We retrospectively reviewed 2500 patients with rotator cuff tears who underwent open or arthroscopic tendon repair, arthroscopic débridement, or a reverse shoulder arthroplasty between 1988 and 2005. We excluded 812 patients with previous surgery on the affected shoulder or incomplete imaging studies. The average age of the remaining 1688 patients at the time of imaging was 57.2 years (range, 18–83 years). Men represented 60% of cases (1005/1688); 72% (1209/1688) involved the dominant side. Of the 1688 patients, 1377 (82%) had a preoperative CT arthrogram and 311 (18%) had a preoperative MRI. Demographic data were obtained for each patient in addition to the details regarding their rotator cuff tear (number of tendons torn, etiology of the tear, time between onset of shoulder symptoms, and diagnosis of rotator cuff tear).
We identified and characterized the rotator cuff tears based upon preoperative CT arthrogram or MRI, ranging from isolated single tendon tears to massive multiple tendon tears [21
]. For the supraspinatus tendon we differentiated between partial thickness (articular side, intratendinous, superficial side) and full thickness tears; for the infraspinatus and the subscapularis tendon, we did not distinguish between partial and complete tears because the study focused on the supraspinatus. There were no teres minor tears in this series. Rotator cuff tears were identified on preoperative images and confirmed intraoperatively, but the observers were not blinded as they had evaluated the preoperative imaging and were present at the time of surgery. The rotator cuff tears involved the supraspinatus in 93% of cases (1573/1688), the subscapularis in 38% (633/1688), and the infraspinatus in 24% (411/1688) (Table ). We did not quantify the size of the tears due to lack of consensus in the literature, no methods to measure tear size and the variability these measurements would introduce into the study.
Type of rotator cuff tears and epidemiology
The etiology of the rotator cuff tear was classified as traumatic (those patients who recalled a specific event), degenerative (those who did not recall a single incident), or mixed. The mixed type was defined as a patient with chronic shoulder symptoms who experienced an acute traumatic event that exacerbated the chronic symptoms and motivated the patient to seek formal treatment. The etiology of the rotator cuff tear was traumatic in 669 of the 1688 cases (40%), degenerative in 903 (54%), and mixed in 116 (7%).
One of us (BM) graded fatty infiltration of the supraspinatus using either the CT arthrogram (soft-tissue sequences) and the classification of Goutallier et al. [11
] or MRI (T1 sequences) and the classification developed by Fuchs et al. [6
]. Both classifications systems compare the ratio of fat to muscle on sagittal oblique images: stage 0–1 is considered normal and has no (0) to trace (1) amounts of fat investing the muscle bellies. Stages 2–4 are considered pathologic: more muscle than fat (2); equal muscle to fat (3); more fat than muscle (4). Lesage et al. reported a moderate interobserver agreement in the analysis of the supraspinatus fatty infiltration according to Goutallier classification on the CT scan [17
]; Williams et al. reported a good intraobserver and a moderate interobserver agreement in the analysis of the supraspinatus fatty infiltration according to Goutallier classification on the CT scan [24
]; Fuchs et al. reported a good interobserver reliability for grading fatty infiltration according to Goutallier classification for the CT-scan and for the MRI [6
]. We did not compare or contrast CT scans with MRI; nevertheless, in our series the delay of appearance of different stages of fatty infiltration and the mean age of patients for each stage of fatty infiltration were similar in CT and MRI groups, and we think that the small series of MRI (only 18%) and the small differences in results have minimal influence on the data. Because the distribution of fatty infiltration through a muscle belly is variable and random, the final stage of supraspinatus fatty infiltration was determined by grading the muscle in an entire set of three imaging planes (coronal, axial, and sagittal) and not in one specific window (Figs. , ). The average of these values was defined as the final stage of fatty infiltration for the supraspinatus muscle. Fatty infiltration of the infraspinatus and the subscapularis was graded using only axial images at the superior (tip of coracoid process) and inferior portion (lowest portion at the glenohumeral joint) of the glenoid as described by Goutallier [11
]. The average of these two values defined the final stage of fatty infiltration for the infraspinatus and subscapularis muscles. For statistical evaluation we classified fatty infiltration as minimal for Goutallier stages 0 and 1, moderate for Goutallier Stage 2 and severe for Goutallier stages 3 and 4 [24
Fig. 1A–C Fatty infiltration of the supraspinatus was graded in soft-tissue sequences of the CT scan according to Goutallier classification . The average of values determined in (A) axial, (B) coronal, and (C) sagittal plane was defined as the final stage of (more ...)
Fig. 2A–C Fatty infiltration of the supraspinatus was graded in T1 sequences according to Goutallier classification adapted by Fuchs to MRI . The average of values determined in (A) axial, (B) coronal, and (C) sagittal plane was defined as the final stage of (more ...)
One of us (BM) also measured muscular atrophy indirectly using the tangent sign on the most lateral cut of the sagittal plane where the spine appears in contact with the scapula on the MRI/CT as described by Zanetti [26
] (Fig. ). A healthy supraspinatus should cross a line drawn for the superior border of the coracoid process to the superior border of the scapular spine; this line is called the “tangent.” Failure of the supraspinatus to cross the tangent is considered a “tangent sign” [26
]. Only 584 had the appropriate “Y” cut on the most lateral sagittal plane where the spine appears in contact with the scapula (as described by Zanetti [26
]) for the atrophy measurement.
If the muscle belly of the supraspinatus fails to intersect a line drawn across the supraspinatus fossa, the tangent sign is Positive (A); if, however, the muscle does cross the line, it is a Negative tangent sign (B).
We used multiple regression analysis to determine the influence of several factors on the appearance of supraspinatus muscle fatty infiltration: increasing patient age, delay between onset of symptoms and diagnosis, etiology, and the number of involved tendons. The values of fatty infiltration were then analyzed two by two and multivariate logistic regression was used to determine the influence of the same factors on the speed of appearance and progression of this phenomenon. To evaluate the speed and progression at discrete time periods, the analysis of variance (ANOVA) with post hoc testing using the Bonferroni method was used or nonparametric test (Kruskal-Wallis, Wilcoxon) when the variance was not equal when analyzing trends with tear size. Fatty infiltration was analyzed by comparing muscle with minimal fatty infiltration to muscle with moderate and severe amount of fatty infiltration, respectively, with chi square of the maximum likelihood test for qualitative variables. Moderate and severe amounts of fatty muscle infiltration were also compared to each other. Finally, regression analysis was used to correlate fatty infiltration and atrophy.