Radiographic view strongly influences detection of humeral component migration after stemless shoulder arthroplasty

Le résumé fourni par la source

Background Stemless humeral implants rely exclusively on metaphyseal fixation, making postoperative detection of humeral component migration important after stemless reverse shoulder arthroplasty. However, migration assessment on standard radiographs may be influenced by humeral rotation, particularly in the absence of a stem serving as a longitudinal radiographic landmark. Purpose The primary objective was to compare the sensitivity of three radiographic views, true anteroposterior in neutral rotation (RN), external rotation (RE), and internal rotation (RI), for detecting humeral component migration after stemless shoulder arthroplasty. Secondary objectives were to assess specificity, observer reliability, the influence of experience, and the minimal reliably detected displacement. Methods An experimental radiographic study was performed using synthetic humeri implanted with a stemless humeral component. Controlled varus migration was sequentially created at the calcar level to predefined depths of 1, 3, 5, 7, and 10 mm. For each displacement condition, fluoroscopic images were obtained using RN, RE, and RI views. Ninety images were anonymized and presented in fully randomized order to six blinded observers, including three junior and three senior shoulder surgeons. Observers performed a binary assessment of migration presence compared with reference images. The evaluation was repeated after 1 month, resulting in 1080 individual assessments. Sensitivity, specificity, interobserver and intraobserver reliability, and subgroup analyses according to observer experience were performed. Results Overall sensitivity differed significantly according to radiographic view. RN demonstrated the highest sensitivity (90.7%), followed by RE (54.0%) and RI (29.0%) (p<0.001). Specificity was high across all views: 98.3% for RN, 98.3% for RI, and 96.7% for RE. Sensitivity increased with displacement magnitude, but reliable detection was mainly achieved for displacements greater than approximately 5 mm. Interobserver reliability was highest for RN (κ=0.69; ICC=0.70), compared with RE (κ=0.55; ICC=0.56) and RI (κ=0.57; ICC=0.58). Intraobserver reliability was high, with mean weighted kappa and ICC values of 0.80. Senior observers showed higher overall sensitivity than junior observers (61.6% vs 54.2%), although view ranking remained unchanged. Conclusion Radiographic detection of humeral component migration after stemless shoulder arthroplasty is highly dependent on projection and has limited reliability for small displacements. A standardized true anteroposterior view in neutral rotation is essential for accurate postoperative assessment, whereas rotational views markedly reduce sensitivity and may contribute to underdiagnosis or overdiagnosis.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.