Glenohumeral joint instability is a complicated subject. Our understanding is greatly attributable to the orthopedic literature, which in recent years has focused on the arthroscopic evaluation of this problem. Despite controversy over accuracy and cost-effectiveness,1,2 the roles of conventional magnetic resonance (MR) imaging and MR arthrography have expanded greatly over the last decade. This article will focus on the common clinical problem of anterior glenohumeral joint instability and its assessment with contemporary MR techniques.
Dr. Cree M. Gaskin is currently a third-year radiology
resident at Duke University Medical Center. He graduated as
co-valedictorian from the University of Florida with a BS in
Basic Medical Science in 1994. He subsequently received his MD
from the University of Florida in 1997 as a member of Alpha Omega
Alpha. Dr. Gaskin will remain at Duke to begin a fellowship in
musculoskeletal/MR imaging in 2002.
What is glenohumeral instability?
The glenohumeral joint is a relatively unstable joint. The
inherent biomechanics that allow for its exceptional range of
motion are equally responsible for it being the most frequently
dislocated joint in the body. Glenohumeral instability is defined
simply as the recurrence of dislocation or subluxation of the joint
with the performance of activity.
Instability is classified by the direction of abnormal movement
of the humeral head relative to the glenoid margin. It can be
unidirectional (anterior or posterior) or multidirectional
(including inferior). Isolated inferior or superior instabilities
are rarely encountered.
Orthopedists often lump shoulder instability into two broad
categories under the acronyms TUBS and AMBRI. TUBS is used to
describe patients with traumatic unidirectional anterior-inferior
instability with a Bankart lesion (generally requiring surgical
intervention). AMBRI refers to patients with atraumatic
multidirectional bilateral instability who can be treated
successfully with rehabilitation or with surgery in the form of an
inferior capsular shift procedure.
This article is largely concerned with the TUBS population as
traumatic anterior glenohumeral joint instability constitutes >
95% of all glenohumeral instabilities encountered in clinical
Virtually all patients with glenohumeral instability have an
unmistakable history of experiencing displacement of the humeral
head. This is most often associated with trauma in the form of
forced abduction and external rotation, such as in a fall on the
outstretched arm. Subsequent to this acute event, the patient may
complain of anterior shoulder pain, joint clicking or catching, or
frank recurrent dislocation.
These symptoms are often exacerbated by overhead activities.
The physical examination for instability includes a multitude of
techniques, including the apprehension, relocation, load and shift,
inferior sulcus sign, and crank tests. The combination of clinical
history and physical examination are considered to be quite
accurate in determining the presence of instability lesions
Liu et al
found physical examination to be 90% sensitive and 85% specific for
labral tears confirmed on arthroscopy. Considering this inexpensive
and relatively accurate preoperative assessment, what is the role
for MR imaging in the evaluation of instability?
The answer to this question is not straightforward. Clearly,
arthroscopy can be used to adequately delineate and treat
abnormalities in patients with clinically obvious instability. Such
patients do not necessarily require diagnostic imaging. However,
preoperative MR (with or without arthrography) may be useful to
determine the presence and type of capsulolabral lesion in order to
aid in the selection of the appropriate surgical procedure.
Preoperative determination that the lesion will require an open
procedure (arthrotomy) prevents edematous saturation of tissues
from arthroscopy prior to conversion to the necessary arthrotomy.
This obviously also saves operating room time and expense.
In contemporary times of managed care and patient convenience, a
great deal of emphasis has been placed upon ambulatory surgery. In
some orthopedic practices, this has essentially mandated
preoperative imaging to determine whether a procedure will be open
or closed. The reason for this is that the ambulatory surgery
center may be remote from the main hospital, restricting access to
the inpatient services needed in cases requiring arthrotomy.
Therefore, in some orthopedic practices, preoperative imaging may
actually determine which surgical suite (main hospital versus
ambulatory surgical center) and accompanying fee schedule are
chosen for an operation.
As explained above, there are varying rationales behind
preoperative imaging in patients in whom instability is clinically
obvious. MR imaging can be even more valuable when the diagnosis is
uncertain. In these cases, MRI can confirm or deny the presence of
instability lesions as well as identify other abnormalities that
may have mimicked the clinical manifestations of instability.
Examples of such pathology include a displaced labrum, a loose
body, or a subluxating tendon of the long head of the biceps.
MR imaging may also uncover unsuspected instability lesions in
patients presenting solely with complaints of shoulder pain.
For example, instability may be the underlying cause of impingement
syndrome in young athletes, although its manifestations may be
obscured by symptoms related to the rotator cuff pathology.
