April 2000, VOLUME 174
NUMBER 4

Recommend & Share

April 2000, Volume 174, Number 4

Musculoskeletal Imaging

High-Field and Low-Field MR Imaging of Superior Glenoid Labral Tears and Associated Tendon Injuries

+ Affiliations:
1 Department of Diagnostic Imaging, Brown University School of Medicine, Rhode Island Hospital, 593 Eddy St., Providence, RI 02903.

2 Department of Orthopaedic Surgery, Brown University School of Medicine, Rhode Island Hospital, Providence, RI 02903.

Citation: American Journal of Roentgenology. 2000;174: 1107-1114. 10.2214/ajr.174.4.1741107

ABSTRACT
Next section

OBJECTIVE. The purpose of this study was to determine the performance characteristics of high-field and low-field MR imaging for the diagnosis of a glenoid superior labral anteroposterior (SLAP) tear.

MATERIALS AND METHODS. High-field (n = 46) or low-field (n = 21) MR imaging was performed on 41 patients with SLAP tears and 26 patients with normal superior labra. The superior labrum was classified into one of four types on the basis of patterns of intralabral signal intensity. The relative frequency of rotator cuff tears and long head of the biceps tendinopathy was also assessed.

RESULTS. For the diagnosis of SLAP tear, the sensitivity of high-field MR imaging was 90% (95% confidence interval = 74%, 98%), specificity was 63% (35%, 85%), and accuracy was 80% (66%, 91%). The sensitivity of low-field MR imaging was 64% (31%, 89%), specificity was 70% (35%, 93%), and accuracy was 67% (43%, 85%). A branched linear or stellate focus of abnormal intralabral signal intensity was associated with a SLAP tear in 86% of patients. Conversely, two other labral patterns correlated with a normal superior labrum in 71% of patients. Abnormal signal intensity in the biceps tendon was seen in 15% of patients with a SLAP tear. Full-thickness (37%) and partial-thickness (31%) rotator cuff tears were often seen.

CONCLUSION. The performance characteristics of high-field MR imaging are superior to those of low-field MR imaging for the diagnosis of a superior labral tear. Rotator cuff tears can be seen in many patients with superior labral tears, but abnormal signal intensity in the biceps tendon in uncommon.

Introduction
Previous sectionNext section

Atear of the superior glenoid labrum, adjacent to the origin of the long head of the biceps tendon, may extend in the anteroposterior direction. Andrews et al. [1] originally described this lesion in 73 overhead-throwing athletes, and subsequent reports have confirmed that many of these injuries are sports-related [2, 3]. However, some superior labral anteroposterior (SLAP) lesions occur in patients with no history of traumatic shoulder injury; in these patients, the lesion may be related to labral degeneration [4]. Patients may complain of shoulder pain that is exacerbated by overhead activities, and many will also have mechanical symptoms such as catching, locking, popping, or snapping. However, an overlap of symptoms exists among patients with a SLAP lesion and those with a rotator cuff disorder or glenohumeral instability [2, 3, 5, 6].

At arthroscopy, four main types of SLAP tears and several variants have been described [2, 5, 7]. The relative frequency of the four major types of SLAP tears in 140 patients was reported by Snyder et al. [3]. Marked fraying of the superior labrum without labral detachment or biceps tendon anchor injury is defined as a type I tear (21%). Detachment of the labrum from the glenoid is a type II tear (55%); a type III tear is a vertically displaced bucket-handle SLAP tear (9%). A type IV tear is a bucket-handle tear that extends to a variable extent into the biceps tendon (10%). Any of these four types may occur with more extensive labrocapsular injuries, such as a Bankart lesion or a tear of the middle glenohumeral ligament (5%).

Although arthroscopy is most often used to diagnose SLAP tears, case reports and small case series support the use of nonarthrographic MR imaging and MR arthrography [7,8,9,10,11,12,13]. However, none of these reports present performance characteristics of MR imaging because they do not include a companion cohort of patients with shoulder pain or instability who did not have superior labral injuries at arthroscopy. Other studies from the orthopedic and medical-imaging literature have questioned the clinical efficacy of MR imaging [9, 14].

The purpose of this study was to determine the performance characteristics of high-field and low-field MR imaging for the diagnosis of SLAP tears. We evaluated specific patterns of intralabral signal intensity and the frequency of concomitant injuries to the rotator cuff and long head of the biceps tendons in patients with SLAP tears.

