Modulation of Vascular Response after Injury in the Rat Achilles Tendon Alters Healing Capacity

Abstract

Tendons are relatively hypovascular but become hypervascular during both injury and degeneration. This is due to the angiogenic response, or the formation of new blood vessels, to tissue injury. The objective of this study was to evaluate the effect of vascular modulation in the rat Achilles tendons during healing. Fischer rats received a bilateral Achilles incisional injury followed by local injections of vascular endothelial growth factor (VEGF), anti-VEGF antibody (B20.4–1–1), or saline (SAL) either early or late during the healing process. Vascular modulation and healing were evaluated using multiple in vivo ultrasound imaging modalities, in vivo functional assessment, and ex vivo measures of tendon compositional and mechanical properties. The late delivery of anti-VEGF antibody, B20, caused a temporary reduction in healing capacity during a time point where vascularity was also decreased, and then an improvement during a later time point where vascularity was increased relative to control. However, VEGF delivery had a minimal impact on healing and vascular changes in both early and late delivery times. This study was the first to evaluate vascular changes using both in vivo imaging methods and ex vivo histological methods, as well as functional and mechanical outcomes associated with these vascular changes.

INTRODUCTION

Tendons are relatively hypovascular but become hypervascular during both injury and degeneration. This is due to the angiogenic response, the formation of new blood vessels, to tissue injury. Blood vessels have numerous roles within tissues to facilitate healing, including the delivery of oxygen and nutrients, the removal of waste products, the transport of regulatory factors, and the control of the immune response.¹ While angiogenesis is necessary for tendon healing, prolonged hypervascularization following tendon injury is not always believed to be beneficial.² An imbalance of pro- and anti-angiogenic factors could lead to abnormal angiogenesis creating vessels with structural and functional deficits. These deficits may lead to insufficient vascular response to enable healing, or conversely they can cause inflammation, inefficient nutrient exchange, and potentially hypoxia despite increased vascular density.^{1; 3} Therefore, there is uncertainty regarding the balance of angiogenic processes in the progression of tendon healing.⁴ While there has been promising research investigating the application of both pro- and anti-angiogenic factor delivery for tendon or ligament healing, the effect on healing outcome remains unclear due to both the lack of either vascular or mechanical evaluations, and to conflicting results between studies.^{2; 3; 5–10} Therefore, it is of great importance to investigate the effects of vascular modulation through the delivery of both pro- and anti-angiogenic factors after tendon injury.

To understand how angiogenic treatments alter the vascular response after tendon injury, there is a need for imaging methodologies that can longitudinally evaluate vascular characteristics. We previously used color Doppler ultrasound and photoacoustics imaging¹¹ to evaluate blood flow velocity, blood oxygenation, and percent area of detected blood flow, however, these measures they are only sensitive to larger vascular changes. To overcome this limitation, non-linear contrast-enhanced ultrasound can be implemented to provide more sensitive measures of both time- and amplitude-based measures of tissue perfusion.¹²

Therefore, the objective of this study was to evaluate vascular modulation in rat Achilles tendons during healing using multiple in vivo ultrasound imaging modalities, in vivo functional assessment, and ex vivo measures of tendon compositional and mechanical properties. We hypothesized that inhibiting the vascular response would decrease the healing response, resulting in reduced scar tissue formation and worsened mechanical properties, while increasing the vascular response would result in the opposite.

METHODS

Study Design

All procedures were performed in accordance to the University of Pennsylvania Institutional Animal Care and Use Committee. A total of 180 male Fischer 344 rats (4 months old, IACUC approved) underwent a bilateral Achilles incisional injury, followed by local injections of vascular endothelial growth factor (VEGF) (Peprotech, Rocky Hill, NJ), anti-VEGF antibody (B20.4–1–1, Genentech, San Francisco, CA), or saline (SAL). Injections were given daily from either days 0–2 (early delivery group) or days 4–6 (late delivery group) after injury. In vivo functional assessments and ultrasound imaging were performed, and animals were sacrificed at 7 (histology only), 14, and 28 days after injury for histological (n=6) and mechanical (n=12) evaluation with contralateral tendons used for separate assays. An image of the overall study design is shown in Figure 1. The tissue from the remaining 6 unused tendons for each experimental group were harvested and stored for future evaluations. Animals randomly assigned to treatment group, sacrifice time point, and evaluation procedure, and were housed 2 per cage with ad libitum access to food and water.

