Abstract
Introduction:
The mechanical stresses and strains surrounding orthopaedic implants can influence bone resorption and formation, micro-fracture, and consequently implant fixation or loosening. Experimental measurement of these internal parameters is generally not feasible. Computational predictions by finite element modeling are promising, but until recently have been limited to assuming the surrounding cancellous bone as a continuous volume, without modeling individual trabeculae. A recent study demonstrated errors in bone-implant stiffness exceeding 100% when using this continuum assumption [1]. Conversely, recently micro-finite element computer models have been built from high resolution imaging of trabecular bone. In the present study we developed such models of central pegs cemented into cadaveric glenoids. We hypothesized that additional applied cement would lead to stronger implant fixation, but less physiologic strains in the trabeculae.
Methods:
Two cadaveric specimens were implanted, with the applied cement volume in the Specimen 2 approximately double that of Specimen 1. The specimens were imaged by micro-computed tomography (vivaCT 40, Scanco, Switzerland) with a resolution of 12 microns. Images were filtered and resampled, then imported in Mimics (Materialise, Belgium) for semi-automated segmentation and 3D reconstruction based on our laboratory's published methods. Finite element models containing 1.7 to 1.8 million elements having sides of 0.1 mm were generated by a direct image voxel-to-element approach [2] (Fig. 1). The material properties of cement and bone were assumed linear elastic (bone: E = 3.5 GPa, cement: E = 3.0 GPa, and implant (UHMWPE): E = 1.3 GPa), and interfaces were assumed fully bonded. All outer walls of the bone were fixed, and a downward force of 250 N was applied to the implant peg. Simulations were run using Abaqus (Simulia, Pawtucket RI) on a 32-core, 1 TB-memory server at PSU's High Performance Computing Systems.
Results:
Specimen 1 had 254 mm3 cement measured in the model, whereas Specimen 2 had 535 mm3. Strain energy density was less for Specimen 2 for bone underneath the implant, but similar between specimens for bone around the implant sides (Figs 2 and 3), providing initial indication of complex effects of cement volume on peri-implant strains. In Specimen 2 a slightly larger volume of cement (8.6 vs. 6.8 mm3) was exposed to von Mises stresses exceeding 10 MPa.
Discussion:
This study is novel in its prediction of stresses and strains down to the level of individual glenoid trabeculae surrounding a cemented implant. In this pilot investigation we found that bone embedded in the cement mantle is subject to low strains, whereas the bone immediately surrounding the cement mantle is subject to abnormally high strains, with both cement technique and trabecular architecture likely influencing results. The study is limited by the lack of application of more complex loads and boundary conditions. Future work includes modeling of additional specimens and statistical analyses, and investigation of the roles of cement stiffness and peg design in dictating peri-implant bone strains.
