Patients receiving reverse total shoulder arthroplasty (RTSA) often have osseous erosions because of glenohumeral arthritis, leading to increased surgical complexity. Glenoid implant fixation is a primary predictor of the success of RTSA and affects micromotion at the bone-implant interface. Augmented implants which incorporate specific geometry to address superior erosion are currently available, but the clinical outcomes of these implants are still considered short-term. The objective of this study was to investigate micromotion at the glenoid-baseplate interface for a standard, 3 mm and 6 mm lateralized baseplates, half-wedge, and full-wedge baseplates. It was hypothesized that the mechanism of load distribution from the baseplate to the glenoid will differ between implants, and these varying mechanisms will affect overall baseplate micromotion.

Clinical CT scans of seven shoulders (mean age 69 years, 10°-19° glenoid inclinations) that were classified as having E2-type glenoid erosions were used to generate 3D scapula models using MIMICS image processing software (Materialise, Belgium) with a 0.75 mm mesh size. Each scapula was then repeatedly virtually reconstructed with the five implant types (standard,3mm,6mm lateralized, and half/full wedge; Fig.1) positioned in neutral version and inclination with full backside contact. The reconstructed scapulae were then imported into ABAQUS (SIMULIA, U.S.) finite element software and loads were applied simulating 15°,30°,45°,60°,75°, and 90° of abduction based on published instrumented in-vivo implant data. The micromotion normal and tangential to the bone surface, and effective load transfer area were recorded for each implant and abduction angle. A repeated measures ANOVA was used to perform statistical analysis.

Maximum normal micromotion was found to be significantly less when using the standard baseplate (5±4 μm), as opposed to the full-wedge (16±7 μm, p=0.004), 3 mm lateralized (10±6 μm, p=0.017), and 6 mm lateralized (16±8 μm, p=0.007) baseplates (Fig.2). The half-wedge baseplate (11±7 μm) also produced significantly less micromotion than the full-wedge (p=0.003), and the 3 mm lateralized produced less micromotion than the full wedge (p=0.026) and 6 mm lateralized (p=0.003). Similarly, maximum tangential micromotion was found to be significantly less when using the standard baseplate (7±4 μm), as opposed to the half-wedge (12±5 μm, p=0.014), 3 mm lateralized (10±5 μm, p=0.003), and 6 mm lateralized (13±6 μm, p=0.003) baseplates (Fig.2). The full wedge (11±3 μm), half-wedge, and 3 mm lateralized baseplate also produced significantly less micromotion than the 6 mm lateralized (p=0.027, p=012, p=0.02, respectively). Both normal and tangential micromotion were highest at the 30° and 45° abduction angles (Fig.2). The effective load transfer area (ELTA) was lowest for the full wedge, followed by the half wedge, 6mm, 3mm, and standard baseplates (Fig.3) and increased with abduction angle.

Glenoid baseplates with reduced lateralization and flat backside geometries resulted in the best outcomes with regards to normal and tangential micromotion. However, these types of implants are not always feasible due to the required amount of bone removal, and medialization of the bone-implant interface. Future work should study the acceptable levels of bone removal for patients with E-type glenoid erosion and the corresponding best implant selections for such cases.

For any figures or tables, please contact the authors directly.

This study examined the regional variations of cortical and cancellous bone density present in superiorly eroded glenoids. It is hypothesized that eroded regions will contain denser bone in response to localized stress. The shift in natural joint articulation may also cause bone resorption in areas opposite the erosion site.

Clinical CT scans were obtained for 32 shoulders (10m/22f, mean age 72.9yrs, 56–88yrs) classified as having E2-type glenoid erosion. The glenoid was divided into four measurement regions - anterior, inferior, posterior, and superior - as well as five depth regions. Depth regions were segmented in two-millimeter increments from zero to 10 millimeters, beginning at the center of the glenoid surface. A repeated-measures multiple analysis of variance (RM-MANOVA) was performed using SPSS statistical software to look for differences and interactions between mean densities in each depth, quadrant, and between genders. A second RM-MANOVA was performed to examine effects of gender and quadrant on cortical to cancellous bone volume ratios. Significance was set at p < 0 .05.

Quadrant and depth variables showed significant multivariate main effects (p 0.147 respectively). Quadrant, depth, and their interaction showed significant univariate main effects for cortical bone (p≤0.001) and cancellous bone (p < 0 .001). The lowest bone density was found to be in the inferior quadrant for cancellous bone (307±50 HU, p < 0 .001). The superior quadrant contained the highest mean density for cortical bone (895±97 HU), however it was only significantly different than in the posterior quadrant (865±97 HU, p=0.022). As for depth, it was found that cortical bone is most dense at the glenoid surface (zero to two millimeters, 892±91 HU) when compared to bone at two to eight millimeters in depth (p < 0 .02). Cancellous bone was also most dense at the surface (352±51 HU), but only compared to the eight to 10 millimeters depth (p=0.005). Cancellous bone density was found to decrease with increasing depth. For cortical-to-cancellous bone volume ratios, the inferior quadrant (0.37±0.28) had a significantly lower ratio than all other quadrants (p < 0 .001)

The superoposterior region of the glenoid was found to have denser cancellous bone and a high ratio of cortical to cancellous bone, likely due to decreased formation of cancellous bone and increased formation of cortical bone, in response to localized stresses. The inferior quadrant was found to have the least dense cortical and cancellous bone, and the lowest volume of cortical bone relative to cancellous bone. Once again, this is likely due to reduction in microstrain responsible for bone adaptation via Wolff's law. The density values found in this study generally agree with the range of values found in previous studies of normal and arthritic glenoids. An important limitation of this study is the sizing of measurement regions. For a patient with a smaller glenoid, a depth measurement of two millimeters may represent a larger portion of the overall glenoid vault. Segments could be scaled for each patient based on a percentage of each individual's glenoid size.