Adhesive capsulitis; evaluation of a recently introduced MRI criterion

all


Introduction
Adhesive capsulitis is a common disorder, defined as the painful limitation of glenohumeral range of motion due to capsular hyperplasia and fibrosis (1). The disorder affects almost all aspects of a man's daily life, resulting in severe pain, anxiety, disability and sleep disorder (2). The symptoms usually persist for two years, however, a lifelong limited range of motion exist in 10% of the cases (1). The clinical course of the disease is divided into four stages: (a) First three months from onset of symptoms, (b) Three to nine months from onset, (c) Nine to fifteen months from onset and (d) 15 to 24 months from onset (3,4). Early-stage diagnosis results in effective conservative treatment-which in turn-reduces the duration of symptoms and patient's morbidity (5,6). The diagnostic criteria of adhesive capsulitis are entirely

Key point
Adhesive capsulitis is a common disorder, defined as the painful limitation of glenohumeral range of motion due to capsular hyperplasia and fibrosis. Clinical criteria for diagnosis of adhesive capsulitis are quite variable and challenging. Different MRI criteria are suggested for the diagnosis up to date, one of which is the anterior capsule thickness. Anterior capsule thickness showed the highest diagnostic performance among other variables. clinical, and include (a) More than 30 degrees limitation in glenohumeral range of motion (in comparison to the contralateral normal shoulder) in at least two planes, and (b) Gradually increasing shoulder pain for at least one month, which exacerbates at rest (7). The abovementioned criteria, however, are mild and variable in the first stages, making the diagnosis challenging; as a result DOI: 10.34172/ipp.2022.34438 we could say no definitive criteria exist for the diagnosis of adhesive capsulitis to date (8). As a consequence magnetic resonance imaging (MRI) -as the gold standard of shoulder imaging-plays a critical role in diagnosis of adhesive capsulitis, in the early stages (9). Studies on histological aspects of adhesive capsulitis suggest the predominance of abnormalities in the anterior joint capsule, which include accumulation of myofibroblasts and collagenous matrix, focal fibrosis, inflammatory cytokines, regenerative neural fibers and increased vascularity in this region, which is interestingly absent in the inferior portion (2). Arthroscopic studies confirm the predominance of capsular thickening in its anterior aspect (7). The anterior predominance of pathologic and arthroscopic abnormalities suggest that the thickness of anterior joint capsule may be a more reliable diagnostic criterion on MRI; however, to our knowledge, only one study has evaluated the significance of this parameter; since, the results of which, may be subject to substantial bias, due to small sample size (7).

Objectives
In the present study, we aimed to evaluate the significance of anterior capsular thickness in diagnosis of adhesive capsulitis on MRI.

Study design
This study was an observational and case-control. Sampling was started upon approval of the study, by the biomedical research ethics committee, Tehran university of medical sciences, Tehran, Iran, on January the 22 nd 2022. Diagnosis of adhesive capsulitis in our institution follows a clinical-radiologic approach. First impression is made based on clinical history and physical examination (which is performed by one orthopedics surgeon and one medical physicist). The clinical criteria include gradually progressing shoulder pain for a minimum period of one month, and more than 30-degree limitation of range of motion at two orthogonal planes in comparison with the contra-lateral normal shoulder. The patient is then referred for shoulder MRI, in order to rule out other possible differential diagnoses; if so, adhesive capsulitis becomes the final diagnosis, based on which the treatment begins. Arthroscopy is reserved for equivocal cases. Once the final diagnosis is made, our musculoskeletal radiologist saves the patient's MRI series on the Picture Archiving and Communication System (PACS, INFINITT Healthcare Co., Seoul, South Korea), inside a specific folder, named "Adhesive capsulitis". All the thirty recorded cases from January the 21 st 2020 until January the 21 st 2022 were selected as the adhesive capsulitis group.
Same number of patients, who were referred for shoulder MRI, for another reason (including shoulder mass, shoulder bony lesions and non-adhesive shoulder pain), in the same period of time, were collected randomly, as the control group.
All cases and controls had undergone the same noncontrast-MRI protocol as follows; The patient was supine, with his shoulder in maximum possible external rotation. The following sequences were obtained, by the same magnetic resonance scanner (Discovery TM MR750 3.0T, 60 cm MRI Scanner, GE Healthcare): Where S1, S2, M1 and M2 were derived from the previous similar study by Park et al (7), and the confidence interval and power were assumed as 95% and 90% respectively.
MRI variables of interest were measured independently, by two blinded radiologists, with 3 and 10 years experience in musculoskeletal imaging, respectively. After calculation of the inter-observer agreement, the radiologists studied the images for the second time, to reach a consensus on the results. The latter data was conducted for the rest of the statistical analysis.
The following quantitative variables were measured in the images; • Anterior capsular thickness, defined as the thickest part of the glenohumeral capsule, at 2-5 o' clock, under the subscapularis tendon, which contains the middle and spiral glenohumeral ligaments, measured on oblique sagittal and axial proton-density weighted (PDW) images ( Figure 1). • Axillary recess capsular thickness, defined as the maximum capsular thickness in the 6 o' clock, measured on oblique coronal PDW images ( Figure 2a). • Glenoid capsular thickness in the axillary recess, defined as the maximum thickness of the glenoid side of axillary recess capsule, measured on oblique coronal PDW images ( Figure 2b). • Humeral capsular thickness in the axillary recess, defined as the maximum thickness of the humeral side of axillary recess capsule, measured on oblique coronal PDW images ( Figure 2b). • Coracohumeral ligament thickness, defined as the thickest part of the ligament measured on the oblique sagittal PDW images ( Figure 2c). • Maximum degree of external rotation, defined as the angle between the line drawn from the center of humeral head along the longitudinal axis of the scapular body, and the line drawn from the center of humeral head to the center of the bicipital groove, measured on the axial PDW images (Figure 2d). The following qualitative variables were assessed in the images: • Anterior capsular abnormal signal intensity, defined as an increased signal of the capsule on fat-suppressed PDW images. • Axillary recess capsular abnormal signal intensity, defined as an increased signal of the capsule at any point of the axillary recess on fat-suppressed PDW images ( Figure 3a). • Glenoid capsular abnormal signal intensity, defined as an increased signal of the glenoid side of axillary recess capsule on fat-suppressed PDW images (Figure 3a). • Humeral capsular abnormal signal intensity, defined as an increased signal of the humeral side of axillary recess capsule on fat-suppressed PDW images ( Figure 3a). • Sub-coracoid fat abnormal signal intensity, defined as any amount of fluid signal in the fat on fat-suppressed PDW images (Figure 3b). • Obliteration of the sub-coracoid fat, defined as complete replacement of the fat with soft tissue signal on T1W images ( Figure 3c). • Rotator interval capsular abnormal signal intensity, defined as an increased signal of the capsule in the rotator interval, on fat-suppressed PDW images ( Figure 3d).