Materials and Methods
Previous sectionNext section
Selection Criteria for Subjects

From the clinical records of a single fellowship-trained orthopedic surgeon, 41 patients with arthroscopic evidence of SLAP lesions (arthroscopic types I-IV) who had MR imaging of the affected shoulder were selected. In addition, 26 patients with a normal superior labrum and biceps anchor at arthroscopy were also selected. These 26 patients presented with clinical signs and symptoms similar to those of patients with SLAP tears (i.e., shoulder pain and physical findings such as clicking, locking, or popping). Both groups of patients had either high-field (n = 46) or low-field (n = 21) MR imaging and arthroscopy between September 1994 and January 1998. Because this was a retrospective study, patients were assigned to high-field or low-field MR imaging primarily on the basis of magnet availability, scheduling needs, or, in a minority of cases, patient preference. The mean time between MR imaging and arthroscopy was 4 months (range, 2 weeks to 10 months).

Description of Subjects

Of the 41 patients with SLAP tears at arthroscopy, 27 men and 14 women had a mean age of 43 years (range, 20-71 years). In this group, the right shoulder was involved in 26 patients and the left shoulder in 15 patients. Sixteen males and 10 females in the group of 26 patients had a normal superior labrum at arthroscopy. The mean age of this group was 45 years (range, 14-70 years; p = 0.36). The right shoulder was involved in 17 patients, and the left shoulder was symptomatic in nine patients (p = 0.07).

Shoulder MR Imaging Technique

High-field (1.5-T) MR imaging of the shoulder was performed on 46 patients with a Magnetom Vision scanner (Siemens, Erlangen, Germany) using a circularly polarized flexible coil. The routine protocol includes axial and coronal oblique turbo spin-echo proton density-weighted and coronal oblique and sagittal oblique turbo spin-echo T2-weighted sequences. The coronal oblique images were oriented parallel to the longitudinal course of the supraspinatus tendon, and oblique sagittal images were oriented orthogonal to those of the coronal oblique plane. The imaging performed had the following parameters: axial proton density-weighted images (TR/TE, 2000/18; echo train length, three; field of view, 18 cm; matrix, 252 × 256; acquisition, one; slice thickness, 3 mm [no gap]; number of slices, 21; scan time, 5.7 min), coronal oblique proton density-weighted images (2000/18; echo train length, three; field of view, 16 cm; matrix, 192 × 256; acquisitions, two; slice thickness, 3 mm [no gap]; number of slices, 21; scan time, 5.4 min), coronal oblique T2-weighted images with frequency-selective fat saturation (3600/96; echo train length, seven; field of view, 16 cm; matrix, 252 × 256; acquisitions, two; slice thickness, 4 mm [no gap]; number of slices, 15; scan time, 5.7 min), sagittal oblique turbo spin-echo T2-weighted images with frequency-selective fat saturation (3500/96; echo train length, seven; field of view, 16 cm; matrix, 252 × 256; acquisitions, one; slice thickness, 4 mm [no gap]; number of slices, 15; scan time, 4.4 min).

Low-field (0.2-T) MR imaging was performed on 21 patients with a Magnetom Open scanner (Siemens) using a multipurpose surface coil. The routine shoulder-imaging protocol includes an axial turbo spin-echo proton density-weighted sequence, both coronal oblique conventional spin-echo T1-weighted and turbo spin-echo sequences, and an oblique sagittal turbo spin-echo T2-weighted sequence. Parameters used were the following: axial proton density-weighted images (2000/24; echo train length, five; field of view, 18 cm; matrix, 250 × 256; acquisitions, two; slice thickness, 5 mm [no gap]; number of slices, 14; scan time, 6.8 min), coronal oblique T1-weighted images (500/26; field of view, 18 cm; matrix, 192 × 256; acquisitions, two; slice thickness, 4 mm [no gap]; number of slices, 11; scan time, 6.5 min), coronal oblique T2-weighted images (3000/102; echo train length, seven; field of view, 18 cm; matrix, 195 × 256; acquisitions, two; slice thickness, 4 mm [no gap]; number of slices, 11; scan time, 5.7 min), and sagittal oblique T2-weighted images (3000/102; echo train length, seven; field of view, 18 cm; matrix, 196 × 256; acquisitions, two; slice thickness, 4 mm [no gap]; number of slices, 11; scan time, 5.7 min).

Evaluation of Superior Glenoid Labrum

All original MR imaging studies were retrospectively reviewed by a board-certified radiologist with 9 years of clinical experience in musculoskeletal MR imaging who had no knowledge of either the patient mix or the arthroscopic diagnoses. MR images of the torn and normal superior labra were mixed and were evaluated in a random fashion.