Accurate diagnosis and treatment of instability in these patients
is required to successfully alleviate the impingement syndrome.
Whatever the case may be, MR imaging can lead to modification in
arthroscopic technique, the choice of an open surgical procedure as
a more appropriate therapeutic approach, or it may deem surgery
One recent study in the orthopedic literature objectively
evaluated the impact of shoulder MRI on clinical decision making.
This study achieved statistical significance in demonstrating that
shoulder MRI resulted in major alterations in clinical management
by orthopedic surgeons.
MRI altered both the primary diagnoses and the types of therapy
(operative versus nonoperative) chosen for shoulder disorders of
multiple etiologies, including anterior instability.
A similar study within the radiologic literature also achieved
statistical significance in demonstrating that shoulder MRI both
changed patient management (62% of cases) and improved clinicians'
Anatomy of anterior shoulder instability
The glenoid fossa of the scapula and the head of the humerus
form a ball-and-socket joint. The osseous glenoid is covered by
articular cartilage that thins centrally to form a shallow fossa.
The glenoid fossa is opposed to only one-third to one-fourth of the
articular surface of the humeral head at any given time. This
allows for great degrees of motion at the unfortunate expense of
stability. To compensate, the joint is reinforced by the rotator
cuff and the capsulolabral complex.
The joint capsule has both an outer fibrous layer and an inner
synovial layer. The capsule exhibits inherent laxity to allow for
wide ranges of motion, while becoming taut at extremes of motion to
provide support for the joint.
The anterior capsular mechanism provides stability during abduction
and external rotation. It is composed of the labrum, subscapularis
muscle and tendon, coracohumeral and glenohumeral ligaments,
scapular periosteum, and capsule.
The posterior capsular mechanism protects against posterior
dislocation. It is composed of the labrum, capsule, inferior
glenohumeral ligament (posterior band), rotator cuff muscles, and
The glenoid labrum (figure 1) is a capsular fold that inserts
onto the periphery of the bony glenoid, serving to deepen the
articular surface. It is composed of fibrous tissue with a
fibrocartilaginous transition zone at its insertion. In addition to
serving as an insertion site for much of the capsule, the labrum
and the adjacent superior glenoid tubercle serve as the origin for
the tendon of the long head of the biceps.
There are three capsular thickenings that provide extra
stability against anterior dislocation during abduction and
external rotation. These structures are known as the superior,
middle, and inferior glenohumeral ligaments. The superior and
middle glenohumeral ligaments originate from the anterior-superior
labrum (and sometimes the glenoid margin) immediately anterior to
the labral-bicipital anchor.
The superior glenohumeral ligament passes anteriorly until it
merges with the coracohumeral ligament in the rotator interval.
The rotator interval is the space between the superior border of
the subscapularis and the adjacent supraspinatus.
The middle glenohumeral ligament courses inferiorly along the
glenoid rim prior to merging with the subscapularis tendon. It
ultimately inserts upon the lesser tuberosity.
The inferior glenohumeral ligament attaches to the inferior
two-thirds of the entire labral circumference. It is also known as
the inferior glenohumeral ligament complex as it has three
components. The axillary pouch is the hammock-like portion
intervening between the ligament's thick anterior and posterior
bands. The inferior glenohumeral ligament is lax in the adducted
position. During abduction, the anterior and posterior bands move
superiorly with respect to the humeral head, ultimately becoming
taut and serving as the primary restraint against anterior and
posterior dislocations at 90 degrees of abduction.
The coracohumeral ligament arises from the lateral margin of the
coracoid process and courses through the rotator interval. It has
two separate insertions onto the greater and lesser tuberosities of
the humerus, forming a tunnel for the biceps tendon.
In addition to stabilizing the biceps tendon,
the ligament also provides support during external rotation
and while the arm is in a dependent position.
It is helpful to consider this anatomy while looking upon the
glenoid labrum from a lateral viewpoint (figure 2). Consider the
labrum as a clock face with 3 o'clock anteriorly, 6 o'clock
inferiorly, and 9 o'clock posteriorly. The important anatomy and
pathology in anterior instability can be grouped into structures
and lesions between the 11 and 3 o'clock positions and those
between the 3 and 6 o'clock positions. The 11 to 3 o'clock interval
includes more structures (superior labrum, biceps-labral complex,
biceps tendon, coracohumeral ligament, and superior and middle
glenohumeral ligaments), but has less pathology (predominantly
superior labral anterior-posterior [SLAP] lesions). Unusual
findings involving this region on MRI are often related to anatomic
variants rather than pathology. Conversely, the 3 to 6 o'clock
position has fewer structures (labrum, subscapularis tendon, and
inferior glenohumeral ligament), but more pathology (Bankart, GLAD,
ALPSA, HAGL, and Perthes lesions).