Figure 1:

Study design schematic. The bilateral Achilles incisional injury surgical procedure was performed on day 0. Angiogenic factors (B20, Saline, or VEGF) were injected daily on either days 0–3 (Early) or 4–6 (Late). Animals were sacrificed for histological evaluation (O) on day 7 (n=6), and for histological and mechanical evaluations (X) on days 14 and 28 (n=12). Contralateral limbs were used for separate assays, and the additional 6 limbs not used for histology were collected for future work. In vivo functional evaluations (F) were performed before surgery and at 3 (early only), 7, 10, and 14 days post-injury. In vivo ultrasound imaging (U) was performed on days 3 (early only), 7, 14, 21, and 28.

Surgical Approach

Animals were anesthetized with isoflurane inhalation and surgery was performed using aseptic technique. A skin incision was made on the medial side of the ankle to isolate the Achilles tendon. Using a 1.5mm flat scalpel blade (#61, MYCO Medical, Apex, NC), a partial-width, full-thickness incisional injury was made in the center of the tendon in the mid-substance region. The tendon was left unrepaired and the skin was sutured closed.

Angiogenic injections

Injections were delivered on 3 consecutive days on either days 0–2 (early) or 4–6 (late) after injury. Each animal received either 5µg VEGF in 20µl saline, 250µg anti-VEGF antibody (B20) in 20µl saline, or 20µl saline-only injected bilaterally intratendinously. Dosages were chosen based on literature values^{3; 6; 13–16} and our previous work.¹¹ Injections were administrated percutaneously in the coronal plane from the medial side of the tendon, with 10µl of the solution injected above and 10µl injected below the injury site.

Ultrasound Imaging

Imaging was performed on days 3 (early only), 7, 14, 21, and 28 (n=12/group) post-injury using a Vevo LAZR ultrasound (VisualSonics, Toronto, ON). Following isoflurane anesthesia, all hair was removed from the left hind limb and the animal was placed on a heated imaging table with the ankle secured at 90° flexion. The transducer was placed to image the sagittal plane, ensuring that the tendon was parallel to the surface of the transducer and the tendon length was in plane with the transducer length. The tendon was centered at a focal zone at 7mm image depth.

B-mode Alignment Analysis:

B-mode images were taken at a center frequency of 40MHz (MS550D transducer) with an effective resolution of 40μm for collagen alignment analysis. A motorized scanner was attached to the ultrasound transducer and consecutive sagittal images were taken every 0.1 mm over 3.5 mm (tendon width < 3mm). Image acquisition settings were constant for all specimens. The central 5 images were analyzed using MATLAB (Mathworks, Natick, MA). Briefly, tendon collagen fascicles appear hyperechoic, whereas the noncollagenous matrix between the fascicles appears hypoechoic, giving the appearance of bands in the images.¹⁷ These bands were analyzed to quantify tendon organization by detecting the fiber angles throughout the region of interest (ROI) and calculating the circular standard deviation (CSD) of the angles as described.¹⁸ CSD and echogenicity were calculated over the entire tendon area as well as within a 3mm² rectangular area over the injury region. Measures for each image segment for a specimen were averaged to obtain a representative value for the tendon.

Color Doppler Ultrasound Analysis:

Spatially sequential color Doppler ultrasound images were acquired every 0.1mm across the tendon over 3.5mm as described.^{11; 19} Image acquisition settings were constant for all specimens, and imaging persistence was used to remove motion artifact from the scanner. The central 8–10 images were analyzed using an IDL program (Harris Geospatial Solutions, Herndon, VA). The mean color level (average blood flow velocity), fractional area (% area of Doppler signal), and color weighted fractional area (weighted velocity, where the weighting factor is the fractional area of vascularity in the region of interest) were quantified over the entire tendon area and within a 3mm² rectangular area over the injury region.²⁰ Measures for each image segment for a specimen were averaged to obtain a representative value for the tendon.

Photoacoustic Analysis:

As described,¹¹ images were taken at two wavelengths (750 and 850 nm)²¹ within the center of each tendon. Blood oxygenation (sO₂ Avg), average hemoglobin (HbT Avg), and relative tissue oxygenation (sO₂ Tot) were quantified over the entire tendon area and within a 3mm² rectangular area over the injury region. Measures for all image segments for a specimen were averaged to obtain a representative value for the tendon.