Statistical analysis
The quantitative variables were reported in terms of mean ± SD, and the qualitative variables in terms of percentage of   frequency. The quantitative and qualitative variables were compared between cases and controls, by independent t-test and chi square, respectively. P-value of 0.05 was chosen as the level of significance. For the significantly different variables, sensitivity, specificity, positive and negative predictive values and accuracy were calculated. Receiver operator characteristic (ROC) curve analysis was conducted on the quantitative variables, to determine the optimum cut-off values and compare their diagnostic performances.
Inter-observer agreement was calculated, by Gamma (Goodman and Kruskal's gamma) Coefficient for quantitative, and Cohen's Kappa coefficient for qualitative variables, respectively. All statistical analyses were performed, using SPSS version20 (Chicago, IL, USA).

Results
Mean ± SD for age of the cases and controls was 51.7 ± 6.6 and 48.9 ± 10.9 years, respectively, and no significant difference existed in the mean age between the two groups (P = 0.23). Cases included 18 females (60%) and 12 males (40%) and controls included 16 females (53.3%) and 14 males (46.7%), which did not show significant difference among the groups (P = 0.23).
All analyzed quantitative and qualitative variables, showed significant difference between the two groups (P < 0.001). Table 1 summarizes the mean and standard deviation for the quantitative variables. Table 2 summarizes the percent of frequency for the qualitative variables. Table 3 shows the sensitivity, specificity, positive and negative predictive values and accuracy of diagnosis of the qualitative MRI criteria. The abnormal signal of anterior capsule, along with the abnormal capsular signal at axillary recess and rotator interval were the most accurate qualitative criteria. The anterior capsular abnormal signal was the most specific criterion; similar to obliteration of the sub-coracoid fat (both with 100% specificity). Abnormal signal of the capsule in axillary recess showed to be the most sensitive.
Optimum cut-off values and diagnostic performances of the quantitative variables, through ROC analysis ( Figures  4 and 5), are summarized in Table 4. Coracohumeral ligament and anterior capsule thicknesses with cutoff points of 2.45 mm and 1.3 mm respectively, showed  (10). Table 5 summarizes the degree of agreement for all the assessed variables. All quantitative variables, showed near perfect agreement between the observers, with the anterior capsular thickness being the mostly agreed upon criterion (gamma coefficient: 0.98). All qualitative variables showed perfect agreement among the observers, except abnormal signal of the subcoracoid fat and glenoid capsule of the axillary recess (with kappa's coefficient: 0.87 and 0.85 respectively).

