This was a prospective, controlled trial study to investigate the efficacy and safety of BoNT-A (http://www.consultingroom.com/treatments/dysport) for the treatment of adhesive capsulitis of the shoulder, which was unresponsive to conventional treatments. The study conformed to the principles of good clinical practice and the Declaration of Helsinki and was approved by the local ethics committees. All patients were required to give a written informed consent before completing any study-related procedures.

Patients were aged 18-70 years with a "painful freezing phase" [4,25] resistant to conventional treatments of the following: physical therapy, nonsteroidal anti-inflammatory drugs (NSAIDs), opioids, exercise, and trigger point injection in predominantly one shoulder lasting 10-36 weeks. The patients also had severe shoulder pain with an intensity of five or higher on a Numeric Rating Scale (NRS; scored from 0-10). Patients were required to have stage II or greater shoulder adhesive capsulitis, defined as symptoms meeting one or more of the following criteria [26]: 1) pain at rest; 2) pain below the elbow; 3) disturbed sleep on the painful side at night; and 4) hard end-feel on lateral rotation. All patients were required to have a significant limitation of passive movements in the glenohumeral joint (more than 30 degrees for at least two of the following three movements: forward flexion, abduction or external rotation) compared with the unaffected side [27]. Patients with previous adhesive capsulitis in the opposite shoulder were accepted, even if the difference between the sides were somewhat less than 30 degrees.

We excluded patients with a significant trauma within 6 months, significant pathology such as bone fracture and nerve damage, pregnancy, lactation, a known allergy or sensitivity to study treatments or severe psychiatric conditions. Patients who were not able to comply with the ROM assessments due to excessive pain or with significant difficulties in relaxing sufficiently to allow the investigator to make adequate recordings were also excluded.

Eligible patients were randomized (1:1) to receive BoNT-A or TA by an intra-articular injection after providing informed consent and undergoing baseline evaluation. The patients who were blinded to the treatments were randomly assigned to a scheme, using a table of random numbers.

The injections were carried out under the fluoroscopic guidance, with the patient in the supine position. The patient's arm was placed as far as possible to the side, with the elbow flexed at 90 degrees and the palm facing up to hold the shoulder in external rotation. The central X-ray was vertically oriented over the joint. Prior to the injection, the skin was cleaned with an iodine solution and the area anesthetized with a local injection of 1% lidocaine. A 21-gauge 2.5-3 inch needle was vertically inserted under the fluoroscopic guidance into the anterior aspect of the joint. Prior to the study treatment, 2-3 mL of the contrast material Omnipaque (Sanofi-Winthrop, Markham, ON, Canada) was injected into the glenohumeral space to confirm the correct intra-articular needle position. This was then followed by an injection of either BoNT-A (200 units in 2 mL of 0.9% saline solution) or 20 mg TA (1 mL) and 1 mL of 0.9% saline solution.

All patients were evaluated using the NRS of pain severity and by the measurement of the ROM at baseline, 30 minutes after injection and at 2, 4, and 8 weeks post-treatment. Pain severity was assessed on a 0-10 NRS, where 0=no pain to 10=worst possible pain. The ROM assessments included an assessment of active flexion and abduction and passive flexion, abduction, external rotation and internal rotation, which were measured using a manual goniometer. Flexion and abduction were examined with the elbow joint extended in a sitting position, and the upper limb movement range was measured in the sagittal and coronal surfaces, respectively. External and internal rotation were defined as the movement range with the shoulder joint abducted and the elbow joint flexed to 90 degrees in a supine position.

The study was planned to have 30 patients completing the study. This would have given the study a statistical power of 0.8, to detect a difference in the pain relief of two points on the NRS at a significance level of 0.05.

A general linear model repeated measurements ANOVA (SPSS ver. 11.5; SPSS Inc., Chicago, IL, USA) with the Huynh-Feldt method was used for the comparison of NRS and ROM (active shoulder abduction and flexion, passive shoulder abduction, external rotation, and internal rotation) between the two groups at each visit. The chi-squared test was used to analyze the association between the categorical variables. The level of statistical significance was set at p<0.05.

In total, 15 patients received the BoNT-A treatment, and 13 patients received the treatment with TA. Two patients in the TA group were excluded, as they did not give consent. The demographic characteristics of the patients are shown in Table 1. There were no significant differences in the diagnosis between the two groups. Five patients in the BoNT-A group and four patients in the TA group experienced mild radiologic abnormalities, including tendinosis and degenerative arthritis on the plain X-ray or ultrasonographic images. All radiologic findings were confirmed by a specialized radiologist.

The mean NRS scores for each treatment group over the course of the follow-up are shown in Fig. 1. The baseline NRS scores were 7.4 in the BoNT-A group and 7.6 in the TA group. At week 8, both treatment groups showed a significant decrease in the NRS scores compared to baseline (mean week 8 score: BoNT-A group, 3.8; TA group, 4.6; both p<0.05). There was no significant difference in the NRS scores between the two groups at baseline or after 8 weeks. The reduced pain intensity score was maintained continuously throughout the 8 weeks post-injection period in both groups.

Improvements in the ROM for each treatment group over the course of the follow-up are shown in Fig. 2. At week 8, both treatment groups showed significant improvements in the active shoulder abduction and flexion as well as the passive shoulder abduction and external rotation compared to baseline (all p<0.05).

There was no significant difference between the two groups in the range of active flexion, passive internal rotation, and passive external rotation. In contrast, the improvements in the range of active abduction (degrees improvement in ROM, 22° vs. 6°) and passive abduction (22° vs. 7°) in the BoNT-A group were significantly greater than those of the TA group (all p<0.05).

No severe adverse events were reported in either treatment group. Adverse events included one case of flu-like symptoms in the BoNT-A group and two cases of flushing in the TA group.