Contrast-Enhanced Ultrasound Analysis:

Following anesthetization, a tail vein catheter was inserted. The Achilles tendon was visualized in non-linear contrast mode using the MS250 transducer (18MHz center frequency). The ultrasound clip was initiated at start of the bolus injection of 100µl of Definity (Lantheus Medical Imaging, Billerica, MA) microbubble contrast agent, followed immediately by a bolus injection of 200µl of saline. A 200 second video clip captured the wash-in and wash-out of the contrast in the tissue. The clip was loaded into a contrast analysis program, VevoCQ (VisualSonics, Toronto, ON).¹² A ROI was traced around the entire tendon area and motion artifact was removed using VevoCQ motion stabilization. Perfusion of the contrast agent was quantified for each frame of the clip. Video data was converted into echo-power data (linearization), which is directly proportional to the instantaneous concentration of contrast agent at each location in the ROI, and processed using a curve-fitting algorithm for a parametric perfusion model.¹²

Perfusion parameters derived from this model (Table 1)¹² were categorized by amplitude, time, and a combination of amplitude and time parameters. All amplitude-based parameters, such as peak enhancement, are expressed as relative echo-power measures. Time-based parameters, such as rise time or mean transit time, are expressed in seconds and define the kinetics of the contrast flowing through the tissue. Combined amplitude- and time-based parameters, such as perfusion index, describe measurements related to blood flow.

Table 1:

Perfusion parameters derived from the fitted curve model

Abbreviation	Parameter Name	Definition	Unit
PE	Peak Enhancement	Difference between maximum amplitude and offset baseline (proportional to relative blood volume)	[a.u]
AUC	Area Under the Curve	Area under the curve to infinite time	[a.u]
mTT	Mean Transit Time	Average amount of time for blood to pass through a region of interest	[s]
WiAUC	Wash-in Area Under the Curve	Area under the curve up to the point that PE is reached	[a.u]
RT	Rise Time	Time to go from baseline to PE	[s]
WiR	Wash-in Rate	Maximum slope of the fitted bolus function (proportional to local blood flow rate)	[a.u]
WiPI	Wash-in Perfusion Index	WiAUC/RT	[a.u]
PI	Perfusion Index	AUC/mTT	[a.u]

Gait Analysis

Hindlimb gait distances and ground reaction forces were measured in the day 14 animals (n=12/group) before injury and at 3 (early only), 7, 10, and 14 days post-injury using an instrumented walkway as described.^{22; 23} All parameters were averaged across walks on a given day for each animal, and all force data was normalized to the body weight of each animal at each time point. Data was collected using LabVIEW (National Instruments, Austin, TX) and parameters were analyzed using MATLAB (MathWorks, Inc., Natick, MA).

Passive Ankle Mechanics

Passive functional ankle joint properties^{23; 24} (n = 12/group) were measured before injury and at 3 (early only), 7, 10 and 14 days after injury. Torque was applied to the ankle with cutoffs at 25 Nmm in plantar-flexion and 35 Nmm in dorsi-flexion. Range of motion and joint stiffness were evaluated using MATLAB (MathWorks, Inc., Natick, MA).

Tendon Histology

Tendon samples from 7, 14, and 28 days post-injury (n=6/group) were dissected, paraffin processed, and sectioned at 5µm in the sagittal plane. Sections underwent hematoxylin-eosin (H&E) staining and immunohistochemical (IHC) staining for vascular endothelial cell marker (CD34), angiopoietin-1 (Ang-1), vascular endothelial growth factor (VEGF), collagen type III (Col III), tumor necrosis factor alpha (TNFα), and matrix metallopeptidase-13 (MMP-13). Processing protocol details are outlined in Table 2.

Table 2:

Immunohistochemistry Protocols and Reagents

Primary Antibody	Antigen Retrieval	Secondary AB/Amplification
Rabbit Anti-CD34 (Abcam, ab81289)	Heat induced at 75°C for 20min in 1mM EDTA, Ph 8.0	Vectastain Elite ABC HRP Kit (Vector Laboratories, PK-6200)
Rabbit Anti-Ang1 (Abcam, ab102015)	Heat induced at 95°C for 10min in 10mM Sodium Citrate, Ph 6.0	Vectastain Elite ABC HRP Kit (Vector Laboratories, PK-6200)
Rabbit Anti-VEGF (Abcam, ab46154)	Digestion in 0.5mg/mL Hyaluronidase for 60min at 37°C	EnVision+ HRP labeled polymer solution (Dako, K4002)
Mouse Anti-Col III (Sigma-Aldrich, C7805)	Digestion in 0.4mg/mL Protease K in 30mM Tris HCl for 4min at RT, 0.5mg/mL Hyaluronidase for 60min at 37°C, and 0.5N acetic acid for 4hr at 4°C	Secondary (BD Biosciences, 550331), ABC Amplification (Vector Laboratories, PK-6200)
Rabbit Anti-TNFα (Novus Biologics, NBP1–19532)	Digestion in 0.5mg/mL Pepsin in 0.1N HCl for 20min at RT	Secondary (Jackson Co, 111–035-003)
Rabbit Anti-MMP13 (Abcam, ab39012)	Digestion in 0.5mg/mL Hyaluronidase for 60min at 37°C	Secondary (BD Sciences, 550338), ABC Amplification (Vector Laboratories, PK-6200)

