Mechanics and differential healing outcomes of small and large defect injuries of the tendon-bone attachment in the rat rotator cuff

ABSTRACT Introduction Rotator cuff tear size affects clinical outcomes following rotator cuff repair and is correlated with the risk of recurrent tendon defects. This study aimed to understand if and how the initial defect size influences the structural and mechanical outcomes of the injured rotator cuff attachment in vivo. Methods Full-thickness punch injuries of the infraspinatus tendon-bone attachment in Long Evans rats were created to compare differences in healing outcomes between small and large defects. Biomechanical properties, gross morphology, bone remodeling, and cell and tissue morphology were assessed at both 3- and 8-weeks of healing. Results At the time of injury (no healing), large defects had decreased mechanical properties compared to small defects, and both defect sizes had decreased mechanical properties compared to intact attachments. However, the mechanical properties of the two defect groups were not significantly different from each other after 8-weeks of healing and significantly improved compared to no healing but failed to return to intact levels. Local bone volume at the defect site was higher in large compared to small defects on average and increased from 3- to 8-weeks. In contrast, bone quality decreased from 3- to 8-weeks of healing and these changes were not dependent on defect size. Qualitatively, large defects had increased collagen disorganization and neovascularization compared to small defects. Discussion In this study, we showed that both large and small defects did not regenerate the mechanical and structural integrity of the intact rat rotator cuff attachment following healing in vivo after 8 weeks of healing.


Introduction
Acute rotator cuff tears are the most common shoulder injuries 1 . These tears generally fail to heal spontaneously, can limit the normal function of the shoulder and often require surgical repair [2][3][4][5][6] . The size of the initial tear prior to repair plays a critical role in clinical outcomes, as larger tears have been correlated with decreased tendon integrity, decreased range of motion, and an increased rate of shoulder osteoarthritis progression 2,7,8 . Small tears are often treated conservatively to eliminate pain 9 . The outcomes of conservative treatments are highly variable, 3,10 and there is a high risk of tear propagation 11 . Larger tears commonly require surgical repair for functional reattachment of the injured tendon to the bone [12][13][14][15][16] , however little is known about how the size of full-thickness tears in a single tendon that propagate into larger, more problematic tears in vivo. Therefore, a gap in knowledge exists related to the potential for tear propagation from partial tears to complete rotator cuff tendon detachment.
The rat is a commonly used animal model to study rotator cuff tears due to anatomical similarities to the human rotator cuff 16 . Previous studies using cadavers or animal models to study rotator cuff tears have shown that strain concentrations at the tendon-bone attachment are elevated with increasing tear size [17][18][19][20][21] . The majority of animal models used to study rotator cuff healing rely on complete tendon tears with sutured reattachment of the tendon to its bony footprint to mimic the clinical scenario of rotator cuff repair [22][23][24][25][26][27][28][29][30][31][32] . Few studies have investigated non-operative healing of small rotator cuff injuries in vivo - 19,20,33 . The simulation of the clinical scenario of rotator cuff tear and healing in animal models has benefits for understanding healing mechanisms of tendon repair. However, several unanswered questions remain regarding the regenerative capacity of the native tendon-bone attachment without surgical repair. We have recently shown that full-thickness injuries of the tendon-bone attachment heal via fibrosis and result in significant reductions in bone quality 20 . We have also shown that non-healed small defects at the tendon-bone attachment lead to decreased strength and the formation of strain concentrations across the remaining intact attachment 19 . While our previous studies explored structural outcomes of small defects with healing and the acute mechanical properties of small rotator cuff defects at the time of injury, we did not evaluate the mechanical properties of the healing attachment in these studies, nor did we assess if and how the size of the defect influences healing outcomes.
Therefore, in this study, we aim to understand if and how defect size differentially affects the healing of the tendon-bone attachment in an established in vivo rotator cuff injury model of full-thickness tears. We created two different sized defects (small and large) in the rotator cuff tendon-bone attachment in the rat shoulder and assessed the biomechanical properties and structural healing of these defects at different healing time points (3-and 8-weeks of healing) compared to acute injury (no healing) as well as intact, un-injured tendon. We hypothesized that small and large defects would both result in fibrotic scar tissue formation, but that large defects would propagate and lead to diminished healing outcomes compared to small defects (i.e., decreased mechanical properties, bone quality, and organization).

