Midterm Outcomes After Arthroscopic Coracoid Decompression Combined With Manipulation for the Treatment of Frozen Shoulder Associated With Subcoracoid Impingement

Shengyang Jin Department of Orthopaedics, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430022 Xiaohong Wang Department of Orthopaedics, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430022 Yu He Department of Orthopaedics, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430022 Wei Huang Department of Orthopaedics, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430022 Ze Wang Department of Orthopaedics, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430022 Chunqing Meng Department of Orthopaedics, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430022 Hong Wang (  wanghwh@126.com ) Department of Orthopaedics, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430022


Introduction
Frozen shoulder (FS) is a disease that often occurs in middle-aged and elderly persons. It is a common clinical condition characterized by progressively increasing pain and decreasing shoulder range of motion (ROM). The cause of FS may be idiopathic, or it may be secondary to a variety of clinical conditions, such as diabetes, thyroid and autoimmune diseases, rotator cuff tear, shoulder fracture, history of surgery, or rheumatologic diseases [1].
Subcoracoid impingement is the symptomatic impingement of the subscapularis tendon between the coracoid process and the humeral head. When the shoulder is in internal rotation, the coracoid process compresses the subscapularis tendon; thus, if the subcoracoid space is narrow, impingement of the subscapularis can lead to edema of the subscapularis bursa or even to subscapularis tendon tears [2][3][4][5][6]. Previous studies have identi ed subcoracoid impingement as a potential source of persistent shoulder pain and limited internal rotation of this joint [3][4][5]. The anatomy of the coracoid process changes with age, and the subcoracoid space is narrower in older individuals than in young people. As such, subcoracoid impingement becomes more prevalent with age [2]. Subcoracoid impingement is a clinical diagnosis supported by magnetic resonance imaging (MRI) ndings: edema of the subscapularis bursa is a sign of subcoracoid impingement [6,7]. In our clinical practice, we found that a proportion of patients with FS have concomitant subcoracoid impingement. We de ne this type of FS as a frozen shoulder associated with subcoracoid impingement (FSASI) for the present study.
FS is a commonly self-limiting disorder with a course of 1-3 years. However, during the disease course, patients are incapacitated and experience excruciating pain, insomnia, and may even feel anxious or depressed. A part of patients with FS continued to have shoulder ROM de ciencies and mild pain for a long time after the onset of initial symptoms compared with the action and sensation of the contralateral shoulder [8,9]. Given these factors, many patients desire an immediate resolution of their symptoms, even if FS may spontaneously resolve in most of the patients [10].
There are many effective conservative treatments for FS, including physical therapy, non-steroidal antiin ammatory drugs, intra-articular corticosteroid injection, and hydrodilation. Surgical options are generally reserved for patients with symptoms refractory to conservative management. These options include manipulation under anesthesia (MUA) and arthroscopic or open capsular release [11]. If the FS is associated with subcoracoid impingement, coracoid decompression is an effective and safe option, which can increase the subcoracoid space, helping to relieve the persistent shoulder pain and promote the recovery of internal rotation function [2,[12][13][14][15][16].
FSASI is rarely reported, and literature on its treatment is limited. Therefore, an investigation into the characteristics and treatment of this type of FS is warranted. In our study, we aimed to introduce coracoid decompression combined with manipulation to treat patients with FSASI and to investigate the midterm e cacy of this surgical approach.

Inclusion criteria
All patients underwent detailed physical examinations, shoulder X-rays, and MRIs for diagnosis. Inclusion criteria were as follows: (1) patients with FS, (2) patients with no positive response to conservative therapy after a minimum of 6 months [17], and (3) patients whose MRI suggested subcoracoid impingement. The exclusion criteria were as follows:(1) patients who were eventually diagnosed arthroscopically as not having subcoracoid impingement, (2) patients with FS secondary to diabetes, rheumatologic diseases, thyroid, and autoimmune diseases, (3) the presence of rotator cuff tear(s) or superior labral anterior and posterior lesions, (4) previous trauma or surgery, and (5) patients with moderate-to-severe glenohumeral osteoarthritis. From April 2015 to May 2018, 177 patients met the inclusion criteria (Fig. 1).

