Due to the configuration of its bony elements, the glenohumeral joint is the most mobile joint of the body, but also an inherently unstable articulation. Stabilization of the joint is linked to a complex balance between static and dynamic soft tissue stabilizers. Because of complex biomechanics, and the existence of numerous classifications and acronyms to describe shoulder instability lesions, this remains a daunting topic for most radiologists. In this article we provide a brief review of the anatomy of the glenohumeral joint, as well as the classifications and the pathogenesis of shoulder instability. Technical aspects related to the available imaging techniques (including computed tomography [CT] arthrography, magnetic resonance imaging [MRI], and MR arthrography) are reviewed. We then describe the imaging findings related to shoulder instability, focusing on those elements that are important to the clinician. J. Magn. Reson. Imaging 2011;33:2–16. © 2010 Wiley-Liss, Inc.
SHOULDER INSTABILITY and the lesions it produces represent one of the main causes of shoulder discomfort and pain (1). Shoulder instability is defined as a symptomatic abnormal motion of the humeral head relative to the glenoid during active shoulder motion (2). The glenohumeral joint is an inherently unstable articulation, due in part to the gross disparity in surface area of the glenoid with respect to the humerus. While this results in a propensity for instability, it affords the advantage of mobility, the shoulder having the greatest range of motion of any joint in the body.
Stabilization of the shoulder is linked to a complex balance between static and dynamic soft tissue stabilizers. Because of complex biomechanics, and the existence of numerous classifications and acronyms to describe shoulder instability lesions, this remains a daunting topic for most radiologists (3).
Recent technological advances allow us to better visualize and characterize these lesions, as well as improve our understanding of their pathogenesis.
After a brief review of the anatomy of the glenohumeral joint, as well as the classifications and the pathogenesis of shoulder instability, we will focus on the available imaging techniques. We will then describe the imaging findings related to shoulder instability, focusing on those lesions that are important to the clinician.
NORMAL ANATOMY AND VARIANTS
As previously mentioned, the extreme range of motion of the glenohumeral joint is largely related to the disparity in surface area of the articular surfaces of the ball-shaped humeral head and the shallow glenoid cavity. Analogies for the joint have been drawn and described as a golf ball on a tee, or a cup on a saucer. Mobility does not come without a cost, the osseous configuration being a major contributor to the inherent instability of this articulation. Additional stabilizing elements are therefore needed and can be divided into static (corresponding to the capsulo-labro-ligamentous complex) and active stabilizers (including the rotator cuff and biceps tendons) (1, 2). The relative stabilizing effect of these elements varies according to the shoulder position, forming a complex biomechanical balance.
In the following paragraphs we briefly review the relevant anatomy of the static stabilizers of the shoulder. The anatomy of the biceps and the rotator cuff tendons are beyond the scope of this review.
STATIC STABILIZERS OF THE SHOULDER
Static stabilizers of the shoulder include the glenoid fossa, the coracoacromial roof as well as the capsulo-labro-ligamentous complex. The static stabilizers of the shoulder are activated in the extremes of motion (4).
The glenoid labrum is a fibrous and fibrocartilaginous structure that rims the glenoid. It increases the surface area and the depth of the glenoid cavity, hence increasing the congruity of the two articular surfaces of the glenohumeral joint. By sealing the joint space, the labrum also participates in the creation of a vacuum mechanism than helps maintain the humeral head position as it moves. It also serves as a point of attachment to other stabilizers of the joint such as the glenohumeral ligament, the biceps tendon, and the capsule.
Many variations of the labrum have been described, mostly located in its anterosuperior segment (between 1 and 3 o'clock) (cf. the paragraph on localizing the lesions) (Fig. 1). At this location the glenoid labrum can be normally detached (forming what is referred to as a sublabral foramen, present in about 10% of subjects) or absent. In both cases they can be associated with a cord-like middle glenohumeral ligament (5–7). The association of an absent anterosuperior labrum and a cord-like MGHL is referred to as a Buford complex, present in about 1.2%–6.5% of subjects (Fig. 2) (5, 8).
An association between these variations with an increased frequency of lesions of the anterosuperior segment of the glenoid rim and surrounding structures has been reported. However, these variations are not associated with shoulder instability per se and should not be considered as a cause of shoulder instability when isolated (5, 6).
