|Home | About | Journals | Submit | Contact Us | Français|
Repeatedly and consistently measuring the mechanical properties of tendon is important but presents a challenge. Preconditioning can provide tendons with a consistent loading history to make comparisons between groups from mechanical testing experiments. However, the specific mechanisms occurring during preconditioning are unknown. Previous studies have suggested that microstructural changes, such as collagen fiber re-alignment, may be a result of preconditioning. Local collagen fiber re-alignment is quantified throughout tensile mechanical testing using a testing system integrated with a polarized light setup, consisting of a backlight, 90 deg-offset rotating polarizer sheets on each side of the test sample, and a digital camera, in a rat supraspinatus tendon model, and corresponding mechanical properties are measured. Local circular variance values are compared throughout the mechanical test to determine if and where collagen fiber re-alignment occurred. The inhomogeneity of the tendon is examined by comparing local circular variance values, optical moduli and optical transition strain values. Although the largest amount of collagen fiber re-alignment was found during preconditioning, significant re-alignment was also demonstrated in the toe and linear regions of the mechanical test. No significant changes in re-alignment were seen during stress relaxation. The insertion site of the supraspinatus tendon demonstrated a lower linear modulus and a more disorganized collagen fiber distribution throughout all mechanical testing points compared to the tendon midsubstance. This study identified a correlation between collagen fiber re-alignment and preconditioning and suggests that collagen fiber re-alignment may be a potential mechanism of preconditioning and merits further investigation. In particular, the conditions necessary for collagen fibers to re-orient away from the direction of loading and the dependency of collagen reorganization on its initial distribution must be examined.
Cyclic preconditioning is a commonly accepted initial component of most tendon mechanical testing protocols. Preconditioning provides tendons with a consistent “history,” and stress-strain results become repeatable, allowing for rigorous evaluation and comparison. Protocols frequently involve the repeated stretching of a sample to a sub-failure load to produce a repeatable mechanical response [1–4]. The internal structure of tendon changes in response to each loading cycle. After applying repeated cycles, a steady state is reached at which no further changes will occur unless the cycling routine is changed .
While it is widely accepted that preconditioning is important, changes that occur during preconditioning are not well understood. Micro-structural alterations, such as re-arrangement of collagen fibers, is one proposed mechanism of preconditioning [5–7]. Recently, a strong correlation between changes in collagen fiber alignment and changes in the mechanical response of ligament during cyclic tensile preconditioning has been reported . The correlation found between reduced force response during preconditioning and change in fiber alignment after preconditioning suggests that viscoelastic effects and microstructural reorganization both contribute to the time-and-history dependence of mechanical properties . However, the dependence of collagen fiber re-alignment during preconditioning on tendon location has not yet been examined. Additionally, collagen fiber re-alignment during stress relaxation or a tensile ramp-to-failure following preconditioning has not been examined in tendon.
Additionally, recent studies have shown that collagen fiber realignment varies by tendon location [8,9]. The tendon-to-bone insertion site of the supraspinatus tendon experiences higher strains and has been shown to have a more disorganized fiber distribution compared to the tendon midsubstance [8,10–12]. Therefore, the objective of this study was to locally measure collagen fiber re-alignment and corresponding mechanical properties throughout tensile mechanical testing to address mechanisms of preconditioning and stress relaxation as well as tissue nonlinearity and inhomogeneity in the rat supraspinatus tendon model. We hypothesized that fiber re-alignment will be greatest in the toe region of the ramp-to-failure test but that a change in circular variance will also occur during preconditioning despite the small loads used in the mechanical testing protocol. Additionally, we hypothesize that the collagen fiber distribution will become more disorganized throughout the stress relaxation test and that the mechanical properties and initial collagen fiber alignment will be greater in the midsubstance location of the tendon compared to the tendon-to-bone insertion site.
