This retrospective study was approved by our Institutional Review Board.
Informed consent was waived due to the retrospective nature of the study.
Inclusion and exclusion criteria
Patients
Between January 2016 and December 2016,
consecutive 219 shoulder MR examinations excluding MR arthrography were performed in our institution.
Medical records of these patients were analyzed.
The initial diagnosis of adhesive capsulitis was based on history and clinical symptoms.
Clinical criteria for the diagnosis of adhesive capsulitis included restricted passive motion of greater than 30 degrees in two or more planes of movement compared to normal contralateral shoulder and gradually increasing shoulder pain that was more severe at rest for at least one month with normal radiographic findings [6,
21].
Among 46 patients who fulfilled the criteria,
17 patients were excluded for the following reasons: bilateral shoulder restricted motion,
tear of rotator cuff tendon,
calcific tendinitis,
rheumatoid arthritis,
severe osteoarthritis,
labral lesion.
The remaining 29 patients (12 males,
17 females; mean age,
51 years; age range,
30–73 years) with a final clinical diagnosis of adhesive capsulitis were identified and their images were evaluated retrospectively.
Control group
Control group composed of consecutive 20 patients (10 males,
10 females; mean age,
49 years; age range,
23–63 years) with normal glenohumeral joint on shoulder MRI,
excluding MR arthrography.
Patients in the control group had been referred for MRI at our institution between January 2016 and December 2016.
They had no restricted shoulder motion or history of adhesive capsulitis.
Control group patients were referred for the following reasons: evaluation of the shoulder pain (12 patients),
soft-tissue masses (lipoma,
5 patients; hematoma,
1 patient),
and bone lesions (2 patients).
Clinical Assessment
All patients,
including control,
underwent physical examination before the MRI examination by one physician (22 years of experience) in the shoulder clinic.
Using a universal goniometer,
range of motion (ROM),
including external rotation,
forward flexion,
and abduction,
was evaluated.
The external rotation was assessed while the shoulder was maintained in 0° of abduction with 90° elbow flexion.
ROM for the internal rotation was measured by noting the highest segment of spinal anatomy reached with the thumb.
Forward flexion was measured as the maximum arm-trunk angle by the elevation of arm forward above the head.
Abduction was measured as the ability to raise both arms from the side to full abduction (180°) above the head.
The mean duration of symptom was 5.5 months (range,
1 - 14 months) in the adhesive capsulitis group.
Mean interval between clinical assessment and MR imaging was 16 days (range,
4 ~ 48 days).
MR Image Acquisition
MRI Protocol
All patient and control group had same MRI protocol with a 3-T MRI (Intera Achieva,
Philips Healthcare) unit with a dedicated shoulder coil.
During imaging,
patients were lying in supine position with arm externally rotated to the maximum extent possible.
The following imaging parameters were used: oblique sagittal fat-suppressed proton density (PD) VISTA sequence with Spectral Attenuated Inversion Recovery (SPAIR) imaging (TR/TE,
2000/18.6; echo-train length,
140; section thickness,
1.2 mm; matrix,
268 × 267; FOV,
160 x 160 mm),
oblique coronal fat-suppressed T2-weighted imaging (TR/TE,
4700/80; echo-train length,
10; section thickness,
3 mm; matrix,
356 ×255; FOV,
160 x 160 mm),
oblique coronal T1-weighted imaging (TR/TE,
530/20; echo-train length,
3; section thickness,
3 mm; matrix,
358 × 258; FOV,
160 x 160 mm),
oblique sagittal T2-weighted imaging (TR/TE,
3800/80; echo-train length,
16; section thickness,
3 mm; matrix,
356 ×256 ; FOV,
160 x 160 mm),
oblique sagittal T1-weighted imaging (TR/TE,
530/20; echo-train length,
3; section thickness,
4 mm; matrix,
356 ×258; FOV,
160 x 160 mm),
axial fat-suppressed PD imaging (TR/TE,
2100/30; echo-train length,
20; section thickness,
3 mm; matrix,
356 × 240; FOV,
160 x 160 mm).
MR Image Analysis
Measurements of parameters were independently performed by two musculoskeletal radiologists (9 years of experience) on a picture archiving and communication system (PACS) workstation (INFINITT,
Infinitt healthcare,
Seoul,
Korea).