Coracohumeral Ligament (CHL)
The CHL is a broad band extending from the lateral border of the coracoid process to the greater tubercle of the humerus, blending with the supraspinatus tendon. It reinforces the upper part of the capsule and stabilizes the intraarticular portion of the biceps tendon in conjunction with the superior glenohumeral ligament (SGHL) in the rotator cuff interval (the space between the anterior aspect of the supraspinatus to the superior aspect of the subscapularis tendon) (Fig. 3) (9).
Genohumeral Ligaments (GHL) and Joint Capsule
Glenohumeral ligaments represent thick infoldings of the glenohumeral capsule. They run from the glenoid rim to the region of the anatomical neck of the humerus. Their normal anatomy includes many variants that can be mistaken for pathologic lesions. The analysis of contiguous slices is very helpful in the identification of normal ligamentous structures, as well as 3D isotropic acquisitions that can be reformatted in the plane of the ligament (Fig. 3).
The SGHL extends from the superior aspect of the glenoid rim, near the base of the coracoid process, just anterior to the origin of the biceps brachii tendon, to a small depression above the lesser tubercle of the humerus. Normal variants of this ligament include a common origin with the biceps tendon and/or the middle glenohumeral ligament (MGHL) (10).
The MGHL can originate from the medial edge of the glenoid rim, at the anterior superior aspect of the labrum, from the SGHL, or from the biceps tendon. It distally inserts on the lower part of the lesser tubercle of the humerus, merging with the subscapularis tendon. It is the most variable of the glenohumeral ligaments and can be absent in up to 30% of individuals (11). It can also have a bifid appearance, which is sometimes difficult to differentiate from a posttraumatic split (Fig. 4).
Although there is no consensus in the literature on the contribution of each of the GHLs to joint stability (which also varies with the shoulder position), the IGHL seems to play a major role and is most frequently injured (10, 12). It consists of a complex formed by two bands (anterior and posterior) and the axillary recess. The anterior and posterior bands run from the glenoid labrum to the anatomical neck of the humerus. With the arm in the abduction external rotation (ABER) position, the anterior band of the IGHL becomes taut and is better depicted (10).
The anterior capsular insertion is variable and three types have been described based on the distance between the insertion and the glenoid labrum. These variants do not seem to have a clear relationship with shoulder instability. However, the posterior capsule always inserts on the glenoid rim and presence of fluid/contrast medial to the rim is indicative of capsular stripping and posterior instability (4).
PATHOGENESIS OF SHOULDER INSTABILITY
The static stabilizers of the shoulder act to restrain mobility of the joint in positions at the extreme of the range of motion, as they are put in tension. As an example, abduction and external rotation of the shoulder will put the anterior band of the IGHL and its insertion sites under tension. Most lesions of shoulder instability occur in this position. Therefore, injury to the static stabilizers is usually due to tensile forces resulting in avulsion. On the other hand, impaction mechanisms can occur due to contact between a dislocating humeral head and the glenolabral rim.
CLASSIFICATIONS OF SHOULDER INSTABILITY
Various classifications are used to describe glenohumeral instability.
Clinically, it can be classified by its causative factors. Shoulder instability can be traumatic or atraumatic. Differentiation between these types can have implications in patient management. Traumatic instability is typically unidirectional, unilateral, and usually requires surgical intervention (13). They are referred to as TUBS (Traumatic, Unilateral, Bankart, Surgery). Atraumatic instability (no history of trauma) can be related to microtrauma (referred to as AIOS for Acquired, Instability, Overstress, Surgery) or be part of a multidirectional instability syndrome (classified as AMBRI lesions (Atraumatic, Multidirectional, Bilateral, Rehabilitation, Inferior capsule shift). AIOS injuries mostly concern athletic patients with repetitive overhead movements. This pattern of injury is also called microinstability. AMBRI injuries correspond to congenital joint laxity, which is diagnosed clinically. The management usually consists of muscle strengthening exercises. If conservative treatment fails, the surgery then consists of an inferior capsular shift.
Shoulder instability can also be characterized by the direction of the movement: the most common are anterior and posterior instabilities, with inferior, superior, and multidirectional instabilities being far less commonly reported.
The degree of instability allows a distinction between dislocations and subluxations. A dislocation corresponds to a complete separation of articular surfaces, whereas a subluxation is a symptomatic translation of the humeral head without complete separation (1, 14).