Using the nonmagnified coronal oblique turbo spin-echo proton density-weighted (high-field MR imaging) and the nonmagnified conventional spin-echo T1-weighted (low-field MR imaging) images, the reviewer classified the intralabral signal-intensity pattern of the superior labrum into one of four patterns. Homogeneous low signal intensity of the labrum was designated as a type A superior labrum (Fig. 1A,1B). A single linear focus of abnormal increased signal intensity extending from the articular surface of the labrum but neither to nor through the bursal side of the labrum was designated a type B superior labrum (Fig. 2A,2B). The type B pattern was included because it is often seen in the normal superior labrum in the region of the origin of the long head of the biceps tendon [15]. A branched linear focus or a stellate focus of abnormally increased signal intensity involving at least one articular surface of the superior labrum was designated as a type C superior labrum (Figs. 3A,3B,4A,4B,5A,5B). An amorphous globular focus of increased signal intensity within the superior labrum was designated as a type D superior labrum (Fig. 6A,6B).

figure
View larger version (167K)
Fig. 1A. —38-year-old woman with type A superior labrum. High-field coronal oblique proton density-weighted MR image (TR/TE, 2000/20) of right shoulder shows diffuse homogeneous low signal intensity (solid arrow) in fibrocartilaginous superior labrum. Hyaline articular cartilage (open arrow), deep in relation to superior labrum, is linear area of hyperintense signal intensity that parallels osseous glenoid. At arthroscopy (not shown), labrum was found to be normal.

figure
View larger version (36K)
Fig. 1B. —38-year-old woman with type A superior labrum. Drawing shows type A superior labrum pattern.

figure
View larger version (158K)
Fig. 2A. —48-year-old woman with type B superior labrum. Low-field coronal oblique T1-weighted image (TR/TE, 767/26) of left shoulder shows linear area of high signal intensity (arrow) in superior labrum. This labral pattern is commonly seen at and anterior to origin of biceps tendon. At arthroscopy (not shown), superior labrum was found to be normal.

figure
View larger version (55K)
Fig. 2B. —48-year-old woman with type B superior labrum. Drawing shows type B superior labrum pattern.

figure
View larger version (144K)
Fig. 3A. —23-year-old man with type C superior labrum. High-field coronal oblique proton density-weighted image (TR/TE, 2000/40) of left shoulder shows oblique L-shaped focus of hyperintense signal intensity (arrow) that extends through articular side of superior labrum. At shoulder arthroscopy (not shown), type II superior labral anteroposterior tear was seen.

figure
View larger version (50K)
Fig. 3B. —23-year-old man with type C superior labrum. Drawing shows oblique L-shaped pattern of intralabral signal intensity.

figure
View larger version (162K)
Fig. 4A. —30-year-old man with type C superior labrum. High-field coronal oblique proton density-weighted image (TR/TE, 2000/20) of right shoulder shows complex branched pattern of abnormal signal intensity (arrow) that involves large area of superior labrum. At arthroscopy (not shown), type III superior labral anteroposterior tear was shown.

figure
View larger version (48K)
Fig. 4B. —30-year-old man with type C superior labrum. Drawing shows complex branched pattern of intralabral signal intensity.

figure
View larger version (151K)
Fig. 5A. —30-year-old man with type C superior labrum. High-field coronal oblique proton density-weighted image (TR/TE, 2000/20) of right shoulder shows stellate focus of abnormal signal intensity (arrow) in superior labrum. At arthroscopy (not shown), type II superior labral anteroposterior tear was seen.

figure
View larger version (50K)
Fig. 5B. —30-year-old man with type C superior labrum. Drawing shows this pattern of increased intralabral signal intensity.

figure
View larger version (210K)
Fig. 6A. —44-year-old man with type D superior labrum. Low-field coronal oblique T1-weighted image (TR/TE, 650/26) of right shoulder shows globular focus of abnormal signal intensity (arrow) that communicates with bursal side of superior labrum. Type II superior labral anteroposterior tear was found at shoulder arthroscopy (not shown).

figure
View larger version (57K)
Fig. 6B. —44-year-old man with type D superior labrum. Drawing shows globular focus of intralabral signal intensity.

To determine the sensitivity, specificity, and accuracy of MR imaging, the diagnostic criteria for a normal or normal variant of the superior labrum was labral type either A or B. The diagnostic criteria for a SLAP tear was the presence of labral type either C or D.

