Tears of the rotator cuff are a substantial source of pain and disability1. While good clinical results have been reported after use of both arthroscopic and open repair techniques2, repairs of massive tears have a high rate of mechanical failure, which has been associated with inferior clinical outcomes compared with those associated with intact repairs3. This failure most commonly occurs at the suture-tendon interface4.
Several surgical solutions have been proposed to correct this problem. Newer double-row techniques are intended to prevent these failures and improve clinical results5,6. Numerous studies have shown double-row reconstructions to be biomechanically superior to traditional single-row techniques, with less gap formation under cyclic loading and higher failure loads7-9. However, several clinical trials have demonstrated equivocal results10-12. Other authors have proposed alternative suture configurations, such as the Mason-Allen repair or massive cuff stitches, to improve the biomechanics of the tendon repair site13-15. These modified techniques have not improved patient outcomes16.
The tendons of the rotator cuff are among many tendons in the body that frequently require repair. Much investigation has been performed regarding the optimal techniques for the repair of various tendons. For example, in the flexor tendons of the hand, greater numbers of sutures crossing the repair site are associated with higher failure loads, less gap formation, and better clinical outcomes17-23. Whether the same is true for rotator cuff repairs has not been investigated, to our knowledge.
Previous time-zero in vitro studies comparing single and double-row repair techniques have shown superior biomechanical properties following double-row repairs. However, to our knowledge, the double-row techniques have always used more sutures (often two or three times more) crossing the repair site than the single-row repairs7-9,14,24-27. Thus, the improved biomechanical performance of the double-row constructs may be related more to the increased number of sutures than to the addition of a second row of suture anchors.
The purposes of our study were to establish whether using greater numbers of sutures would improve the biomechanical properties of a rotator cuff repair model, and whether using equal numbers of sutures would lead to equivalent results between single and double-row configurations. We hypothesized that using more sutures crossing the repair site of a sheep infraspinatus tendon would lead to less gap formation with cyclic loading and allow the constructs to withstand higher loads to failure. We further hypothesized that single-row repairs would be similar to double-row repairs in terms of gap formation and load to failure when the same number of sutures was used in both groups.
Specimen Preparation
Nineteen matched pairs of fresh-frozen sheep shoulders were used. The sheep infraspinatus tendon is an established rotator cuff model28. Specimens were thawed at room temperature for twenty-four hours before dissection. All tissues were removed except the humerus and the infraspinatus tendon. Each infraspinatus was sharply removed from its insertion on the humerus with a number-10 blade. The distal stump of each tendon was then sharply resected back to an area of intact full-thickness tendon to simulate tendon preparation during rotator cuff repair. Normal saline solution was used to keep the tendons moist throughout the dissection and the remainder of the experiment. The infraspinatus muscle was carefully removed from the tendon to allow secure fixation to the testing apparatus. The width and thickness of the tendons were carefully measured with digital calipers to calculate the cross-sectional area.
Repair Technique
Specimen pairs (the left and right shoulders from a single animal) were assigned randomly to two groups: (1) two-suture single-row repair for one shoulder and six-suture single-row repair for the other (ten pairs) or (2) four-suture single-row repair for one shoulder and four-suture double-row repair for the other (nine pairs). This was done to reduce specimen variability for the comparison between the four-suture single-row and four-suture double-row groups. Within these two groups, left and right shoulders were assigned randomly to each of the four subgroups. The infraspinatus tendon of one specimen was damaged on arrival; this specimen had been assigned to the two-suture single-row group but could not be used, leaving only nine specimens in that group.
Two-Suture Single-Row Repair
The tendons were repaired with use of two 5.5-mm metal corkscrew suture anchors placed 10 mm apart from each other in the center of the infraspinatus footprint in the coronal plane (Fig. 1-A). For this group, the anchors were single-loaded with number-2 FiberWire (Arthrex, Naples, Florida). Two mattress sutures were placed at equal intervals 10 mm from the cut edge of the tendon. The sutures were tied with two sliding half hitches followed by alternating half hitches for a total of five throws (Fig. 2-A).
Four-Suture Single-Row Repair
The tendons were repaired with the same anchors, placed in the same positions, but the anchors were double-loaded. Four mattress sutures were placed at equal intervals 10 mm from the cut edge of the tendon. The sutures were tied with use of two sliding half hitches followed by alternating half hitches for a total of five throws (Fig. 2-B).
Four-Suture Double-Row Repair
The tendons were repaired with four single-loaded 5.5-mm metal corkscrew suture anchors. Two single-loaded anchors were placed, 10 mm apart from each other, along the articular margin of the infraspinatus footprint. Two more single-loaded anchors were placed, 10 mm apart from each other, at the lateral edge of the infraspinatus footprint directly adjacent to the medial two anchors (Fig. 1-B). The medial anchors were used to place two mattress sutures at equal intervals 10 mm from the cut edge of the tendon. The lateral two anchors were used to place two mattress sutures at regular intervals 7.5 mm from the medial row of sutures and 2.5 mm from the cut edge of the tendon. The sutures were tied with two sliding half hitches followed by alternating half hitches for a total of five throws (Fig. 2-C).
