Biomechanical and histological changes secondary to aging in the human rotator cuff: A preliminary analysis

The high failure rate of rotator cuff repair surgeries is positively correlated with age, yet the biomechanical changes to the tendons of the rotator cuff with age have not been described. As such, we sought to benchmark and characterize the biomechanical and histopathological properties with the accompanying gene expression of human rotator cuff tendons as a function of age and histopathological degeneration. All four rotator cuff tendons from fresh human cadaver shoulders underwent biomechanical, histopathological, and gene expression analyses. Following cadaver availability, samples were grouped into Younger (i.e., less than 36 years of age, n = 2 donors) and Aged (i.e., greater than 55 years of age, n = 3 donors) as a means of characterizing and quantifying the age‐related changes exhibited by the tendons. Biomechanical testing and subsequent computational modeling techniques revealed both differences in properties between tendons and greater Young's moduli in the Younger tendons (supraspinatus 3.06x, infraspinatus 1.76x, subscapularis 1.25x, and teres minor 1.32x). Histopathological scoring using the semi‐quantitative Bonar scoring scheme revealed a positive correlation with age across all tendons (r = 0.508, p < 0.001). These data contextualize the biomechanical and histopathological changes to tendons that occurs naturally with aging, highlighting the innate differences in biomechanical properties of all four rotator cuff tendons, as well as the difference in their degenerative trajectories. Additionally, the histopathological scoring revealed moderate signs of degeneration within the Younger supraspinatus tendons, suggesting tissue quality may decrease in this specific tendon in patients less than 40 years old, before clinical symptoms or tears.


| INTRODUCTION
There is a high prevalence of rotator cuff tendon tears in the general population, with tear rates increasing markedly with age. 1,2Considering rotator cuff repair (RCR) failure rates are still documented to be as high as 90% depending on risk factors (e.g., age), 3,4 there continues to be a clear need to develop new therapeutics and treatment devices that can drive improved RCR healing outcomes.Current treatment approaches have been aimed at either utilizing implantable scaffolds/patches at the bone-tendon interface to either improve time zero repair strength, 5,6 facilitate and guide the formation of tendinous tissues, 7 and/or to deliver biologically active agents. 8However, persisting failure rates necessitate alternative treatment strategies.Recent work has illustrated the importance of proper mechanobiological cues to resident cell populations to maintain healthy tissue homeostasis, collagen turnover, and gene expression [9][10][11][12][13][14] ; this concept has been further exemplified through utilization of overuse tendinopathy models that have shown breakdowns in tendon collagen structure and inferior mechanical properties secondary to overuse regimes in rats. 15,16th this in mind, it can be imagined that by deliberately tailoring scaffold mechanical properties to match those exhibited by healthy tendons, tenocyte function, and behavior could be optimized with the goal of improving scaffold integration and repair integrity.
Unfortunately, the current biomechanical properties of healthy human rotator cuff tendons suitable for the design of such a scaffold have not yet been reported.Specifically, there is a need to benchmark those properties when tendons are healthy (e.g., in younger population with low tear prominence) and diseased (e.g., in aged populations where tears are more prominent).Additionally, understanding that each tendon has unique biomechanical properties as a result of anatomical arrangement and function, [17][18][19] it would follow that scaffold properties may need to be specifically tailored to match the target tendon(s), especially in cases of massive rotator cuff tendon tears involving multiple tendons.Unfortunately, although some researchers have documented age-related changes to tendon biomechanical properties in animal models, 17 changes to all four rotator cuff tendons in humans secondary to aging have not yet been documented.
With this in mind, the goal of this study was to benchmark and characterize the biomechanical and histopathological properties with the accompanying gene expression of human rotator cuff tendons as a function of age and histopathological degeneration.Additionally, the biomechanical properties of these tendons were to be modeled using quasilinear viscoelastic theory as a means of providing the necessary material properties for the future design of repair scaffolds with tailored mechanical properties.Due to the difficulty of obtaining fresh cadaver samples, this work reports a preliminary data set.It was hypothesized that the four tendons of the rotator cuff would exhibit deleterious changes with age.

| MATERIALS AND METHODS
All human donor tissues were deidentified with only age and sex as indicators (Donor Alliance).As such, the Colorado State University Institutional Review Board (IRB) deemed approval unnecessary.