Histology images were taken in the injury region of the tendon at 50x magnification for CD34 (to view vascular structure) and 100x magnification for all other stains. H&E and CD34 were semi-quantitatively graded by three blinded investigators. H&E was graded for cell shape (1=spindle to 3=round shape) and cellularity (1=less cells to 3=more cells), and CD34 was graded for vessel density (1=less to 4=more dense) and vessel size (1=small to 4=large diameter). All other IHC stains were quantitatively analyzed for percent area of positive stain using a cluster analysis in MATLAB (Mathworks, Natick, MA) to separate out blue (hematoxylin cell nuclei), white (slide or tissue background), and brown (DAB positive stain) colors.

Tendon Mechanics

Tendons from 14 and 28 days post-injury (n=12/group) were prepared for tensile testing. The Achilles tendon was fine dissected, leaving the calcaneus insertion and foot intact. Verhoeff stain was applied to the tendon for optical strain measurement of the full tendon and the injury region. Tendon cross-sectional area was measured using a laser-based device.²⁵ The proximal side of the tendon was fixed between sandpaper at the 12mm stain line, and the foot was secured in polymethylmethacrylate. The specimen was positioned so the foot and the tendon were oriented perpendicular and submerged in a 37°C phosphate-buffered saline bath. Testing was completed using an ElectroPuls E3000 (Instron, Norwood, MA) with a 250N load cell, and consisted of (1) preloading (0.15N), (2) preconditioning (0.5% to 1.5% strain at 0.25Hz for 30 cycles), (3) stress-relaxation (6% strain for 10 minutes), (4) dynamic frequency sweep (0.125% strain amplitude at 0.1, 1, 5, and 10Hz, for 10 cycles each), and (5) ramp to failure (0.1% strain/sec). Images for optical strain measures were captured. Tendon viscoelastic and dynamic properties of percent relaxation, dynamic modulus (|E*|), and the tangent of the phase shift between stress and strain (tan(δ)) were computed. Quasi-static properties of stiffness, elastic modulus, max force, displacement, and stress were computed from the linear component of the ramp to failure test. Full tendon and injury site modulus calculations were performed by using the optical strain and cross-sectional area measurements of either the full tendon or injury regions.

Statistics

Sample sizes were determined through power analyses based on historical data using similar methods. As our hypotheses were only concerned with changes in the treatment groups relative to the saline group, changes over time and between Early and Late delivery groups were not evaluated. Statistical evaluations were performed on raw, non-normalized data for all study outcomes. Normally distributed data sets (ultrasound, mechanics, quantitative histology, and functional analyses) were analyzed using a 1-way ANOVA followed by Bonferroni multiple comparisons post-hoc tests with all comparisons made to saline control within each time point. The data was displayed as bar plots showing the mean and standard deviation. Non-normally distributed data (semi-quantitative histology grading for H&E and CD34) was analyzed using Kruskal-Wallis 1-way ANOVA followed by Dunn’s multiple comparison post-hoc tests with comparisons made to saline control within a time point. The data was displayed as box plots represent median and interquartile range, with raw data points plotted over the box plots. Significance was set at p≤0.05 (indicated by solid bars) and trends at p≤0.1 (indicated by dashed bars).

RESULTS

For all results reported, statistical comparison statements are all made relative to the saline control group within a given time point stated.

Late Delivery of Angiogenic Factors

Ultrasound:

Color Doppler and photoacoustic analysis representative images for the late delivery group are shown in Figure 2A. The B20 group had a decrease in fractional area (FA) and color weighted fractional area (CWFA) at day 7, and a decrease in mean color level (MCL), FA, and CWFA at day 14 in the full tendon region of interest (ROI). However, all three parameters were increased at day 21 (Figure 2B). For the VEGF group tendon ROI, MCL increased at day 21 (Figure 2B). Similar trends were observed when evaluating MCL in the injury ROI (Figure S1). No significant changes were observed in photoacoustic imaging parameters (Figure 2C and S1B).