Finally, shoulder instability can be acute (within 48 hours after the injury) or chronic.
In practice, when evaluating shoulder instability we try to give a description of the lesions including their localization and their acuity. The causative factor and the direction of the movement can be determined based on these elements, in association with the clinical history.
Indications for Imaging Techniques
Plain radiographs are performed to make the diagnosis of shoulder dislocation and to detect secondary bony lesions of the humeral or glenoid structures (Hill Sachs and Bankart, respectively) related to shoulder instability. In the acute setting, an anteroposterior view of the shoulder (Grashey view) is obtained, as well as a transcapular (scapular “Y” view) view. The Garth view (obtained by orienting the x-ray beam 45° caudally from a standard AP view) is very useful in the acute setting. It nicely demonstrates the anteroinferior margins of the glenoid as well as the posterosuperior aspect of the humeral head (location of Bankart and Hill Sachs lesions, respectively), but does not require an abduction of the arm as with the axillary view. Whenever the mobility of the patient allows, five views may be obtained: three PA views (Grashey view, internal rotation, external rotation), as well as a transcapular and axillary view (3, 15). In our practice, in specific cases (ie, when the clinical examinations is difficult), we sometimes add dynamic stress radiographs to these views to objectively document the (often multidirectional) instability of the shoulder.
CT Arthrography vs. MR Arthrography
MR arthrography is considered to be the method of reference for the study of the labrum and glenohumeral ligaments. In a study by Chandnani et al,(16) MR arthrography was shown to be more sensitive than CT arthrography for the detection of labral tears and labral degeneration (96 vs. 52% and 56 vs. 24%, respectively). However, in more recent studies, CT arthrography showed to be a valuable method for the study of the labrum (17). In a recent study comparing CT arthrography performed on a 16-channel scanner to 1.5 T MR arthrography, Oh et al (18) showed similar performance of CT arthrography and MR arthrography for labral lesions (SLAP, Bankart and Hill-Sachs lesions): sensitivity of 86 vs. 72, 86 vs. 90, and 93 vs. 75, respectively; specificity of 90 vs. 95, 95 vs. 100, 90 vs. 98, respectively. However, that study was limited by the fact that the CT arthrography and MR arthrography were performed on two different groups of patients, hindering any direct comparisons, and by the absence of interobserver and intraobserver reproducibility.
CT arthrography may be a good alternative to MRI whenever the latter cannot be performed (either because of its availability or because of contraindications, claustrophobia, or the presence of metallic prosthesis, which is a frequent occurrence in cases of recurrent instability in athletes) (3).
MRI vs. MR Arthrography
MR arthrography is the method of choice for the diagnosis of shoulder instability and labral lesions (16, 19, 20). The intraarticular injection of contrast material distends the joint, enters labral tears, and outlines tears and detachments. However, in the setting of an acute trauma, conventional MRI without intraarticular injection is usually performed, as the joint effusion plays the role of the contrast agent and distends the capsule (3).
The advent of 3T scanners with high-resolution MRI has raised the question of the added value of MR arthrography compared to conventional MRI. In a recent study on 150 consecutive shoulder examinations from patients 50 years or younger, both 3T MRI and MR arthrography were performed and correlated with arthroscopy (21). MR arthrography was shown to have a statistically increased sensitivity for the detection of anterior labral tears (98% vs. 83%) and SLAP tears (98% vs. 83%), altering management in 15 cases of labral pathology. There was no significant difference in the sensitivity in detection of posterior labral tears, nor in the specificity for any of the lesions.
Type of Contrast Material
Dilution of the contrast product can be performed with local anesthetics or saline to avoid beam-hardening artifacts. However, the dilution mainly depends on the radiologist preferences (24–29). In our practice, we use 10 mL of ionic contrast material (320 mg of iodine per milliliter), diluted with 5 mL of local anesthetic.
