Testing apparatus
A validated robotic system was used to generate automated motion trajectories for a cadaveric torso [5, 6]. The system consists of a lower frame holding the cadaveric torso and an upper frame to which the upper limb is attached (Fig. 1a and b). The lower (torso) frame can generate three translational degrees of freedom (DOF) along the x, y, and z-axes and one rotational DOF around the z-axis, while the upper (limb) frame has three translational DOFs along the x, y, and z-axes. Any desired motion trajectory within the range of the system is generated by linear and rotary closed loop actuators and controlled via a programmable central controller with high reproducibility and accuracy [5, 6].
Cadaveric torsos and surgical procedures
Six shoulders from three fresh-frozen human cadavers were acquired (Medcure, Inc., Portland, OR, USA). The cadaveric torsos originated from three Caucasian males with an age of 55 ± 4 years, height of 190 ± 4 cm, and body mass index of 27.1 ± 1.85 kg/m2. The range of motion and laxity of the each shoulder was tested prior to testing procedures to see if the motion was normal without any reduction during the arc of abduction. Specimens were allowed to thaw at room temperature and tested immediately thereafter. The experimental protocol was performed sequentially, allowing each specimen to serve as its own control. Four conditions were compared:
-
1.
Intact-A baseline for the data was established using the intact specimen.
-
2.
Opened-The RI was cut, separating the leading edge of the supraspinatus from the superior edge of the subscapularis. The incision extended from the base of the coracoid to the humeral head (approximately 2–3 cm) in line with fibers of the rotator cuff.
-
3.
Repaired-The RI was repaired with a running suture (Ethibond #1, Ethicon, Somerville, NJ, USA) to create no overlap or tightening. The arm was abducted, in a resting position, in twenty degrees of external rotation to prevent overtightening.
-
4.
Tightened-the running suture was removed. The arm was allowed to rest in the neutral position. A new suture was then used to tighten the rotator interval using a horizontal mattress technique (Ethibond #1, Ethicon, Somerville, NJ, USA). The two passes were made with the needle separated by one centimeter. The needle entered one edge of the RI one centimeter from the cut edge and exited the other side one centimeter from the cut edge. The second pass was made in the opposite direction and the two suture limbs were tied, “tightening” the RI by two centimeters.
-
5.
The running suture was removed and the RI was re-approximated with a horizontal mattress suture. The suture needle was passed one centimeter from the free edge of the tissue edge to tighten the RI by 2 cm.
Simulation of abduction motion
Torsos were mounted on a rod fixture (Fig. 1a and b) and held in place with expanding foam [6, 7]. The hand was disarticulated at the distal radioulnar joint, and the arm was secured directly to the upper frame using a Schanz. The skin and the deltoid muscle were removed. Passive retro-reflective marker clusters were placed in the humeral shaft, the posterolateral acromion, and the sternum [6, 7]. To protect the specimens, testing was performed at a reduced speed (duration of motion, 28.6 s), in accordance with previous studies [6, 8]. The arm was raised in the coronal plane from 30 to 150° of abduction for three repetitions.
Motion analysis
Five Qualisys Pro Reflex (Qualisys AB, Göteborg, Sweden) high-speed cameras (120 Hz) were used to record the motion of the passive retro-reflective bone-embedded marker clusters. The clusters were placed into the humeral shaft, the sternum, and the acromion (Fig. 2). Before testing, the cameras were subject to multi-aspect calibration [9]. Anatomic scapular, humeral, and thoracic landmarks were calibrated with respect to these technical (bone-embedded) markers, as defined by the International Society of Biomechanics (AC joint (AC), the posterolateral edge of the acromion (AA), the coracoid process (PC), the inferior angle of the scapula (AI), the root of the spine of the scapula (TS), the spinous process of the seventh cervical vertebra (C7) and eighth thoracic vertebra (T8), the xiphoid process, the suprasternal notch (IJ), and the medial and lateral epicondyles (EM and EL)) (Fig. 3) [9]. The calibrated scapular and humeral landmarks were analyzed per Meskers et al. to determine the instant center of rotation of the GH joint within the scapular reference system [10]. For all displacements, the x-axis, y-axis, and z-axis correspond to anterior-posterior (AP; coronal), superior-inferior (SI; sagittal), and medial-lateral (M; transverse) planes, respectively. Displacements were quantified in mm.
In this system, the z-axis is a line connecting the TS and AA points; the x-axis originates from the AA point and is perpendicular to the plane formed by the AI, the AA, and the TS points; and the y-axis is the common line perpendicular to the x- and z-axes.
Statistical analysis
GH translation was recorded continuously throughout the abduction motion from 30 to 150°. For each condition, the average of three repetitions of GH translation was plotted over time to calculate the total translation and the area under the curve (AUC) for each motion segment. Absolute GH translation was calculated for Baseline, Opened, Repaired and Tightened conditions. A mixed model analysis of variance (ANOVA) was used to compare GH translations for each condition on each axis. AUC was calculated using the trapezoidal rule to appropriately assess the path-dependent motion (MATLAB version 12, MathWorks, Natick, MA, USA). The Wilcoxon signed–rank test was used to compare the AUC.
Six specimens allowed for the detection of a difference of greater than 1.0 mm in GH translation and 85 % power to detect mean differences of greater than 1.2 mm translation using ANOVA with a compound symmetry correlation structure to handle the paired specimens.
Cadaveric studies do not require Institutional review Board consideration or approval at our institution. Statistical analysis was conducted using SPSS (version 21.0, IBM-SPSS, Armonk, NY, USA). Two-tailed p-values less than 0.05 were considered significant.