Josef K. Eichinger, MD, FAOA and Joseph W. Galvin, DO, FAAOS
Posterior shoulder instability is becoming increasingly recognized in young, athletic populations, especially in the military.1–3 Compared to anterior shoulder instability, posterior instability can be more challenging to diagnose both clinically and radiographically. Patients often do not experience frank posterior dislocation events such as that with anterior shoulder instability and more commonly develop attritional lesions. As a result posterior shoulder instability may present with vague shoulder pain, and a clinical examination is less demonstrative than anterior shoulder instability and may therefore be more difficult to diagnose. Imaging studies therefore are an important adjunct to the diagnosis and treatment of posterior shoulder instability. However, imaging studies do not always demonstrate obvious pathologic findings and thus a nuanced approach to the interpretation of x-rays, computed tomography (CT), and magnetic resonance imaging (MRI) is necessary to elucidate and identify subtle findings that can enable the clinician to make the correct diagnosis. In this chapter we will review imaging findings of posterior instability on standard radiographs, CT scan, MRI, and magnetic resonance arthrogram (MRA), and 3-dimensional (3D) reconstruction CT and 3D MRI, which assist in the diagnosis and treatment of symptomatic posterior shoulder instability.
Plain radiographs in patients with posterior shoulder instability are an important and critical adjunct to making the diagnosis of posterior shoulder instability. Biplanar radiographs should always be obtained when evaluating patients with suspected shoulder instability. An anteroposterior (AP) Grashey image (also known as a “true AP” view because the beam is oriented perpendicular to the scapula, which is oriented 30 degrees anterior to the coronal plane) (Figure 17-1) along with an axillary x-ray (Figure 17-2), are the minimum radiographs that should be obtained. Although x-ray findings are typically normal, they must be scrutinized to avoid errors of diagnosis such as missed posterior dislocations. Diagnosis of a locked posterior humeral dislocation can be avoided by recognizing on the AP Grashey radiograph the presence of the “lightbulb sign” (Figure 17-3A), which is the humeral head taking on a rounded appearance similar to the shape of a lightbulb because of fixed internal rotation secondary to a posterior glenohumeral dislocation.4 In addition to recognizing the lightbulb sign on an AP Grashey radiograph, an axillary x-ray will confirm the diagnosis of a locked posterior dislocation (Figure 17-3B). In addition to aiding in the recognition of a locked posterior dislocation, the axillary radiograph is necessary to a complete an orthogonal radiographic analysis. If the patient is unable to abduct the arm, then a Velpeau view is an alternate orthogonal radiograph (Figure 17-4). The axillary radiograph is also helpful in the traumatic scenario for identifying a posterior glenoid rim fracture or a reverse Hill-Sachs lesion. X-rays also demonstrate evidence of glenoid dysplasia (increased retroversion and hypoplasia), arthritic changes, and posterior humeral head subluxation or decentering of the humeral head. Other radiographic lesions that may be associated with posterior labral pathology and instability include the Bennett lesion, which is an extra-articular posterior ossification of the posterior inferior glenoid. Bennett lesions are more commonly found in overhead athletes, typically baseball players, and can be visualized on axillary radiographs.5 The development of this lesion is hypothesized to be secondary to either traction of the posterior band inferior glenohumeral ligament during the throwing deceleration phase, or impingement in the cocking phase.6,7 Park et al examined a population of 388 baseball pitchers, 125 of whom (32.2%) had Bennett lesions. When comparing the 2 groups, they found that 12% of patients in the Bennett group had a posterior labral tear on MRI, whereas only 6.8% of patients in the non-Bennett group had a documented posterior labral tear, although the results were not statistically significant.8 Therefore, although Bennett lesions are typically not associated with posterior shoulder instability, it is important to recognize these lesions because they can be associated with posterior labral tears.
Additionally, a recent study by Meyer et al9 highlighted the importance of x-rays in evaluation of posterior shoulder instability. The authors found that specific acromial morphology on scapular-Y x-rays is significantly associated with the direction of glenohumeral instability. In shoulders with posterior instability, the acromion is situated higher and is oriented more horizontally in the sagittal plane than in normal shoulders and those with anterior instability. Future larger studies are needed to confirm these findings.