The contrast material of choice for MR arthrography is gadolinium-based (gadolinium-DTPA), providing a better contrast-to-noise ratio than other types of contrast media such as saline combined with T2-weighted images (30–32). It is possible to perform either indirect (less invasive intravenous gadolinium-DTPA injection) or direct MR arthrography (intraarticular gadolinium injection). In our institution we classically favor the intraarticular injection of the contrast material for the study of shoulder instability because it allows joint distension and a better delineation of labral lesions. With the increasing utilization of 3T MRI, the question of the performance of indirect arthrography, considered to be less invasive than direct arthrography, has been revisited. In a recent study, Jung et al (33) found similar performance of direct and indirect MR arthrography at 3T for the diagnosis of labral lesions (for anterior labral tears the sensitivity was 100% for both techniques, the specificity was 100% for direct MR arthrography, and ranged from 83%–100% for indirect MR arthrography). However, this study was performed on a small number of patients (19 with arthroscopic confirmations).
Many studies have focused on determining optimal gadolinium-DTPA concentrations and studied the temporal behavior of intraarticular contrast after injection. At 1.5T a concentration of 2 to 2.5 mM is considered best for imaging to be performed within about an hour after injection (34, 35). At 3T a slightly greater dilution may be useful (36). Aspiration of joint effusion before injection can prevent excessive dilution of contrast material but this is usually not a problem in clinical practice (36, 37).
It has been shown that iodinated and gadolinium-based contrast material can safely be mixed, and combined MR arthrography and CT arthrography examinations have successfully been obtained for comparison of both studies (38–41). However, at 3T the presence of iodinated contrast agents has to be minimized because signal-to-noise ratio (SNR) peak levels for iodinated contrast dilutions are lower at 3T than at 1.5T (36).
Volume of Contrast Material
The volume of injected contrast product necessary for proper capsular distension is the same for all arthrographic techniques. About 12 cc of contrast material generally provides adequate distension as indicated by increased resistance to injection or retrograde flow of contrast product into the needle after disconnection of the syringe (42). The injection should be stopped in case of pain.
The injection is usually performed under fluoroscopic guidance but many other injection techniques have been described, using CT, ultrasound, MR guidance or even by using surface landmarks (26, 43–51). The choice relies on the radiologist's preference and on the equipment available at one's institution. The injection technique follows standard arthrographic procedures, which have been widely described in the literature (52–54). The puncture approach is ideally adapted with the patient symptoms, avoiding the site of instability (posterior approach in case of anterior instability and vice versa) (55).
Time Delay Between the Injection of Contrast Media and Imaging
Once injected in the joint the concentration of contrast products rapidly decreases by diffusion into the cartilage and synovium, resorption and fluid influx into the joint (56). It is recommended to perform the CT acquisition within 30 minutes and the MR within an hour after the contrast injection (35, 39, 57, 58). The addition of epinephrine to the injected contrast material (for instance, by mixing 1 mL of a 0.1% solution containing 1 mg of epinephrine with 10 mL of contrast product) has been used to slow the resorption of the intraarticular contrast (59, 60). However, the use of epinephrine may increase postarthrographic pain (23). Use of epinephrine is usually not necessary with MR arthrography (23, 61).
It has been shown for the shoulder that exercise has no beneficial or detrimental effect for MR arthrography (62). However, gentle active and passive full range articular motion after the injection allows the contrast material to completely cover cartilage surfaces, while the increased risk of capsular effraction is usually not a problem in our experience (63).
The CT acquisition parameters include narrow collimation, low pitch values, and a high milliampere-second value to obtain high resolution isotropic multiplanar reformats (MPR) (64). The reconstructions use bone algorithms providing high spatial resolution images and bone windowing are used to view the images. Posttreatment of these high-resolution isotropic images may include curved and maximum intensity projection (MIP) reformatting. One of the indications to perform CT arthrography rather than MRI is the presence of metallic artifacts in the setting of postoperative patients. Metallic artifacts usually remain mild on new generation CT scanners compared to MRI (22, 65). They can be diminished and, more generally, SNR can be increased by retrospectively increasing the thickness of the reformats and using soft tissue algorithms; however, at the expense of spatial resolution (66, 67).
There is no consensus as to which MRI sequences to perform. MR arthrography classically includes spin echo T1-weighted sequences with frequency-selective fat suppression in three planes (axial, coronal-oblique, sagittal-oblique), including at least one fluid sensitive sequence to allow detection of bone marrow edema and extraarticular fluid collections (68). We also perform one non fat-suppressed coronal-oblique T1 sequence to assess fatty infiltration of muscles and marrow content.
