Analysis of Glenoid Component Shift and Osteolysis Following Anatomic Total Shoulder Arthroplasty

Study evaluates glenoid component position and radiolucency using 3D CT imaging

By Eric T. Ricchetti, MD; Bong-Jae Jun, PhD; Jason C. Ho, MD; Thomas E. Patterson, PhD; Kathleen A. Derwin, PhD; Joseph P. Iannotti, MD, PhD

Advertising Policy

Cleveland Clinic is a non-profit academic medical center. Advertising on our site helps support our mission. We do not endorse non-Cleveland Clinic products or services Policy

While total shoulder arthroplasty (TSA) is the preferred surgical treatment for advanced glenohumeral arthritis, with substantial improvement in pain and function, a subset of patients does not experience clinical improvement or sustains a complication. Glenoid component loosening is the most common long-term complication of anatomic TSA and a common reason for revision surgery. While an association between glenoid component loosening on plain radiographs and worse clinical outcomes has been shown at longer follow-up, the significance of early radiographic changes has not been established, nor have the factors associated with glenoid component loosening been well defined.

Our group has developed and validated three-dimensional (3D) computed tomography (CT) imaging methods with metal artifact reduction (MAR) techniques for post-operative analysis after anatomic TSA that allow for precise and accurate determination of implant position of a polyethylene glenoid component, with the ability to detect subtle changes in component position or loosening over time. Although such movement may be clinically silent in the short-term, its presence may be predictive of later, more obvious component loosening and/or premature clinical failure. Therefore, the purpose of this study was to evaluate glenoid component position and radiolucency using our 3D CT imaging analysis methods with minimum two-year follow-up.


One hundred patients who underwent anatomic TSA with a polyethylene anchor peg glenoid component were prospectively enrolled for sequential CT scanning and analysis; including preoperative CT (CT1), early postoperative CT within three months of surgery (CT2), and postoperative CT performed at minimum two-year follow-up (CT3). All patients were also evaluated with plain radiographs and Penn Shoulder scores (PSS) at minimum 2 year follow-up. On the postoperative CTs, the location of the glenoid and humeral head components were detected based on 4 metal markers embedded in the pegs of the glenoid component and a volumetric center fit to the humeral head component using 3D image analysis software. This technique was used to determine glenoid component version, inclination, and joint line position; and humeral head alignment. Evidence of glenoid component central anchor peg osteolysis (CPO) or radiolucency was assessed on postoperative CT3. Glenoid component shift was defined as a change in component position of ≥3° version and/or inclination from CT2 to CT3. Factors associated with CPO and/or shift were assessed with univariate analyses.   

Results and implications

Forty-eight percent of patients showed evidence of glenoid component shift on CT3; 25 with increased inclination alone, 14 with both increased inclination and retroversion or anteversion, six with increased retroversion alone, and three with increased anteversion alone. See Figure 1.

Advertising Policy

Figure 1: Digital templates of the position of the glenoid and humeral head components on the immediate post-operative CT (Red) and 2 year follow-up CT (Green) are superimposed on the 2 year follow-up CT in 2 patients (A-C, D-F). In the first patient (A-C), the glenoid component has shifted into increased inclination (A) and retroversion (B), and central anchor peg osteolysis is seen on the 2 year follow-up CT after the digital templates are removed (C). In the second patient (D-F), the glenoid component has shifted into increased inclination (D) with stable version (E), and bone integration around the central anchor peg is seen on the 2 year follow-up CT after the digital templates are removed (F).

CPO was present on CT3 in 14 patients, with only 12/48 (25 percent) patients with component shift having CPO. Four groups of patients were, therefore, identified: 1. No shift and no CPO (N = 50), 2. No shift and CPO (N = 2), 3. Shift with CPO (N = 12), 4. Shift without CPO (N = 36). Pre-operative joint line medialization (= 0.003), glenoid component absolute rotational (combined version and inclination) shift (< 0.0001), and glenoid component medialization (medial translational shift) (= 0.0002) were significantly different across the 4 groups, with cases having both shift and CPO associated with greater preoperative joint line medialization and greater component shifts (Table 1, Figure 2). Walch classification was also significantly different across the 4 groups (= 0.02), particularly with respect to CPO. A2 (3/13, 23%) and B3 (6/15, 40%) glenoids had higher CPO rates than A1 (1/32, 3%) and B2 (2/29, 7%) glenoids. PSS was not significantly different across the 4 groups at minimum 2 year follow-up (= 0.51).

Postoperative 3-D CT imaging analysis demonstrates that shift of a polyethylene glenoid component commonly occurs following anatomic TSA, with increased inclination the most common direction of shift. This cohort demonstrates the novel finding that most cases of glenoid component shift at minimum 2 years post-operatively show bony integration around the central anchor peg without osteolysis (36/48, 75 percent), consistent with a stable implant. In contrast, cases of glenoid component shift with CPO (12/48, 25 percent) have larger absolute shifts and significantly more implant medialization over time, which suggests early implant loosening and subsidence that raise concern for eventual implant failure. CPO was found to be associated with preoperative joint line medialization, assessed both by millimeters of bone loss and Walch type (A2, B3 glenoids).

Advertising Policy

Figure 2: Cases having both glenoid component shift and central peg osteolysis at 2 years (n=12/100) were associated with significantly greater baseline joint line medialization, and greater component medialization and combined rotational shift (version and inclination). Joint line data are means, component data are medians.

These findings are not associated with a decrease in patient reported outcomes or increased revision surgery across patient groups at current short term follow-up. Follow-up and larger patient cohorts are needed to confirm if the combination of early glenoid component shift and CPO place an implant at risk for clinically relevant loosening, while early shift in the absence of CPO does not.