COMPUTED TOMOGRAPHY SCAN
Following plain radiographs, a CT scan is another useful imaging modality to evaluate the bony morphology of the glenoid including retroversion, glenoid dysplasia, and glenoid bone loss (GBL), and to further characterize the size and location of a reverse Hill-Sachs lesion. A CT scan is typically performed to evaluate posterior bone loss due to either a reverse bony Bankart lesion or attritional bone loss, and to assess degree of retroversion and glenoid dysplasia, and is performed in revision scenarios. Also, it allows preoperative planning if a posterior bone block procedure is planned. Glenoid retroversion has been shown to be a risk factor for posterior shoulder instability.3 In a prospective study of 714 West Point cadets who were followed for 4 years, 46 shoulders had a documented glenohumeral instability event, 7 of which (10%) were posterior instability. Glenoid retroversion was significantly associated with the development of posterior shoulder instability (P < .001). Similarly, Bradley and colleagues found that in a cohort of 100 shoulders that underwent arthroscopic capsulolabral repair, patients with posterior instability had significantly greater chondrolabral injury and osseous retroversion in comparison with controls.10 The measurement of glenoid retroversion on 2-dimensional CT scan is performed by using Friedman’s method, which has been validated and accepted (Figure 17-5).11 It is generally accepted that normal glenoid version is between 4 to 7 degrees of retroversion. Although increased glenoid retroversion is a risk factor for posterior shoulder instability, there is little evidence to support the claim that increasing glenoid retroversion is associated with worse outcomes following posterior labral repair.12 Hurley et al found that patients with symptomatic posterior instability and glenoid retroversion of greater than 9 degrees had higher recurrence rates after open soft-tissue procedures.13 Conversely, Bigliani and colleagues performed CT scans for 16 of 35 shoulders prior to an open posterior capsular shift and found the average retroversion was –6 degrees.14 Their surgical cohort had an 80% success rate but they did not attribute their failures to osseous anatomy. Mauro et al found increased retroversion in a cohort of 118 patients who were operatively treated for posterior instability in comparison with a group of normal controls, but the authors did not attribute retroversion as a risk factor for failure. They did find that smaller glenoid width was a risk factor for failure.12
Glenoid Dysplasia and Hypoplasia
Severe glenoid dysplasia or hypoplasia is a rare condition due to either brachial plexus birth palsy or a developmental abnormality with lack of stimulation of the inferior glenoid ossification center. These terms are interchangeable because there is underdevelopment of the posterior inferior aspect of the glenoid. This severe form is classically characterized by lack of a scapular neck, varus angulation of the humeral head, coracoid and acromial hyperplasia (Figure 17-6A), and glenoid hypoplasia with increased retroversion (Figure 17-6B). With increased advancements in CT and MRI, more subtle forms of glenoid dysplasia have been recognized. Edelson was the first to define the incidence of subtle forms of glenoid dysplasia by studying scapular specimens from several museum collections.15 Posteroinferior hypoplasia was defined as a “dropping away” of the normally flat plateau of the posterior part of the glenoid beginning 1.2 cm caudad to the scapular spine (Figure 17-7). Glenoid dysplasia/hypoplasia occurred in 19% to 35% of specimens.15,16 Additionally, several studies have identified that subtle posteroinferior glenoid deficiency and hypoplasia are significantly associated with posterior labral tears and symptomatic posterior shoulder instability.17–19 Weishaupt et al18 used CT arthrograms to determine the incidence and severity of glenoid dysplasia in a population of patients with atraumatic posterior shoulder instability. They developed a classification system in which a “pointed” glenoid on axial imaging sequences is a normal-appearing glenoid without dysplasia, a “lazy J” has a rounded appearance of the posterior inferior glenoid, and a “delta” glenoid is a triangular osseous deficiency. These are depicted in Figure 17-7. Harper and colleagues17 similarly developed a classification scheme with normal, mild, moderate, and severe glenoid dysplasia. Also, although better visualized on MRA imaging, a hypertrophied posterior glenoid labrum is evident in patients with glenoid dysplasia (Figure 17-8). Despite multiple studies documenting a clear significant association between subtle glenoid dysplasia and posterior labral tears with associated posterior shoulder instability, there is little evidence demonstrating an association with worse outcomes following surgical intervention. Galvin et al performed a retrospective comparative outcomes analysis of 37 patients, mean age 28 years, who underwent arthroscopic posterior labral repair for symptomatic posterior shoulder instability with a mean follow-up of 3.1 years. Comparison between 18 patients with glenoid dysplasia and 19 patients without dysplasia revealed no significant difference in outcomes between the 2 groups.20