The MRI findings of these lesions and anatomic variants will be
discussed following an introduction to the technical aspects of
shoulder MR imaging.
There is little agreement regarding a specific MR imaging
technique to evaluate glenohumeral instability. Techniques vary,
although with frequent overlap, secondary to individual
preferences, the type of MR imaging device available, and the
specific clinical question that has been posed. Various techniques
and their rationales will be discussed.
Patient positioning and surface coils
--Patients are generally imaged supine, going head-first into the
scanner. Their comfort level is optimized with padding and
sometimes pharmacologic sedation in order to reduce motion. The arm
is generally at the patient's side in a neutral to slightly
externally rotated position (thumb pointing up or slightly
lateral). Elbow support or arm board fixation can help to reduce
Thin slice (3 mm), small field of view (14-18 cm),
high-resolution images are desired. Surface coils are required to
ensure adequate signal-to-noise ratios to achieve these results. A
single loop coil is commonly used in an anterior coronal oblique
position. Alternatively, two flat circular coils can be placed both
anterior and posterior to the shoulder to be used as a Helmholtz
pair. Recently, further improvements in signal-to-noise ratios have
been achieved by more contemporary quadrature and phased-array
Throughout the exam, the coil is generally held in place with
the use of restraining straps. Very bright signal, known as coil
burnout, may be problematic in tissues immediately adjacent to the
coil. This artifact may be alleviated by placing a thin pad between
the coil and the skin.
--A set of scout coronal images is obtained with a large
field-of-view and a short repetition time (TR) spin-echo or
gradient-echo sequence. Utilizing an imaging sequence based on
physician preference, a series of axial images is then obtained
from the acromioclavicular joint through the inferior aspect of the
glenoid. This set of axial images subsequently serves as the
localizer for the oblique coronal and oblique sagittal images. The
oblique coronal images are prescribed parallel to the supraspinatus
tendon, whereas the oblique sagittals are prescribed perpendicular
to this tendon. Coronal coverage should extend from the
subscapularis muscle anteriorly through the infraspinatus-teres
minor complex posteriorly. Sagittal images should be obtained from
the coracoid process medially to the lateral cortex of the humerus.
--The precise imaging sequences employed vary from institution to
institution. Typical protocols include conventional and fast
spin-echo, as well as two- and three-dimensional gradient echo
sequences. Some protocols also utilize fat-suppression with fast
spin-echo T2-weighted sequences.
Conventional spin-echo images are widely used for shoulder MR
imaging as they are robust, widely available, and familiar to most
radiologists. T1-weighted images provide excellent anatomic
definition and are necessary to evaluate the marrow space and the
peribursal fat planes. T2-weighted sequences are necessary for the
detection of pathology.
Fast spin-echo T2-weighted se-quences are widely used as they
offer considerable time savings over conventional spin-echo. This
time savings may be utilized simply to shorten acquisition time or
it may be reinvested to increase spatial resolution (increased
matrix) or signal-to-noise ratio (increased number of signal
averages). Fast spin-echo images display similar contrast to those
of conventional spin-echo, except for having brighter fat signal.
Fat suppression is required in order to avoid obscuring small
pathologic lesions adjacent to fat. Satisfactory fat suppression
may be achieved with frequency selective techniques or with short
tau inversion recovery (STIR). Fast spin-echo sequences also result
in some loss of edge definition relative to conventional spin-echo.
Gradient echo images are widely used because of short
acquisition time and excellent visualization of the labrum.
These sequences allow improved delineation of the low-signal labrum
because of relatively increased signal intensity of the adjacent
articular cartilage and synovium secondary to their water content.
MR arthrography and provocative positioning of the
--To improve lesion detection when instability is questioned, many
institutions will perform MR arthrography. Additional modifications
in technique may also include imaging the shoulder in provocative
positions, such as the ABER (abduction external rotation) and
apprehension test (requires the lateral space of an open magnet)
The success of these techniques will be discussed later.
Briefly, MR arthrography can be performed by direct injection of
saline or gadolinium into the joint space, or by indirect
(intravenous) injection of gadolinium. A major advantage of direct
injection of gadolinium over saline is that it allows determination
of the origin of extra-articular fluid collections (i.e.,
inflammation versus abnormal communication with the joint)
encountered upon imaging.
