously described [30,31], and the results were then analysed using a nonlinear FEA software (MSC Marc 2017, MSC Software, Newport Beach, CA).
Testing and analysis
Biomechanical testing was conducted using synthetic composite femurs (Sawbones Worldwide). Composite bones were placed in a bench-mounted vice grip, and then neck osteotomy, trochanteric reaming, and rasping were performed. Polymethyl methacrylate (PMMA) cement (Simplex P, Stryker) was pressurised into composite bone, and an Exeter hip prosthesis (Stryker) was manually inserted. Stem alignment was checked using X-ray (data not shown). To provide additional fixation to the proximal fragments, two cerclage cables (Depuy Synthes) were used with a tension of 400 N (Fig. 3A). The strain gauge (KFG-2N-120-C1, Kyowa, Chofu, Japan) was attached to the surface of the LCP-DF, parallel to the plate axis, and at the defect level (Fig. 3B). The distal end of the composite bone was placed in an 80-mm-wide threaded steel pipe and fixed with two steel bolts for anti-rotation. The constructs were further fixed by pouring the cement into the steel pipe, and the fracture fixation models were made with the mechanical test equipment (AGS-H, Shimadzu, Japan). To achieve maximum vertical load directly on the head of the prosthesis, the mounting platform was placed to facilitate biaxial translation of the specimen (Fig 3C). For the axial loading test, a sequentially vertical loading test was performed on the prosthetic head at a velocity of 5 mm/min up to 1,500 N. The test was repeated thrice for each construct. The maximum displacements and strains were calculated based on the load-displacement and load-strain curves generated by the static compression tests.
Data analysis and statistics
Statistical analysis was performed using Student’s t-test to compare the differences between two independent groups, and the results were considered significant when P < 0.05. Data are presented as means ± standard error.