More recently, 3D gradient echo sequences have been compared to conventional 2D MR arthrography for the diagnosis of labral lesions of the shoulder (69). A single submillimeter isotropic acquisition allowing multiplanar reformatting showed the same diagnostic performance as the conventional 2D sequences, in a shorter acquisition time (5 min 32 sec compared to 16 min 40 sec). However, the power of this study was limited and further studies are needed to validate this method.
The shoulder is usually positioned in neutral position. However, positioning the shoulder in such a way that the shoulder stabilizers are put into tension in the same fashion as at the time of injury can help to depict subtle lesions. Many groups acquire a T1-weighted fat suppressed sequence with the shoulder in ABER position (Fig. 5) (70). By loading in tension the anterior stabilizers, the ABER position can reveal subtle lesions at the anteroinferior labrum, as well as articular-sided supraspinatus and infraspinatus tears. Others include internal/external rotation positions of the shoulder in their protocol to better depict posterior/anterior labral lesions, respectively (3) (Fig. 6).
As with any other arthrographic procedures, CT arthrography and MR arthrography present risks linked to the injection (mainly infectious risk) and to the injected contrast material (allergic reactions) (71). However, the risk of infection is quite low (with 1 infection out of 25,000 arthrograms according to Berquist (72) and 3 cases of iatrogenic septic arthritis out of 126,000 arthrographic procedures according to Newberg et al (71)). The risk of severe systemic allergic reactions is also low, although minor reactions can occur (72). A meta-analysis of 112 published studies found gadolinium-DTPA to be a safe and efficient technique for diagnosing internal derangement of joints (34).
To our knowledge, there has been no report of nephrogenic systemic fibrosis after the intraarticular injection of gadolinium-based contrast product.
Moreover, as with other arthrographic techniques, there is a risk for vasovagal reactions and postarthrographic pain. The best prevention for vasovagal reactions is good communication with the patient and preparation of the injection material out of the patient's sight (73). In a recent study evaluating pain and other side effects of MR arthrography, Saupe et al (74) concluded that mild postarthrographic pain is most pronounced 4 hours after the procedure, and disappears within 1 week. This chronology helps to differentiate it from infectious complications in which pain appears later, typically after a few days. The origin of postarthrographic pain is debated in the literature (23, 24, 75–77). However, in general, the arthrographic procedure is generally well tolerated by patients (75).
In addition to those risks, there is patient radiation exposure with CT arthrography and not with MR arthrography, especially due to the position of the thyroid glands. Care will be taken to keep the radiation dose to the minimum, at the expense of SNR.
The role of the radiologist in shoulder instability is to determine which of the stabilizing elements are compromised, to localize the lesions, and to determine the acuity of the lesions. Associated lesions also have to be carefully reported.
Localizing the Lesions (Fig. 1)
The lesions are localized by dividing the glenoid rim into segments. This can be done by using the clock face analogy (12 o'clock being at the top of the glenoid, near the coracoid process, 6 o'clock the bottom, 3 o'clock corresponding to the anterior aspect of the glenoid, and 9 o'clock to the posterior aspect, no matter the shoulder side). It can also be divided into six regions as shown in Fig. 1. When analyzing the glenoid labrum, it is helpful to keep in mind that most normal anatomical variants occur in the anterosuperior segment, and that isolated lesions of the labrum are uncommon in this segment. On the other hand, if a labral abnormality (absence, morphological abnormality, or signal intensity abnormality) is located in the anteroinferior or the posterior segments, it is likely to be pathological.
Injury to the Stabilizing Elements (Figs. 7, 8)
The main goal of imaging in the evaluation of shoulder instability is to identify and characterize bony or soft tissue lesions. Since capsular and bony anatomy share common features around the glenohumeral joint, the patterns of injury have similar elements but variable location depending on the direction of humeral head displacement. Anterior and posterior instability, which represent 99% of the shoulder instabilities, have parallel bony and soft tissue patterns of injury (20, 78).
The impact of the humeral head with the glenoid rim during shoulder dislocation may give rise to bone contusions and fractures. Bony defects in the humeral head or the glenoid have a complementary role in shoulder instability, and must be evaluated and treated together with other lesions (79).
