We conducted a retrospective study on prospectively collected data from January 2013 to April 2017. All patients with large-sized or massive rotator cuff tears, which were determined by pre-operative MRI, were included in this study. Our institute classified the rotator cuff tears with modified Millstein and Snyder classification [13, 14]. A large-sized rotator cuff tear was defined as a complete supraspinatus tendon tear. Massive rotator cuff tear was defined as at least two tendons involvement. The tear size was confirmed under arthroscopic examination. Patients with rotator cuff re-rupture or with contraindications for surgery were excluded. Standard plain radiographs and MRI were performed in all patients.
Baseline characteristics including age, gender, duration of symptoms, and traumatic event were collected. Furthermore, radiographic parameters from plain X-ray and MRI including rotator cuff arthropathy, proximal migration of humerus, rotator cuff tear size, fatty infiltration, and muscle atrophy were gathered preoperatively. All parameters were collected twice by each of the two independent observers, in order to compare intra- and inter-observer reliability. However, we used the data from experienced orthopaedist only to analyze for the prognostic factors.
The arthroscopic surgery was performed by three experienced surgeons after the patient was sedated with general anesthesia and set in the beach chair position. In cases of adhesive capsulitis, the capsular release and manipulation was done in the same setting. Acromioplasty was performed if an acromial spur was defined. After the tear size was measured with a probe to confirm the size measured from the MRI, the surgeon released any adhesion surrounding the tendon. The tendon then was mobilized in order to cover its footprint as much as possible. If necessary, the interval sliding technique was also performed. The transosseous-equivalent double-row technique was chosen to be the first line of treatment. If the double-row technique were not feasible, a single-row technique or partial functional repair technique was chosen instead.
Complete rotator cuff repair was defined as the tendon covering up at least 50% of the anatomical footprint, while partial repair was defined as the tendon covering up less than 50% of the original footprint (Fig. 1) [12, 15]. All patients were underwent passive motion exercises on the first post-operative days. An arm sling was applied for 6 weeks. A progressive active-assisted passive motion exercises were commenced at the third to fourth week for muscle strengthening. This research was approved by Research and Ethic Committee from Queen Savang Vadhana memorial hospital.
Radiographic parameters
Pre-operative plain radiograph focused on three parameter measurements to evaluate rotator cuff arthropathy which include osteoarthritic change, acromiohumeral distance, and inferior glenohumeral distance. Further pre-operative MRI evaluation focused on mediolateral (ML) tear size, anteroposterior (AP) tear size, tendon retraction, fatty infiltration and muscle atrophy. Radiographic parameters were evaluated by one experienced orthopedist and one senior orthopedic resident. A 1.5-T MRI (Siemens Healthcare, Germany) was used in this study. All measurements were made on a PACS workstation using Agfa IMPAX 6 (Waterloo, Canada) software technology—this software uses data within the DICOM (NEMA, VA, USA) header on all MRI (whether from our institution or from outside institutions) to allow referenced measurements to be made.
Rotator cuff Arthropathy
The rotator cuff arthropathy was graded by Hamada’s classification [5], which showed good inter- and intra-observer reliability from a previous study [16]. The studied population were classified into 5 grades: grade 1, the AHI is ≥6 mm; grade 2, the AHI is ≤5 mm; grade 3, AHI is ≤5 mm with acetabulization of the acromion; grade 4, glenohumeral narrowing; and grade 5, humeral head collapse [7].
Superior Migration of the Humeral Head; Acromion-Humeral Interval (AHI) & Inferior Glenohumeral Distance (IGHD).
Acromion-humeral interval (AHI) was one of the parameter to evaluate superior migration of the humeral head (SMHH). It was measured on an anteroposterior view of the plain radiograph, as the distance between inferior border of acromion and superior aspect of the humeral head. Inferior glenohumeral distance (IGHD) was another parameter to evaluate SMHH measuring the distance between inferior glenoid margin and the inferior humeral head margin at just the intersection of the humeral head and the humeral neck [12] (Fig. 2).
Tear size
Tear size of supraspinatus tendon was collected from T2-weighted MRI in both coronal oblique and sagittal oblique views which gave mediolateral (ML) tear size and anteroposterior (AP) tear size respectively. ML tear size was a straight distance from the tendon edge to the lateral cortex of the greater tuberosity in the coronal oblique cut, which the tendon most retracted medially. AP tear size was a longest straight distance from the anterior tendon edge to the posterior tendon edge in the sagittal oblique cut [17]. The frayed tissues at the tip of tendon edge were not included in any measurement (Fig. 3).
Tendon retraction
From the coronal oblique view of T2-weight MRI, we use the slice which showed the most medial tendon retraction. The tendon retraction was classified using the classification described by Patte [18]. In stage 1, the tendon stump was close to the bony insertion. In stage 2, the tendon stump was retracted and lied at the level of humeral head between the footprint and glenoid. In stage 3, the tendon stump was seen at the level of the glenoid or beyond.
Fatty infiltration
Using the sagittal oblique view of T1-weight MRI, fatty infiltration in all rotator cuff muscles were classified by the modified Goutallier staging system described by Fuch et al. [19]. Based on the lateral image on which the scapular spine was in contact with scapular body (Y-shaped view), the patients were classified into 5 grades: grade 0, normal muscle without fatty streak; grade 1, some fatty streaks in the muscle; grade 2, fatty infiltration is present, but more muscle than fat; grade 3, equal amount of fat and muscle; and grade 4, more fat than muscle.
Muscle atrophy
We used the sagittal oblique view of T1-weighted MRI to evaluate the rotator cuff muscle atrophy by selecting the most lateral image on which the scapular spine was in contact with scapular body (Y-shaped view). First, the Zanetti’s tangent line extended from the superior aspect of the coracoid to the superior border of the scapular spine and the other two lines were extended from each scapular process to the tip of scapular body (Fig. 4). These lines divided muscle atrophy into 4 groups: none or no muscle atrophy, the muscle crossing over the tangent line; mild, border of the muscle touching the tangent line; moderate, a concave curve of the muscle beside the tangent line; and severe, the muscle bundle atrophy nearly touching the scapular spine and scapular body [20].
Statistical analysis
As the data were not normally distributed, we selected the Chi-square test to compare proportions of categorical variables between complete and partial repair group. These variables included gender, traumatic event and tendon involvement. A Mann-Whitney U test was used to compare between continuous data variables in the complete and partial repair groups. These include age, duration of symptom, AHI and IGHD, ML and AP tear size, grading of fatty infiltration and muscle atrophy. Pearson-correlation analysis and Spearman rank-order correlation took a role in determining association between categorical and continuous variables, respectively. These variables, which determined the independent factors affecting reparability, were analyzed in a stepwise multivariate logistic regression analysis. We created the receiver operating characteristic (ROC) curve of significant predicting factors which correlated with reparability of rotator cuff tendons. The cut-off levels of each factor were defined from the ROC curve. The likelihood ratios of each factor were then calculated to weight the importance to rotator cuff tendon reparability.
Intra-observer and inter-observer reliabilities were evaluated with the Kappa analysis and intraclass correlation coefficient for categorical and continuous data respectively.
Statistical significant was set at p-value < 0.05. We use the IBM SPSS statistics software version 20 for all statistical analysis.