The assessment of rotator cuff muscle abnormalities is one of the main determinants in clinical decision-making in the management of patients with rotator cuff tear, because the severity of presurgical fatty degeneration is especially relevant to a tendency toward retear and worse functional outcomes after treatment (1). For the qualitative assessment of fatty degeneration, the Goutallier classification and modified Goutallier classification have become widely accepted as a standard five-grade scoring system in current practice (2). However, they are shown to be highly observer-dependent, and inter- and intra-observer reliability are not high (3). The objective of this study was to quantify fatty degeneration of the supraspinatus muscle by using a 2-point Dixon technique at 3.0-T MRI for patients with rotator cuff tears, and to evaluate the inter- and intra-observer reliability of quantitative analysis of fatty degeneration in comparison with the qualitative modified-Goutallier classification.

Two hundred patients including 100 partial-thickness rotator cuff tears and 100 full-thickness tears (mean: 66.1 years, range: 34 to 91 years, 86 males and 114 females) were selected retrospectively, who underwent MRI studies on 3.0-T unit, including 2-point Dixon sequence for quantification of fatty degeneration within supraspinatus. The study was approved by IRB, and the informed consent was waived because of the study’s retrospective nature. All MR exams were performed on a 3.0-T unit (Magnetom Verio, ver. VB17, Siemens AG, Erlangen, Germany) by using a 4-channel flex coil. In addition to standard routine clinical protocols used at our institution, we also performed three-dimensional (3D) 2-point Dixon VIBE sequence in oblique sagittal plane. The acquisition parameters were as follows. Image-matrix = 128*128mm, Slice thickness = 2.5mm, TR/TE = 6.5/1.225, 2.4msec, Flip angle = 10 degrees, FOV = 196mm, one acquisition, acquisition time = 2min30sec. From the 3D 2-point VIBE sequence, water-only, fat-only, in-phase (water and fat), and out-of-phase (water minus fat) images were produced. The quantitative analysis of fatty degeneration was performed by two radiologists independently, measuring signal intensity values of in phase and fat image within supraspinatus, and calculating fat fraction as S(Fat)/S(In) (Fig.1). We used the intra- and inter-class correlation coefficients (ICCs) to assess the intra- and interobserver reliability of the ROI measurement. The qualitative evaluation of all subjects using modified-Goutallier classification was performed by two radiologists and three orthopedic surgeons independently (Fig.2). The Siegel’s kappa analysis was performed for evaluating the interobserver reliability in scoring by the modified-Goutallier classification. One-way analysis of variance and the Steel-Dwass post hoc test were used to compare the relationship between fat fraction value and the modified-Goutallier classification. Figure 3 shows the relationship between the severity of rotator cuff tears and modified Goutallier classification. Severity of rotator cuff tears is correlated with higher modified-Goutallier grade. As the modified-Goutallier stage increases, the fat fraction value becomes larger (p<0.01) (Fig.4). The interobserver agreement of the modified-Goutallier classification was considered moderate (k=0.44), while the inter-class correlation coefficient regarding fat fraction value quantified in 2-point Dixon sequence was excellent (0.847, 95% confidence interval [CI], 0.751-0.908). The intra-class correlation coefficient of the measurement of fat fraction value was also excellent (0.908, 95% CI, 0.829-0.951). Figure 5 shows the result of rating based on modified-Goutallier classification in each reader.Fat quantification in the rotator cuff muscles for severity of tears using a 2-point Dixon technique at 3.0T-MRI is accurate and highly reproducible in comparison with the qualitative evaluation using modified-Goutallier classification.