Abstract
Introduction
Glenoid loosening, still a main complication in shoulder arthroplasty, could be related to glenohumeral orientation and conformity, cementing techniques, fixation design and periprosthetic bone quality [1,2]. While past numerical analyses were conducted to understand the relative role of these factors, so far none used realistic representations of bone microstructure, which has an impact on structural bone properties [3]. This study aims at using refined microFE models including accurate cortical bone geometry and internal porosity, to evaluate the effects of fixation design, glenohumeral conformity, and bone quality on internal bone tissue and cement stresses under physiological and pathological loads.
Methods
Four cadaveric scapulae were scanned at 82µm resolution with a high resolution peripheral quantitative computer tomography (XtremeCT Scanco). Images were processed and virtually implantated with two anatomical glenoid replacements (UHMWPE Keeled and Pegged designs, Exactech). These images were converted to microFE models consisting of nearly 43 million elements, with detailed geometries of compact and trabecular bone, implant, and a thin layer of penetrating cement through the porous bone. Bone tissue, implant and cement layer were assigned material properties based on literature. These models were loaded with a central load at the glenohumeral surface, with the opposite bone surface fully constrained. Effects of glenohumeral conformity were simulated with increases of the applied load area from 5mm-radius to a fully conformed case with the entire glenoid surface loaded. The models were additionally subjected to a superiorly shifted load mimicking torn rotator cuff conditions. These models were solved and compared for internal stresses within the structures (Figure 1) with a parallel solver (parFE, ETH Zurich) on a computation cluster, and peak stresses in each region compared by design and related to apparent bone density.
Results
Peak cement stresses were generally located at the interface with bone rather than implant (p<0.05), and peak bone stresses occurred around the cemented region. A larger trabecular load share was predicted with the Pegged compared to the Keeled design (Figure 2a). Superior load shift reduced this ratio but resulted in slight stress increase in the cement and implant, with the Keeled design less sensitive to this shift (Figure 2b). These effects were more pronounced with decreased overall bone density (Figure 2c). Increasing conformity significantly affected peak stresses in the cement and implant for both Keeled and Pegged designs (Figure 3) (p<0.041), but only significantly changed bone stresses for the Keeled design (p<0.047). Generally higher peak cement and trabecular bone stresses were predicted for the Pegged design.
Discussion
Our detailed microFE analyses suggest that implant fixation design affects the sensitivity of internal stresses to glenohumeral load shifts, in particular within the cement region and through alterations in load sharing in the periprosthetic bone. Future steps including reverse replacements and more physiological loading conditions, combined with experimental validation tests in dynamic loading, will provide improved insights into the clinical incidences of glenoid loosening.