This study was approved by the University of Pennsylvania IACUC. Twenty-two Sprague–Dawley rats were sacrificed, and supraspinatus tendons (SST) were removed for mechanical testing. All soft tissue was removed from around the tendon, leaving the supraspinatus muscle-tendon unit attached to the humerus. The supraspinatus muscle was removed. Verhoeff stain lines were placed on the tendons denoting the insertion site and tendon midsubstance for optical strain analysis to denote regions for alignment analysis and to define the gauge section. The first stain line was placed at the tendon-to-bone insertion site on the humerus as described previously , and the remaining stain lines were placed at 2, 4, 7, and 9 mm from the insertion site. The cross-sectional area at each location was measured using a noncontact laser device with a resolution of 2 µm . The humerus was embedded in a holding fixture with the use of polymethylmethacrylate (PMMA). After the PMMA had set, the humeral head was sanded down to prevent the bone from impeding light from passing through the tendon insertion site for polarized light analysis. A second coating of PMMA was applied to prevent failure at the growth plate. The holding fixture was inserted into a custom testing fixture. The proximal end of the tendon was glued between two pieces of sandpaper and placed in custom grips for tensile testing. All samples had an initial gauge length of 7 mm and had an average width and thickness of 3 mm × 0.4 mm.
Samples were placed in a 37 deg phosphate buffered saline bath and loaded in a tensile testing system (Instron, Norwood, MA) integrated with a polarized light setup, consisting of a linear backlight (Dolan-Jenner, Boxborough, MA), 90 deg-offset rotating polarizer sheets (Edmund Optics, Barrington, NJ) on either side of the test sample, and a digital camera (Basler, Exton, PA) (Fig. 1) . Prior to testing, the encoder embedded in the stepper motor (Fig. 1) (Lin Engineering, Santa Clara, CA) that rotates the polarizer sheets was initialized by resetting the encoder value with the polarizer sheets set at a position corresponding to 0 deg of angular rotation. To determine biomechanical properties, tensile testing along the long axis of the tendon was performed using a 100 N load cell with a resolution of 0.001 N for all tests.
This study utilizes an established uniaxial mechanical testing protocol for rat rotator cuff tendon mechanical testing to examine and compare tendon mechanical properties at specific locations and in response to different treatments . At several points throughout the mechanical testing protocol, sets of 14 images were acquired as the polarizers rotated through a 125 deg range for measurement of fiber alignment during loading as previously described . Initially, samples were preloaded to a nominal load (0.01 N) and then preconditioned under load control for 10 cycles between 0.1 N and 0.5 N. Preconditioning cycles were performed at an average frequency of 0.22 ± 0.03 Hz and cycled between 0.98 ± 0.06% and 1.13 ± 0.03% grip-to-grip strain. Preconditioning is performed before the stress relaxation and ramp-to-failure tests to provide all samples with a consistent loading history to allow for more consistent and simplified mathematical interpretation of mechanical properties. The preconditioning loads used in this study were selected to ensure that the tendons would not be damaged during preconditioning and were selected based on pilot studies with intact and injured and repaired rat SSTs. The loads are low to enable the preconditioning protocol to be applied for both injured and intact tendon to compare their mechanical properties across studies. One 14 image alignment map was acquired both before preconditioning (but after the preload was administered) and immediately following the 10th cycle of preconditioning to determine if the spread of the collagen fiber distribution changed during preconditioning (that collagen fiber re-alignment occurred) (Fig. 2; points 1 and 2, respectively).
After returning to the lower load limit of 0.1 N following the 10th cycle of preconditioning, a 300 s hold was applied to allow the tissue to equilibrate before the stress relaxation test. In addition to the alignment map taken immediately following the 10 preconditioning cycles, an additional alignment map was acquired at the end of the 300 s hold to determine if collagen fiber alignment changed over the 300 s time course (Fig. 2; point 3). Next, to measure viscoelastic properties, samples were then subjected to a relative ramp of a 0.42 mm grip-to-grip displacement at a rate of 0.35 mm/s, followed by a 600 s hold to reach an equilibrium load. A series of alignment maps was taken every 5 s during the stress relaxation test to determine if the initial 0.42 mm grip-to-grip displacement as well as the subsequent 600 s hold affected the collagen fiber distribution.
