Correlations between clinical features and MRI findings in early adhesive capsulitis of the shoulder: a retrospective observational study

Adhesive capsulitis (AC) of the shoulder is characterized by shoulder pain and limited active and passive range of motion (ROM) in the shoulder [1‐3]. In the past, the diagnostic terminology for this entity, such as “frozen shoulder,” was ambiguous and based on clinical features and symptoms [3, 4]. However, the disease presents with characteristic pathophysiological features, including capsular thickening and fibrosis due to chronic inflammation of the joint capsule, which may lead to capsular adhesion [1, 5].

Given that other diseases of the shoulder, such as rotator cuff tear, bursitis, and calcified tendinitis, may have similar clinical symptoms [1, 6], magnetic resonance imaging (MRI), ultrasound, and magnetic resonance (MR) arthrography are useful for differentiating AC from other diseases [2, 7‐10]. MRI offers better resolution and soft tissue contrast than other methods and is a key modality for differentiating shoulder disease [2]. Based on previous studies that used MRI, the key diagnostic findings for AC include capsular thickening, a hyperintense T2 signal and contrast enhancement in the axillary capsule and rotator interval, thickening of the coracohumeral ligament (CHL), and obliteration of the subcoracoid fat triangle [2, 11, 12]. These MRI findings have an important role in the diagnosis of early AC when clinical features are atypical and in shortening the duration of joint stiffness by allowing timely physical therapy and intra-articular steroid injection, which could help to reduce the morbidity rate [13, 14]. Although diverse structural abnormalities are known to be associated with AC, there is limited literature on the association between radiologic findings and clinical features, and the few studies available have assessed several MRI findings with some ROMs (external rotation and abduction) [9], or some MRI findings with several ROMs [15‐17]. Particularly it is known that, in pathophysiology, the capsule thickening and hypervascularization gradually progress in the early stages of AC disease [9, 15]. We hypothesized that these changes in AC were measured by MRI and could be related to the severity of clinical findings such as the patient’s ROM and shoulder pain. Accordingly, the objective of the present study was to investigate the association between various clinical features and the MRI findings for early AC that are known to date.

The study protocol was approved by our Institutional Review Board. The requirement for informed consent was waived in view of the retrospective nature of the study.

Table 1 shows the clinical characteristics of the study population, which comprised 12 men and 17 women of mean age 51.2 years. The mean interval between physical examination and MRI was 16 (range, 4–48) days. The mean duration of symptoms was 5 (range, 1–9) months, meaning that only patients with early stage AC were included. The VAS pain score tended to increase in the order of pain at rest, pain at night, pain during motion, and worst pain. Comparison of ROM between the affected and unaffected sides showed a significant decrease in ROM of the affected shoulder in all directions.

The retrospective study analyzed the correlations between MRI findings and clinical features (ROM, pain, and duration of symptoms) in patients with AC. Axillary capsular thickness and hyperintensity on MRI were negatively correlated with duration of symptoms and ROM in some directions.

Reduced ROM and shoulder pain have a variety of causes, including rotator cuff tear, bursitis, and calcified tendinitis. Therefore, it is important to be able to differentiate between these based on radiological findings [1]. Rotator cuff tear shows hyperintensity on T2-weighted images due to the space created by a partially or completely torn tendon being filled by a watery component, such as joint fluid [25]. Bursitis shows expanded findings due to fluid build-up in the subacromial subdeltoid bursa [26]. Calcified tendinitis shows radiopaque calcified deposits on tendons on plain radiographs and computed tomography scans and T2 hyperintensity is found because of nearby inflammation [27]. If the aforementioned findings are absent on radiologic examination, primary AC with capsular abnormality could be differentiated and diagnosed.

