Rotator cuff tears are one of the most common degenerative diseases of the upper extremity and are seen in up to 30% of the population aged older than 65 years.1, 2 One can assume that the number of rotator cuff tears will continue to increase as the population grows older. Along with conservative treatment options, surgical procedures, ranging from open repair to arthroscopic rotator cuff repair techniques focused on increasing strength at the repair site in order to prevent re-rupture, have been further developed and evolved over the two decades.3, 4 These techniques demonstrate good short-term results in pain reduction and functional outcome.5, 6 However, several studies show that a high percentage of rotator cuff tears do not heal.5–11
Because the healing process is influenced by a number of factors, which to date have been reduced to the biomechanics of suture techniques, research now puts more attention on the biological aspects of tendon healing to find ways of biologic augmentation to improve tendon-to-bone healing after rotator cuff repair.2, 12, 13
Some recent studies suggest that one reason for the long-term failure of rotator cuff repair might be the preload of the tendon-bone-interface which is particularly increased in old and/or large tears.14–16 Gimbel et al.17 demonstrated in a chronic rotator cuff model in rats that the tension needed to reattach the supraspinatus tendon to the footprint increases with time after detachment and is positively correlated with failure. Davidson14 concluded, after evaluating repair tension in 67 patients, that high-tension repairs are associated with poor subjective and objective outcomes. Lowering repair tension surgically however is limited in releasing adhesions and is subject to restrictions.
Botulinum neurotoxin A (BoNtA), one of seven neurotoxin subtypes produced by the bacteria Clostridium botulinum, acts by binding presynaptically to high-affinity recognition sites on the cholinergic nerve terminals. It thereby decreases the release of acetylcholine causing inhibition of neurotransmission resulting in muscle paralysis. BoNtA has the potential to temporarily paralyze the rotator cuff muscle decreasing muscle tension and therefore lowering the tension at the tendon-to-bone healing site. To our knowledge, Tuzuner et al.18 were the first to introduce BoNta as a neoadjuvant in surgery as they induced forearm flexor paralysis in children less than 6 years old with zone 2 flexor tendon repairs. Tuzuner concluded that, due to a reduction in spontaneous activity of the fingers, a better rehabilitation was possible. Ma et al.19 tried to protect a sutured Achilles tendon in an rat model with the use of BoNtA and found significantly better results in terms of rupture force 3 weeks after surgery in the BoNtA-groups. A study published by Galatz et al.20 also showed an increase of mechanical properties in tendon healing after paralyzing the supraspinatus muscle. However, histologically the BoNtA-groups had inferior properties compared to the non-BoNtA control groups. In contrast, Hettrich et al.21 found no difference in mechanical properties but an increase in collagen fiber organization.
These three studies had two major flaws:
(a)BoNtA was always injected intraoperatively even though it unfolds its full effect after 1 week.22 During that first week, one cannot assume a proper bioprotection of the healing site.
(b)In all three studies, the tendons were reattached right away, simulating a fresh lesion with normal repair load.
The hypothesis of the present study is that with a pre-operative injection of BoNtA, consequently lowering the tension at the tendon-to-bone healing site, structural improvements, such as a higher level of collagen 1 or a fiber organization, can be promoted. The attachment strength at the healing tendon-to-bone site should therefore be improved. The specific aim of the study was to compare rotator cuff lesions with normal and an increased repair load. Furthermore, we investigated if immobilization after rotator cuff repair has superior effects, for example, less inflamed tissue, a better structural fiber organization or a higher ultimate load at failure, to free range of motion.
Study Design and Experimental Groups
As prior recommended by Soslowsky23 and Derwin et al.24 we used a rat model because of the anatomical similarities. After obtaining approval from the Institutional Animal Care and Use Committee (# 55.2-1-54-2531-194-09), Sprague–Dawley rats were obtained with a mean weight of 250 g. There was no difference in weight at the time of operation between the groups.
In order to correctly infiltrate the supraspinatus muscle in a blinded fashion, six animals were harvested in a pre-study. Colored saline solution was injected into the supraspinatus muscle by the author (AF). Success of the infiltration was then evaluated by exposure of the muscle.
All other 160 rats were randomly assigned to either the BoNtA or the control group (80 animals per group). Subgroups were established to evaluate possible differences between normal and increased load repairs and post-operative immobilization. Figure 1 demonstrates all groups. Post-operatively, the rats were housed in pairs for 2 days and afterwards in groups of four. The animals were killed at 8 weeks and the tissues were randomly assigned to histologic or biomechanical analysis. Specimens that were used in biomechanical analysis were damaged/destroyed and could not be investigated histologically.
