Abstract
Introduction
Mechanical property relationships used in the computational modeling of bones are most often derived using mechanical testing of normal cadaveric bone. However, a significant percentage of patients undergoing joint arthroplasties exhibit some form of pathologic bone disease, such as osteoarthritis. As such, the objective of this study was to compare the micro-architecture and apparent modulus (Eapp) of humeral trabecular bone in normal cadaveric specimens and bone extracted from patients undergoing total shoulder arthroplasty.
Methods
Micro-CT scans were acquired at 20 µm spatial resolution for humeral heads from non-pathologic cadavers (n=12) and patients undergoing total shoulder arthroplasty (n=10). Virtual cylindrical cores were extracted along the medial-lateral direction. Custom-code was used to generate micro finite element models (µFEMs) with hexahedral elements. Each µFEM was assigned either a homogeneous tissue modulus of 20 GPa or a heterogeneous tissue modulus scaled by CT- intensity. Simulated compression to 0.5% apparent strain was performed in the medial-lateral direction. Morphometric parameters and apparent modulus-bone volume fraction relationships were compared between groups.
Results
Comparing morphometric parameters, arthroplasty patients had significantly larger bone volume fractions (p = .023) and mean trabecular separation (p = .031), but no significant differences in mean trabecular thickness (p = .060) or trabecular number (p = .178). Variations were observed in the fit curves between normal and arthroplasty cases, with normal bone being best fit by power relationships, and arthroplasty bone exhibiting a more linear relationship. There was no significant difference in mean apparent modulus for homogeneous tissue moduli (p = .060) but was a significant difference for heterogeneous tissue moduli (p = .038).
DISCUSSION
Consistent with previously developed relationships that map apparent mechanical properties, normal cadaveric bone was best fit by a power relationship with an exponential coefficient over 2. However, the apparent modulus- volume fraction relationship in the arthroplasty patient bone exhibited a more linear relationship. These results suggest that the architectural and mechanical properties of normal cadaveric and arthroplasty patient trabecular bone are not equal. Since these relationships are used to map apparent mechanical properties to computational models, these preliminary results suggest that relationships derived from cadaveric normal bone may map the apparent mechanical properties differently than patients who undergo arthroplasty. Additional samples added to this dataset will allow for mechanical property relationships to be developed that account for these bone mechanical property variations. This has the potential to greatly improve the computational modeling of patients undergoing arthroplasty procedures and computational models that are used to design and improve shoulder arthroplasty components.