In the event of a shoulder dislocation the impact frequently generates a bony defect of the humeral head (in up to 71% of acute anterior dislocations, and 86% of acute posterior dislocations) (80–82). These lesions are located either at the posterosuperior margin of the humeral head in case of anterior dislocation (referred to as Hill-Sachs lesions) or at the anterior margin of the humeral head (reverse Hill-Sachs lesions, also called trough or McLaughlin's fractures) (Fig. 9). Hill-Sachs lesions have to be differentiated from a normal humeral groove. The latter is located distal to the typical impaction site. Hill-Sachs lesions are usually found at the most cephalic part of the humeral head, above the level of the coracoid process (83). Other differential diagnoses of Hill-Sachs lesions at MRI include humeral head cysts and reactive marrow changes. These findings can be related to rotator cuff pathology, internal posterosuperior impingement, or be incidental in nature (20, 84). Large Hill-Sachs lesions may create an abnormal area of glenohumeral contact when the shoulder is in the same position in which the dislocation occurred. This so-called engaging lesion was described with anterior instability and can mimic the symptoms of instability as patients present with a catching and popping sensation associated with pain (85). Different methods and classifications exist to quantify the Hill-Sachs lesion and assess the probability of engagement; however, none of them is uniformly accepted (79, 85, 86).
On the glenoid side, bony defects created by the shoulder dislocation are called Bankart (located anteroinferiorly) or reverse Bankart (located posteroinferiorly) lesions (Fig. 10). These can represent compressive fractures generated by the shoulder dislocation or avulsive injuries at the attachment site of the labroligamentous complex. Compressive fractures have a higher association with recurrent instability (87). Quantification of these lesions is clinically important. A bone loss of more than 25% at arthroscopy (which typically gives the glenoid an inverted pear shape) requires open bone grafting to prevent recurrence of shoulder instability or dislocation (85, 88). Patients without significant glenoid bone loss can benefit from an arthroscopic Bankart repair with a recurrence rate under 5%. 3D CT has been used to calculate the glenoid index (ratio of the maximum inferior diameter of the injured glenoid compared to the maximum inferior diameter of the uninjured contralateral glenoid). A threshold of 0.75 was showed to correlate with the arthroscopically derived glenoid index and could be used to preoperatively determine which patients would benefit from a bone-grafting procedure (89, 90). However, this technique requires irradiation of both shoulders. In practice, the bone loss is roughly estimated on the axial and sagittal views, preferably with CT (Fig. 10). More measurements may be made whenever there is a specific request from a surgeon.
Along with Hill-Sachs and Bankart lesions, greater tuberosity and superior humeral head fractures can also accompany shoulder instability and have to be recognized.
Congenital bone alterations may also contribute to shoulder instability. Glenoid dysplasia comprises a spectrum of abnormalities (hypoplasia of the glenoid neck of the scapula and glenoid rim deficiency, seen in 18%–25% of individuals) in which various degrees of bone deficiency in the posterior glenoid are associated with cartilage/labral hypertrophy (4). These abnormalities are better depicted with CT rather than MRI, showing a blunting and convexity of the posterior glenoid rim in the axial plane (the lazy “J” sign) (78, 91) (Fig. 11). Care should be taken not to confuse this pattern with the normal rounding of the glenoid rim present in every patient in the most distal axial images. Patients with this condition have a tendency for posterior instability with a high incidence of posterior labral tears (91). Glenoid retroversion may also play a role in posterior shoulder instability; however, there is no agreement as to the importance of this finding (92).
Injuries to the Capsulo-Labro-Ligamentous Complex
The labroligamentous structures can be compromised at numerous sites. An accurate diagnosis and localization of these lesions is clinically important because this information may have implications for the surgical management (93).
Lesions of the Capsulo-Labro-Ligamentous Complex at Its Glenoid Attachment (Fig. 7).
The most frequent site of soft tissue compromise in shoulder instability is at the glenoid attachment (94). Many different patterns of injury can occur at this site, and with recurrent instability these lesions may progress or become chronic (95).
