Posterior capsule edema in adhesive capsulitis: comparison with established non-contrast MRI findings and multivariable analysis

ObjectivesTo evaluate posterior glenohumeral capsule edema compared to other MRI findings in adhesive capsulitis (AC).MethodsThis study was approved by the local Institutional Review Board and it is HIPAA compliant. A retrospective search identified subjects who received fluoroscopically guided intra-articular corticosteroid injections for AC and had an MRI within 6 months prior to injection. The study group was compared with an age-, sex-, and side-matched control group who underwent the same procedures but did not have AC. MRIs were evaluated for edema of posterior capsule, anterior capsule, axillary pouch, coracohumeral ligament (CHL) and rotator interval (RI), thickness of axillary pouch and CHL, thickness of anterior capsule, RI and subcoracoid fat replacement, and teres minor atrophy and edema. Multivariable analysis was performed.ResultsA total of 57 subjects with AC and 57 matched controls were studied: mean age 52 & PLUSMN; 7 (range 31-71) years, 37 female and 20 male, 22 right and 35 left. Posterior capsule edema was more common in the AC group vs. control group (66.7 vs 17.5%, p < 0.001). Multivariable analysis showed posterior capsule edema, CHL edema, and axillary pouch (glenoid) thickness (optimum cutoff = 4 mm) were significant independent predictors of AC. Simplified analysis using these three variables had an area under the curve of 0.860 (95%CI: 0.792-0.928). With all three variables present, the sensitivity and specificity for AC were 32% and 98%, respectively.ConclusionsPosterior joint capsule edema may be helpful to confirm AC. Posterior capsule edema, CHL edema, and axillary pouch (glenoid) thickness produce a strong model for distinguishing AC from controls.

Automated summary

This study evaluated posterior glenohumeral capsule edema in adhesive capsulitis (AC) compared to other MRI findings. The results showed that posterior capsule edema, CHL edema, and axillary pouch (glenoid) thickness were significant independent predictors of AC. When all three variables were present, the sensitivity for AC was 32% and specificity was 98%, providing an effective way to distinguish AC from controls.