Surgical techniques
All patients underwent a standardized procedure under general anesthesia in the lateral decubitus position. Anatomic structures and arthroscopic portals were marked before the operation. A gentle MUA was rst performed to improve a part of the ROM of the shoulder. The posterior viewing portal was created approximately 2 cm medial and inferior from the posterolateral corner of the acromion. An anterolateral portal was then made approximately 1.5 cm lateral to and in line with the acromion border ( Fig. 2.A) [18]. We inserted a 30° arthroscope through the portals to examine the joint and rotator cuff tendons and recorded the pathological ndings. An anterior portal was created approximately 1 cm lateral to the coracoid process under arthroscopic supervision ( Fig. 2.A). Before creating the anterior portal, a 20-mL syringe needle was used as a guide to provide access to the rotator interval ( Fig. 2.B).
Hyperemic, hyperplastic, and scar tissues around the subscapularis were removed using a 90°r adiofrequency pulse (Fig. 3.A-B). The coracoid process was then exposed. Impingement of the coracoid against the subscapularis tendon and lesser tuberosity was con rmed by manipulating the arm in a combination of forward exion, adduction, and internal rotation [19]. After coracoid impingement has been con rmed, coracoid decompression was performed until external and internal rotation ranges of motion of the arm demonstrated adequate decompression ( Fig. 3.C-D). After the operation, the patient's shoulder was gently manipulated again in all directions to ensure that its ROM was the same as that of the healthy shoulder ( Fig. 3.E).

Rehabilitation program
The rehabilitation protocol [9] was initiated on the rst day after the procedure. Patients were encouraged to perform active and passive ROM exercises in all directions. The range and frequency of exercise increased gradually, according to each patient's recovery progress. Progressive resistance exercises were encouraged to enhance muscle strength. An arm sling and ice compress treatment were used at rest (Fig. 3.F). A month after the operation, patients were permitted to remove the arm sling and resume daily activities.

Evaluation
The abbreviated Constant-Murley score (ACMS; full score, 75) and the modi ed University of California at Los Angeles Shoulder Rating Scale (MUCLA; full score, 35) were used to evaluate the shoulder condition of each patient [15,20]. The ACMS and MUCLA assessments were performed preoperatively, at 1, 3, and 6 months postoperatively, and at the nal follow-up (May 15, 2020). The assessments were documented by two doctors who were blinded to the study details. (Please refer to the supplementary materials for the assessment forms used.)

Institutional review board approval
This study was conducted following the principles of the Declaration of Helsinki. The data reported from this study comply with the Consolidated Standards of Reporting Trials (CONSORT) statement.

Statistical analysis
All quantitative data were presented as means with 95% con dence intervals (95% CI). Student's t-test was used to analyze differences between ndings. Changes in preoperative and postoperative scores were tested using the paired t-test. p < 0.05 was considered statistically signi cant. All statistical analyses were performed using IBM SPSS Statistics version 21.0 (IBM Corp., Armonk, NY, USA), and the graphs were generated with GraphPad Prism version 7.04 (GraphPad Software Inc., CA, USA).
Preoperative MRIs showed edema of the subscapularis bursa near the coracoid process in the patients. In some patients, the subscapularis muscle was compressed and showed morphological changes. Bone edema at the impingement site was also observed (Fig. 4.A). During the operation, we found synovial membrane hyperemia and hyperplasia around the subscapularis tendon, accompanied by scar hyperplasia (Fig. 3.A-B). At the 6th month after surgery, 20 patients underwent MRI reexaminations. The MRIs showed that the edema around the subscapularis tendon had signi cantly reduced or resolved ( Fig. 4.B-C).

