| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Musculoskeletal Imaging |
1 From the Department of Radiology, University of Wisconsin Hospital and Clinics, 600 Highland Ave, E3/311, Madison, WI 53792-3252. From the 2001 RSNA scientific assembly. Received May 7, 2001; revision requested June 8; revision received September 20; accepted October 22. Address correspondence to M.J.T. (e-mail: mjtuite@facstaff.wisc.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: We reviewed transverse MR images in 50 consecutive patients with a sublabral foramen or Buford complex at arthroscopy. Images were randomly mixed with those of 58 patients with either a normal labrum (n = 20) or an anterior labral tear (n = 38) at arthroscopy. MR imaging was fat suppressed fast spin echo intermediate or T2 weighted (repetition time msec/effective echo time msec, 1,800–3,000/30–102). Two radiologists evaluated by means of consensus the anterior labrum while blinded to patient history and arthroscopic results. Transverse images obtained through the glenoid fossa were totalled to determine the midpoint. Sensitivity, specificity, and accuracy of MR for depicting a sublabral foramen or Buford complex were calculated along with 95% CIs, by using surgical findings as the reference standard.
RESULTS: The sensitivity of MR for diagnosing a sublabral foramen or Buford complex was 0.94 (47 of 50 patients, 95% CI: 0.87, 1.00), specificity was 0.80 (16 of 20 patients, 95% CI: 0.62, 0.97), and accuracy was 0.90 (63 of 70 patients, 95% CI: 0.82, 0.97). The anterior labrum was abnormal on the first transverse section inferior to the midpoint in nine (18%) patients. The labrum was also abnormal on the second section below the midpoint in three (6%) patients. Because of the anterior tilt of the scapula, the midpoint was near the anterior glenoid notch at about the position between 2- and 3-o’clock.
CONCLUSION: The labrum may be unattached or absent on the first two transverse images obtained below the midpoint.
© RSNA, 2002
Index terms: Shoulder, anatomy, 41.481 • Shoulder, injuries, 41.481 • Shoulder, MR, 41.121411, 41.121415
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONREFERENCES |
|---|
The anatomic region of the anterior labrum that is abnormal is one of the important criteria for distinguishing a normal variant from a labral tear on MR images. In patients with anterior instability after glenohumeral dislocation, the detached or absent labrum usually involves the anteroinferior quadrant (8). It has been reported that it is important to look for the tear involving the inferior half of the anterior labrum on conventional MR images because the portion of a Bankart tear above the midpoint may be more obvious at initial inspection, thus mimicking a normal variant (9). This is particularly important in patients with a chronic Bankart tear, in whom there are few additional findings such as edema or hemorrhage on MR images and in whom the tear is minimally displaced.
Superior labral anterior to posterior (SLAP) tears can also involve the anterosuperior labrum because they extend anteriorly and inferiorly, but the tear involving the 12-o’clock region should be visible on the coronal oblique images (10–12). Therefore, most authors (5,6,13) consider an unattached or absent labrum isolated to the upper half of the anterior glenoid rim with a normal superior labrum as a normal variant, while a similar MR appearance involving the lower half of the labrum represents a tear.
Not all authors agree that the sublabral foramen and Buford complex extend down to but not below the 3-o’clock position that divides the anterior labrum into superior and inferior halves. Williams et al (7) and Stoller and Wolf (3) have reported that the normal variants should not extend below the anterior glenoid notch (between the 2- and 3-o’clock positions). The anterior glenoid notch is an indentation in the glenoid rim at the anterior edge of the physeal line that roughly divides the upper one-third of the glenoid from the lower two-thirds. Alternatively, Palmer et al (14) state that the labral variants occur between the MGL and the anterior band of the inferior glenohumeral ligament (IGL). If this is true, the anterosuperior labral variant in some people might still be visible on a transverse section at or slightly below the 3-o’clock position, because the anterior band of the IGL can occasionally originate at the 4-o’clock position (15).
It is important when interpreting a shoulder MR study to distinguish between the anterosuperior labral variants and a labral tear. Patients with pain, instability, and a labral tear benefit from surgery, while surgery is unnecessary for the anterosuperior labral variants. It has been reported (7) that if the MGL of a Buford complex is mistakenly surgically attached to the glenoid rim, patients can have painful restriction of humeral rotation and elevation.