| Tendon dissection procedure
Shoulders were obtained within 16-h postmortem delivered on wet ice.Samples were immediately dissected with care to isolate each tendon and associated muscle of the cuff (i.e., supraspinatus, infraspinatus, subscapularis, and teres minor).Muscle biopsies (n = 1, 5 mm × 5 mm × 5 mm) were harvested from the central muscle belly of each tendon for histopathologic analysis before full exposure of the tendons.Upon visualization and isolation of each tendon, a small (i.e., 2 mm wide × 4 mm long × full thickness depth) tendon biopsy was taken from the lateral aspect with a sterile blade and flash frozen in liquid nitrogen for the purpose of RNA sequencing (associated Methods and Results are presented in Supplementary Materials due to limited sample size).Biopsies were sized so as to minimize impact to tendon mechanical analysis.Following isolation of each tendon/muscle group and bulk skeletonization of the humerus (leaving the tendon attachment sites intact), the muscle tissue was carefully removed from each tendon.

| Tendon biomechanical testing procedures
Following tendon isolation, the humerus was mounted (SmoothCast 321, Smooth-On) within a potting sleeve to allow for biomechanical testing. 20,21This potting sleeve was installed in a custom fixture to ensure tendon testing occurred in anatomically appropriate directions (i.e., aligned with the natural direction of loading) on the servohydraulic load frame (Model 858, MTS Corp.).The tendons were gripped using a custom cryo-grip maintained between −10°C and −15°C, and a warm physiologic saline spray was used to ensure no tendon freeze artifact while ensuring adequate gripping strength. 20gh-speed video (120 FPS) was captured for all tests to ensure tendons did not slip within the grips.3][24][25][26] Before initiation of biomechanical testing, a static preload of 5 N was applied to each tendon for a duration of 2 min, at which point a reference gage length was measured and recorded for purposes of strain application.Tendons were then subjected to two 100-s duration stress relaxation tests, first at 6%, and then at 8% global stretch.Subsequently, two cyclic tests at 1 Hz and 0.1 Hz at 10% global stretch were applied randomly.
A 1000 s rest period was enforced between all tests to minimize the viscoelastic history of previous test cycles. 25Force and displacement data were sampled at 100 Hz.Tendon hydration was preserved with physiological saline spray at all times during dissection and testing.
The stress relaxation tests at both stretch magnitudes were subsequently analyzed to generate the following outcome parameters: peak force, peak stress, and percent relaxation.Following biomechanical testing, each tendon was transected from the humerus at the insertion for area measurements and post hoc histological processing.Photogrammetry techniques were utilized to measure the cross-sectional area (CSA) of each tendon adjacent to the bony insertion for the purposes of transforming force into stress. 26,27

| Viscoelastic modeling
9][30] As proposed by Fung, 31 soft tissue response (i.e., induced stress) to uniaxial loading can be modeled by: where ε(t) denotes the engineering strain applied to the tissue, G(t) denotes the reduced relaxation modulus (capturing the viscous component of tissue response), and σ e (ε) denotes the nonlinear elastic behavior (capturing the instantaneous hyperelastic component of tissue response).A Prony series was implemented to model the reduced relaxation modulus as follows 26,28 : To ensure unique solutions, the Prony series terms were subjected to the following constraints: all , and τ i = [0.1 s, 1 s, 10 s, 100 s]. 28A 1-term Ogden strain energy function was implemented as a means of capturing the hyperelastic behavior of the tendons: In this formulation, μ and α are experimentally derived material constants that provide for the strain-dependent material stress. 32is specifical model was selected for ease of implementation in commercially available finite element analysis packages while maintaining similar predictive accuracy as compared to other strain energy function models (i.e., Mooney Rivlin and Neo-Hookean; data not shown).By recasting Equation (3) in terms of stretch then subsequently differentiating with respect to stretch (λ), the formulation of Young's Modulus as a function of stretch can be obtained: Implementation of this modeling approach as defined in Equations (1-3) resulted in a total of seven experimentally fitted terms: μ, α, G 1 , G 2 , G 3 , G 4 , and G ∞ .Cyclic biomechanical testing data for each sample were independently fit through implementation of Matlab's (R2020a, Mathworks) fmincon function with the optimization goal of reducing rootmean-squared errors (RMSE). 27The fitted terms derived from both cyclic tests (i.e., 1 Hz, 0.1 Hz) were then averaged to generate one set of seven fitted terms for each specimen.To validate the model fitting results and ascertain their predictive capabilities, these parameters were utilized to predict the stress-relaxation responses.Specifically, the mean fitting parameters from each respective group were utilized to predict the mean experimental stress-relaxation response of that same group.Predicted peak stress was compared to experimental peak stress, and RMSE values were reported as outcome parameters for the predictive testing.As a means of further visualizing the age-related changes to tendon biomechanical properties, Equation ( 4) was utilized to calculate the stretch-dependent Young's modulus values at discrete intervals (i.e., 1%) over the experimental stretch range (i.e., 1.00 ≤ λ ≤ 1.10) for each tendon/age group.