ACMS and MUCLA ratings
All subjects demonstrated improved ACMS and MUCLA scores after surgery ( Table 2). Figure 5.A-B show that the trend of shoulder score improvement is consistent between the ACMS and MUCLA scores. Improvement was signi cant during the rst 3 months. The magnitude of improvement then decreased, and the rating scores remained stable between the 6th month after surgery and nal follow-up. As shown in Fig. 5.C-G, the recovery of forward exion and abduction progressed synchronously, and almost full scores were achieved at the nal follow-up (9.96 ± 0.08 and 9.92 ± 0.11 points, respectively). A full score ranges from 151° to 180° [20]; thus, a full score does not always mean that the patient recovered completely. Some patients (n = 8) still had complaints of forward exion and abduction ranges that were not the same as those of their healthy shoulders: their forward exion and abduction ROM at the nal follow-up were 163.13°±7.04° and 160.00°±8.65°, respectively. At the nal follow-up, some patients still had limited external and internal rotation ROMs: 4 patients had a score of 8 in external rotation, 12 patients had a score of 8 in internal rotation (thumb pointing to the 12th dorsal vertebra), and 2 patients had a score of 6 in internal rotation (thumb pointing to the waist). However, the degree of loss was slight and did not affect their activities of daily living; all patients were satis ed with the ROM of their affected shoulders at the nal follow-up.

Pain and sleep quality
We used the pain score subitems of the ACMS and MUCLA scales to assess pain. The ratings ranged from 0 to 15 points and from 0 to 10 points, respectively. In the ACMS pain rating, the score improved from 1.12 ± 0.68 (preoperative) to 9.69 ± 0.74 (1st month; p < 0.001), 11.73 ± 0.70 (3rd month; p < 0.001), 12.76 ± 0.73 (6th month; p < 0.001), and 13.06 ± 0.71 (last follow-up; p < 0.001), respectively. In the MUCLA pain rating scale, the score improved from 1.67 ± 0.27 (preoperative) to 5.96 ± 0.66 (1st month; p < 0.001), 8.04 ± 0.49 (3rd month; p < 0.001), 8.78 ± 0.44 (6th month; p < 0.001), and 8.94 ± 0.43 (last follow-up; p < 0.001), respectively. Figure 5.H-I show that the trend of pain improvement was consistent in both pain scales, which suggested that pain relief was gradual and that most of the improvement occurred within the rst month after surgery. Nocturnal pain is the leading cause of sleep disturbance in patients. With the resolution of pain, the patient's sleep state also gradually improved. Forty-ve patients complained of sleep disturbances preoperatively, and the number decreased to 9 at the 1st month and to 1 at the 3rd month postoperatively. From the 6th month postoperatively, none of the patients had insomnia due to shoulder pain.

Patient satisfaction, complications, and recurrence
Forty-seven patients were satis ed with the operation, whereas the remaining two patients were dissatis ed. The latter two complained that the shoulder ROM recovery was slow. They did not follow our recovery plan because they experienced pain while performing it. They had limited improvement in shoulder ROM at 1 month after surgery. They were asked to comply with their prescribed rehabilitation exercises, and nally, they recovered their ROM at 2 and 2.5 years, respectively). However, they both went through a long and painful recovery process. At the nal follow-up, no complications or recurrence occurred in any of the patients.

