Subjects
This study recruited patients with MRCTs involving at least two tendons, including the supraspinatus (SSP) and infraspinatus (ISP), with or without the subscapularis (SSC). MRCTs were confirmed by magnetic resonance imaging (MRI) in all patients. Exclusion criteria included a concurrent neuromuscular disorder, a history of surgery on the shoulder joint, a score greater than 3 on the numerical pain rating scale during arm elevation, and an inability to elevate the arm by at least 140°.
A total of 15 patients (15 shoulders, mean age 76.1 years) were divided into two groups: 10 shoulders in the SSP and ISP with SSC tears (Torn SSC group; mean age, 75.0 ± 7.4 years), and 5 shoulders in the SSP and ISP tears (Intact SSC group; mean age, 78.4 ± 2.3 years). The demographic data for the two groups are shown in Table 1. The Institutional Review Board approved the study protocol, and all subjects gave their written informed consent before participation.
Image evaluation
T1-weighted and T2-weighted MR images were obtained (3.0-T, X-series; Philips Healthcare, Best, Netherlands). in the coronal oblique, sagittal oblique, and axial planes. The tear sizes were measured using MRI. For the SSP and ISP, we used the classification of DeOrio and Cofield.[24] A greater than 5 cm retraction in the coronal plane was defined as a massive tear. For the SSC, we used the modified Lafosse's classification[25] as follows: type I, a partial tear of the upper one-third of the SSC; type II, a complete tear of the upper one-third of the SSC; type III, a complete tear of the upper two-thirds of the SSC; and type IV, a complete tear of the entire width of the SSC. Fatty infiltration of the SSP, ISP, and SSC muscles was graded using the 5-point semiquantitative scale described originally by Goutallier et al.[10] and modified for MRI analysis by Fuchs et al.[26] as follows: 0, normal; 1, some fat streaks; 2, fatty degeneration of less than 50% but still more muscle than fat; 3, fatty degeneration of 50% (equal fat and muscle); and 4, fatty infiltration of more than 50%. Moreover, the radiologic evaluation for cuff tear arthropathy was classified into six types according to Hamada et al.[11]: Grade 1, acromiohumeral interval (AHI) ≥ 6 mm; Grade 2, AHI ≤ 5 mm; Grade 3, AHI ≤ 5 mm, with acetabulization; Grade 4A, glenohumeral arthritis, without acetabulization; Grade 4B, glenohumeral arthritis, with acetabulization; Grade 4A, humeral head collapse, which is characteristic of cuff tear arthropathy. The imaging evaluation data for the two groups are shown in Table 1.
Image acquisition and 3D modeling
Scapular plane abduction was recorded using a flat panel radiography/fluoroscopy (R/F) system (SONIALVISION Safire, Shimadzu, 0.286 × 0.286 mm/pixel) and fluoroscopic images were acquired in a single anterior-posterior direction. Patients elevated the arm in the scapular plane (30° anteriorly to the frontal plane) from a natural hanging position to a maximum elevation over 3 seconds, with the elbow joint extended while standing. The distance from the tube of the flat panel R/F system to the target shoulder was 1500 mm, and the sampling rate was 7.5 frames/second.
CT was then used to obtain 0.5 mm tomographic images of the humerus and scapula. A 3D bone model of the humerus and scapula was created from the tomographic images using segmentation software (3D-Doctor, Able Software Corp., Lexington, MA, USA). The 3D bone models were converted to a polygonal surface model, and a smoothing process was applied using a 3D mesh processing software (MeshLab; www.meshlab.net/). A single experienced researcher embedded the local coordinate system of the glenoid and humerus onto the 3D bone models using the 3D-Aligner software (GLAB Corp., Higashihiroshima, Japan). Humerus coordinates were set with their origin at the center of the humeral head, a y-axis parallel to the humeral shaft, and an x-axis passing through the center of the intertubercular groove.[27] Scapular coordinates were set with their origin at the center of the scapular glenoid cavity, a y-axis parallel with a line connecting the topmost and lowermost edges of the glenoid cavity, and a z-axis parallel to a line connecting the anterior-most and posterior-most edges of the glenoid cavity.[27]
Model-image registration
JointTrack (open-source software; www.sourceforge.net/projected/jointtrack) was used to match the completed 3D bone model with the fluoroscopic images. The 3D bone model was matched to each fluoroscopic image. Outlines in the 3D bone model were matched to outlines in the fluoroscopy images. We used the greater tubercle, lesser tubercle, humeral head, and humeral shaft as landmarks when matching the humerus. We used the acromial process, coracoid process, glenoid cavity, scapular spine, superior angle, medial margin, and inferior angle as landmarks when matching the scapula (Fig. 1).
Data processing
The 3D shoulder kinematics were obtained using the 3D-Joint Manager software (GLAB Corp., Higashihiroshima, Japan). For the 3D joint orientation, the position of the distal bone in the local coordinate system of the proximal bone was calculated by the Euler angle.[28] Humeral elevation was defined as rotation about the z-axis. Scapular motion was defined as anterior-posterior tilt about the x-axis, internal-external rotation about the y-axis, and upward-downward rotation about the z-axis. Internal-external humeral rotation relative to the scapula was defined as rotation about its y-axis. The humeral head translation (in the superior-inferior, anterior-posterior, and medial-lateral directions) was calculated as the position of the humeral head center relative to the glenoid center. All kinematics data were measured from the beginning to the end of arm elevation. We also measured translation on each axis three times and calculated the root-mean-square (RMS) error to investigate measurement error. The RMS error observed in this study was an in-plane error of 0.12 mm and an out-of-plane error of 0.61 mm, which are comparable to previous validation studies.[23]
Statistical Analysis
Image evaluation and kinematics results were compared between the intact and Torn SSC groups. The Mann–Whitney U test was used to compare age, fatty infiltration, and glenohumeral and scapular rotation angles at the beginning and end of arm elevation. Chi-square tests were used to analyze categorical data such as gender and rotator cuff tear arthropathy. The effect of the subject group (Torn SSC group and Intact SSC group) on the GH kinematics in the three directions of translation of the humeral head were analyzed using a two-factor linear mixed-effects model. When a significant interaction between the subject group and arm elevation angle was observed, posthoc Bonferroni correction was used for further significance testing. The software used for statistical processing was SPSS Statistics 24 (IBM, Japan), and the significance level used was 5%.