To evaluate muscle recovery from atrophy, we also infiltrated another 12 rats unilateral (right shoulder) with BoNtA and saline solution (left shoulder), performed no operation and harvested three rats each at 4, 8, 16, and 32 weeks.
One week prior to surgery, all animals underwent infiltration of the supraspinatus muscle which was done only by AF. Eighty animals were assigned to the BoNtA group and were infiltrated with 6 U/kg botulinum neurotoxin A (Allergan Pharmaceuticals, Ettlingen, Germany).21, 22 The other 80 animals were infiltrated with saline solution in a volume equal to the BoNtA group (0.05 ml saline solution/100 g). The rats were anesthetized with isoflurane followed by an injection of 60 mg/kg ketamine and 9 mg/kg xylazine administered intramuscularly. Oxygen was given via nose-cone. All operations were performed by sterile technique with the rat in the lateral position. A deltoid-splitting incision was made and the acromioclavicular joint was divided to visualize the rotator cuff. The supraspinatus tendon was identified and sharply detached off the greater tuberosity. All supraspinatus muscles that were infiltrated with BoNtA showed a distinct atrophy. There was no such atrophy in the saline group. In the group where a lesion with increased repair load was simulated, 2 mm of the supraspinatus tendon was resected25 prior to tendon refixation. For tendon refixation a 5-0 double armed Prolene suture (Ethicon, Somerville, NJ) was passed through the supraspinatus tendon via a Mason–Allen suture. Zero point 5 mm bone tunnels were made 2 mm from the articular surface on the greater tuberosity. The suture ends were passed through the bone tunnels and tied over the humeral cortex, reconnecting the supraspinatus to its footprint on the greater tuberosity. The deltoid split was closed with 3-0 Ethibond, and the skin was closed with 3-0 Vicryl subcutaneous suture (Ethicon). Rat shoulders which were assigned to an immobilization group were put in casts for 219 days using a technique described previously by Gimbel et al.26 In short, immobilization was achieved by molding plaster around one shoulder including the elbow and the chest. The left arm was left free allowing the rat to ambulate and feed (see Supplementary Material).
Post-operative animal care was administered by the authors of this article (A.F., T.-K.H., M.S.). Animals were monitored according to the guidelines provided by the Institutional Animal Care and Use Committee for discomfort, distress and pain. Rats which exhibited signs of distress, such as dyspnea or decrease in body weight, were administered buprenorphine (0.01–0.05 mg/kg) subcutaneously every 12 h.
A total of 48 specimens underwent histologic analysis, six from each group.20, 21 The specimens were fixed in 4% neutral-buffered formalin (Micros GmbH, Garching, Germany) for 48 h. After fixation the tissues were decalcified in EDTA-4Na 20% citric acid (pharmacy of the hospital) for 14 days. Dehydration and embedding of the preparations in paraffin was fully automated by a tissue processor (Hypercenter XP, Thermo Scientific Fisher, Schwerte, Germany). Three-micrometer-thick sections were cut in the coronal plain and then stained with hematoxylin-eosin and Safranin-O. Furthermore immunohistochemical staining of Collagen 1 (Novus Biologicals, Littleton, CO), Collagen 2 (CIIC2, Development Studies Hybrid Iowa, IA) and Collagen 3 (Novus Biologicals) was made. Three slides were made for each specimen. The specimens were evaluated by three investigators (S.M., A.F., M.S.) who were blinded with regard to specimen group.
Images were created with a Zeiss Axioskop 40 light microscope (Zeiss, Göttingen, Germany) that was interfaced to a Zeiss Axio Cam MRc5 video camera (Zeiss) mounted on an eyepiece tube. For all specimens, the images were taken under identical conditions of magnification and illumination.
The organization of collagen tissue, vascularity at the tendon-bone interface, fibrocartilage at the tendon-bone interface, inflammation, tendon diameter, and collagen fiber continuity between tendon and bone tissue were evaluated. Furthermore, an area measurement based on immunhistochemical staining was used to show the relationship between collagens 1 and 32, 27 and the size of the fiber cartilage at the tendon-bone interface (Zeiss Visio Release 4.5).
Biomechanical testing was performed by the authors (T.-K.H., C.S., A.F.) who were blinded with regard to specimen group. Rat shoulders were kept in an −80°C freezer until biomechanical testing was performed. The specimens were then thawed at room temperature, and the humerus, with attached supraspinatus tendon and muscle, was carefully dissected from the scapula and surrounding tissues. The supraspinatus muscle was then stripped from the supraspinatus tendon and set aside for determination of weight. The cross-sectional area of the supraspinatus tendon was measured with a micrometer in the midsubstance of the tendon.
