Magnetic resonance arthrography in patients with multidirectional instability: could inferior capsulsar width be considered the cornerstone in the diagnosis of non-traumatic shoulder instability?

Shoulder macro-instability includes different clinical entities [1]. These are classified into two groups following the etiopathogenetic features and possible therapeutic options:

Multidirectional shoulder instability (MDI) is characterized by generalized instability at least in two planes of motion (anterior, posterior, or inferior) due to capsular redundancy. The features of MDI were first described by Neer and Foster in 1980 [3, 4]. Diagnosis is made clinically and strongly depends on the patient’s history: patients may present with a sulcus sign (two or more axes), positive apprehension, load and shift, and hyperabduction tests. Signs of generalized hypermobility may also be present including elbow or metacarpophalangeal joint hyperextension, genu recurvatum, patellar instability, and the ability to rest the thumb on the ipsilateral forearm, as assessed by Beighton’s criteria: if > 4/9 patient is considered hyperlax [5].

Unlike patients with traumatic shoulder instability, patients with MDI are more likely to experience episodes of recurrent dislocation [6].

Imaging may be useful in the diagnosis of MDI; in particular magnetic resonance arthrography (MR-a) may demonstrate an increased capsular volume defined by the glenocapsular ratio [1, 7], although these measures are difficult to reproduce [8].

Previous studies have demonstrated that rotator interval and axillary recess width correlate with MDI while anterior or posterior capsular redundancy shows no correlation [1, 7]. Furthermore, glenoid bone loss and version are known to be instability factors in the development of shoulder instability [9] and discrepancy in size between the small glenoid fossa and the humeral head also plays an important role [10].

The objective of our study is to demonstrate whether there is a quantitatively measurable anatomical difference between patients with and without clinically diagnosed multidirectional instability with specific attention to size of the recess at the rotator interval, inferior joint capsule recess size, and glenoid perimeter size.

Among the retrieved examinations, after applying the exclusion criteria, 37 studies were selected as described in Fig. 2.

Of these 37 patients, 20 were clinically diagnosed with multidirectional instability, while 17 received MR-a for other reasons.

The patients’ average age was 32.68 ± 14.22 years (range: 14–60 years, 6 female) in the case group and 33.69 ± 13.77 years (range 15–55 years, 5 female) in the control group. Regardless of gender, the average measurements of axillary recess, rotator interval, and glenoid circumference in MDI patients were, respectively, 1.86 cm (1.89 ± SD 0.3), 0.8 cm (0.8 ± SD 0.2), and 7.3 cm (7.18 ± SD 1.2) (Table 2).

The same measurements in the control group were 1.53 cm (1.58 ± SD 0.3), 0.6 cm (0.7 ± SD 0.18), and 7.5 cm (7.86 ± SD 1.5).

The Shapiro–Wilk test for normality proved positive in all the considered groups. Given the normal distribution of all datasets, the two groups were compared using Student’s t-test, which yielded the following results: the average axillary recess width in the MDI group was significantly greater than in the control group (t(33) = 3.15, p = 0.003); no significant differences were found in rotator interval width or glenoid circumference measurement (t(35) = 1.75, p = 0.08 and t(30) = 0.51, p = 0.6, respectively).

Using a cut-off value of 1.89 cm, the sensitivity, specificity, and overall performance of the test based on axillary recess amplitude are respectively 0.579, 0.941, and 0.802 (95% CI 0.656–0.948) (Fig. 3).

Multidirectional instability of the shoulder is a complex pathology to diagnose and requires experience from the clinician. Starting from cadaveric studies, it is largely accepted that capsular redundancy is one of the key points in the development of MDI [11, 12]. MRA can provide useful information to the clinician about the actual origin of clinically evident MDI, whether due to predisposing anatomical variants, unexpected injuries, or actual capsulo-ligamentous laxity [13].

Different methods have been previously reported; some authors have shown how an increased capsular volume, expressed as the three-dimensional capsular volume with respect to glenoid surface, and an increased sagittal cross-sectional capsular area are related to MDI [8, 14]. However, they also observed that the glenoid surface area is not significantly different in patients with or without atraumatic instability. This result confirms one of the findings of our study: glenoid circumference is not significantly different between MDI patients and control group patients. These results suggest that glenoid dimension is not linked to MDI, with the exception of the presence of a bony Bankart lesion, which, instead, is strongly correlated with traumatic instability [9, 13].

The width of the rotator interval has been previously studied on both traumatic [15] and atraumatic shoulder. Patients with chronic anterior traumatic instability have been proven to have an increased rotator interval height, area, and index [13]; also, patients with multidirectional atraumatic instability have increased width and depth of rotator interval and superior capsular elongation, compared to patients without instability [12, 14, 16]. We do not find statistically significant difference in terms of rotator interval width between patients with and without atraumatic shoulder instability, in line with a previous study by Provencher et al., who found no difference of rotator interval dimension expressed as the shortest distance between the anterior edge of the supraspinatus tendon and the superior edge of the subscapularis tendon [17]. On the contrary, they found that the long head of the biceps tendon assumed a more anterior position relative to the supraspinatus tendon in patients with posterior instability. An explanation for these results might lie in the limited number of patients in both the studies, although they agreed to say that there is a relationship between MDI and rotator interval dimensions [18].

As inferior instability is the main component of MDI of the shoulder, previous authors experimented different methods to measure capsular redundancy, such as gleno-capsular ratio or labro-capsular distance and they all agree that increased axillary recess depth is correlated to shoulder instability [8, 10]. Lee et al. and Kim et al. also found significant correlation between inferior capsular redundance and MDI [19, 20]. Our results are in line with these previous studies as we observed that the width of the axillary recess at its largest point is significantly increased in patients with clinically diagnosed MDI, compared with patients without instability.

Our study has some limitations: first its retrospective nature and the limited number of patients, but MDI patients were carefully selected as well as control group patients by applying rigorous exclusion criteria, listed above.

Second, capsular volume and capacity varied from patient to patient; therefore, the amount of distention of the joint was not well controlled. However, the injection was performed by the same expert musculoskeletal radiologist to avoid another potential bias due to the variation of contrast material injection. Lastly, our results are related to patients with symptomatic instability, but not with asymptomatic hyperlaxity.

Finally, since this was a retrospective study, it was not possible to obtain an asymptomatic control group. Consequently, we do not know the measures present in completely asymptomatic patients.

In conclusion, our results confirm that axillary recess width may be used in complex clinical situations where a pattern of multidirectional instability may pre-exist or overlap with other clinical conditions. Any corrective surgery should in fact take into account the possible presence of multidirectional instability due to capsular laxity. On the contrary, glenoid circumference is not related to MDI. The relationship between rotator interval width and instability remains debated as statistical significance was not achieved in our case when comparing the two groups.