Six-Suture Single-Row Repair
The tendons were repaired with use of two anchors (Fig. 1-A), which were triple-loaded. Six mattress sutures were placed at equal intervals 10 mm from the cut edge of the tendon. Sutures were tied with use of two sliding half hitches followed by alternating half hitches for a total of five throws (Fig. 2-D).
Biomechanical Testing
A servohydraulic testing system (MTS, Eden Prairie, Minnesota) was used for biomechanical testing. The infraspinatus tendon was fixed to a custom clamp fitted with sandpaper grips to prevent tendon slippage. This clamp was then fixed to the load cell of the test system. The humerus of each specimen was potted in automotive body filler and secured to the base of the test system. Care was taken to mount the specimens so that the direction of applied tensile load recreated the normal anatomic relationship of the infraspinatus tendon to the humerus. Gap formation was measured with a linear displacement sensor (M-DVRT-6; MicroStrain, Williston, Vermont). One end of the sensor was held fixed with the humerus. The other end was placed in the tendon 2 mm proximal to the most proximal sutures (Fig. 3).
Specimens were pretensioned at 10 N for one minute. They were then cycled from 10 to 180 N for 200 cycles at 0.2 Hz. This protocol was chosen to simulate the forces placed on a rotator cuff during the early phase of rehabilitation and has been used by other investigators7,25. Gap formation across the repair site was defined as the difference between the initial distance (measured by the linear displacement sensor after pretension) and the distance after each cycle. Gap formation was recorded after cycles 1, 50, 100, 150, and 200. Following cyclic loading, specimens were loaded to failure at a rate of 1 mm/s. Failure was defined as a gap of 20 mm or complete discontinuity of the specimen. Twenty millimeters was chosen because this was the maximum displacement measurable by the linear displacement sensor.
Statistical Analysis
Repeated-measures analysis of variance (ANOVA) tests were used to compare the average gap formation among the two, four, and six-suture single-row groups after cycles 1, 50, 100, 150, and 200, and to compare the average load to failure among these groups. Bonferroni post hoc analyses were then performed for comparisons between individual groups.
The Student t test was used to compare the average gap formation between the four-suture single-row and four-suture double-row groups at each interval as well as to compare the average load to failure between these two groups.
Repeated-measures ANOVA tests were used to compare tendon cross-sectional area and width between groups.
Significance was set at p ≤ 0.05.
Source of Funding
No external funding was used for this study. Suture anchors were donated by Arthrex (Naples, Florida).
Cyclic Testing
There were significant differences among the two, four, and six-suture single-row groups at cycles 50, 100, 150, and 200 (Table I). Bonferroni post hoc analyses showed significant differences between the two-suture single-row group and both the four and six-suture single-row groups at each of these cycles, but the four and six-suture groups were not significantly different from each other at any point in the cyclic testing. Gap formation in the two-suture single-row group was 5.1 mm at cycle 1 (37% and 42% greater than that in the other groups), and it increased throughout the cyclic testing, reaching 13.6 mm (97% and 109% greater than that in the other groups) at cycle 200 (Table I).
The four-suture single-row and four-suture double-row groups showed similar gaps at each recorded cycle, with no significant differences in cyclic gap formation at any point in the testing. The greatest difference in gap formation between these groups was 0.5 mm (Table I).
Three specimens in the two-suture single-row group failed at the suture-tendon interface during cyclic testing (one at cycle 11, one at cycle 69, and one at cycle 152). All other specimens withstood cyclic testing.
Load-to-Failure Testing
The four-suture single-row specimens failed at a 32% higher average load than the two-suture single-row group. The six-suture single-row specimens failed at a 58% higher average load than the four-suture single-row group and a 109% higher load than the two-suture single-row group (Table II). A repeated-measures ANOVA test showed that these differences were significant (p < 0.0001). Bonferroni post hoc analyses showed that the load to failure of the six-suture group was significantly different from that in both the two and the four-suture single-row group but that the difference between the failure loads of the two and four-suture single-row groups did not reach significance independently.
No significant difference (p = 0.578) in the average failure load was found between the four-suture single-row and the four-suture double-row group (Table II).
In no specimen did the gap reach 20 mm at any point during testing without complete failure and discontinuity. One specimen from each of the four groups failed at the tendon-grip interface during load-to-failure testing, and these were excluded from the load-to-failure analysis. None were from the same matched pair. All other specimens failed at the suture-tendon interface, with sutures pulling through the tendon. No anchors pulled out of bone, no sutures broke, and no knots failed.
Tendon Dimensions
No significant differences in the average infraspinatus tendon cross-sectional area (p = 0.87) or width (p = 0.58) were found among the groups (Table III). Tendon width was used as a measure of tear size, as the entire width of the tendon was detached from the tuberosity. Linear regression analysis performed with Microsoft Excel (Microsoft, Redmond, Washington) demonstrated poor correlations between failure load and tendon cross-sectional area and width, with R2 values ranging from 0.03 to 0.40.
