Good correlation of goutallier rating of supraspinatus fatty changes on axial and reformatted parasagittal computed tomographic images

Abstract

UNLABELLED Traumatic tendon tear of the rotator cuff occurs frequently and leads to fatty muscle infiltration. With regard to the outcome, fatty infiltration of the rotator cuff muscles constitutes a major negative predictive factor after rotator cuff surgery. In 1989, Goutallier et al established his classification system for assessment of fatty infiltration of the rotator cuff muscles. He used computed tomographic (CT) images in the axial plane. Today, the assessment of rotator cuff muscles on axial CT images has been widely replaced by assessment on parasagittal acquired magnetic resonance images. This change raised 2 important questions. First, there is a controversy whether the Goutallier classification can simply be adopted for magnetic resonance imaging. The second question is whether the muscle assessment in the axial plane is interchangeable with that in the parasagittal plane. We hypothesize that the assessment of fatty muscle infiltration is the same on reformatted parasagittal CT images as on axial CT images METHODS Three independent readers, 2 radiologists and one shoulder surgeon, rated fatty changes of the supraspinatus muscle on CT scans of 91 shoulders. Goutallier grades were assessed on axial and reformatted parasagittal CT images in 2 separate reading sessions. The paired t test was used to find differences between grading results on axial and reformatted parasagittal images. The Pearson correlation coefficient and weighted kappa statistics were used to quantify linear correlation, intrareader, and interreader agreement. RESULTS Mean (SD) Goutallier grading among all readers was 0.80 (1.16) (range, 0-4) on axial images and 0.89 (1.05) (range, 004) on parasagittal reconstructions. We detected a trend toward a slightly higher Goutallier grading on parasagittal reconstructions; however, this result was not significant (P = 0.07). The Pearson correlation coefficient was 0.702 (P < 0.001). Weighted kappa statistics indicated a moderately good to good intrareader (range of weighted kappa, 0.53-0.62) and interreader (weighted kappa, axial images, 0.55; reformatted parasagittal images, 0.65) agreement. CONCLUSION Grading of fatty infiltration of the supraspinatus muscle on parasagittal CT images is comparable with the standard Goutallier grading on axial images and is characterized by a moderately good to good intrareader and interreader agreement. Assessment of parasagittal images is characterized by a slightly higher interreader agreement and may therefore be the preferable modality.