Footprint coverage comparison between knotted and knotless techniques in a single-row rotator cuff repair: biomechanical analysis

A total of 24 fresh frozen human shoulders with a mean age of 67.2 years ranging from 50 to70 years, including 12 female and 12 male specimens were tested. Every shoulder was checked for macroscopic evidence of rotator cuff pathology and significant osteoarthritis, and exchanged if present. After thawed out for 24 h at room temperature, a bone mineral density evaluation via micro-CT (Lunar DXE, Madison, WI) was taken to assess bone quality at the site of suture anchor insertion at the greater tuberosity in a consistent manner of a 1x1cm area. All specimens were prepared by removing all superficial soft tissues followed by disarticulation at the glenohumeral joint to isolate the supraspinatus muscle and its tendinous insertion for better access and visualization. The individual muscles of the rotator cuff were dissected free from the joint capsule. The humeri were cut 15 cm distal from the greater tuberosity and potted in a 1.5-in.-diameter polyvinyl chloride pipe with plaster of Paris. The rotator cuff tendon was sharply dissected from its insertion on the greater tuberosity and was left in continuity with the individualized supraspinatus muscle. The bony footprint was then marked and measured using a MicroScribe digitizer (Immersion, San Jose, CA) to trace the outline of the attachment [3, 15]. The proximal end of the supraspinatus muscle was sutured to a polyester tape, using No. 5 non-absorbable suture (FiberWire, Arthrex Inc., Naples, FL) with an interlocking whipstitch in order to apply various tendon loads [3, 16]. A physiological saline solution was used to keep specimens moist during all phases of dissection, preparation, and testing.

Prior to the repair, the footprint area was macroscopically cleaned and prepared. The shoulder specimens were randomly divided into one of the three repair groups. Eight specimens were assigned for each repair technique. Utilizing a pen and a digital caliper, the distance from the lateral tendon edge was kept equivalent for all suture passages through the tendon during the repair in all specimens. Every repair construct consisted of two anchors performed by a single surgeon (JSF).

Starting 5 mm posterior to the bicipital groove, two holes were punched as far lateral as possible while still remaining on top of the footprint, thus maximizing the potential tendon contact area on the bony insertion. Each anchor was placed close to the lateral edge, 15.0 mm apart in the anterior to posterior direction. Two double-loaded anchors (4.5-mm Bio-Corkscrew FT® Arthrex Inc., Naples, FL) were used, and they were double loaded with #2 non-absorbable suture (FiberWire®). Simple suture configurations were used for this Knotted technique. The suture was passed 10.0 mm apart from one another for each given anchor, and 10.0 mm medial to the lateral edge of the simulated tear, with 5.0 mm separating both repair systems. All knots were tied with a Samsung Medical Center (SMC) sliding knot [17] followed by alternating 3 simple half-hitches, for a total of 4 throws using a knot-pusher (Fig. 1a).

Two inverted mattress stitches with a 2 mm suture tape (Fibertape®; Arthrex Inc) sutures were passed 10.0 mm apart from one another and 10.0 mm medial to the lateral edge of the simulated tear, with 5.0 mm separating both repair systems. Starting 5 mm posterior to the bicipital groove, two lateral holes were made close to the lateral edge of the greater tuberosity and centered 15.0 mm apart. Two knotless anchors (4.75-mm Bio-Composite SwiveLock®; Arthrex) were used to fix the tape down into the tuberosity holes, after the surgeon tensioned it laterally over the tendon edge (Fig. 1b).

Two inverted mattress sutures with 2-mm suture tape (Fibertape®; Arthrex Inc) were passed 10.0 mm medial to the lateral edge of the simulated tear, approximately 10.0 mm apart. After that, a cinch suture (FiberLink®; Arthrex) was passed just medial to the inverted mattress suture tape. This step with the cinch suture was repeated in same fashion for the second anchor. Two lateral holes were punched in the same manner as for the Groups A and B. After the surgeon tensioned it laterally over the tendon edge, two knotless anchors (4.75-mm Bio-Composite SwiveLock®; Arthrex) were used to fix the suture tape and the cinch suture down into the tuberosity holes (Fig. 1c).