Although sublabral foramen and Buford complex variants are often clearly visible on transverse MR images obtained through the superior portion of the anterior labrum, some normal variants are not as well seen on every image through the anterosuperior labrum. Because of the uncertainty in the literature (3,5–7, 13,14) as to the maximum inferior extent of the normal variants, we did not know the point below which we could confidently call a labral abnormality a tear. The purpose of our study was to determine the inferiormost extent of the anterosuperior labral variants on conventional transverse MR images.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONREFERENCES |
|---|
All of the patients who underwent MR imaging and arthroscopy during the same period and had an arthroscopically proven anterior labral tear (31 patients) or a sublabral foramen and anterior labral tear (seven patients), were combined with 20 randomly selected patients who underwent MR imaging during the same period and had an arthroscopically normal labrum. These 58 patients made up the control group. There were no patients with both an anterior labral tear and a Buford complex. There were 47 men and 11 women in the control group, with an average age of 35 years (range, 16–66 years). The primary postoperative diagnoses of the patients with a normal labrum were impingement or rotator cuff tear in 14, multidirectional instability in four, and bursal adhesions and adhesive capsulitis in one each.
All MR images were obtained with a 1.5-T scanner (GE Medical Systems, Milwaukee, Wis) with a phased-array shoulder coil (Medical Advances, Milwaukee, Wis). Patients underwent scanning while supine, with the arm in slight external rotation. Fast spin-echo intermediate-weighted or T2-weighted transverse images were obtained, with a repetition time msec/effective echo time msec of 1,800–3,000/30–102, an echo-train length of 5–6, matrix of 256 x 192–256, three to four signals acquired, and frequency-selective fat suppression. The field of view was 14 cm, and the section thickness was 3 mm, with a 1-mm intersection gap. The transverse images were proscribed from a coronal oblique image and were perpendicular to the tabletop and to the long axis of the imager bore. Coronal oblique and sagittal oblique images (2,356/102 [effective]) also were obtained but were not evaluated for this study.
The transverse MR images were randomly mixed and the anterior labrum evaluated by two musculoskeletal radiologists (M.J.T., D.G.B.) blinded to patient history, arthroscopic results, and relative proportion of the various arthroscopic findings. All readings were done by means of consensus. Transverse images through the glenoid fossa were summed to determine the middle transverse section. If an even number of sections was counted through the glenoid, the midpoint between equal numbers of sections was identified. We then determined on which transverse sections the anterior labrum was abnormal and recorded their location relative to the midaxial section or midpoint. The labrum was considered abnormal if high signal intensity extended to either free surface, if a portion of the labrum was separated from either the rest of the labrum or the glenoid rim, or if the labrum was absent (16). We did not attempt to distinguish an unattached or detached labrum from an absent labrum. The institutional review board at our institution did not require its approval or informed consent for review of patient medical records or images.
We randomly selected 10 patients from the study group and in each patient localized the transverse sections on the sagittal oblique image that contained the glenoid fossa of the scapula. One author (M.J.T.) then determined the transverse section that passed through the anterior glenoid notch and the transverse section through the 3-o’clock position. These sections were again recorded relative to the midpoint or midsection. The anterior glenoid notch is the normal indentation in the contour of the anterosuperior glenoid rim, as seen on sagittal oblique images, usually at about the position between 2- and 3-o’clock. The 3-o’clock position was determined by first drawing a line down the longest axis of the roughly oval-shaped glenoid fossa and then a perpendicular line from the midpoint of that line to the intersection with the anterior glenoid rim. Because of the anterior tilt of the scapula, the line down the longest axis of the glenoid fossa was oriented slightly anterosuperior to posteroinferior.