| Histomorphological analysis
Following the nondestructive biomechanical testing, the removed tendons were placed in 10% neutral buffered formalin for 24 h then were processed using standard formalin-fixed paraffin embedding procedures. 33Two 5 µm thick slides were produced from each specimen displaying the entire length and width of the tendon, with one slide stained with hematoxylin and eosin (H&E) and one slide stained with Picro-sirius red.Muscle biopsies were processed similarly, with only one slide for each specimen being produced and stained with H&E.
Polarized light microscopy was utilized to image the picro-sirius red stained slides and assess collagen organization. 20Briefly, tendon sections were imaged at 100X magnification using a polarized light scanning microscope with a manual exposure of 40 ms (Olympus BX61VS).A machine learning algorithm (Image-Pro Premier v9.2, RRID:SCR_007369) was utilized to quantify the percent area of the organized/disorganized collagen within the tendon section.
The presence of intramuscular adipose content has long been correlated with decreased RCR outcomes and increased degenerative tendon pathology. 34As such, the intramuscular adipose content within the rotator cuff muscle groups was quantified using histomorphometry techniques, 21 with the intention of correlating changes to tendon biomechanical properties at different stages of associated muscle fatty atrophy.Adipose area and total tissue area were quantified (Image-Pro Premier v9.2, RRID:SCR_007369).To normalize for differing section areas, percent intramuscular adipose area was calculated.The reported percent intramuscular adipose content value reported for each sample was the mean of three randomly located equally sized analysis regions within each specimen.

| Histopathological analysis
Tendon slides stained with H&E were graded by a blinded veterinary pathologist according to the Bonar tendinopathy scale. 35The semi-quantitative Bonar scoring system has previously been implemented as a means of assessing the degenerative pathology of tendon samples. 36Specifically, the tendon specimens were evaluated on a semi-quantitative scale from 0 (normal/healthy) to 3 (markedly degenerated) across four categories: (1) tenocyte reactivity, (2) angiogenesis, (3) tendon bundle organization and polarization, and (4) deposition of ground substance. 36The summation of the four category scores was used to provide an overall score for each tendon ranging from 0 (histopathologically normal) to 12 (histopathologically degenerated).Following completion of semi-quantitative scoring, the pathologist was unblinded to specimen information (i.e., tendon and age) and made observational comments regarding structural and tissue component features to provide additional histopathological descriptions and comparisons across tendon and age groups.

| Statistical analysis
Cadavers were divided into two groups: younger (<36 years old) and aged (>55 years old).Age groups were determined based on cadaver availability.Statistical comparisons within a specific tendon and across age groups were made using Welch's t-test (GraphPad Prism v9.2.0, GraphPad Software).A significance threshold (α) of 0.05 was used for all statistical comparisons.Sample numbers for each grouping are outlined in Table 1.Sample sizes were verified as adequate as demonstrated by a post hoc power test based on biomechanical properties of the tendons (infraspinatus hyperelastic µ, power = 0.913).Correlations between all outcome parameters were assessed to illuminate patterns between histopathological features, gene expression, and biomechanical properties; Pearson correlation coefficients and associated p-values were reported.Correlation strength was characterized as follows: slight (0-0.2),fair (0.21-0.4), moderate (0.41-0.6), strong (0.61-0.8), and very strong (0.81-1.0). 37| RESULTS

| Tendon biomechanical testing
The rotator cuff tendons exhibited differentially altered biomechanical properties with age.Specifically, the infraspinatus tendons exhibited a 48.8 (p = 0.007) and 52.6% (p = 0.001) decrease in peak stress at stretch magnitudes of λ = 1.06 and 1.08, respectively (Figure 1A,B).Similarly, the subscapularis tendons demonstrated a 47.8% (p = 0.002) and 44.6% (p = 0.020) decrease in peak stress at both stretch magnitudes (Figure 1A,B).While the infraspinatus tendons exhibited increased percent relaxation across both stretch magnitudes, only the change at λ = 1.06 reached statistical significance (p = 0.005, Figure 1C).Comparably, the subscapularis tendons also demonstrated increased percent relaxation across both stretch magnitudes; however, only the change at λ = 1.08 reach statistical significance (p = 0.049, Figure 1D).Differences between the younger and aged supraspinatus and teres minor tendons did not reach the level of statistical significance (Figure 1A-D).