Figure 2:

Late-delivery Doppler and photoacoustic imaging in the full tendon region of interest. (A) Representative images of Doppler (left) and photoacoustic (right) scans. Quantification of (B) Doppler imaging (with zoomed in frames for days 14–18 for Fractional Area and Color Weighted FA) and (C) photoacoustic imaging. There was an early reduction followed by a late increase in vascular parameters in the B20 group.

For non-linear contrast-enhanced ultrasound imaging, averaged curves are shown for days 7, 14, and 28 in Figure 3A. The B20 group injury region analysis demonstrated a decrease in peak enhancement (PE) at day 7 and decreases in PE, wash-in rate (WiR), area under the curve (AUC), and perfusion index (PI) at day 14 (Figure 3B, D, E, and G). The B20 group also had increased rise time (RT), indicating slower flow into the tissue, at day 14. However, by day 28 these findings were reversed and there was a decrease in RT and an increase in WiR in this group (Figure 3C–D). The full tendon analysis detected similar changes to the injury region analysis, however, fewer significant differences were observed (Figure S2).

Figure 3:

Late-delivery contrast-enhanced ultrasound in the injury site region of interest. (A) Representative curves at days 7, 14, and 28. (B–G) Quantification of time- and amplitude-based parameters demonstrate reduced vascularity at early time points and increased vascularity at late time points in the B20 group.

For the VEGF group, the contrast analysis of the injury region showed decreased peak enhancement and wash-in rate at day 7, along with increased rise time at days 7 and 14 (Figure 3B–D). While mean transit time was also decreased at day 14, it increased by day 21 (Figure 3F). Similar to the B20 group, the VEGF group full tendon analysis showed similar changes compared to the injury analysis (Figure S2).

B-mode ultrasound alignment analysis representative images are shown in Figure 4A. There was decreased circular standard deviation (CSD) of collagen fiber orientation, indicating more aligned tissue, at day 14 in the B20 group injury region (Figure 4D). Additionally, B20 increased echogenicity at days 7, 14, and 28 in the injury region (Figure 4E), and at day 28 in the tendon ROI (Figure 4C).

Figure 4:

Late-delivery B-mode ultrasound circular standard deviation (CSD) and echogenicity measures. (A) Representative b-mode images showing the full tendon (quantified in B–C) and injury site (quantified in D–E) regions of interest. B20 delivery caused a reduction in CSD, indicating increased collagen alignment, and an increase in echogenicity in the injury site.

Gait Analysis:

There were no significant differences in ground reaction forces (Figure S3) or paw placements (Figure S4).

Passive Ankle Mechanics:

There were no significant differences in range of motion (ROM) measures (Figure S5A).

Histology:

Representative images for H&E staining for cellular properties and CD34 staining for vascular endothelial cells at day 7 are shown in Figure 5A. There was an increase in vascular size in the VEGF group at day 7 (Figure 5D). CD34 staining demonstrated a decrease in vascular density in the B20 group at day 14 (Figure 5C). Immunohistochemical results are shown in Figure S7 and S8.

Figure 5:

Late-delivery H&E cellular and CD34 vascular histological analysis (scale bar 200 µm). (A) Representative images of H&E and CD34 at day 7. Histological images were graded semi-quantitatively for (B) H&E cell shape, (C) CD34 vessel density, and (D) CD34 vessel size. B20 reduced vessel density at day 14, whereas VEGF increased vessel size at day 7.

Mechanics:

The B20 group had a decrease in failure load, max stress, and tendon modulus at day 14 (Figure 6F–H). However, there was an increase in injury site modulus in the B20 group at day 28 (Figure 6I). Dynamic mechanical properties were not significantly different between groups at any time point (Figure S9).

Figure 6:

Mechanical evaluation of the late-delivery groups. B20 caused a reduction in mechanical properties at day 14, but an increase in injury site modulus at day 28.

Early Delivery of Angiogenic Factors

Ultrasound:

Early delivery group color Doppler and photoacoustic analysis representative images are shown in Figure 7A. In the B20 group, there was a decrease in FA, CWFA, and blood oxygenation at day 7 for the full tendon ROI (Figure 7B and C). However, there was an increase in blood oxygenation at day 14, and an increase in tissue oxygenation at day 28 (Figure 7B–C). The injury ROI results were similar (Figure S10). In the VEGF group, there was a decrease in blood oxygenation in both ROIs at day 3, and a decrease in FA in the injury region at day 7 (Figure 7C and Figure S10A–B)

Figure 7:

Early-delivery Doppler and photoacoustic imaging in the full tendon region of interest. (A) Representative images of Doppler (left) and photoacoustic (right) scans. Quantification of (B) Doppler imaging and (C) photoacoustic imaging demonstrates an early reduction followed by a late increase in vascular parameters in the B20 group.