Surgical procedure and postoperative care
Mature, Long Evans female rats (N = 32 rats total; former breeders, 299 ± 54 g) were used under the accordance with the University of Delaware Institutional Animal Care and Use Committee. For survival surgeries (n = 18 rats total), animals underwent a bilateral surgical procedure under anesthesia (isoflurane carried by oxygen) to model small-and large full-thickness tears of the infraspinatus tendon-bone attachment using 0.3 mm (small) and 0.75 mm (large) punch biopsies (Robbins, Chatham, NJ USA) 19,20 . The outer diameters of the small and large biopsy punches were measured using digital calipers at 0.75 mm and 1.10 mm, respectively, which were visually equivalent to approximately half or two-thirds, respectively, the width of the adult rat infraspinatus attachment. A 2-cm incision was made craniolateral to the glenohumeral joint and was followed by a horizontal incision to detach the deltoid from the acromion. A suture was passed under the acromion to elevate the scapula and expose the rotator cuff tendon-bone attachments. The infraspinatus tendon-bone attachment was located following internal shoulder rotation, the joint was stabilized, and the defect was created. Small or large punch biopsies were placed at the midline of the tendon-bone attachment to create consistent, full-thickness defects, which also removed the fibrocartilaginous attachment while also keeping intact the superior and inferior attachment sites. The shoulder (left or right) that received a small or large defect was alternated between rats using controlled randomization. The muscle layer and skin were closed using simple interrupted resorbable sutures (5-0 Vicryl, Ethicon Inc., Somerville, New Jersey). Rats were given subcutaneous buprenorphine (0.05 mg/kg) as analgesia and local bupivacaine hydrochloride (0.05 mg/kg) as an anesthetic pre-operatively and postoperatively, respectively. The animals were closely monitored during recovery and twice daily for 3 days post-operatively. Rats were housed in cages (two rats maximum per cage) with enrichment and unlimited access to food and water and were euthanized via carbon dioxide asphyxiation and thoracotomy at 3-weeks post-surgery or 8-weeks post-surgery and entire shoulder complexes were dissected to visualize the infraspinatus attachment.

Gross morphology and biomechanics
Gross morphology images were taken using a digital camera (DSLR, Nikon). The 3-week time point was used to represent the inflammatory stage of the injury, and the 8-week time point was used to represent the remodeling stage ( Figure 1, study design). Mechanical testing was performed to compare mechanical outcomes associated with two different defect sizes, both with no healing (t 0 , n = 14 rats with bilateral defects) and following a duration of healing (8-week time point, We compare the biomechanical properties and structural outcomes of paired small and large defects in the infraspinatus tendon-bone attachment of Long Evans rats in vivo. Biomechanical properties of small and large defects were assessed with no healing (t 0 ) and after 8-weeks of healing. Structural outcomes from gross morphology, micro-computed tomography (microCT), and histology were assessed after 3-weeks and 8-weeks of healing. n = 10 rats with bilateral injuries). In the no healing (t 0 ) group, rats were euthanized prior to injury and their shoulders were immediately dissected, at which time the punch defect was placed as described for in vivo punch injuries. Stereoscopic images were taken of the lateral and inferior sides of the attachment (DSLR, Nikon) and ImageJ was used to measure the width and thickness of these sides, respectively. The remaining intact attachment width was calculated by subtracting the defect diameter from the total attachment width for non-healed injuries. For non-healed injuries, the attachment was assumed to be rectangular to approximate the cross-sectional area (CSA). For healed attachments, the CSA was assumed to be circular and measured from the lateral attachment width because there was a large circular scar around the healed attachments rather than rectangular "borders" that were more easily visualized in the non-healed injuries. The humeral head was secured using steel wire (32AWG) to prevent growth plate failures. Distal humeri were potted in 1.5 mL centrifuge tubes using polymethyl methacrylate (Ortho-Jet BCA, Lang Dental) or DP100 epoxy (3 M, Scotch-Weld). Infraspinatus muscles were removed, and tendons were affixed in thin-film grips lined with PBS-soaked tissue paper (Kim wipes, Kimberly-Clark). Tendons were positioned in a materials test stand at 0-degree shoulder abduction. A 0.1 N tare load was applied, and gauge length was recorded before uniaxial tensile tests began (Instron 5943, Norwood, MA). The test protocol was as follows: 10 preconditioning cycles from 0.025 to 0.075 mm at a rate of 0.01 mm/sec, followed by 90 s of rest, and ramp to failure at 0.01 mm/sec. Ultimate load, stiffness, ultimate stress, strain at ultimate stress, tangent modulus, area under the curve (AUC) at ultimate stress, and AUC of the last five load-displacement preconditioning cycles (energy loss) were computed (MATLAB, Natick, MA) 19 .

