Three-dimensional characterization of glenoid defects in failed shoulder arthroplasties

Glenoid vault defects are often observed in revision shoulder arthroplasty, and improper management of glenoid bone loss during revision to reverse total shoulder arthroplasty (rTSA) increases the risk of complications. However, there is limited understanding of the types of glenoid vault defects that occur in patients. Therefore, the purpose of this study was to three-dimensionally characterize the underlying glenoid morphology of patients with failed shoulder arthroplasties subsequently undergoing revision arthroplasty to rTSA. Computed tomography scans of 100 patients, preoperative to revision arthroplasty to rTSA (revision group) and 35 healthy cadaveric control scapulae (control group), were three-dimensionally reconstructed. A negative projection glenoid vault was created and residual screw or peg holes were removed and smoothed. Glenoid defects were quantified using various size, area, volume, and geometric measures of the negative projection and the glenoid. A cluster analysis was performed on the revision group to group patients based on glenoid shape and find patterns in the underlying glenoid morphology using principal measures: normalized distance (height; width; maximum projection thickness; and smoothed maximum residual vault thickness), angular (anterior tilt; inclination; and version), and radius of curvature (ROC) (superior-inferior ROC; anterior-posterior ROC; and vault-sphere ROC). When compared to the control group, the measures of the revision group showed more variation with higher standard deviations. The cluster analysis formed 2 clusters, between which 6 of the principal measures had statistically significant differences. Of the principal measures, cluster A had larger height and maximum projection thickness, more retroversion, and smaller ROCs when compared to cluster B. Of the other measurements, projection volume, lateral surface area, circumferential surface area, and height-width ratio were larger in cluster A than in cluster B. Glenoid morphology after failed arthroplasty is highly variable and may require novel defect management strategies. Additionally, the formation of clusters based on glenoid morphology indicates that patterns exist in the types of glenoid defects, highlighting a need to further investigate a three-dimensional classification system and potentially new standardized revision implant component designs for revision to rTSA cases.