The humerus was embedded in a custom-made cylinder using polymethylmethacrylat. The specimen was then placed into the Zwick Universal Testing Machine (model Z010/TN2A, Zwick, Ulm, Germany). Force was measured with a transducer (model HBM Z6FD1, Zwick) with a measurement range up to 100 N and an uncertainty of measurement of 0.2%. Measurements and data evaluation were recorded with testXpert V5.0 (Zwick). The end of the tendon was secured in between two aluminium clamps using sandpaper and cyanoacrylat (Pattex Ultra Gel; Henkel, Düsseldorf, Germany).26 Testing was performed with a simulated 90° of shoulder abduction.20
To define a standard “zero load”, a load state where all samples begin from, a preload to 0.2 N was performed. The preload was then followed with five cycles of preconditioning at 5% grip-to-grip strain at a rate of 0.1%/s to define a consistent load history for each sample. A constant strain rate experiment consisting of a constant strain rate test to failure was afterwards performed.20
For biomechanical testing, group size was calculated with 14 animals using an a priori power analysis. Groups were compared using non-parametric Mann–Whitney U-test (GraphPad Prism version 5.0d for Mac, GraphPad Software, San Diego, CA). The level of significance was set at p < 0.05. Histology-based results were qualitative in nature and were not statistically compared except for the collagens 1–3 quotient. Six animals per group were chosen according to several other publications with similar histologic investigation.20, 21, 28
Two animals died post-operatively because of anesthesia-related problems. There were no other post-operative complications. Rats were not restricted in normal gait or food intake (i.e., because of pain or cast immobilization). Those rats whose right shoulder was immobilized for 2 days quickly returned to a normal gait and resumed foreleg function after the cast was removed. We did not find any differences in body weight between the groups after 8 weeks at the time of shoulder harvest.
Muscle Weight and Tendon Diameter
After 8 weeks, we found a significant difference in supraspinatus muscle weight between the BoNtA and saline groups (p = 0.0286). This effect was reversible after 16 weeks. Data can be seen in Figure 2. A significant difference (p = 0.0036) in tendon diameter was found between the BoNtA and saline group (BoNta 0.32 cm ± 0.1 vs. NaCl 0.27 cm ± 0.17) as well as when compared with the untreated contralateral shoulder (p = 0.0015; 0.19 cm ± 0.04; Fig. 3).
After 8 weeks, specimens from the BoNtA-group, which were treated with an increased repair load, showed less cellularity, less vascularity and more organization in the interface tissue than compared to the saline control group. In addition, we found that the collagens 1–3 quotient in the BoNtA specimen was significantly (p = 0.0051) higher than in the control group (Fig. 4). The area of fibrocartilage was significantly higher in the BoNtA treated specimen.
On the other hand, specimens that were treated with a normal repair load, did not show as many differences. Cellularity was slightly higher in the BoNtA groups and the organizational properties did not differ from the saline control group. A tidemark, indicating better-organized interface tissue,21 was present in both the BoNtA and saline groups but was not as clear as within the increased repair load groups. The collagens 1–3 quotient did not significantly (p > 0.05) differ between the BoNtA and the saline control group (Fig. 5). The area of fibrocartilage was also not significantly different from the control group. Representative histologic images can be seen in Figures 6–9.
We did not see any significant differences between the mobilized (no cast) and immobilized (cast) specimen.
After 8 weeks, a total of 90 specimens were available for biomechanical testing. Three of 45 specimens in the BoNtA- and 3 of 45 specimens of the saline-groups failed due to breakage of the humeral head. The rest of the specimens failed at the tendon to bone interface. The load to failure was 4.6 ± 2.0 N in the BoNtA-group with an increased repair load with free mobilization and 6.6 ± 3.3 N in the saline control group (p = 0.1007). For specimens immobilized for 2 days in a cast, the results were 5.0 ± 2.3 N in the BoNtA and 5.0 ± 3.2 N in the saline group (p = 0.7350). There was no significant difference between the groups as seen in Figure 10.
Ultimate load failure in the groups with normal repair tension was 5.3 ± 2.3 N (BoNtA, no cast) and 6.5 ± 2.8 N (saline, no cast; p = 0.2687). For groups which had been in a cast for 2 days, the ultimate load at failure was 5.5 ± 2.6 N (BoNtA, cast) and 7.1 ± 2.5 N (saline, cast) respectively (p = 0.0980). This data demonstrates no significant difference between the two groups. Results are shown in Figure 11.