Microcomputed tomography
Immediately following euthanasia and dissection, shoulders allocated for histology and micro-computed tomography (micro-CT) from 3-and 8-week time points (n = 8 rats total with bilateral injuries; four rats per time point) were fixed in 4% paraformaldehyde for 24-48 h then scanned in air using micro-CT (1276 Skyscan; Bruker, 55kV source voltage, 200 μA source current, 10.6 µm image pixel size, 1200 ms exposure). MicroCT images were reconstructed and aligned with DataViewer (Bruker). Bone morphometric properties for the defect area, greater tubercle, and proximal epiphysis were measured using Dragonfly software (Object Research Systems Inc., Figure 4A and S1). The defect area was defined following alignment of the X-and Y-planes for a normal view of the defect in the Z-planar view. Using the 2D painter tool in Dragonfly, a circle of diameter equal to the diameter of the punch biopsy was digitally painted on the first slice where the full defect was observed in the z-plane ( Figure S2). A cylinder-shaped region of interest (ROI) was then filled into the defect using a consistent cylinder height (370 µm). Split and Otsu functions were used to separate mineral (bone) and non-mineral within this cylindrical ROI. Bone volume (BV) was divided by the total volume (TV) of the cylinder to calculate (BV/TV) ratio. Epiphyseal ROIs (total epiphysis and greater tubercle) were isolated using consistent anatomical landmarks, and the bone analysis function was used to fill holes smaller than 300 μm to obtain the TV of the epiphysis. Trabecular and cortical bones were separated using the Kohler method (maximum trabecular thickness = 150 μm) 34 . Total, trabecular, and cortical bone ROIs were isolated and analyzed using the Bone Analysis tool in Dragonfly ( Figure 4A).

Histology
After microCT, the same samples were decalcified in ethylenediamine tetraacetic acid (EDTA), paraffin embedded, and sectioned at 6 µm thickness. Sections were stained with nuclear stain 4,'6-diamidino-2-phenylindole (DAPI) and imaged in three non-overlapping fields of view within the injured attachment (40× magnification; 0.024 mm 2 /image) to quantify cell number, aspect ratio (nAR), minor axis, major axis, equivalent diameter, perimeter, solidity, eccentricity, orientation and area of nuclei using a custom MATLAB code ( Figure 5A-C) 35,36 . Image replicates (3 per sample) were averaged per sample and then used for statistical comparisons between defect groups. Hematoxylin & Eosin (H&E) was used to assess fibrosis, cellularity, and vascularity, and Picrosirius Red was used to assess collagen organization.

Statistical analyses
One-and two-way ANOVA with repeated measures and Sidak's multiple comparison tests were used to compare mechanical outcomes, bone morphometry from microCT, and cell nuclear shape of paired small and large defects at t 0 (mechanics only), 3-or 8-weeks of healing. A mixed-effects analysis was used to compare Intact, t 0 , and 8-weeks of healing mechanical outcomes with the size of the defect. All statistics were performed using Prism (version 8, GraphPad).
Mechanical properties of intact attachments from Locke et al. were used for comparisons 19 . Data presented graphically represent mean ±95% confidence interval; data described in-line in the text represent mean ± standard deviation.

