Clinical accuracy of humeral and glenoid component placement in total shoulder arthroplasty using ASTRA patient-specific guides

Background: Malalignment of the humeral and glenoid components in total shoulder arthroplasty is associated with complications such as instability, implant loosening, and restricted motion. While the accuracy of patient-specific instrumentation (PSI) for glenoid component orientation is well documented, the precision of PSI for humeral osteotomy has not yet been published for a noncadaveric environment. This study aims to calculate the deviation between the planned and achieved orientations of humeral and glenoid components using PSI in a clinical, in vivo setting. 

Methods: Forty-six patients were enrolled. One patient was excluded due to an intraoperative change to humeral component selection. The gender, etiology for arthroplasty (osteoarthritis or cuff tear arthropathy), severity of glenoid wear based on the Walch classification, and procedure (anatomic or reverse) were recorded. Computed tomography (CT) scans were obtained preoperatively for 3-dimensional reconstruction, from which the target humeral and glenoid component orientations for each patient were selected by the surgeon in the ASTRA surgical planning platform (Enovis, Austin, TX, USA). Humeral and glenoid PSIs were designed and 3-dimensionally printed to execute the planned alignments. A postoperative CT scan was acquired and used to calculate the deviation between the planned and achieved humeral version, humeral neck-shaft angle, humeral osteotomy height, glenoid version, and glenoid inclination. Humeral version was able to be calculated for 22 patients (48.9%) with a CT scan which extended distally beyond the humeral epicondyles.

Results: A total of 8 outliers (17.8%) were identified across all orientation parameters: 5 outliers (11.1%) for humeral orientation and 3 (6.7%) for glenoid orientation. The average deviation was 2.9°, 2.2°, 1.3 mm, 1.7° and 1.4° for humeral version, humeral neck-shaft angle, humeral osteotomy height, glenoid version, and glenoid inclination, respectively. The mean deviation in glenoid version was greater in males than females (P = .007). However, no other statistically significant difference was found between the mean or variance in deviation when comparing cohorts stratified by etiology of arthroplasty, severity of glenoid wear, or surgical procedure. 

Conclusion: PSI generated from CT-based planning can achieve precise execution of preoperative plans in 89% of humeral and 93% of glenoid components in both anatomic and reverse procedures across a variety of patient morphologies. The humeral guide, when used in a more challenging in vivo environment, demonstrates accuracy comparable to those observed in existing ex vivo studies.