The results presented in this study support our hypothesis that botulinum neurotoxin A (BoNtA) can lead to an improvement of structural properties at the tendon-to-bone interface in rotator cuff tears with an increased load tension at the repair site. These lesions are often old and/or large rotator cuff tears. However, we were not able to demonstrate superior attachment strength in this group. Furthermore, we did not find any significant difference between the mobilized and immobilized groups.
As expected, we found a rapid decrease of muscle mass in all supraspinatus muscles that were infiltrated with BoNtA when compared to the saline injected groups. This effect was fully reversible after 16 weeks. We did not find any persistent changes in muscle morphology, fatty infiltration or changes in tendon quality. Similar results were presented by Barton et al.29 who induced muscle atrophy by detaching rat supraspinatus tendons without repair. Barton found that the rat supraspinatus muscle can reverse atrophic muscle changes with loads as little as those induced via adhesions.
Our results contrast with a study by Galatz et al.20 who examined post-operative regimens (cast immobilization or free range of motion) after supraspinatus tendon repair using botulinum neurotoxin A. Animals were killed after 3 or 8 weeks for biomechanical testing and 1, 2, 3, or 8 weeks for histological analysis. Galatz found that complete removal of load is detrimental to rotator cuff healing regarding both histological and biomechanical properties. The effects were even stronger when BoNtA and cast immobilization were combined. Contrary, Gimbel et al.26 showed that a decreased activity level due to shoulder immobilization alone would lead to positive effects on biomechanical and compositional properties.
From our own experience and according to Thomopoulos,30 we can tell that cast immobilization in a rat model only prevents gross movement of the shoulder joint thereby protecting the healing site from excessive load but still allowing micro-motions. Apparently, combining both methods, cast immobilization and muscle paralysis via BoNtA, was too much for the investigated lesions with normal repair load. Maybe that is why we only found significant differences between the normal and increased repair load groups. We therefore conclude that a certain level of tendon tension at the tendon-to-bone healing site is required for superior healing properties. The exact amount of residual tendon tension after BoNtA infiltration and tendon repair will have to be investigated.
We think that we did not find any differences between the immobilized and mobilized groups because the 2 day immobilization was too short. In contrast to our study, and the study conducted by Ma et al.,19 Gimbel et al.,26 and Galatz et al.20 performed an immobilization up to 8 weeks. Although Gimbel and Galatz did not report about major problems immobilizing rat shoulders for 8 weeks we chose not to handicap our animals for that long period, maybe compromising our results.
The results from Hettrich et al.21 who found an improved collagen fiber organization, support our findings. The fact that Hettrich had these improvements after 4 but not after 8 weeks might be explainable due to the missing BoNtA protection during that first week until onset (Hettrich infiltrated BoNtA intraoperatively). This has led us to the pre-operative BoNtA infiltration in order to have the BoNtA protection beginning at the start of the repair. On the other hand, Hettrich, as in all other comparable studies, did not use a rotator cuff tear model to simulate lesions with increased repair load as in old, retracted tears. Therefore, we included a rotator cuff model with increased repair load as previously described by Gimbel.17 As our data demonstrates, it is this group of rotator cuff tears that might benefit the most from a pre-operative injection of botulinum neurotoxin A. We assume that the remaining tendon tension at the tendon-to-bone interface sustains a minimal load promoting intrinsic molecular signals.20, 29 Still it is unclear why we found improved structural properties without any differences in tendon attachment strength. A possible explanation is that healing in the rat model occurs no matter what surgical techniques have been performed. Therefore the possible positive effects of tendon paralysis may be outweighed by the intrinsic healing potential and may be more pronounced in humans where healing does usually not occur by itself. Another aspect may be that our structural measurements are more sensitive and cannot be detected by our mechanical tests.
This study has several limitations, including clear differences between rodent and human muscle properties and rotator cuff healing. In humans, the supraspinatus muscle degeneration is irreversible and fatty. In rodents, muscle atrophy is completely reversible and rotator cuff healing is fast and more likely to produce a long-term successful outcome.7, 8, 21 We did not find any re-tears of the sutured rotator cuff which one might have expected especially in the saline groups with increased repair load. This might be due to the fact that the ability to heal is much more pronounced in rats than it is in humans. Therefore, the next step now has to be to transfer our results to a larger animal model where re-tears take place.
The results of this study suggest that partial removal of tendon tension in rotator cuff repair by preoperative injection of botulinum neurotoxin A in rotator cuff lesions with an increased preload, for example, retracted, large and/or old lesions, has the potential to significantly improve structural properties. Further studies are needed to investigate the amount of remaining tendon tension after BoNtA injection and tendon repair which is needed to promote intrinsic healing.