Mechanical properties increased with healing
All animals tolerated bilateral surgery. Biomechanical data from four no-healing (t 0 ) rats and one rat at 8-weeks of healing were excluded due to errors in data acquisition (e.g., software failure during test), damage during dissection, and/or abnormal defect placement (non-healed t 0 only); therefore, n = 10 shoulders per defect size were analyzed in the no-healing group, and n = 9 shoulders per defect size were analyzed for the 8-week healed group. Two additional rats in the healed group were excluded from energy loss comparisons, but not load to failure comparisons, due to data acquisition errors during preconditioning.
Without healing, the attachment CSA was reduced with both defects compared to uninjured attachments (no healing), and CSA measurements were significantly lower with no healing for large but not small defects compared to intact (Figure 2A, gray horizontal line; one-way ANOVA p = 0.0029; intact = 2.04 ± 0.55 mm 2 ; small no healing = 1.72 ± 0.82 mm 2 , large no healing = 1.12 ± 0.33 mm 2 ). With healing, attachment CSA for both small and large defects was significantly increased compared to intact CSA (p < 0.0001 for both; small with healing = 3.82 ± 0.89 mm2; large with healing = 4.12 ± 0.92 mm 2 ) and respective no healing (t 0 ) CSA (p < 0.0001 for both) and, surprisingly, no difference in CSA was observed between small and large defects after 8 wk of healing (Figure 2A; p = 0.9766). As expected, ultimate load and stiffness increased with healing compared to t 0 regardless of defect size (p < 0.0001; ultimate load: intact = 39.5 ± 8.5 N; small with healing = 29.1 ± 6.8 N; large with healing = 29.3 ± 10.3 N; stiffness: 43.5 ± 8.9 N/mm; small with healing = 28.1 ± 6.9 N/mm; large with healing = 25.8 ± 9.5 N/mm). The energy loss was lower in with healing compared to t 0 regardless of defect size (p = 0.0216 for small defects; p = 0.0004 for large defects). Additionally, the ultimate load and stiffness were reduced in both non-healed defect groups compared to intact attachments (p < 0.0001, Figure 2B-C; ultimate load: small no healing = 14.3 ± 3.0 N; large no healing = 10.9 ± 3.5 N; stiffness: small no healing = 14.8 ± 3.7N/mm; large no healing = 11.8 ± 4.2N/mm), corroborating previous findings from our group 19,20 . Small defects also had a higher ultimate load compared to large defects without healing (p = 0161). After 8-weeks of healing, both small and large defects had increased ultimate load and stiffness compared to non-healed (t 0 ) defects; however, healed defects did not regain comparable properties to native, uninjured attachments ( Figure 2B-C). Ultimate stress was significantly reduced for all defects regardless of healing ( Figure 2D; intact = 21.6 ± 8.2 MPa; small no healing = 10.6 ± 6.1 MPa; large no healing = 9.8 ± 5.6 MPa; small with healing = 7.9 ± 2.5 MPa; large with healing = 7.6 ± 3.3 MPa). Energy loss was reduced for healed defects compared to defects with no healing. No differences were observed for tangent modulus, ultimate stress, AUC at ultimate stress, or strain at ultimate stress between defect sizes (Table S1).

Gross morphology and bone properties were affected by defect size and healing duration
From gross morphology, no tear propagation into complete tears or tendon retraction was observed for either defect size ( Figure 3A-D), although increased neovascularization and fibrosis at the defect site were found for large compared to small defects at both 3-and 8-weeks ( Figure 3A-D, white arrows). Large defects had increased fibrosis at 8-weeks compared to 3-weeks ( Figure 3A-D). Additionally, we qualitatively observed less cortical bone for large defects compared to small defects at 3-weeks ( Figure 3E-F).
The amount of bone within the defect increased over time for both defect sizes ( Figure 4) and bone volume fraction (BV/TV) was higher in large compared to small defects especially at 3-week ( Figure 4B and S1, p = 0.02). Conversely, the trabecular epiphyseal BV/TV ( Figure 4D) decreased with increased healing time but did not vary between defect sizes. In the greater tubercle, trabecular separation ( Figure 4E, p < 0.01) increased and trabecular thickness ( Figure 4F, p < 0.01) decreased from 3-to 8-weeks but did not differ between defect sizes.

Tissue cellularity and structure were mildly different between defect groups
Within the defect, cellularity was greater for large defects compared to small defects overall ( Figure 5D, p = 0.02). Large defects had significantly increased nuclear aspect ratio ( Figure 5E) and eccentricity (Table S2) compared to small defects overall and at 8-weeks of healing ( Figure 5E). The nuclear area ( Figure 5F), equivalent diameter ( Figure 5G), major axis, minor axis, and perimeter (Table S2) increased from 3-to 8-weeks of healing. The nuclear solidity and orientation were not affected by defect size or healing time (Table S2).
Both defect sizes had evidence of vascularity within the injured attachment ( Figure 6A-B), and more vessels were observed for large defects compared to small defects at  Figure 6B). Additionally, at 3-weeks, both defects had rounded nuclei within the injury region rather than elongated tendon fibroblasts, indicating the presence of macrophages or other cell populations ( Figure 6A-B). From 3-to 8-weeks, both small and large defects had evidence of fibrosis surrounding the defect region and reduced appearance of vessels ( Figure 6C-D). Qualitatively, we observed that small and large defects had mixed collagen organization after 3-weeks of healing ( Figure 6E-F), while large defects resulted in more unorganized collagen compared to small defects after 8-weeks of healing ( Figure 6G-H).

