Supraspinatus tendon repair using anchors: a biomechanical evaluation in the rabbit

One hundred twelve adult female New Zealand white rabbits weighing 3.0 ± 0.3 kg were used. The experimental procedures were approved by the Institutional Animal Care Committee and Research Ethics Board (Protocol ME-2479). All rabbits had one shoulder randomly selected to undergo a two-time surgery, first a detachment of the SSP tendon, followed by reattachment surgery 1 week later. During detachment surgery, the footprint of half of the rabbits underwent channeling (channeling group) and the footprint of the remaining rabbits was untouched (no channeling group). The 7-day delay between channeling and repair was selected (1) based on preliminary histological data on rabbit footprint channeling, wherein 7 days after channeling corresponded to the greatest subenthesial cellular proliferation, and (2) to replicate an ongoing clinical trial investigating channeling 7 days prior to repair (CTI: NCT01706978, clinicaltrials.​gov). One hundred twelve contralateral shoulders served as controls. All rabbits were housed individually at 22 °C on a 12-h light/dark cycle with access to water and standard chow and allowed weight bearing after surgery. Rabbits were euthanized at 0, 1, 2, or 4 weeks following reattachment surgery in groups of n = 16 (Fig. 1). The 4-week maximum postoperative follow-up was based on a previous study showing complete restoration of rabbit SSP mechanical strength and stiffness between weeks 2 and 6 after repair [10]. Sample size was calculated based on previous experiments: detection of a difference in the primary outcome measure (load at failure) of 33% with a power = 0.80 and alpha = 0.05 required 15 rabbits per group [10]. Allowing n = 1 for technical failures resulted in a sample size of n = 16.

Detachment surgery consisted of sharply detaching the SSP insertion from the greater tuberosity, simulating a complete tear, as detailed in a previous investigation [22]. The distal, free end of the tendon was wrapped in a polyvinylidene membrane (5 μm, Durapore; Millipore, Bedford, MA) to prevent spontaneous reattachment. Following detachment, half of the rabbits underwent bone channeling. Channeling involved partitioning the SSP humeral footprint into four quadrants and drilling a 1-mm-diameter hole at the center of each quadrant. The holes were drilled to a depth of approximately 10 mm to ensure deep communication with the red bone marrow of the epiphysis. The other half of the operated animals was not subjected to channeling. The deltoid was then closed followed by skin closure. The rabbits received fentanyl and buprenorphine for 3 days postoperatively and were allowed to roam freely in cages with unlimited access to food and water.

Reattachment surgery was performed 1 week later and was the same for the channeling and no channeling groups. The incision was reopened. A curette was used to clear any tissue or fluid that had accumulated at the drill sites and shallowly decorticate the footprint. The free distal SSP tendon stump was mobilized and the Millipore wrapping removed. A single 3-mm Bio-FASTak® anchor with #2 FiberWire sutures (Arthrex, Naples, FL) was inserted lateral and distal to the footprint in the cortical bone, and the tendon was reapproximated to the footprint using a horizontal mattress stitch as described by Boileau et al. [23, 24]. The wound was closed as after the first surgery.

All rabbits were euthanized with a pentobarbital overdose at the specified times. Both the operated and contralateral shoulders were harvested from each animal and individually wrapped in saline-soaked gauze to avoid dehydration. The scapula, rotator cuff muscles, and proximal humerus were harvested en bloc and immediately frozen at − 20 °C.

The specimens were thawed gradually to room temperature. The SSP and proximal humerus were isolated to ensure that only the SSP attached to the humeral head contributed to the mechanical evaluation. The humerus was fixed in a bone clamp which compressed the humeral head to prevent premature tensile failure of the bone. The bone clamp was then potted in an adaptor cup using a low-melting bismuth alloy. The SSP muscle was clamped in a cryogenic fixation unit (CFU) that uniformly transfers load to the tendon by external freezing of the muscle using liquid nitrogen of the myotendinous junction embedded in a saline solution [25]. The humeral head to myotendinous junction test length was standardized to 24 mm.

Both fixtures were mounted on an electro-mechanical load frame with a 2.5-kN load cell (MTS Sintech 1G; MTS Systems Corporation, Eden Prairie, MN, USA) with the CFU attached to the crosshead and the bone clamp attached to the base (Fig. 2). Petroleum jelly was applied to the exposed tendon to prevent drying, and a heater surrounded the tendon to ensure that the enthesis and tendon test length remained at room temperature. The tendons were positioned along their anatomic direction of pull, at an angle of 45° to the longitudinal axis of the humeral shaft. The specimens were preconditioned for 12 cycles from a preload of 5 N to a peak load of 50 N at a loading rate of 15 N/s. This was followed by a tensile load to failure at a rate of 1 mm/s where a 50% drop in tensile strength was defined as the breaking point. The load and displacement data were collected, and the mode of failure was noted. The load at failure was determined using TestWorks 4 software (MTS Systems Corporation, Eden Prairie, MN, USA). Stiffness was calculated by fitting a regression line to the linear portion of the load–displacement curve for each specimen. We defined three sites of failure: footprint failure, bone avulsion, and mid-substance tendon tear (Fig. 2).

Load at failure increased significantly with increased durations of healing (0, 1, 2, or 4 weeks; F(3,198) = 32.1, p < 0.001; Fig. 3a). Load at failure was also significantly different between the three shoulder groups (channeling, no channeling, contralateral; F(2,198) = 21.3, p < 0.001; Fig. 3a). Finally, there was a significant interaction between shoulder groups and duration of healing (F(5,198) = 7.22, p < 0.001). Post hoc testing revealed that loads at failure were significantly lower in repaired shoulder groups (both channeling and no channeling) than in the contralateral shoulders at weeks 0, 1, and 2 (all p < 0.01; Fig. 3a). Both repaired shoulder groups had reached contralateral loads at failure by week 4 (p > 0.05). There was no significant difference in load at failure with channeling compared to no channeling at any time point (all p > 0.05; Fig. 3a).

Similarly, stiffness increased significantly with increased durations of healing (F(3,198) = 29.4, p < 0.001; Fig. 3b) and was significantly different between the shoulder groups (F(2,198) = 65.9, p < 0.001; Fig. 3b). There was a significant interaction between shoulder groups and duration of healing (F(5,198) = 7.96, p < 0.001). Stiffness was significantly lower in both repaired shoulder groups than in the contralateral at weeks 0, 1, and 2 (p < 0.01; Fig. 3b). Both repaired shoulder groups had reached contralateral stiffness by week 4 (p > 0.05). There was no significant difference in stiffness with channeling compared to no channeling at any time point (all p > 0.05; Fig. 3b).

All contralateral shoulders across all time points failed through mid-substance tendon tear near the myotendinous junction (Fig. 4). The site of failure for the two repair shoulder groups combined, varied with duration of healing (p < 0.001). At week 0, all repaired tendons failed at the footprint due to suture pullout. The sites of failure shifted from a mix of a rupture at the footprint and bone avulsion at 1 and 2 weeks toward bone avulsion and mid-substance failures by 4 weeks (p < 0.05; Fig. 4). Failure at the footprint was associated with the lowest load at failure, bone avulsions were associated with higher loads at failure, and mid-substance tears were associated with the highest loads at failure (p < 0.05; Fig. 4). There were no statistically significant differences in site of failure between channeling and no channeling groups across all time points (p > 0.05).