A study by Hannafin et al. [18] used a four-stage classification system based on progression of AC and defined the first stage (0–3 months) and second stage (3–9 months) with reduced ROM as early stages of AC. However, one of our study inclusion criteria was symptoms for ≥1 month to exclude patients with transient symptoms that were not associated with AC and would resolve spontaneously. Although the exact etiology of AC has not been identified, the pathophysiology likely involves capsular inflammation, which subsequently leads to capsular fibrosis [1, 6]. Therefore, AC shows a pattern of pain being dominant in the early stage followed by gradual reduction in ROM due to synovial and capsular inflammation and fibrosis. Capsular fibrosis becomes most severe in the late stage as the inflammation subsides. Although the symptoms of AC are known to improve naturally over time, complete recovery may take up to 2 years [2, 6, 20, 28]. There are various treatments available, including physical therapy, corticosteroid injection, manipulation under anesthesia, and surgical capsular release [13, 18, 29]. The decision regarding such treatment modalities may vary depending on the clinical stage, accurate diagnosis of AC, and accurate identification of AC stage and could ultimately shorten the clinical course [13, 18, 20].

Ahn et al. [15] reported a significant correlation between axillary capsular thickness and ROM in external rotation. The present study also found a negative correlation of axillary capsular thickness with ROM in external rotation and in other directions (internal rotation, forward flexion, and abduction). However, there was an only significant negative correlation between axillary capsular thickness and internal rotation after the adjustment for multiple comparisons, This discordance may be attributed to the fact that the study by Ahn et al. [15] did not differentiate different stages of AC whereas the present study analyzed only patients with early AC. A study by Sofka et al. [17] reported that axillary capsular thickness was significant greater in the early clinical stage (stage 2) than in any other stage but that there were no significant differences between other stages. Therefore, capsular thickening occurred together with capsular inflammation in the early stage of disease, suggesting that such findings may be correlated with clinical features. Given that the present study included only patients with early AC, correlations between capsular thickness in ROM for internal rotation may have appeared, unlike in other studies. Moreover, ROM in internal rotation tended to show a relatively higher correlation than other directions of ROM in the present study. This may have been due to the scoring system used for the degree of internal rotation based on each location in the spine, which may have shown correlations that could not be identified in previous studies [15]. Furthermore, a study by Park et al. [9] reported that hyperintensity in the axillary recess was correlated with ROM in external rotation, and the present study also found correlations between hyperintensity in the axillary recess and ROM in various directions. After the correction of multiple comparisons, there were only significant negative correlations between hyperintensity in axillary recess and abduction/forward flexion. A negative correlation was also found between hyperintensity in the humeral capsule and duration of symptoms, which is determined to be the effect of increased blood flow and edema associated with capsular inflammation in the early stage of AC, similar to results in previous studies [9, 17, 18]. As in previous studies [15, 22], this study also found no correlation between severity of pain and capsular thickness.

According to previous studies, CHL thickening and obliteration of the subcoracoid fat triangle are important diagnostic criteria for AC [2, 30]. While the present study found CHL thickening and obliteration of the subcoracoid fat triangle in most patients, these findings were not correlated with clinical features, as reported previously [9, 15].

This study had some limitations, including the following. First, the number of patients with AC was relatively low, which may have introduced a degree of bias. However, the results are similar to those of previous studies because the study population was limited to patients with early AC and the sample size was large enough to derive statistically valid conclusions. Second, correlations between MRI findings and clinical features at different stages could not be evaluated because no subgroup analysis based on different clinical stages could be performed. Third, patients with AC were defined based on clinical criteria not by diagnosis based on arthroscopy or tissue biopsy. However, because arthroscopic treatment is not recommended as first-line treatment for AC, a histological diagnosis may have been unethical. Therefore, we selected those patients suspected of having AC based on clinical symptoms, in the same way that AC is diagnosed in clinical practice and eliminated the possibility of another disease by MRI findings.

In conclusion, axillary capsular thickening and hyperintensity on MRI in patients with early AC correlates with limited ROM. MRI findings in patients with AC could be used for indicators of clinical symptoms.