Different types of arthroscopic SLAP tears are designated by roman numerals, I-IV, whereas the four types of superior labrum based on MR imaging signal-intensity patterns are indicated by capital letters, A-D. Because the number of patients with arthroscopic SLAP tear types III and IV is small compared with the number of types I and II tears in this study, we did not attempt to distinguish among the different arthroscopic types of SLAP tear on MR imaging.

Assessment of the Rotator Cuff and Biceps Tendon

Morphology and signal intensity of the rotator cuff tendons and biceps tendon were evaluated on oblique coronal or oblique sagittal T2-weighted images. The primary diagnostic sign for a full-thickness tear was either a discrete tendon gap or abnormally high signal intensity within the tendon that was isointense to fluid and that extended through the entire width of the tendon [16]. The primary diagnostic sign for a partial-thickness tear was abnormally high signal intensity (not necessarily as hyperintense as fluid) that did not involve the entire thickness of the rotator cuff or biceps tendon [16,17,18] (Fig. 7A,7B). A secondary sign of a full-thickness rotator cuff tear was fluid in the subacromial or subdeltoid bursae or tendon retraction [16].

figure
View larger version (200K)
Fig. 7A. —61-year-old man with superior labral anteroposterior (SLAP) tear and long head of biceps tendinopathy (anterior is to reader's left). High-field oblique sagittal T2-weighted image shows coracohumeral ligament (arrowhead).

figure
View larger version (199K)
Fig. 7B. —61-year-old man with superior labral anteroposterior (SLAP) tear and long head of biceps tendinopathy (anterior is to reader's left). Contiguous oblique sagittal image shows focus of increased signal intensity in biceps tendon (arrow) just inferior to coracohumeral ligament (arrowhead).

Statistical Analysis

The sensitivity, specificity, and accuracy of both high-field and low-field MR imaging for the diagnosis of SLAP tears were determined, with arthroscopy as the gold standard. Sensitivity was defined as the quotient of true-positive diagnoses of SLAP tears on MR imaging (numerator) and both true-positive and false-negative diagnoses (denominator). Specificity was defined as the quotient of true-negative diagnoses of SLAP tears on MR imaging (numerator) and both true-negative and false-positive diagnoses (denominator). Accuracy was calculated as the proportion of all patients who were correctly classified, or the proportion of patients who were either true-positive or true-negative. Positive predictive value was defined as the quotient of true-positive diagnoses of SLAP tears on MR imaging (numerator) and all positive diagnoses (both true-positive and false-positive diagnoses [denominator]). Negative predictive value was defined as the quotient of true-negative diagnoses of SLAP tears (numerator) and all negative diagnoses (both true- and false-negative diagnoses [denominator]).

Because some of the expected frequencies were small (i.e., fewer than five), a two-sided Fisher's exact test was used to test for a statistically significant difference. The null hypothesis was rejected with a p value of less than 0.05.

Results
Previous sectionNext section
Sensitivity of MR Imaging for Arthroscopic SLAP Tear Types I-IV

At shoulder arthroscopy, 41 patients had a SLAP lesion, including 11 type I tears, 25 type II tears, three type III tears, and two type IV SLAP tears. The sensitivity of high-field and low-field MR imaging for diagnosis of the four arthroscopic SLAP types is summarized in Table 1.

TABLE 1 Sensitivity of High-Field and Low-Field MR Imaging for the Diagnosis of Arthroscopic Superior Labrum-Anteroposterior (SLAP) Tear Types I-IV

Performance Characteristics of High-Field MR Imaging

For the diagnosis of SLAP tear, the sensitivity of high-field MR imaging was 90% (95% confidence interval [CI] = 74%, 98%), specificity was 63% (95% CI = 35%, 85%), and accuracy was 80% (95% CI = 66%, 91%).

The distribution of labral patterns in the group of 27 true-positive diagnoses included 20 type C and seven type D cases. Of the true-negative diagnoses, the distribution of labral pattern types was three type A cases and seven type B cases. Three false-negative diagnoses were made. On MR imaging, all three had type B labral patterns. At arthroscopy, two missed tears were type II, and one was a type I SLAP tear. Six false-positive diagnoses included two type C and four type D labral patterns.

With a type A or B labral pattern on high-field MR imaging, the negative predictive value for a SLAP tear was 77% (10/13). With a type C or D labral pattern, the positive predictive value for a SLAP tear was 82% (27/33).