| Histomorphological assessment of collagen fiber alignment and intramuscular adipose content
Collagen organization was quantified within each tendon across both age groups through utilization of Picro Sirius red staining.No significant differences were noted in percent organized collagen between the aged and younger groups in any of the four tendons (Figure 4A).Intramuscular adipose content as measured through histomorphometry techniques was markedly increased in the aged infraspinatus (8.28x, p = 0.043) and teres minor (13.76x, p = 0.033) muscle samples.No differences were noted in the intramuscular adipose content between the aged and younger groups in the supraspinatus (p = 0.688) and subscapularis (p = 0.637) groups (Figure 4B).

| Histopathology results
Typical degenerative histopathological characteristics were semiquantitatively assessed through Bonar scoring which revealed increased degenerative pathology in the aged tendons as compared with the younger tendons in the infraspinatus and subscapularis tendons.Specifically, the Bonar scores were increased by 180.6% (p < 0.001) and 237.1% (p = 0.067) in the aged infraspinatus and subscapularis tendons, respectively (Figure 4C).No significant differences were noted between the younger and aged supraspinatus and teres minor tendon scores, with both groups exhibiting moderate characteristics of chronic degeneration (Figure 4C).As outlined previously, the overall Bonar score is generated by summing the scores of four different histopathological categories (i.e., tenocytes, ground substance, collagen, and vascularity).The scores for all ages groups and tendons in each of these four categories are provided in    Tenocyte nuclei were more rounded and closer spaced with regional progression toward prominent cytoplasmic expansion (Figure 5, †).
Inter-bundle vasculature was often more prominent, though still thinwalled, with low to moderate numbers of mononuclear cells infiltrating the peri-vascular loose connective tissue.Rarely vessels were present within collagen bundles in areas of replacement by disorganized collagen.Vessels were most prominent in healthy and diseased tendon sections at the myotendinous junction.Here there was extensive interdigitation of the tendon fiber bundles with skeletal muscle.
Interspersed throughout the skeletal muscle were variably sized aggregates of adipocytes which were frequently associated with blood vessels.Additionally, biopsies of each muscle were examined.In these sections there was occasional extension of adipose into myofiber bundles (Figure 5, ‡), consistent with fatty degeneration.These features were noted across most sections; however, histomorphology assessment of intramuscular adipose content only revealed significantly altered changes between the younger and aged groups in the infraspinatus and teres minor tendons.

| DISCUSSION
As expected, the degradation of tendon biomechanical properties with age did not manifest similarly in all four rotator cuff tendons.
Specifically, these data demonstrate changes were most magnified in the supraspinatus tendon (3.06x difference) followed by the infraspinatus (1.76x difference), as evidenced by the difference in Young's modulus vs stretch curves between the younger and aged groups, especially at higher stretch levels.This finding is particularly interesting considering the increased prominence of supraspinatus tendon tears noted clinically. 1 Additionally, these data illustrate large differences in biomechanical properties between the four tendons of the human rotator cuff.Previous work has investigated changes to rotator cuff biomechanics secondary to aging in animals.Future work should include the utilization of these computational modeling data to design scaffolds with appropriate mechanical properties to better facilitate patient healing and improve RCR repair success rates.As mentioned previously, the sample size is the main limitation of this study.However, it is the belief of the authors that the sample size is adequate for the goal of illustrating differences between the four tendons comprising the rotator cuff and their changes with age.While sample size is limited, the utilization of nondestructive biomechanical testing with subsequent same-sample histological and gene expression analyses provides novel information on the aging and degeneration process of human tendons.Of note, this testing methodology required collection of a small tendon biopsy for the purpose of RNAseq analysis before biomechanical testing, which may have impacted said results.Additionally, while strain levels were confined to those replicative of in vivo loading, it is not possible to ensure biomechanical testing did not impact histology samples.
Finally, demographic data on donors (e.g., history of tendon pathology, etc.) were not known for specimens; as such, they are not provided in this work.Finally, while H&E-stained slides have been used to assess ground substance for use in the Bonar score, the measurement relies on inferences by the trained pathologist and should be interpreted as such.

| CONCLUSIONS
The purpose of this study was to characterize the changes to biomechanical, histopathological, and gene expression properties of all of the human rotator cuff tendons secondary to aging.
T A B L E 4 Categorical Bonar degeneration scoring results.