Direct injection is preferred over indirect injection because it
provides the capsular distension necessary to optimally visualize
important structures, particularly the capsular ligaments.
Anterior instability lesions and the efficacy of
--The Bankart lesion is the most common injury sustained from
anterior shoulder dislocation. Classically, this lesion refers to
an avulsion of the labral-ligamentous complex from its attachment
to the anterior-inferior glenoid in association with disrupted
scapular periosteum. The lesion may occur with or without a
fracture of the bony glenoid.
The hallmark finding of a Bankart lesion on MRI is a band of
abnormally increased signal intensity separating the
anterior-inferior capsulolabral complex from the osseous glenoid
margin (figures 3 and 4).
In the acute setting, the lesion may be more conspicuous with the
additional findings of a joint effusion, bony fracture, and edema
within the bone marrow and surrounding soft tissues. The labral
tear is often large enough that it extends upward to involve the
middle and sometimes the superior anterior labrum. If the lesion
does not heal correctly following dislocation, the shoulder will
remain unstable. Partial healing, including fibrosis and
resynovialization, may make detection of the chronic Bankart lesion
difficult with conventional MRI.
Multiple studies have evaluated the abilities of conventional
MRI and MR arthrography to detect anterior labral tears in patients
with instability. Liu et al
evaluated conventional MRI, reporting a sensitivity of 59% and a
specificity of 85%. Palmer and Caslowitz
evaluated MR arthrography, reporting 92% for both sensitivity and
specificity. In a smaller series offering direct comparison of the
two techniques, MR arthrography compared favorably with
In this study, conventional MRI failed to detect six of nine labral
tears found at arthroscopy, while all nine lesions were
successfully identified with MR arthrography.
The additional use of the ABER position with MR arthrography has
been shown to improve labral tear detection when compared to MR
arthrography performed in the standard neutral position alone.
Cvitanic et al
demonstrated improvement in sensitivity (from 48% to 89%) and
specificity (from 91% to 95%) with the addition of ABER positioning
to their standard MR arthrographic protocal.
Bankart variant lesions
--In the Bankart lesion, the anterior scapular periosteum ruptures
as the anterior-inferior labroligamentous complex is avulsed from
the glenoid. This results in anterior displacement of the detached
labroligamentous structure. In the Bankart variant lesions, the
stripped scapular periosteum remains attached to the torn
labroligamentous complex (figure 4). In one such lesion termed
ALPSA (anterior labroligamentous periosteal sleeve avulsion)
(figures 4 and 5), the stripped periosteum rolls up like a long
shirt sleeve causing the detached labroligamentous structure to
displace medially and rotate inferiorly toward the scapular neck.
In the chronic state, the detached structure becomes fibrosed and
resynovialized in this abnormal position. It can look surprisingly
normal at arthroscopy and on MR imaging, although it is a cause of
The Perthes lesion, another Bankart variant, also has persistent
attachment of the stripped scapular periosteum to the torn
labroligamentous complex (figure 4). Unlike the ALPSA lesion, the
stripped periosteum does not become rolled up, generally allowing
the torn labroligamentous structure to reapproximate its normal
position. Like the ALPSA lesion, once resynovialized, the Perthes
lesion may appear deceptively normal at surgery and on MR
Superior labral anterior-posterior tears
--SLAP tears were first described by Snyder et al
in a series of 27 patients who presented with shoulder pain and/or
joint clicking. This series divided SLAP tears into four categories
based on arthroscopic findings: type I (11%), fraying of the free
edge of the superior labrum; type II (41%), avulsion of the
labral-bicipital complex from the superior glenoid; type III (33%),
bucket handle tear of the superior labrum; and type IV (15%),
bucket handle tear of the superior labrum with extension into the
At least three less common variations of SLAP tears have
subsequently been described in the literature.
These variations generally represent more extensive tears.
Two mechanisms of injury have been shown for SLAP lesions. The
first mechanism involves a compressive force from a fall on an
outstretched arm. The second mechanism involves traction on the
arm, either as a result of a sudden pull or secondary to repetitive
overhead use, such as in baseball pitchers, swimmers, and tennis
and volleyball players.
When isolated to the labrum, SLAP tears are not unstable in the
classical sense. However, they may mimic instability by presenting
with anterior shoulder pain, clicking, and occasionally mechanical
locking of the joint.
When SLAP lesions involve the biceps anchor, stress maneuvers on
physical exam may reveal instability. Surgical reattachment of the
superior labrum is indicated when it is associated with avulsion of
the biceps anchor. Isolated superior labral tears are treated with
surgical debridement alone.