Performance Characteristics of Low-Field MR Imaging

For the diagnosis of a SLAP tear, the sensitivity of low-field MR imaging was 64% (95% CI = 31%, 89%), specificity was 70% (95% CI = 35%, 93%), and accuracy was 67% (95% CI = 43%, 85%).

The distribution of labral patterns in the group of seven true-positive diagnoses included type C in four patients and type D in three patients. Of the seven true-negative diagnoses, the distribution of labral pattern types was type A in two patients and type B in five patients. Four false-negative diagnoses were found on low-field MR imaging. A type B labral pattern was present in three patients, and the type A labral pattern was seen in the other false-negative patient. At arthroscopy, three of the missed tears were type I SLAP tears, and one was a type II tear. Three false-positive diagnoses included two type C and one type D labral patterns.

With a type A or B labral pattern on low-field MR imaging, the negative predictive value for a SLAP tear was 64% (7/11). With a type C or D labral pattern, the positive predictive value for a SLAP tear was 70% (7/10).

Diagnosis of SLAP Tear Based on Specific Labral Patterns

A normal superior labrum at arthroscopy was often associated with glenoid labral patterns A or B on high-field and low-field MR imaging (Table 2). The ratios of a true-negative to false-negative diagnosis of SLAP tear for labral patterns A and B were 5:1 and 2:1, respectively. The negative predictive values for labral type A and for type B were 83% (5/6) and 67% (12/18), respectively. With either type A or B, the negative predictive value of MR imaging was 71% (17/24). Conversely, a tear of the superior labrum was often predicted by labral patterns C or D on high-field and low-field MR imaging (Table 3). The ratios of a true-positive to a false-positive diagnosis of a SLAP tear for labral patterns C and D were 6:1 and 2:1 respectively. The positive predictive values for labral type C and type D were 86% (24/28) and 67% (10/15), respectively. Given labral type either C or D, the positive predictive value of MR imaging was 79% (34/43) for the diagnosis of a SLAP tear.

TABLE 2 Negative Predictive Value of Labral Types A and B for SLAP Tears on High-Field and Low-Field MR Imaging

TABLE 3 Positive Predictive Value of Labral Types C and D for SLAP Tears on High-Field and Low-Field MR Imaging

Rotator Cuff and Long Head of the Biceps Tendons

Among the 41 patients with SLAP tears at arthroscopy, 15 had full-thickness rotator cuff tears, 13 had partial-thickness tears, and 13 had normal-appearing rotator cuff tendons on MR imaging. In the 26 patients with a normal superior labrum at arthroscopy, nine had full-thickness cuff tears, six had partial tears, and 11 had normal rotator cuff tendons on MR imaging (p = 0.38). All fullthickness tears and articular-side partial rotator cuff tears were confirmed at arthroscopy.

Abnormally high signal intensity in the biceps tendon was noted in six patients (15%) with arthroscopically proven SLAP tears. Abnormal signal intensity was found in only one (4%) patient who had no evidence of a SLAP lesion at arthroscopy (p = 0.01). On both high-field and low-field MR imaging, abnormal signal intensity in the biceps tendon was most often shown on sagittal oblique images of the shoulder. On these images, it was often difficult to distinguish the origin of the biceps tendon from the anterior margin of the rotator cuff. However, we found that the coracohumeral ligament provided an important landmark for the origin of the long head of the biceps tendon on these images (Fig. 7A,7B).