The sensitivity, specificity, and accuracy of MR for depicting a sublabral foramen or Buford complex were calculated, along with 95% CIs, by using the surgical findings as the reference standard. Sensitivity was defined as, for those patients with a sublabral foramen or Buford complex, the fraction that had at least one transverse section above the midpoint or midsection graded as abnormal. Specificity was calculated by determining the fraction of patients with an arthroscopically normal anterior labrum whose anterior labrum above the midpoint or midsection was graded as normal on MR images.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONREFERENCES |
|---|
Of the 47 patients with arthroscopically proven anterosuperior labral variants and labral abnormalities on MR images, the labral abnormality was seen only superior to the midpoint or midsection in 38 (81%) patients (Table). In the other nine (19%) patients, the labral abnormality also extended below the midpoint or midsection (Fig 1). Six (13%) patients had an abnormal labrum that extended to the first section inferior to the midpoint or midsection. In three (6%) patients, one with a sublabral foramen and two with a Buford complex, the labrum was graded as abnormal on the first two sections below the midpoint or midsection. In none of the patients was the labrum abnormal below the second section inferior to the midpoint of the glenoid.
|
|
|
Anterior Labral Tear Only
Of the 31 patients with an anterior labral tear, the labrum was graded as abnormal on at least one transverse section below the midpoint or midsection in 27 patients. The abnormality extended superiorly to involve at least one section above the midpoint or midsection in 13 of these 27 patients (Fig 2). Of the other 14 patients, one had a normal-appearing intervening first section above the midpoint, with an abnormal labrum on the second section above the midpoint. The anterior labrum in this patient would have been considered as having a Bankart tear and a (false-positive) sublabral foramen. In the other 13 patients, the labrum was detached or absent on only MR images obtained below the midpoint or midsection. Of the 20 patients with a normal anterior labrum, all had an anteroinferior labrum graded as normal.
|
|
Anterior Labral Position of the Middle Transverse Section
There was slight variability in the location where the middle transverse section or midpoint intersected the anterior labrum in the 10 selected patients. The midpoint or midsection passed through the anterior glenoid notch in four patients, half a section width above the notch in one patient, within a section below the notch in four patients, and two sections below the notch in one patient. Because of the anterior tilt of the scapula, the 3-o’clock position was intersected by the first section inferior to the midpoint or midsection in six patients, between the first and second inferior sections in one patient, and crossed by the second section below the midpoint or midsection in three patients.
One of these 10 patients had a sublabral foramen; the patient’s anterior labrum was interpreted as abnormal on the two sections below the midsection (Fig 3). The midsection passed within one section below the anterior glenoid notch, and the second section below the midsection crossed the 3-o’clock position.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONREFERENCES |
|---|
After counting the total number of transverse sections through the glenoid fossa, we also found that the middle transverse section crosses the anterior labrum above the 3-o’clock position. The middle section does not cross the anterior labrum at a uniform position, although it is usually within one section width of the anterior glenoid notch. Similarly, either the first or second section below the midpoint or midsection intersects the 3-o’clock position when 3-mm-thick sections with a 1-mm gap are used. This variability appears to be due to slight differences in the amount of anterior tilt of the scapula and the size of the glenoid fossa. These findings are based on an analysis of a subset of 10 patients, however, which is a limitation of these conclusions.
In this study, we decided to count the total number of transverse sections through the glenoid fossa and use these as our reference points for localizing the abnormal portions of the labrum. We chose this technique because it was objective and easy to perform on conventional MR images. Although a clock face is the most common method of specifying a location on the labrum, our transverse sections are localized from a coronal oblique image, and we do not routinely obtain a transverse localizer superimposed on a sagittal oblique image through the glenoid fossa. In addition, the anterior glenoid notch and the 3-o’clock position can sometimes be difficult to determine in a nondistended joint. Sagittal oblique MR arthrographic images show the outline of the glenoid fossa much better, and we have found it easier to specify the clock-face location of transverse sections on images from these studies. Also, it is fairly simple to specify a cross-reference image when interpreting images at a soft-copy reading station. Localizing transverse images on a sagittal oblique image through the glenoid fossa during soft copy MR image interpretation may be an additional useful technique for helping to distinguish a normal variant from a chronic labral tear.
We did not attempt to determine the site of origin on the labrum of the anterior band of the IGL because this is often difficult on conventional MR images. MR arthrographic depiction of the variable origin of the anterior band of the IGL is much better and is therefore superior for determining whether the unattached or absent labrum occurs above the origin of the ligament, as is typically seen in the normal variants, or involves the origin, as is seen in most tears. The ability of MR arthrography to demonstrate the anterior band of the IGL would be particularly helpful in patients whose normal labral variant extends two sections below the midaxial section and for those with chronic Bankart tears that extend above the 3-o’clock position.