F
I G U R E 1 Stress-relaxation testing results.(A) Peak stress (λ = 1.06).The aged infraspinatus and subscapularis samples exhibited significantly decreased peak stress as compared with their respective younger age group samples.(B) Peak stress (λ = 1.08).The aged infraspinatus and subscapularis samples exhibited significantly decreased peak stress as compared to their respective younger age group samples.(C) Percent relaxation (λ = 1.06).The aged infraspinatus samples exhibited significantly increased percent relaxation as compared with the younger infraspinatus samples.(D) Percent relaxation (λ = 1.08).The aged subscapularis samples demonstrated significantly increased percent relaxation as compared to the younger subscapularis samples.Statistical test: Welch's t-test.Box plot indicates median and interquartile range; whiskers indicate maximum and minimum values.

F I G U R E 2
Young's modulus of younger versus aged groups across all four rotator cuff tendons.Across all rotator cuff tendons, the younger groups exhibited increased Young's Moduli values as compared to the aged groups.(A) The younger supraspinatus tendons exhibited moduli 3.06x greater than the aged supraspinatus tendons (95% CI = 3.03-3.10).(B) The younger infraspinatus tendons exhibited moduli 1.76x greater than the aged infraspinatus tendons (95% CI = 1.71-1.81).(C) The younger subscapularis tendons exhibited moduli 1.25x greater than the aged subscapularis tendons (95% CI = 1.19-1.30).(D) The younger teres minor tendons exhibited moduli 1.32x greater than the aged teres minor tendons (95% CI = 1.28-1.35 vascularity was observed in all aged tendons, with relative score increases of 55.6%, 56.5%, 48.1%, and 105.4% for the supraspinatus, infraspinatus, subscapularis, and teres minor tendons, respectively.Histologic sections of each tendon stained with H&E and examined under bright field microscopy appeared as parallel bundles of consistently staining collagen fibers interspersed with paucicellular elongate nuclei with little to absent cytoplasm, consistent with tenocytes.Between large bundles were rare, small caliber, thin-walled blood vessels supported by loose connective tissue.Degeneration was characterized by a loss of orientation and distinction (Figure5, *), with hyalinization of the bundles and/or loss of bundle distinction with replacement by disorganized collagen.This was best observed using polarized light where fiber disarray was demonstrated by a loss of regular, patterned transillumination in relation to fiber orientation.

F I G U R E 4
Histomorphometry and histopathology results.(A) Percent organized collagen across aged and younger groups in all four tendons.No significant differences were noted between age groups in any tendons.(B) Intramuscular adipose content.The aged infraspinatus and teres minor muscle biopsies exhibited significantly increased intramuscular adipose content as compared with their respective younger groups.(C) Bonar degeneration scores.The aged infraspinatus samples demonstrated significantly increased pathological degenerative characteristics as compared to the younger infraspinatus samples.Statistical test: Welch's t-test.Box plot indicates median and interquartile range; whiskers indicate maximum and minimum values.tears have found few, if any, tears to male patients through their 40s, 1,40 the existence of these histopathological changes suggest a much earlier onset of tissue deterioration before clinical symptoms.

Table 4 .
Individual category scores were increased in all four categories for all four tendons in the aged tendons as compared to score increases of 108.7%, 177.4%, 183.9%, and 102.2% for the supraspinatus, infraspinatus, subscapularis, and teres minor tendons, respectively.Tendon bundle organization and polarization was decreased in all aged tendons, with relative score increases of 60.0%, 38.3%, 44.4%, and 70.0% for the supraspinatus, infraspinatus, subscapularis, and teres minor tendons, respectively.Increased Best-fit values for Young's modulus versus stretch curves for all age groups and tendons.
). CI, confidence interval.T A B L E 2Note: Best-fit values were generated by fitting one line to group means (Y = B0 + B1*X + B2*X 2 ).
Note: Data are presented as means ± SD.F I G U R E 5 Representative histology micrographs.(A) Representative healthy tendon section from younger group.(B) Representative degenerated tendon section from aged group.(C) Representative healthy muscle section.(D) Representative muscle biopsy illustrating high level of intramuscular adipose.Decreased collagen organization (*) and increased tenocytes ( †) were noted in the degenerated tendon specimens.Increased intramuscular adipose content ( ‡) was noted in some samples.Scale bars: 100 µm.