Discussion
Previous sectionNext section

The value of MR imaging for the diagnosis of SLAP tears has been suggested in several small case series [7, 9, 11, 12, 14]. However, the clinical efficacy of MR imaging has been questioned because other studies have reported the sensitivity of MR imaging for the diagnosis of SLAP tears to be as low as 17% [9, 14]. Most recently, Synder et al. [3] reported that 75 MR imaging studies obtained from multiple imaging centers suggested an injury to the superior labrum in only 26% of patients with arthroscopic evidence of SLAP tears. We report the sensitivity, specificity, and accuracy of high-field MR imaging to be 90%, 63%, and 80%, respectively; and for the arthroscopic type II SLAP tear, the most common type encountered clinically, the sensitivity of high- and low-field MR imaging was 88%. In this study, low-field (0.2-T) MR imaging had a sensitivity and specificity of 64% and 70%, respectively. Differences in imaging technique and in particular differences in spatial resolution may explain the lower sensitivity of low-field MR imaging. The low-field MR imaging technique that was used in this study included T1-weighted oblique coronal images; field of view, 18 cm; and contiguous slices, 4 mm. The high-field MR imaging protocol includes proton density-weighted oblique coronal images; field of view, 16 cm; and contiguous thin slices, 3 mm. In the only other study of low-field MR imaging to our knowledge, Yoneda et al. [6] reported a specificity of 86% but a sensitivity of only 44% in 43 patients with SLAP tears. The minor difference between our results and those of Yoneda et al. may also be attributed to imaging technique. In that study, abnormal signal intensity in the superior labrum was best shown on 7-mm, oblique coronal, T2*-weighted gradient-echo images. On low-field MR imaging, we found that 4-mm thin T1-weighted images were better than T2-weighted images for defining intralabral signal intensity abnormalities. This finding is similar to findings on MR imaging of the knee meniscus, in which short TE sequences show abnormal signal intensity in the fibrocartilage more accurately than T2-weighted sequences [19]. Thinner slices should also increase the sensitivity of MR imaging for detecting tears of the glenoid labrum. Given the performance characteristics of low-field MR imaging, radiologists need to be aware of ancillary findings that are associated with SLAP tears such as rotator cuff tears and abnormal signal intensity in the long head of the biceps tendon. Radiologists also need to investigate technique-related factors that may improve detection of labral abnormalities. Technical innovations that increase spatial resolution or image contrast between synovial fluid and cartilaginous structures (e.g., modified gradient-echo pulse sequences such as dual-echo steady state) will probably improve the performance of low-field MR imaging for the detection of labral tears.

In other studies of high-field MR imaging, the torn superior labrum has been correlated with several different patterns of abnormal intralabral signal intensity. The more common patterns that have been previously described include linear signal intensity that transgresses both the bursal and articular surfaces of the labrum, complex linear patterns of increased intralabral signal intensity, and a large globular focus of increased signal intensity [7,8,9, 11, 12, 14]. Our study confirms the value of specific labral patterns for predicting the presence or absence of a SLAP tear on MR imaging. The positive predictive value for labral types either C or D is 79%, and in particular, the type C pattern is associated with a SLAP tear in 86% of 28 patients. Conversely, the negative predictive value for labral types either A or B is 71%. False-negative and false-positive diagnoses can occur, particularly with labral types B and D, respectively. The type B labrum was designated as normal because the sublabral recess and the origin of the biceps tendon may appear as a linear focus of increased signal intensity in the anterior and middle part of the superior labrum [4, 11, 13, 14, 20, 21]. However, in one third of patients in this study, the type B pattern was associated with a SLAP tear at arthroscopy. In most of these false-negative findings, the linear focus of abnormal signal intensity was located posterior to the long head of the biceps tendon. Therefore, as has been mentioned by others [15, 21], the type B labral pattern should be considered a SLAP tear if it is seen posterior to the biceps tendon. Conversely, a false-positive diagnosis of SLAP tear was made in one third of patients in whom a globular focus of abnormal signal intensity was seen in the superior labrum (type D pattern). In these false-positive findings, we postulate that abnormal labral signal intensity may be attributed to fibrovascular tissue and mucoid or eosinophilic degeneration of the labrum [21, 22]. Finally, three of the four missed SLAP tears on low-field MR imaging were arthroscopic type I SLAP tears. The type I SLAP tear—marked fraying of the superior labrum without labral detachment—is the most difficult to diagnose on MR imaging and MR arthrography [7, 9, 23]. Some controversy exists as to whether this lesion is truly an injury or a degenerative lesion [3, 4].

Rotator cuff tears and long head of the biceps tendinopathy may accompany a superior labral tear. In our series, rotator cuff tears on MR imaging were seen in 68% of patients with SLAP tears, compared with other studies that have reported cuff tears in as many as 55% of patients with SLAP lesions [2, 3, 5, 6]. Rotator cuff tears are more common in older patients with SLAP lesions, and in these patients, the superior labral lesion may be degenerative, not traumatic [4]. Both chronic overuse tears of the posterosuperior labrum and articular-side partial tears of the supraspinatus and infraspinatus tendons may result from internal impingement in the overhead position [24, 25]. Abnormal signal intensity in the biceps tendon was identified in 15% of patients with SLAP tears; it occurred significantly more often than in those with a normal superior labrum. Contiguous T2-weighted images in the oblique sagittal plane show the origin and proximal part of the long head of the biceps tendon in cross-section. The most medial and proximal part of the long head of the biceps tendon can be identified just inferior to the coracohumeral ligament at the point where this ligament merges with the supraspinatus tendon [26] (Fig. 7A,7B). This part of the long head of the biceps tendon is most often injured in patients with SLAP tears [1,2,3].