Discussion
Regarding its prevalence and debilitating nature, adhesive capsulitis has long been a subject of study; with many of them being focused on MRI criteria for diagnosis of the disease.
A meta-analysis performed by Suh et al on 2019 ultimately suggested that thickening of the coracohumeral   ligament was the most specific, while enhancement of the axillary recess and rotator interval were the most sensitive MRI criteria of adhesive capsulitis. The latter two criteria were assessed in contrast-enhanced images, which were not included in our study, due to the fact that contrastenhanced MRI is not usually performed in the approach to shoulder pain in practice (18).
Rotator interval of the shoulder was first introduced by Neer in 1970 as a shoulder stabilizer (19). Since then, many studies were focused on changes of this space and its structures in the course of adhesive capsulitis. The relative complexity of its anatomy and small size of the structures, make it difficult to evaluate the rotator interval on imaging and arthroscopy. The coracohumeral ligament and subcoracoid fat are the only easily recognizable structures of this space on non-contrast MRI, and that is the reason why studies based on conventional shoulder MRI, were focused on these two criteria (20).
From 2019 on, some authors proposed new quantitative criteria, aiming to suggest a way of more accurate and practical evaluation of the rotator interval, on noncontrast images, which include the width of rotator interval, coracohumeral ligament area, thickness of rotator interval soft tissue and the superior glenohumeral ligament thickness (21)(22)(23).
In 2019, Park et al introduced the concept of anterior capsule of the shoulder for the first time, and evaluated its thickness and signal intensity in a case-control study, the results of which showed a significant difference of both variables between the groups (7).
Anterior capsule of the shoulder was defined as the glenohumeral capsule from 2-5 O' clock, which is supported by the middle and spiral glenohumeral ligaments. The spiral glenohumeral ligament is attached to the lesser tubercle of humerus proximally and the infraglenoid tubercle distally. In the middle of its course the spiral glenohumeral ligament, passes in close proximity of the middle glenohumeral ligament and it blends with the inferior glenohumeral ligament at the end (7).
In the present study, thickening and abnormal signal of anterior capsule, both were significantly more frequent in adhesive capsulitis patients, in comparison with the controls. Given the cut-off point of 1.3 mm, anterior capsular thickness was 86.7% sensitive and 96.7% specific. With the area under curve equal to 0.95, anterior capsular thickness showed the highest diagnostic performance only after coracohumeral ligament thickness. One of the patients had normal axillary recess capsule and coracohumeral ligament thicknesses, while showing an anterior capsule thicker than the cut-off (5.7 mm). This means that if we would not have measured the anterior capsular thickness for this patient, the diagnosis of adhesive capsulitis was not made on MRI ( Figure 6).
Our results were in congruence with the study by Park et al. The only significantly different finding was the mean of the anterior capsular thickness in patients, between Figure 6. A 55-year-old female, with the final diagnosis of adhesive capsulitis. a. Axial proton-density weighted MRI shows thickening and edema of the anterior capsule of glenohumeral joint (calipers). b. Thickening and edema of the anterior joint capsule are also evident in this oblique sagittal proton-density weighted image (arrowhead). Note "within normal limit" thickness and signal intensity of the anterior (short arrow) and posterior (long arrow) bands of the inferior glenohumeral ligament.
the two studies (2.5 mm in ours versus 3.99 mm in the study by Park et al). The different imaging protocol and measuring method may be the cause; additional possible reasons include (a) Difference in disease pathogenesis course between the genetically different study populations. (b) Difference in the stage of the disease in the study population (7).
The abnormal signal of the anterior capsule, in conjunction with obliteration of the sub-coracoid fat and glenoid capsule abnormal signal showed to be 100% specific, which means that if either criterion exists, the diagnosis of adhesive capsulitis will be made confidently on MRI.
Another noteworthy point was the near perfect interobserver agreement of anterior capsular thickness-actually the highest among our quantitative variables -which suggests that this criterion is acceptably clear for the inexperienced radiologist, to measure reliably.
Similarly, all other studied variables were significantly different between cases and controls, among which the coracohumeral ligament thickness was the most sensitive. This means that a thickness less than 2.45mm on MRI excludes adhesive capsulitis almost confidently. These findings are similar to those of Chi et al (24).

Conclusion
Anterior capsule thickness and signal, as two reliable and reproducible criteria, are of significant value to be added to our routine MRI measurements, for diagnosis of adhesive capsulitis. Future studies with large sample volumes, clinical sub-categorization of the patients and multivariate analysis are recommended to more accurately define the role of anterior capsule, in MRI-based diagnosis of adhesive capsulitis.

Limitations of the study
The current study faced with some limitations: (a) Small sample size, as most of our adhesive capsulitis patients did not undergo MR examination. (b) We did not sub-categorize our patients based on the clinical stage, due to their small number. Studies with larger sample volumes are necessary to correlate the anterior capsular thickness with the clinical stage. (c) The gold standard of adhesive capsulitis was clinical-radiologic diagnosis, just the way it happens in practice. As most of the patients with adhesive capsulitis undergo conservative treatment, we did not have histopathologic prove to their diagnosis, and this makes our results prone to possible bias.

Ethical issues
The research followed the tenets of the Declaration of Helsinki. The Biomedical Research Ethics Committee, Tehran University of Medical Sciences approved this study (IR.TUMS.IKHC. REC.1400.008). This study was extracted from "Medical specialist degree" thesis of "Samin Khoei" at this university. Besides, this study was an observational, case-control, with informed consent being waived, due to its retrospective nature.