Two musculoskeletal radiologists who were blinded to the clinical information independently evaluated all variables on MR images.
Quantitative and qualitative MRI findings for the diagnosis of adhesive capsulitis were defined as those described in the literature [3,
5,
11,
22,
23].
Prior to the measurement,
training session for both readers was performed using images that were different from the analysis data.
After two readers measured the parameters separately,
all parameters were re-evaluated under the consensus and statistical analysis was performed with these data.
Quantitative analysis
In quantitative analysis,
the following parameters were measured on MR image with magnification: anterior capsular thickness,
humeral and glenoid capsular thickness in axillary recess,
maximal axillary capsular thickness,
coracohumeral ligament thickness,
and degree of external rotation.
Anterior capsular thickness was measured on the thickest portion of the anterior portion of the glenohumeral joint capsule,
below the subscapularis tendon,
including the middle glenohumeral ligament and spiral glenohumeral ligament,
which seemed to have relatively low signal intensity compared to the subscapularis tendon at the level of the glenohumeral joint.
This measurement was performed on both oblique sagittal 3D PD VISTA SPAIR and axial fat-suppressed PD images (Fig.
1,
2) [24-26].
In the axillary recess,
humeral and glenoid capsular thickness were measured on oblique coronal T2-weighted MR images of both humeral and glenoid regions after magnification at the thickest portion.
The maximal axillary capsular thickness was then determined to the larger value of humeral and glenoid capsular thickness (Fig.
3).
The maximal coracohumeral ligament thickness was measured on oblique sagittal T2 images (Fig.
4).
The degree of external rotation on the axial MR image was measured by drawing a line from the center of the humeral head to the longitudinal axis of the scapular body with a second line from the center of the humeral head to the bicipital groove of the humeral head (Fig.
5) [6].
All measurements were recorded to two decimal places.
Qualitative analysis
The following findings were evaluated as present or absent: anterior capsular edema,
humeral and glenoid capsular edema in axillary recess,
edema and obliteration of the subcoracoid fat triangle.
If there was edema on either side of the humeral or glenoid capsule of the axillary recess,
the axillary capsular edema was considered present.
Edema of joint capsule and subcoracoid fat triangle was evaluated on oblique coronal fat-suppressed T2-weighted MR images (Fig.
6,
7).
Obliteration of the subcoracoid fat triangle was defined as low signal intensity of fat on T1-weighted images with respect to subcutaneous fat on oblique sagittal T1-weighted images.
Both partial and complete obliteration were considered as signs of adhesive capsulitis [11,
12,
23] (Fig.
8).
Statistical Analysis
Demographic and various MR findings were compared between the adhesive capsulitis group and the control group.
Fisher’s exact test or Mann–Whitney U-test was used to compare demographic data and imaging parameters between the two groups.
A correlation study was also performed to evaluate the effect of external rotation degree on MRI findings.
Binary multiple logistic regression analysis was performed to determine the relative contribution of different MR imaging findings.
Characteristics with a p-value of less than 0.05 at univariate analysis were used as independent input variables for multiple logistic regression analysis.
To eliminate multicollinearity,
multivariate analysis was conducted separately in quantitative and qualitative findings.
To evaluate diagnostic utilities of various parameters,
we performed receiver operating characteristic (ROC) analysis to determine sensitivities,
specificities,
and cut-off values.
Inter-class correlation coefficient (ICC) was calculated to assess the extent of agreement between the two readers in terms of measurement of four parameters in quantitative analysis.
To evaluate inter-observer variability for qualitative analysis,
Cohen kappa statistics were also calculated.
ICC or kappa value was interpreted as follows: 0,
poor agreement; 0.01–0.20,
slight agreement; 0.21–0.40,
fair agreement; 0.41–0.60,
moderate agreement; 0.61–0.80,
good agreement; and 0.81–1.00,
excellent agreement.
All statistical analyses were performed using one of two computer software programs: SPSS version 20 (SPSS,
Chicago,
IL,
USA),
and MedCalc version 16.2.1 (MedCalc Software,
Ostend,
Belgium).