Posterior-inferior glenoid rim shapes by MR imaging

Our purpose was to try to apply a CT classification of glenoid rim shapes to MR images and to determine the reliability of the subsequent MR classification by testing observer variability. Shoulder MR imaging exams of 54 consecutive patients were reviewed retrospectively by two musculoskeletal radiologists independently. Posterior-inferior glenoid rim shape was categorized, based on reported CT criteria, as triangle, "lazy-J" or "delta" using the most caudal MR section that unequivocally showed articular cartilage. The same studies were reviewed again one month later to test interobserver and intraobserver variability. Final determination of glenoid rim type was made by consensus. There were 31 males and 23 females with an average age of 42 years, range 14-75 years. Forty-one patients were referred for evaluation of nonspecific shoulder pain, 9 for suspected rotator cuff tear and 4 for instability. The most common shape was "lazy-J" with 32 patients (59%) having this type. The least common was delta shape, with 7 patients (13%) having this form. Fifteen patients (28%) had a normal triangle shape. There was significant interobserver variability in determining the shape of the glenoid rim with 17/54 (31%) discordant readings. For the "lazy-J" shape, the kappa value was .02, for the "delta" type, kappa value was .60 and for the triangle shape, kappa was .32. Intraobserver variability was 35% for radiologist A and 28% for radiologist B. Kappa values for intraobserver variability ranged from .53 to .85. Four patients had posterior instability, 3 were judged to have normal triangular inferior glenoid rims and 1 a "lazy J" type by Radiologist A. Radiologist B judged 2 of the instability patients as normal triangular and 2 as "lazy J". A CT classification of posterior-inferior glenoid rim shapes can be applied using MR imaging, however, there is significant observer variability that limits the applicability of this classification scheme with MR imaging.