Pressurized contact area and contact pressure were measured using a Tekscan model 4205 sensor (Tekscan Inc., South Boston, MA). Tekscan sensors have the ability to continuously collect data points in real time. The matrix dimensions for the sensor we used was 41.9 mm by 45.7 mm. This working area surpasses the average footprint area, thereby allowing us to incorporate the sensor into the repair without losing sensitivity. The reliability and accuracy of the Tekscan sensors on a curved surface, such as the greater tuberosity, were ensured in previously published studies [3, 18]. Before every test, the sensor was precalibrated to a force and pressure consistent with previous rotator cuff repair studies [3, 7]. To incorporate the sensor, in all groups, two 4.0 mm holes were created by use of a sharpened leather punch into the Tekscan sensor, 12.0 mm apart on the lateral side. Those holes were made in order to introduce the anchors and sutures through the sensor to allow for interposition between the footprint and the cuff during all repairs. The sensors were sealed between 2 layers of clear tape to prevent moisture from entering and causing delamination of the sensor. During the repairs, the Tekscan sensor was placed between the supraspinatus tendon and the greater tuberosity footprint.

After the tendon was finally fixed, we determined the footprint area coverage by compressing the outline of the reconstructed area on the sensor. This marked the region of interest for further analysis. All contact variables were measured at 0° and 30° of abduction with 0 N, 30 N and 50 N of supraspinatus load, respectively.

Biomechanical setup and testing was performed in accordance with previously published protocols [3, 19]. The potted cadaveric specimens were then mounted on a custom shoulder platform and inserted into the mechanical testing system (MTS, Eden Prairie, MN). The humeri were centered in a cylinder attached to a jig that allowed free motion in the X-Y-Z planes to adjust for rotation and abduction angles. A goniometer was used to ensure neutral humeral rotation and 0° of abduction. After fixation within the MTS machine, the humeri were locked in neutral rotation by aligning the supraspinatus footprint dimension (anterior-to-posterior) perpendicular to the loading vector [20]. The loading vector line of action was defined by a No. 5 FiberWire® whip-stitched interlocking suture and a polyester tape aligned with the center of the muscle and parallel to the baseplate. The straps were locked to the actuator of the system to transfer loads. (Fig. 2) To evaluate the effect of increasing supraspinatus loads onto the footprint, contact pressure and contact area measurements were recorded at time zero after repair and with the following supraspinatus loads: 0 N, 30 N and 50 N. Measurements were taken in neutral humeral rotation at 0° and 30° of glenohumeral abduction for all repairs. For the unloaded measurements, a 10-s time frame was recorded directly after the repair. In order to capture the values for the loaded state, each specimen was tested following the listed protocol. First, preloading with a constant load of 30 N for 5 min was performed, to act as a pre-condition. This was followed by a force-controlled ramped loading up to 50 N and held for 30 s, recording contact pressure and area. Then a step down to 30 N for more 30 s recording contact pressure and area was carried out followed by unloading. Altogether, this simulates a physiological load that may be experienced postoperatively, as has been described in previous studies [21, 22]. The relative low load applied (50 N as a maximal simulation) was used as it best simulates the in vivo early postoperative load situation.

There is no predicate for determining a relevant difference in footprint coverage. A difference in coverage of 10% between the repaired groups with an assumed standard deviation of 5 to 6% equates to an effect size of 1.75. A sample of 8 shoulders per group provides 80% power to detect a 10% difference in footprint coverage at an alpha level of 0.05. A Monte Carlo simulation of 1000 ANOVAs with an estimated 5.5% standard deviation among the groups resulted in 87% power to detect a 10% difference in coverage.

Descriptive statistics to characterize the study groups were calculated using mean and standard deviation. Differences between the groups were analyzed with a one-way ANOVA. When statistically significant, pairwise differences between the repair groups were analyzed with independent t-tests along with a Bonferroni adjustment. The alpha level for all analysis was set at 0.05. All statistical analyses were performed using Stata 12 (StataCorp. 2011. Stata Statistical Software: Release 12. College Station, TX: StataCorp LP).

There are several limitations to this study. This is a cadaveric study with macroscopically intact supraspinatus tendons, individualized and isolated detached from the greater tuberosity, may not mimic chronic and degenerative changes as expected in vivo. The use of the Tekscan sensor is susceptible to artifacts because of crinkling as well as edge effects. Ideally, it should be calibrated using surface topography places on a flat surface. All repairs should be performed tension-free, which may not be the case in clinical practice. In addition, none of the repairs were loaded to failure or cyclic loaded. Although both contact pressure and contact area may be influenced by cyclic loading and load to failure test, they have not been performed because of the sandwiched Tekscan sensor interposed between the bone and the tendon limiting the biomechanical rational.