SLAP lesions can be identified on MR images as abnormalities in
signal and morphology involving the labrum and possibly the biceps
anchor (figure 6). Tuite et al
reported a sensitivity of 65% and a specificity of 84% in the
detection of SLAP tears utilizing conventional (nonarthrographic)
MR imaging. With MR arthrography, Bencardino et al
observed 89% sensitivity and 91% specificity in the detection of
SLAP tears confirmed at arthroscopy.
--The acronym HAGL represents humeral avulsion of the glenohumeral
ligament. Like other causes of anterior instability, this uncommon
lesion occurs in the setting of anterior dislocation, although it
typically occurs in a patient whose first dislocation occurs after
the age of 40. Tirman et al
described the MRI findings of seven patients with HAGL lesions.
They found heterogeneity or frank disruption of the anterior
capsule at its humeral insertion (6/7 patients), tears of the
subscapularis tendon (6/7), dislocation of the biceps tendon (4/7),
and Hill-Sachs deformities (4/7). MR arthrography demonstrated
extravasation of contrast through the capsular defect in two of
The BHAGL lesion is a HAGL lesion in which a bone fragment has
avulsed from the humeral insertion of the inferior glenohumeral
--The GLAD (glenolabral articular disruption) lesion is not a cause
of instability; however, it is related to the anterior instability
lesions in terms of mechanism, clinical presentation, and findings
on MR imaging. At arthroscopy, the lesion is identified as a
superficial anterior inferior labral tear seen in association with
an anterior inferior glenoid articular cartilage injury. This
infrequent cause of anterior shoulder pain results from forced
adduction from the abducted and externally rotated position. The
treatment of choice is arthroscopic debridement.
This subtle lesion is best visualized by MR arthrography.
Normal variants and other imaging pitfalls
Knowledge of the multiple variations of the normal
labral-capsular-ligamentous complex is important to avoid the
misdiagnosis of pathologic abnormalities. One such variant is the
Buford complex which consists of a cord-like middle glenohumeral
ligament and an absent anterior-superior labrum (figure 7).
This entity may be mistaken for a pathologic lesion by both the
radiologist and the arthroscopist. Recognition of the Buford
complex on MRI can avoid the false-positive diagnosis of a glenoid
labral tear, possibly avoiding unnecessary surgery.
This variant was detected on arthroscopy in 1.5% of cases
and on MR imaging in 2% of cases.
Another variant of the anterosuperior labrum is a normal focal
labral detachment seen anterior to the bicipital anchor. This
variant, termed a sublabral hole or foramen, allows fluid to
accumulate between the detached labrum and the underlying glenoid
(figure 8). The sublabral foramen is seen in 7% to 12% of the
normal population and can be mistaken for an anterior labral tear.
Linear high signal, mimicking a tear, may be seen adjacent to
the labrum at the insertions of the glenohumeral ligaments. Care
should be taken to follow these ligaments to their insertions in
order to avoid this diagnostic pitfall.
Classically, a triangular shape is said to characterize the normal
morphology of the anterior-superior labrum; however, normal labral
morphology is variable. Young asymptomatic patients have been shown
to have cleaved (11%) or notched (3%) anterior labra, distinct from
the sites of ligament insertions.
It is uncertain whether these findings represent subclinical tears
or normal variants.
The magic angle phenomenon has been shown to cause areas of
increased T1 and proton density signal in the posterosuperior and
anteroinferior labrum. However, this is easily distinguished from a
true labral tear as the abnormal signal will not approach that of
fluid on the T2-weighted images.
A final potential pitfall is caused by undercutting of the labrum
by the hyperintense transitional zone between the fibrous labrum
and the hyaline cartilage covering the bony glenoid. This anatomy
can also mimic a labral tear.
Anterior shoulder instability is a complex problem encountered
frequently by both radiologists and their referring clinicians. MR
imaging of the shoulder plays an important role in surgical
planning and clarifies diagnostically difficult scenarios by
demonstrating both intra- and extra-articular anatomy and
pathology. Optimization of MR techniques, including the use of
arthrography, will increase the likelihood of accurate diagnosis.
Additionally, knowledge of ana-tomic variants and other imaging
pitfalls is essential to successfully discern pathology of the
The author thanks Clyde A. Helms, MD, for reviewing the
manuscript and providing both images and insight. The author is
also grateful for the technical assistance of Scott Faber. Special
thanks is extended to Theodore A. Dorsay, MD for his generosity in
providing MR images.