Our results demonstrate that increasing the number of sutures crossing a rotator cuff repair improves the performance in both cyclic loading and load-to-failure testing. During cyclic loading, the two-suture single-row group performed significantly worse than all other groups. The four-suture single-row, four-suture double-row, and six-suture single-row groups showed similar gap formation under cyclic loading, with no significant differences among them, suggesting that a plateau may exist after which additional sutures provide no additional benefit toward preventing cyclic gap formation. However, this cannot be definitively determined without testing a broader range of sutures.
In load-to-failure testing, as the number of sutures crossing the repair increased, the failure load increased between each of the three single-row groups. As was the case for cyclic gap formation, a plateau may exist after which additional sutures provide inconsequential additional benefit but, if so, this plateau lay outside the range of our experiment.
The four-suture single-row and four-suture double-row groups showed similar gap formation at each recorded cycle, and similar failure loads, with no significant differences between them. These findings suggest that the number of sutures crossing the repair is more important than the row configuration at time zero.
Tendon width and cross-sectional area were both measured. Tendon width in this model is analogous to tear size, as the entire tendon was detached from the humeral footprint. While the number of sutures was highly associated with higher failure loads, tendon width and cross-sectional area were not. In several cases, the larger tendons failed at much lower loads than the smaller tendons in the same group. We believe that this was the case because at time zero all load is carried only through the sutures, whereas beyond time zero, as healing begins to occur, tendon size may make a contribution to repair strength.
While many studies examining the biomechanics of rotator cuff repair have been published, the majority were investigations of suture type and row configuration. Until now, to our knowledge no study has isolated the number of sutures crossing the repair as the sole independent variable in determining the performance of rotator cuff repair. We looked specifically at the differences among two, four, and six sutures crossing the repair, with all other experimental variables controlled. We found that, as the number of sutures crossing the repair increases, so does the strength of the repair.
We also found that the biomechanics of a four-suture single-row repair are no different from those of a four-suture double-row repair in either cyclic loading or load-to-failure testing. Many studies in the current literature have shown a biomechanical advantage to double-row reconstruction, but all were performed with use of more sutures—frequently more than twice as many—in the double-row constructs7-9,14,24-27. We believe that the biomechanical advantages shown in these studies are related more to the number of sutures than to the anchor or row configuration.
In the clinical outcomes literature, most studies have shown no difference between single and double-row repairs29-33. Only one study showed any clinical difference in outcomes between these two techniques. Park et al.34 performed single-row repair in their first forty patients and double-row repair in their next thirty-eight patients. They found that, of patients with a massive tear of >3 cm, those with a double-row repair had significantly higher American Shoulder and Elbow Surgeons (ASES) and Constant scores. The authors did not report the exact numbers of sutures used in each construct but stated that, in massive tears, three anchors were used for single-row repairs and five, six, or seven anchors were used for double-row repairs. More sutures were probably used in the double-row repairs, which may have contributed to the better outcome scores in the second group of patients.
Other investigators included the number of sutures as a variable in their analysis. Cummins et al.35 used an ovine model similar to ours to study the best combination of anchors and suture techniques to repair torn rotator cuff tendons. They found that increasing the number of sutures per anchor and increasing the number of anchors (both effectively increasing the number of sutures crossing the repair) significantly increased the load to failure. They did not, however, examine the strength of the repair under cyclic loading. Barber et al.36 compared single-row constructs using triple-loaded suture anchors with various double-row constructs and found that the triple-loaded single-row repairs were more resistant to cyclic gap formation. However, the double-row repairs employed lateral push-in anchors capturing the medial row's sutures over the end of the tendon, with no additional sutures through the tendon (suture bridge/transosseous equivalent). This effectively left the double-row constructs with fewer sutures through the tendon compared with the number used in the triple-loaded single-row repairs, and this may explain the poorer performance of the double-row groups.
Aside from the proposed biomechanical advantages, double-row techniques were developed to recreate the anatomic footprint of the rotator cuff, suggesting that compressing a greater surface area of tendon to the humeral tuberosity would be beneficial for healing. Several techniques have been proposed to achieve this goal, including traditional double-row, suture-bridge, and transosseous repairs37-41. Our study offers information only on the initial biomechanical performance of a repaired rotator cuff; it does not provide information as to whether anatomic footprint reconstruction improves the biology of healing.
Our study has limitations. We used an in vitro, ovine model. Although this is an established model, its anatomy differs from that of human rotator cuff tissue. Additionally, we employed a single-tendon-tear model in specimens from young, healthy animals. Double-row repairs are proposed to have the greatest benefit in the treatment of massive, degenerative cuff tears, which were not replicated in this experiment.
We believe that, while there may be biologic benefit to anatomic footprint reconstruction, using a second row of suture anchors provides no biomechanical advantage. This second row increases cost, operative time, and the technical difficulty of the procedure. Surgeons should focus on maximizing the number of sutures crossing the tendon, thereby decreasing gap formation during early rehabilitation and increasing the ultimate strength of the repair.
In conclusion, increasing the number of sutures leads to decreased gap formation under cyclic loading and increased load to failure in an ovine rotator cuff repair model. Double-row repair offers no biomechanical advantage when the number of sutures is kept constant.