Do magnetic resonance imaging and computed tomography provide equivalent measures of rotator cuff muscle size in glenohumeral osteoarthritis?

Abstract

Rotator cuff muscle volume is associated with outcomes after cuff repair and total shoulder arthroplasty. Muscle area on select magnetic resonance imaging (MRI) slices has been shown to be a surrogate for muscle volume. The purpose of this study was to determine whether computed tomography (CT) provides an equivalent measurement of cuff muscle area to a previously validated MRI measurement. We included 30 patients before they were undergoing total shoulder arthroplasty with both preoperative CT and MRI scans performed within 30 days of one another at 1 institution using a consistent protocol. We reoriented CT sagittal and MRI sagittal T1 series orthogonal to the scapular plane. On both CT and MRI scans, we measured the area of the supraspinatus, infraspinatus-teres minor, and subscapularis on 2 standardized slices as previously described. We calculated intraclass correlation coefficients and mean differences. For the 30 subjects included, when MRI and CT were compared, the mean intraclass correlation coefficients were 0.989 (95% confidence interval [CI], 0.976-0.995) for the supraspinatus, 0.978 (95% CI, 0.954-0.989) for the infraspinatus-teres minor, and 0.977 (95% CI, 0.952-0.989) for the subscapularis. The mean differences were 0.2 cm2 (95% CI, 0.0-0.4 cm2) for the supraspinatus (P = .052), 0.8 cm2 (95% CI, 0.1-1.4 cm2) for the infraspinatus-teres minor (P = .029), and -0.3 cm2 (95% CI, -1.2 to 0.5 cm2) for the subscapularis (P = .407). CT provides nearly equivalent measures of cuff muscle area to an MRI technique with previously validated reliability and accuracy. While CT underestimates the infraspinatus area as compared with MRI, the difference is less than 1 cm2 and thus likely clinically insignificant.