Discussion
The presented data represent a significant advance over what is already known by demonstrating the robust healing response of the rat rotator cuff attachment when subjected to a punch injury. Previous studies by our group and others have shown that punch defects in tendons and their attachments often lead to fibrotic scar formation and bone loss and that small defects result in placement-dependent changes in biomechanical properties 37,38 . However, how the mechanical properties of the injured attachments change with varying size defects has not been previously described. In our study, we showed that the size of full-thickness rotator cuff defects did not play a significant role in biomechanical properties after 8 weeks of healing, which is a common (or even longer) time point used in previous studies to compare tendon healing in small (A) DAPI stained sections from within the injured attachment were imaged at 40x. Three technical replicates per biological sample were imaged and (B) a custom MATLAB code was used to digitize and auto-segment the nuclei. The code was used to calculate and average (C) nuclear properties for each technical and biological replicate. (D) Cellularity, (E) nuclear aspect ratio, (F) nuclear area, and (G) equivalent (Eq) diameter were measured and compared between defect sizes and healing durations. Bars: significant difference (p<0.05). Dashed lines: paired data points. Data are presented as mean ± standard deviation. animal models following injury 22,[39][40][41] . The size of the defect did affect the gross morphology, bone morphometry, cellularity, and structure of the tendon-bone attachment. Surprisingly, and counter to our hypothesis, we observed similar mechanical properties of both defect sizes after 8-weeks of healing. Our mechanical testing approaches were sensitive to subtle changes in baseline (no healing) properties, but the healing process resulted in high variability within groups. Neither defect group regained the native mechanical properties of the healthy attachment by 8 weeks post-injury, with both reaching approximately 75% of the load to failure of the intact tendon. Thus, the attachment does not fully regenerate its structural properties by 8-weeks following injury. It is possible that a longer duration of healing could have led to improved mechanical properties comparable to the intact tendon-bone attachment, yet it was nonetheless surprising that many of our findings were similar between the two defect sizes with healing. Additionally, the two defect sizes chosen may elicit similar responses to the remaining intact tissue, which likely underwent remodeling and healing in response to the injury. Although we did not find a mechanical difference between defect sizes, we did observe several spatiotemporal differences in attachment bone properties, cellularity, and structure between defect sizes. We show that the size of fullthickness rotator cuff defects is in part associated with morphologic healing outcomes.
Our model of full-thickness injury is similar to crescent-shaped tears that are commonly present in the clinic and lead to tear propagation 1,2 . Tear propagation is correlated with increased strain at the defect site 20 . In our injury model, although we observed decreased mechanical properties between non-healed (t 0 ) small and large defects compared to intact attachments and have previously observed strain concentrations at the defect site 18 , we did not observe tear propagation in either defect size following 8-weeks of healing in vivo under normal physiologic loading. However, increased mechanical loading at the attachment (e.g., overuse) may increase the likelihood of tear propagation even for small defects 42,43 . In our study, we did not challenge the injured attachment with overuse beyond the initiation of a defect, but we did include bilateral injuries in our study design that we assumed would lead to comparable loading between sides. Physiologic loading in this study (e.g., reduced mobility; shoulder unloading) may have resulted in a protective healing response that prevented tear propagation into a complete tear and supported a similar increase in mechanical properties of both defects during healing 44 . Future studies could investigate the role of acute or repetitive loading on attachment damage and tear propagation by modulating muscle loading and attachment mechanobiology (e.g., treadmill running or muscle stimulation) 45 . Additionally, future studies could explore if and how functional gait is restored, specifically in the context of shoulder function, throughout the healing process.
Rotator cuff tears commonly lead to fibrosis and neovascularization 20,25 , and, in this study, we similarly observed a qualitative increase in fibrotic tissue and vessel formation with increased defect size. Additionally, various dynamic cell behaviors, such as migration, proliferation, and differentiation, as well as abnormal cell phenotypes, such as rounded cells, have been observed following tendon injury 40,46 . Notably, we observed a dynamic increase in cell number and structural characteristic of the attachment, including fibrosis and collagen organization, that were dependent both on defect size and healing time, concurring with Lemmon et al. but contrasting with other injury models 20,31,32 . A decrease in cell number and increased collagen organization may be observed with further healing duration; nonetheless, comparable tendon remodeling has been observed between 8-and 12-weeks 47 , suggesting that 8-weeks is an appropriate time point to assess attachment remodeling. These findings suggest that the rat attachment may adapt to the increased tear size in an attempt to remodel the larger injury.
Previous studies have shown that tendon injury leads to an acute decrease in nAR compared to healthy tendons and that nAR increases with increased healing duration, indicating a mechanobiological response to injury and healing 40 . Interestingly, in this study, we observed changes in nAR of local fibroblasts that were dependent on injury size. Whether large defects result in increased tissue-scale strain compared to small defects and if mechanical differences between defect sizes lead to differences in nAR remain unknown 19 . Increased nAR in larger defects may suggest focal differentiation of native or extrinsic stem cells into fibroblasts 40,48 . The dynamic nuclear changes observed in this study may have developed from increased tissue-scale strain with increased defect size and/or altered mechanobiological force transmission with increased tissue remodeling. A further understanding of attachment mechanobiology may help to clarify these cellular adaptations to attachment defect size in vivo 45,49 .