Of the 31 patients in our study with an anterior labral tear and no anterosuperior labral variant, only one had a tear that was seen exclusively above the midpoint. This was a small tear near the midglenoid notch that probably occurred after a fall and was not associated with an anterior dislocation or instability. Other authors (6,7,17) have reported that tears isolated to the anterosuperior labrum are uncommon but can also be seen in athletes who throw. Most anterior labral tears that require surgery occur after an anterior glenohumeral joint dislocation. The anteroinferior direction that the humeral head must travel to pass below the coracoid process and the typical site of attachment of the anterior band of the IGL cause most anterior labral tears to at least partly involve the anteroinferior labrum.
Large SLAP tears can also extend down to involve the anterosuperior labrum but are usually also seen on coronal oblique images at the 12-o’clock position. Although some SLAP tears are not seen on coronal oblique conventional MR images, we are unaware of any reports in which a tear was seen on transverse sections in the anterosuperior labrum but not on the coronal oblique images. MR arthrography is probably a more sensitive technique than conventional MR for identifying the labral tear at the 12-o’clock position and therefore may be more accurate for distinguishing a SLAP tear from a normal variant in some patients (18,19). Several authors (5,13,20) have noted that the superior sublabral recess, another normal variant of the glenoid labrum, is often large in patients with a sublabral foramen. The superior recess curves medially over the top of the glenoid and does not involve the posterior third of the superior labrum (21); this finding helps distinguish a large superior recess associated with an anterosuperior labral variant from a SLAP tear.
Although the anatomic region of a labral abnormality is an important MR criterion for distinguishing a labral variant from a chronic Bankart tear, there are other MR findings that can be helpful. In addition to a deep superior sublabral recess, the normal variants are often associated with a thick cordlike MGL (7). Alternatively, Bankart tears will have an associated Hill-Sachs lesion in about 75% of patients (22). In addition, a history of a prior anterior dislocation or physical examination findings of anterior instability makes a Bankart tear more likely.
There were several limitations of this study. First, we did not mask the MR images in the regions of potential findings such as a Hill-Sachs lesion or a cordlike MGL. Although we attempted to analyze only the anterior labrum, we cannot be certain if the presence or absence of obvious secondary findings on the images may have influenced which sections were graded as abnormal. Second, we included only patients who underwent subsequent arthroscopy, which introduces a selection bias. Third, as mentioned earlier in this article, the transverse images were proscribed perpendicular to the tabletop and not to any anatomic landmarks. Although this is a common method for proscribing transverse images, this results in some variability in the position of the transverse midsection relative to the 3-o’clock position. Finally, we could not determine interobserver variability or error data because the MR images were read by means of consensus.
In conclusion, the anterosuperior labrum in most patients with a sublabral foramen or Buford complex is unattached or absent on only the transverse MR images obtained above the midpoint. However, in 6% of patients, the labrum is still abnormal on the first two transverse MR images obtained below the midpoint. If an anterosuperior normal labral variant is not well seen on the more superior transverse MR sections, other MR findings such as a thick MGL, a prominent superior sublabral recess, or the absence of Hill-Sachs lesion may be helpful in distinguishing a variant from an extensive chronic tear of the anterior labrum.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, M.J.T.; study concepts, M.J.T., M.S., J.F.O.; study design, M.J.T., M.S.; literature research, M.S., A.J.Z.; clinical studies, M.J.T., D.G.B., J.F.O.; data acquisition, M.J.T., D.G.B., A.J.Z.; data analysis/interpretation, M.J.T., D.G.B.; statistical analysis, M.J.T.; manuscript preparation, M.J.T., M.S.; manuscript definition of intellectual content, editing, revision/review, and final version approval, M.J.T.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. V. R. Mohana-Borges, C. B. Chung, and D. Resnick Superior Labral Anteroposterior Tear: Classification and Diagnosis on MRI and MR Arthrography Am. J. Roentgenol., December 1, 2003; 181(6): 1449 - 1462. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