Discussion
In our present study, we introduced a particular type of FS (FSASI) and a feasible method to treat the disease. FSASI was diagnosed in this study according to the symptoms of the patient, MRI, and arthroscopy examinations. We describe our experience with 169 patients with FSASI who required surgical management after conservative therapy alone failed. Over an average of 38 months after surgery, the patients improved signi cantly in key areas such as pain and functionality. Our results showed that arthroscopic coracoid decompression combined with manipulation is an alternative method for FSASI. The procedure had few complications, the rates of recurrence were low, and patient satisfaction was high. This study contributes to the limited literature on therapeutic interventions for FSASI.
Arthroscopic capsular release is a conventional, minimally invasive method for the treatment of FS. However, the optimal method for capsular release remains controversial [11]. Although the "360° total capsular release" allows for a complete release of the contracted capsule and provides immediate improvements in ROM, this technique may lead to axillary nerve injury [21][22][23][24][25]. Studies have shown that the rotator interval is the primary site of pathology in FS. The release of the rotator cuff space, combined with the inferior capsule release, provides a signi cant improvement in ROM [26]. Some other studies have indicated that the lower capsule release should be avoided to prevent axillary nerve injury. Many surgeons release only the rotator cuff interval and the contracted coracohumeral ligament; they report positive outcomes [27][28][29]. In our study, we performed a nite release of the rotator interval: we cleaned the pathological tissues covering the subscapularis and rotator interval as well as a part of the anterior articular capsule and the medial glenohumeral ligament, which was performed before the coracoid decompression. After the nite release, we found that it was insu cient for restoring full ROM; hence, we followed this procedure with MUA.
MUA is often used to treat FS when conservative treatments have failed. MUA can provide improvement of shoulder function [30,31]. However, sometimes MUA is subject to great resistance because of joint adhesion, and aggressive manipulation may result in iatrogenic lesions, including hemarthrosis, superior labral anterior and posterior lesions, rotator cuff tears, and humeral or glenoid fractures. Arthroscopic release before MUA may decrease the incidence of these complications [32,33]. We found a gentle manipulation could easily restore the patient's ROM after arthroscopic coracoid decompression.
In summary, our treatment method is a combination and re nement of previously reported techniques [27][28][29][31][32][33]. The axillary nerve injury resulting from a wide range of capsule release and iatrogenic injuries resulting from aggressive manipulation are avoided with our method. Meanwhile, coracoid decompression can increase the subcoracoid space, helping to relieve shoulder pain and promote the recovery of internal rotation function [19,34].
In our study, we used ACMS and MUCLA scales to evaluate the patients. The ACMS scale excluded the strength subitem of the Constant-Murley score because the strength score was based on the weight of pull that a patient could resist in 90° of abduction, and this subitem does not apply to patients with FS [20,35,36]. As a supplement, we used the MUCLA scale to evaluate strength by testing forward exion with manual muscle testing. In addition, the MUCLA scale evaluates pain and function according to different evaluation criteria and considers patient satisfaction. The MUCLA and ACMS scales complement each other and together re ect the symptoms and progress during recovery.

Limitations
The evaluation method we used has limitations. First, the full scores of forward exion and abduction represent 151-180°; thus, a full score does not necessarily mean that patients recover completely. Moreover, the external rotation outcome in the ACMS re ects a functional movement that combines external rotation, forward exion, and abduction. Pure external rotation was not assessed [37]. Although our results indicate that our surgical approach was effective and that most patients were satis ed with the rehabilitation process and the eventual outcome, the recovery of ROM was time-dependent, and a slight degree of loss in ROM may occur. As mentioned in the results, two patients were not satis ed with the operation. The lesson we learned from these two patients was that because we performed a nite release, postoperative rehabilitation exercise is essential, especially in the 1st month after surgery. Otherwise, ROM can be di cult to recover.
In addition, although the clinical manifestations of coracoid impingement and its association with FS are well established [1], the causal relationship between FS and subcoracoid impingement remains controversial [5,38,39]. Some authors propose that progressive osteoproliferation of the coracoid process caused by constant friction during the internal rotation leads to collagen bers' thickening. They described coracoid impingement as a potential etiology of shoulder pain, limited ROM, and subscapularis tears [3][4][5]. Conversely, some authors argue that subcoracoid impingement is a consequence rather than a cause of FS, and they proposed that rotator interval lesions and the contracture of the capsule reduced the subcoracoid space, resulting in further impingement in patients with FS [40,41]. This remains an important topic for future study.
Lastly, although we introduced a novel method to treat FSASI and found the procedure was effective, we did not use a randomized controlled trial to prove superiority over other methods. In future studies, randomized controlled trials are needed to compare our surgical approach with other treatment methods.

Conclusion
Arthroscopic coracoid decompression combined with MUA was an alternative method to treat FSASI when coupled with a rehabilitation program, and the satisfactory results were maintained through midterm follow-up. Randomized controlled trials are required to determine the treatment regimen that is superior for patients with FSASI.

Consent for publication
Written informed consent for publication of images was obtained from the participants.

Availability of data and materials
The datasets supporting the conclusion of this article are included within the article. Upon request, raw data can be provided by the corresponding author.    Magnetic resonance imaging (MRI). (A, B) Typical preoperative MRI manifestations of the frozen shoulder associated with subcoracoid impingement. Red star, subscapularis bursa edema near the coracoid process; red arrow, subscapularis muscle being compressed and changed in shape; yellow arrow, bone edema at the impingement site. (C) MRI performed 6 months after surgery of a patient whose preoperative MRI manifestation is presented in B.