Immediately following the 600 s hold, samples were returned to zero displacement by being displaced 0.42 mm at a rate of −0.35 mm/s. To measure toe-region properties during the ramp-to-failure, samples were returned to zero-displacement via displacement control; therefore some samples maintained a nominal load during the 60 s hold. Pilot studies demonstrated that a return to zero displacement was more consistent and repeatable than a return to zero load. An additional alignment map was taken following the return to zero displacement (Fig. 2; point 6). Following the 60 s hold, a ramp to failure was applied at a rate of 0.21 mm/s. Alignment map images were taken every 5 s during the ramp-to-failure. Images were also obtained every 5 s for optical strain analysis during the ramp-to-failure.
A custom MATLAB program (Matlab, Natick, MA) was used to optically track strain lines during the ramp-to-failure as previously described . Stress was calculated as force divided by initial area. A pilot study examined the best model to fit the load-to-failure data and determined the most accurate way to compare the optical strains measured to fiber re-alignment. A structural fiber recruitment model , an exponential model, and a bilinear fit model were all used on a subset of data. The pilot study determined that the bilinear fit model provided the most consistent and accurate representation of the data and therefore was implemented for this study. A bilinear curve fit was applied to the optical stress-strain data to quantify optical transition stress, optical transition strain, and the moduli in the toe and linear regions from the optical stress-strain data. Following the bilinear fit analysis, points representing the toe and linear regions of the optical stress-strain curve were selected to examine how the collagen fiber distributions changed during the toe and linear regions of the optical stress-strain curve. Alignment maps from the optical transition strain (calculated from the bilinear fit) and a point in the linear-region were selected for fiber alignment analysis.
Fiber alignment was calculated from the image sets as described . Briefly, images of the tendon surface were divided into rectangular areas (30-wide × 30-long = 900 areas). Pixel intensities were summed by area per image and plotted against angle of polarizer rotation. A sine wave was fitted to the intensity-angle data to determine the angle corresponding to minimum pixel intensity, which represents the average direction of the area’s collagen fiber alignment. A limitation of the crossed polarizer method is that fiber angles can only be calculated within a 90 deg range (±45 deg to the predominant fiber direction) rather than the entire possible range of orientations. To overcome this limitation, fibers were assumed to re-align in the direction of loading as previously described . Circular variance (VAR), a measure of the distribution of collagen fiber alignment (from Eq. 26.17 in Zar et al.), was calculated at several points throughout the mechanical test [17,18]. VAR was calculated for fiber distributions before preconditioning (BP), immediately after preconditioning (AP), following a 300 s hold before stress relaxation (300h), immediately after the SR displacement (SRdisp), at the end of the SR test (SR), after the SR test following a return to zero displacement (zero), at a point representing the toe-region of the optical stress-strain curve, and at a point representing the linear-region of the optical stress-strain curve (Fig. 2). Fiber re-alignment throughout the mechanical test was evaluated by comparing VAR values at two mechanical testing points. Collagen fiber re-alignment was said to occur during that region of the mechanical test if the difference of fiber distributions was found to be statistically significant. Fiber re-alignment during preconditioning was evaluated by comparing VAR values before and after preconditioning. Similar methods were used to determine fiber re-alignment during the 300 s hold following preconditioning, during the SR, after a return to zero displacement following SR, as well as in the toe and linear regions of the stress-strain curve. Mean angle values were calculated for collagen fiber distributions at each location for all mechanical test regions examined.
Shapiro–Wilk tests indicated non-normally distributed data for VAR values. As a result, nonparametric statistical tests were used for evaluating fiber realignment. Changes in fiber alignment (Friedman test) were compared for tendon location (midsubstance versus insertion) and for the mechanical test region (preconditioning, stress-relaxation, toe and linear region). Bonferroni corrections were used for multiple comparisons to remain conservative in our data interpretation and significance (P < 0.0125 = 0.05/4) was determined using Wilcoxon signed-rank post hoc tests. VAR data are presented as median ± interquartile ranges, and statistics are paired comparisons. Mechanical parameters were evaluated with parametric statistics. Changes in parameters were compared for tendon location (midsubstance versus insertion). Data are presented as means ± standard deviation.