A detachment of the labroligamentous complex from the glenoid with a disrupted periosteum is termed a soft tissue Bankart lesion (Fig. 12). This lesion was originally described at the anteroinferior aspect of the glenoid rim. The posterior instability counterpart to this lesion is a reverse soft tissue Bankart affecting the posteroinferior margin of the glenoid (95, 96). Contrast or joint fluid is seen between the detached labroligamentous complex and the glenoid, allowing the diagnosis by MR in 80% of the cases (93). Because the labroligamentous complex is completely separated from the glenoid in the setting of a Bankart lesion, it may migrate from its normal site of attachment, usually to a superior position. With time, this displaced tissue may scar into a round shape, not to be confused with intraarticular bodies, a pattern known as glenoid labrum ovoid mass (GLOM) (14).
Different variants to the Bankart lesion have been described in which its periosteal attachment remains continuous. A displaced labral detachment with a stripped periosteum that maintains an attachment, albeit abnormal, is termed labroligamentous periosteal sleeve avulsion (LPSA). This lesion can occur anteriorly or posteriorly, corresponding to the acronyms ALPSA and POLPSA, respectively (Figs. 12, 13) (97, 98). At MRI, contrast material or intraarticular fluid can be traced between the detached labrum and the glenoid. Since the periosteum is still intact, the labral displacement can be small, making the diagnosis sometimes challenging (14). In the chronic setting, fibrous scar tissue forms around the labral detachment appearing as an amorphous hypointense mass that migrates inferomedially to the region of the glenoid neck (Fig. 12). Superimposed calcifications can sometimes be found. Posteriorly, periosteal stripping may be one of the possible explanations for the origin of the Bennett lesion (97) (Fig. 14). This chronic posterior capsular avulsion is classically described in athletes due to stress to the IGHL in the deceleration phase of the throwing motion and is usually asymptomatic (99). Crescent-shaped calcifications or ossifications in this location can be seen with plain films. They are better depicted with CT compared to MRI (78, 100).
The other variant of the Bankart lesion consists of a pattern of minimally displaced labral detachment with continuous periosteum (which can be stripped). In the anterior labrum this pattern is called a Perthes lesion after the German physician who described it (101). Recently, a posterior counterpart of this lesion was described in the arthroscopy literature, called the Kim lesion (102). The Perthes lesion is difficult to diagnose with MRI, having a sensitivity of only 50% for its identification (93). This lesion differs from the acute ALPSA in that it is occult in the neutral adducted position and is unmasked when the ligament is placed under tension through the ABER position (Fig. 5) (20, 93). Loss of labral height is another sign that can be associated with Kim lesion (78, 102).
A periosteal stripping at the attachment of the anterior band of the GHL without labral detachment has been described at the anterior labroligamentous complex (anterior ligamentous periosteal sleeve avulsion [ALIPSA]) (14). No reports of this pattern of lesion in other regions of the glenoid were found. It is still debated whether this pattern represents a real lesion or only a rare variation at the insertion of the anterior band of the IGHL.
Pure Capsular Tears (Fig. 8).
Pure capsular injuries are the next most common abnormality seen in anterior shoulder instability. These injuries are particularly common in the anterior pouch. All types of capsular failure are related to capsular stretching and laxity (94). Imaging evaluation of these findings is usually difficult. In the acute setting MRI depicts discontinuity and irregularity of the capsular tissues, with the main diagnostic clue being an extravasation of joint fluid or contrast media into the periarticular region. In the chronic stage capsular thickening may be the only imaging finding (103, 104). MR arthrography may be able to accurately depict and objectively quantify capsular laxity (105).
Lesions of the Capsulo-Labro-Ligamentous Complex at Its Humeral Attachment (Fig. 8).
Lesions of the anterior band of the inferior GHL are termed humeral avulsion of the glenohumeral ligament or HAGL lesions (Fig. 15). If the posterior band of the inferior GHL is affected the words posterior or reverse are added forming the acronyms PHAGL or RHAGL (106). These injuries are seen more frequently in athletes, with a marked male predominance, and tend to occur in patients older than 30 years (10). They can be purely ligamentous or be associated with a humeral bone avulsion (BHAGL lesion), in which case the lesion can be suggested in the plain films or CT (107). The MR diagnosis is made by identifying changes in the configuration of the axillary pouch with extravasation of joint fluid or contrast media (which represents the main diagnostic clue). In the normal shoulder the axillary pouch is “U”-shaped in the oblique coronal plane, whereas in the setting of an HAGL lesion it assumes a “J”-shaped configuration (20). The accuracy of MR arthrography in the detection of these lesions is not yet determined, and false positives may occur when there is contrast/fluid extravasation from other sources such as other capsular injuries, soft tissue trauma, or even iatrogenic (104). However, it is important to recognize these lesions because they can be missed at arthroscopy and change the surgical approach (open repair vs. arthroscopic) (4).