Our study had several limitations. First, it was a retrospective study that included a large number of SLAP tear patients in the cohort. Therefore, bias was introduced in favor of diagnosing SLAP tears on MR imaging even though the reviewer was unaware of the final arthroscopic diagnosis. Second, this study was subject to verification or workup bias: when a study is restricted to patients who require an invasive test to verify disease, then the study population may be biased toward patients with more severe disease. In this study, the reference standard was shoulder arthroscopy. Patients with negative findings on MR imaging would be less likely to undergo arthroscopy, and this tendency may result in a report of relatively higher sensitivity and lower specificity for MR imaging than is true because the number of negative tests (both false-negative and true-negative) is decreased. Third, although we made an attempt to match groups of patients with and without SLAP tears, the number of patients in those groups was not equal. In addition, the number of patients who had low-field MR imaging was small compared with the number who had high-field studies. Finally, only one radiologist evaluated the MR imaging studies; therefore, variability among reviewers for the classification of labral patterns was not assessed. Furthermore, because only one interpretation was evaluated, its reproducibility could not be addressed in this study.

In conclusion, the performance characteristics of high-field MR imaging were superior to those of low-field MR imaging for the diagnosis of SLAP tears. We defined patterns of intralabral signal intensity that correspond to the normal and to the torn superior labrum. In particular, a branched linear pattern of abnormal intralabral signal intensity is most consistent with a SLAP tear. T2-weighted oblique sagittal images should be evaluated for abnormal signal intensity in the long head of the biceps tendon, which was seen in 15% of patients with SLAP tears. Finally, partial-thickness and full-thickness rotator cuff tears are often associated with superior labral tears.

Address correspondence to G. A. Tung.