VAR values and mean angle were examined at several stages of the mechanical test. Similar results were found for re-alignment behavior and mean angle changes at each location. During preconditioning, VAR values demonstrate significant re-alignment under tension (decreasing VAR). Additionally, the mean angle shifted an average of 4 deg in the midsubstance and 5.5 deg at the insertion site location during preconditioning. During the 300 s hold following preconditioning, no significant changes in re-alignment were found at either location. After the initial displacement was performed for the SR test, a decrease in VAR (increase in organization) was found at both locations. However, no significant changes in re-alignment were found throughout the 600 s stress relaxation test for either location (Fig. 3). Following the stress relaxation test, the sample was returned to zero displacement, and a more disorganized collagen fiber distribution was found at both locations. Finally, the collagen fiber distribution became more organized (decrease in VAR) throughout the ramp to failure. Significant differences in collagen fiber distributions were found at both the toe and linear regions at both locations (Fig. 3).
Locally, VAR values were significantly different (less organized at the insertion site compared to the midsubstance) at all mechanical testing points. Linear-region moduli were significantly greater in the midsubstance than in the insertion site (1.6 × greater, Fig. 4), and a trend was present for the toe-region moduli. Additionally, moduli values demonstrated that both the insertion site and midsubstance locations were highly nonlinear (~10 × linear/toe-region ratio). Optical transition strain (determined from the optical stress-strain data) was higher for the insertion site than for the midsubstance location (Fig. 5).
This study found a correlation between cyclic preconditioning and early re-alignment of collagen fibers regardless of location. Contradictory to our hypothesis, the largest amount of collagen fiber re-alignment occurred during preconditioning (Figs. 3 and and6).6). Additionally, the largest shift in mean angle also occurred during preconditioning. While previous work has noted some collagen fiber re-alignment during preconditioning, the large shift in alignment seen at such small loads in the present study was surprising [1,19]. These findings suggest that the re-alignment of collagen fibers may be an underlying mechanism of preconditioning as well as a potential explanation for the increase in tendon strength seen after preconditioning in highly aligned tendons. This result supports previous studies examining the post-preconditioning mechanical response of patellar and Achilles tendons [6,20]. The midsubstance location of the rat SST consists mainly of parallel oriented fibers that can be gradually re-oriented toward the direction of loading during preconditioning cycles. Additionally, a vector correlation algorithm has recently been used in ligament to detect microstructural changes in collagen fiber alignment during preconditioning and found a strong correlation between collagen fiber rotation and changes in force . The mean angle shift toward 90 deg found in the present study during preconditioning supports the concept that fibers re-align in the direction of loading and could account for the changes in mechanical response noted post-preconditioning [1,21–23]. Additionally, no changes in alignment were found following the 300 s hold applied post-preconditioning (Fig. 3). These findings indicate that holding the tendon at a constant load for 300 or less seconds may not signal a decrease in collagen fiber alignment with short recovery times. It is possible that the response of collagen fibers depends not only on the amount of relaxation time but also the initial displacement and loading history. After the 300 s hold was performed, a 0.42 mm grip-to-grip displacement was performed. Following the displacement, VAR values decreased, demonstrating an increase in collagen fiber organization at both locations.
Next, a 600 s stress relaxation test was performed. Contrary to our hypothesis, no change in collagen fiber re-alignment was found during the stress relaxation test (Figs. 3 and and6).6). Interestingly, crimp analysis in rabbit medial collateral ligament revealed no changes in collagen fiber crimp behavior before and after stress relaxation tests, while fibers were recruited during creep tests , supporting the idea that the microstructural mechanisms of creep and stress relaxation are different . This finding together with the lack of collagen fiber re-alignment during stress relaxation in the present study suggests that a shift in the structural organization of collagen fibers may not be responsible for stress relaxation.