Intrasubstance Labral Tears
Intrasubstance labral tears are associated with shoulder instability (Fig. 11) (20, 78, 81, 103, 109). Differentiating this type of injury from labral detachments may imply changes in the surgical technique (109). However, the imaging differentiation between these lesions can be difficult. Detection of these injuries relies on altered morphology or signal intensity of the labrum, and should be confirmed in at least two different imaging planes (103). These lesions should be assessed for their size, extension around the glenoid rim, presence of displaced labral tissue (flaps, bucket handle), extension to tendinoligamentous structures (glenohumeral ligament, biceps anchor), and for the presence of peri-labral cysts (110). However, it has to be kept in mind that the normal MR and CT arthrography appearance of the labrum is highly variable. In a study by Zanetti et al (111, 112), abnormal MRI signal and morphology was seen in half of arthroscopically normal labra, making the diagnosis of subtle lesions quite difficult to interpret.
Finally, it is important to acknowledge that about 20% of the anterior labroligamentous complex lesions cannot be classified either by MRI or by arthroscopy (93).
Assessing the Acuity of the Lesions
The age of the injury is also important to assess because it can influence the treatment outcome (113). Fluid collections around the joint, bone contusions and soft tissue edema are the hallmarks of an acute injury. On the other hand, thickening and distortion of the labroligamentous complex, including calcifications, and signs of degenerative joint disease suggest chronicity (Fig. 9) (93). Perilabral cysts, which have been shown to be associated with glenoid labral tears and shoulder instability, are also a sign of a more long-standing process (Fig. 6) (70). The differentiation between acute and chronic lesions can at times be extremely challenging, particularly in the setting of recurrent instability when acute and chronic injuries are superimposed.
Reporting Associated Findings of Shoulder Instability
Albeit rare, a nondisplaced labral injury may be accompanied by a glenoid chondral lesion (glenolabral articular disruption [GLAD]) in the setting of shoulder instability. These patients usually complain of persistent shoulder pain after injury in the ABER position, not explained by rotator cuff pathology. These symptoms are related to cartilage delamination, fissuring, or erosion, which can be demonstrated by MR or CT arthrography, but there is classically no association with shoulder instability (114). These lesions have recently been described with posterior shoulder instability (Fig. 10) (78). Traumatic focal cartilage lesions have also been described in the humeral head, associated or not with shoulder instability (115). In this context it is important to look for intraarticular bodies and mention them in the report. In case of chronic instability, degenerative joint disease can develop, and radiological signs of osteoarthritis have to be reported to the clinician (Fig. 6).13
There is a positive association between shoulder instability and rotator cuff tears, especially in patients over 40 years old (Fig. 6). Whether the cuff tears are directly secondary to trauma or related to the acquired instability is not known. However, the diagnosis of these tears has obvious clinical implications (116). Traumatic myotendinous injuries are also related to shoulder instability. Subscapularis strains can occur with anterior or posterior dislocations (20, 117). Teres minor injuries also may occur, probably due to its intimate association with the posterior capsule (118). Additionally, isolated high signal intensity and fluid collection in the area of the infraspinatus myotendinous junction and the posterior aspect of the scapula has been recently described in association with trauma and may progress to fatty atrophy in this muscle (119). Posterior entrapment of the long head of the biceps tendon can be associated with anterior shoulder dislocation. This injury usually occurs in association of rotator cuff tears and may prevent complete reduction of the humeral head (120).
The role of the radiologist in assessing shoulder instability is to provide the surgeon with a description of all the lesions of shoulder stabilizers, as well as their location, extent, and age. The status of the periosteal attachment of the capsulo-labro-ligamentous complex (whether it is disrupted or not) is important to report, as it can change the management. Detailed knowledge of the anatomy of the structures involved in shoulder stabilization is therefore fundamental, more than the names and acronyms associated with these lesions. Using an appropriate technique is also important, and the injection of intraarticular contrast material, along with appropriate positioning of the shoulder (including the ABER position) are helpful in practice both for the depiction and characterization of these lesions.