Humeral Bone Loss in Revision Total Shoulder Arthroplasty: the Proximal Humeral Arthroplasty Revision Osseous inSufficiency (PHAROS) Classification System

Background Humeral bone loss is commonly encountered during revision shoulder arthroplasty and anticipating humeral bone defects can help the revision surgeon make appropriate plans to achieve adequate fixation and stability. No validated classification system exists to characterize humeral bone loss in the setting of revision shoulder arthroplasty. Questions/purposes The purposes of this study were (1) to create a classification system for humeral bone loss in revision shoulder arthroplasty; (2) to determine the classification system's reliability; and (3) to determine whether humeral bone loss type is associated with intraoperative humeral-related reconstruction characteristics. Methods This was a comparative retrospective radiographic study. First, six surgeons from five centers collaborated to create a classification by consensus. Second, two surgeons from two other centers who had fellowship training in shoulder and elbow surgery, who were blinded to each other's grades and all patient details other than plain radiographs, and who were not involved in creation of the system, classified true AP, AP, and lateral (axillary and/or scapular-Y) radiographs from 108 revision (413 radiographs) from one center that were performed between November 15, 2006, and January 4, 2018. Interobserver reliability was calculated by comparing those two reviews and determining Cohen's kappa. In addition, one reviewer repeated his assessments twice, 4 months apart, to determine intra-observer reliability using Cohen's kappa. Third, we performed a retrospective chart study of these same revisions to determine intraoperative humeral-related reconstruction characteristics such as the use of greater tuberosity fixation, stem length, humeral bone grafting, and the use of proximal humeral replacement or total humeral replacement; at the center where these revisions were performed during that timespan, no attempt to classify bone loss was made. During that period, the general indications for greater tuberosity fixation included the absence of a stable osseous connection between the greater tuberosity and the shaft of the humerus with a tuberosity amendable to repair; the general indications for use of longer stems were inability to obtain a minimum of two cortical widths of overlap between the implant and the humeral diaphysis and/or loss of the greater tuberosity; and the general indications for proximal and total humeral replacement were bone loss that was felt to be too severe to allow reconstruction with allograft. Results The classification system consists of three types of humeral bone loss: Type 1 is loss of the epiphysis with subtypes for loss of the calcar and loss of the greater tuberosity; Type 2 is loss of the metadiaphysis above the deltoid attachment with a subtype for cortical thinning; and Type 3 is bone loss extending below the deltoid attachment with a subtype for cortical thinning. We studied 108 revisions: 38 (35%) without bone loss, 34 (31%) Type 1, 27 (25%) Type 2, and nine (8%) Type 3. For reliability, interrater kappa was 0.545 and in 71% (77 of 108) of revisions, the two raters agreed on a numeric type. Intrarater k was 0.615 and in 77% (83 of 108) of revisions, the rater agreed with himself as to the numeric type. Stem length increased with class type (Type 1 median [range] 130 [70-210], Type 2 150 [70-210], Type 3 190 [70-240], p = 0.005). Most greater tuberosity fixation for intraoperative fracture was in Types 1 and 2 (13 of 18 compared with the five of 18 of greater tuberosity fixation that was within Types 0 and 3, p = 0.043). Most bone grafting was in Types 2 and 3 (eight of 13 compared with five of 13 of bone grafting was in Types 0 and 1, p = 0.044). Most proximal humeral and total humeral replacements were in Type 3 (three of four compared with one of four, p < 0.001). Conclusions We developed the Proximal Humeral Arthroplasty Revision Osseous inSufficiency (PHAROS) system, which has adequate, if imperfect, reliability to classify humeral bone loss in the setting of revision shoulder arthroplasty. This classification system may be useful to anticipate the complexity of humeral reconstruction. Further validation incorporating advanced imaging and further evaluators will be necessary. Level of Evidence Level III, diagnostic study.