Following the stress relaxation test, samples were returned to zero displacement, and a decrease in collagen fiber organization (increased VAR) was found compared to the distribution imaged at the end of the stress relaxation test at both the midsubstance and insertion site locations. This finding suggests that returning to a reference point (marked by a decrease in both load and displacement) may cause a decrease in collagen fiber alignment at both locations allowing the tendon’s collagen fibers to re-orient away from the direction of loading. This supports previous work in which a tissue-equivalent demonstrated a recovery of fiber alignment during a stress-free return to zero-displacement but not during a return to zero-force, suggesting that fiber re-alignment away from the direction of loading occurred only after a return to an un-stretched length during unloading . While the majority of fiber re-alignment occurred during preconditioning in our study, significant re-alignment was also found throughout the tensile ramp-to-failure test in both the toe and linear regions of the optical stress-strain curve for both the midsubstance and insertion site locations (Fig. 3). Additionally, no significant differences in collagen fiber organization were found between the two “rest” points (after the 300 s hold and following the return to zero displacement) at either locations (data not shown).
Further, this study reports the inhomogeneous, nonlinear mechanical properties and fiber alignment of the rat SST. After examining collagen fiber re-alignment throughout the entire mechanical testing protocol, similar patterns in collagen fiber realignment were found for fiber distributions between the midsubstance and insertion site locations at all mechanical testing points, indicating that the tendon functions as a continuous unit. Furthermore, local differences in mechanical properties and collagen fiber alignment, at all regions of the mechanical loading profile, demonstrate that the rat SST is inhomogeneous. The lower moduli (Fig. 4) and higher VAR values (signifying increased disorganization) found at the insertion site compared to the midsubstance location demonstrate that the multiaxial loads experienced at the SST insertion may affect both the structure and mechanics of the tissue. Additionally, a higher strain was required for the insertion site to reach the linear region than the tendon midsubstance (Fig. 5). This suggests that the more disorganized insertion site may require more microstructural changes before it is able to transition to the linear region of the stress-strain curve, such as collagen fiber crimp, which is believed to explain the toe-region of the stress-strain curve [25–29].
This study is not without limitations. First, the crossed polarizer method used in this study can only measure fibers ±45 deg of the tendon long axis (90 deg total range), instead of the full 180 deg range that would contain all possible fiber orientations. While other methods are available to determine collagen fiber orientation, we selected the crossed polarizer method as it is the simplest method to provide the data necessary for analysis to test our study hypotheses. An angle value correction was applied based on the assumption that fibers must reorient toward the direction of loading. Second, the observations of collagen fiber re-alignment behavior made in this study were for one specific mechanical testing protocol. Despite this limitation, this study provides insight into the understanding of tendon fiber re-alignment throughout the mechanical testing protocol in addition to examining potential mechanisms of preconditioning and stress relaxation. While collagen fiber re-alignment at the low loads seen during the applied preconditioning protocol may be a function of the mechanical testing protocol, imposed uniaxial boundary conditions, or removal of the native tension in the tendon, it is still important to understand what effects or influence the mechanical testing protocol and experimental design set up itself has on the measured mechanical properties because these are commonly used in the biomechanical testing literature and interpreting such data for clinical relevance. Preload and specific preconditioning protocols are not consistently reported in tendon literature. The results of this study suggest that the mechanical testing protocol itself may have implications on the consistency and repeatability of the mechanical properties measured; therefore, it is important to report the potential effects of preconditioning protocols, their implications on interpretations of mechanical properties, and relating future ex vivo work to the physiologic condition. Further, this study provides valuable data regarding the specific mechanical and organizational properties of the rat SST.
Future studies in our laboratory will investigate if the amount of collagen fiber re-alignment is dependent on the preconditioning protocol as well as studies to investigate if local changes in collagen fiber crimp are present to account for the differences noted in transition strain. It is possible that the insertion site is more crimped, requiring an increased strain to reach the linear region as noted in this study. Further studies are necessary to evaluate structural changes throughout a variety of mechanical testing protocols to determine if the changes observed are a result of the strain rate, amount of load versus the type of loading (load control, displacement control). Finally, the amount of time and conditions necessary for collagen fiber orientation to return to its original orientation is currently unknown. Future studies incorporating longer rest intervals are necessary to determine what conditions are necessary for fiber alignment to return to re-align away from the direction of loading.
The authors would like to acknowledge Elizabeth Feeney, David P. Beason, and Joseph J. Sarver for assistance and the NIH/NIAMS for financial support.