Limitations
A limitation of this work is that we only used two different defect sizes to test if and how attachment defects can lead to poor healing and, potentially, tear propagation. While these defects did lead to marked structural impairments and reduced functional quality of the healing attachment, we were unable to determine if and how the infraspinatus attachment in rats can resist failure via propagation with inclusion of a punch defect. It is possible that a larger defect size than what was used could elicit tear propagation, but this was not found in our current study. Another limitation of this work is that our injury model included a bone defect at the enthesis, which may not perfectly mimic the clinical state. However, the approach to creating a defect using a biopsy punch allowed for consistent placement and maintenance of some residual tendon at the interface to study nascent enthesis repair without the need for surgical reattachment. This approach also resulted in consistent and repeatable biomechanical changes to the enthesis that can be tracked over time during the healing process. Other models of rotator cuff injury, including full tendon transection with and without surgical repair, have larger variations in reported outcomes (e.g., biomechanical properties and bone quality) 22,44 as well as an elevated risk of retear 39 . This model, although limited, will still provide a platform for studying enthesis regeneration as it can be useful for future studies using biomaterials or cell-based strategies for enthesis healing. Importantly, because this model is consistent, repeatable, and does not fully regenerate the enthesis, it will be useful for preclinical testing of potential therapeutics aimed at regenerating the gradient morphology of the enthesis.
The use of this bilateral model allowed us to minimize the potential for abnormal loading on healing outcomes. However, a limitation of this study is that we did not assess limb preference during gait in the post-injury period. Our model was beneficial as we utilized a bilateral defect that allowed us to pair our comparisons using biologically identical shoulders. Additionally, while decreased bone quality following acute tendon injury is common 26 , the nature of our model likely contributed to a decrease in bone volume as the punch biopsy permeated through cortical bone 19,20 . Finally, the use of an animal model to study rotator cuff tears has limitations when translating to humans due to differences in joint loading and healing. Although rats are a commonly used rotator cuff model, the use of rats in our study could potentially be a limiting factor. Rats have an increased propensity to heal with limited scar formation and are also quadrupeds that weight-bear on their forearms unlike humans. Although they perform overhead reaching and grasping, this is not identical to human biped motion. This healing response results in indistinguishable outcomes for more subtle/sensitive healing metrics, such as biomechanical properties. This healing response has also been described in studies by others, using "excisional" (biopsy punch) or "incisional" (blade) defects in mice wherein biomechanical properties after post-defect healing were comparable between the two defect types even though the size of these defects was likely different (due to differences in amount of tissue lost) using Achilles and other injuries 50 .
Understanding how defect size can influence the mechanical and structural outcomes of attachment healing will aid in the design and development of regenerative therapies for regenerating the enthesis following injury. The use of animal models to study the mechanical consequences and healing outcomes dependent on defect size(s) may also improve our preclinical understanding of early rotator cuff disease. This defect model can serve as a tool to understand the native healing process of full-thickness tears and as a testbed for preclinical testing of biomaterials that could be used to mitigate or delay surgical intervention of early rotator cuff disease.