References
Previous sectionNext section
1. Andrews JR, Carson WG Jr, Mcleod WP. Glenoid labrum tears related to the long head of the biceps. Am J Sports Med 1985; 13:337-341 [Google Scholar]
2. Snyder S, Karzel R, Del Pizzo W, Ferkel RD, Friedman MJ. SLAP lesions of the shoulder. Arthroscopy 1990; 14:274-279 [Google Scholar]
3. Snyder SJ, Banas MP, Karzel RP. An analysis of 140 injuries to the superior glenoid labrum. J Shoulder Elbow Surg 1995; 4:243-248 [Google Scholar]
4. DePalma A, Gallery G, Bennett G. Variational anatomy and degenerative lesions of the shoulder joint. In: Instructional course lectures: American Academy Orthopedic Surgery, vol. 6. St. Louis: Mosby,1949: 225-281 [Google Scholar]
5. Maffet MW, Gartsman GM, Moselely B. Superior labrum-biceps tendon complex lesions of the shoulder. Am J Sports Med 1995; 23:93-98 [Google Scholar]
6. Yoneda M, Izawa K, Hirooka A, Hayashida K, Wakitani S. Indicators of superior glenoid labral detachment on magnetic resonance imaging and computed tomography arthrography. J Shoulder Elbow Surg 1998; 7:2-12 [Google Scholar]
7. Cartland JP, Crues JV, Stauffer A, Nottage W, Ryu RK. MR imaging in the evaluation of SLAP injuries of the shoulder: findings in 10 patients. AJR 1992; 159:787-792 [Abstract] [Google Scholar]
8. Garneau RA, Renfrew DL, Moore TE, el-Khoury GY, Nepola JV, Lemke JH. Glenoid labrum: evaluation with MR imaging. Radiology 1991; 179:519-522 [Google Scholar]
9. Hodler J, Kursunoglu-Brahme S, Flannigan B, Snyder SJ, Karzel RP, Resnick D. Injuries of the superior portion of the glenoid labrum involving the insertion of the biceps tendon: MR imaging findings in nine cases. AJR 1992; 159:565-568 [Abstract] [Google Scholar]
10. Hunter JC, Blatz DJ, Escobedo EM. SLAP lesions of the glenoid labrum: CT arthrographic and arthroscopic correlation. Radiology 1992; 184:513-518 [Google Scholar]
11. McCauley TR, Pope CF, Jokl P. Normal and abnormal glenoid labrum: assessment with multiplanar gradient-echo MR imaging. Radiology 1992; 183:35-37 [Google Scholar]
12. Monu J, Pope TL Jr, Chabon SJ, Vanarthos WJ. MR diagnosis of superior labrum-anterior to posterior (SLAP) injuries of the glenoid labrum: value of routine imaging without intraarticular injection of contrast material. AJR 1994; 163:1425-1429 [Abstract] [Google Scholar]
13. Tuite MJ, Orwin JF. Anterosuperior labral variants of the shoulder: appearance on gradient-recalled-echo and fast spin-echo MR images. Radiology 1996; 199:537-540 [Google Scholar]
14. Smith AM, McCauley TR, Jokl P. SLAP lesions of the glenoid labrum diagnosed with MR imaging. Skeletal Radiol 1993; 22:507-510 [Google Scholar]
15. Smith DK, Chopp TM, Aufdemorte TB, Witkowski EG, Jones RC. Sublabral recess of the superior glenoid labrum: study of cadavers with conventional nonenhanced MR imaging, MR arthrography, anatomic dissection, and limited histologic examination. Radiology 1996; 201:251-256 [Google Scholar]
16. Stoller DW, Fritz RC. Magnetic resonance imaging of impingement and rotator cuff tears. Magn Reson Imaging Clin N Am 1993; 1:47-63 [Google Scholar]
17. Rafii M. Shoulder. In: Firooznia H, Golimbu C, Rafii M, eds. MRI and CT of the musculoskeletal system. St. Louis: Mosby, 1990: 526-527 [Google Scholar]
18. van Leersum M, Schweitzer ME. Magnetic resonance imaging of the biceps complex. Magn Reson Imaging Clin N Am 1993; 1:77-86 [Google Scholar]
19. Kaplan PA, Nelson NL, Garvin KL, Brown DE. MR of the knee: the significance of high signal in the meniscus that does not clearly extend to the surface. AJR 1991; 156:333-336 [Abstract] [Google Scholar]
20. Yeh L, Kwak S, Kim YS, et al. Anterior labroligamentous structures of the glenohumeral joint: correlation of MR arthrography and anatomic dissection in cadavers. AJR 1998; 171:1229-1236 [Abstract] [Google Scholar]
21. Kreitner KF, Botchen K, Rude J, Bittinger F, Krummenauer F, Thelen M. Superior labrum and labral-bicipital complex: MR imaging with pathologic-anatomic and histologic correlation. AJR 1998; 170:599-605 [Abstract] [Google Scholar]
22. Loredo R, Longo C, Slonen D, et al. Glenoid labrum: MR imaging with histologic correlation. Radiology 1995; 196:33-41 [Google Scholar]
23. Bresler F, Blum A, Braun M, et al. Assessment of the superior labrum of the shoulder joint with CT-arthrography and MR-arthrography: correlation with anatomical dissection. Surg Radiol Anat 1998; 20:57-62 [Google Scholar]
24. Walch G, Liotard JP, Boileau P, Noel E. Postero-superior glenoid impingement: another impingement of the shoulder [in French]. J Radiol 1993; 74:47-50 [Google Scholar]
25. Jobe CM. Superior glenoid impingement. Orthop Clin North Am 1997; 28:137-143 [Google Scholar]
26. Ho CP. MR imaging of rotator interval, long biceps, and associated injuries in the overhead-throwing athlete. Magn Reson Imaging Clin N Am 1999; 7:23-37 [Google Scholar]

Recommended Articles

High-Field and Low-Field MR Imaging of Superior Glenoid Labral Tears and Associated Tendon Injuries

Full Access, ,
American Journal of Roentgenology. 2003;181:1449-1462. 10.2214/ajr.181.6.1811449
Citation | Full Text | PDF (48998 KB) | PDF Plus (2194 KB) 
Full Access, ,
American Journal of Roentgenology. 2000;174:1099-1106. 10.2214/ajr.174.4.1741099
Abstract | Full Text | PDF (7256 KB) | PDF Plus (7341 KB) 
Full Access, ,
American Journal of Roentgenology. 2011;197:604-611. 10.2214/AJR.11.6575
Abstract | Full Text | PDF (1097 KB) | PDF Plus (948 KB) 
Full Access, , , , ,
American Journal of Roentgenology. 2004;182:1271-1278. 10.2214/ajr.182.5.1821271
Abstract | Full Text | PDF (7662 KB) | PDF Plus (7787 KB) 
Full Access,
American Journal of Roentgenology. 2006;187:1448-1452. 10.2214/AJR.05.0338
Abstract | Full Text | PDF (2049 KB) | PDF Plus (2126 KB) 
Full Access, , , , , , ,
American Journal of Roentgenology. 1993;161:1229-1235. 10.2214/ajr.161.6.8249731
Abstract | PDF (1252 KB) | PDF Plus (696 KB)