Contrast-enhanced ultrasound averaged curves are shown in Figure 8A. The B20 group had increases in PE and WiR in both ROIs, and an increase in PI in the tendon ROI, at day 3 (Figure 8B and D; Figure S11A, C, and F). This was followed by increases at days 21 and 28 (Figures S11C–F). The VEGF group showed increases in PE and PI in the tendon ROI and increased in AUC in both ROIs at day 3 (Figure 8E and Figure S11A, D, and F).

Figure 8:

Early-delivery contrast-enhanced ultrasound in the injury site region of interest. (A) Representative curves at days 7, 14, and 28. (B–G) Quantification of time- and amplitude-based parameters demonstrated that both the B20 and VEGF groups showed some increases in vascularity at day 3.

B-mode ultrasound alignment showed a decrease in circular standard deviation (CSD) of collagen fiber orientation in the B20 group on days 14–28 in the injury site, and days 7 and 28 in the tendon (Figure 9B and D). Decreases in CSD indicate more aligned collagen than the saline group. This group also showed an increase in echogenicity at day 7 and 14 in both ROIs (Figure 9C and E).

Figure 9:

Early-delivery b-mode ultrasound circular standard deviation (CSD) and echogenicity measures. (A) Representative b-mode images showing the full tendon (quantified in B–C) and injury site (quantified in D–E) regions of interest. B20 delivery caused a reduction in CSD, indicating increased collagen alignment, and an increase in echogenicity.

Gait Analysis:

There was a decrease in breaking force in the B20 group at day 3 (Figure S12C). No other parameters showed significant changes with treatment.

Passive Ankle Mechanics:

Both treatment groups had decreased plantarflexion linear stiffness at day 10 (Figure S14E).

Histology:

Representative images for H&E staining of cellular properties and CD34 staining of vascular endothelial cells at day 7 are shown in Figure 10A. There were no significant changes in H&E measures of cellularity or cell shape for any group at any time point (Figure 10B and Figure S15). Immunohistochemical quantitative analysis showed a significant increase in TNFα in the B20 group at day 14 (Figure S16E).

Figure 10:

Early-delivery H&E cellular and CD34 vascular histological analysis (scale bar 200 µm). (A) Representative images of H&E and CD34 at day 7. Histological images were graded semi-quantitatively for (B) H&E cell shape, (C) CD34 vessel density, and (D) CD34 vessel size. No significant changes were observed.

Mechanics:

When evaluating the early delivery group Achilles tendon mechanics, the only significant difference was a decrease in percent relaxation in the VEGF group at day 28 (Figure 11C). All other parameters were unchanged with treatment.

Figure 11:

Mechanical evaluation of the early-delivery groups. VEGF caused a reduction in percent relaxation at day 28.

DISCUSSION

This work demonstrated that alterations in vascular response after injury impact tendon healing outcomes. The late delivery of anti-VEGF antibody, B20, caused a temporary reduction in healing capacity at the time when vascularity was also decreased, but an improvement in healing capacity at a later time point when vascularity was increased. However, the early delivery of B20 did not cause mechanical property changes, and VEGF delivery had minimal impact on healing and vascular changes in general.

There were more drastic changes observed with the delivery of B20 than the delivery of VEGF. As hypothesized, B20 delivery decreased vascularity initially, which is supported by studies of anti-VEGF treatments in tumors.^26–30 However, the B20 group caused increases in vascularity at later time points, similar to our previous study.¹¹ Specifically, the vascularity decrease in the late group was sustained through day 14, which likely caused the reduction in mechanical properties observed at that time point. This decrease is supported by previous studies. In an ACL healing study, the expression of a VEGF antagonist caused a significant reduction in biomechanical strength.³¹ Additionally, the delivery of an anti-VEGF drug slowed cellular proliferation and delayed epithelial healing in the cornea.³²

Both the early and late B20 groups showed improvements in collagen alignment, which is supported by a tendinopathy study where an anti-VEGF antibody drug resulted in more organized collagen.¹⁰ This data suggests that B20 may change the mechanism in which the injured tissue is remodeling. The late group only showed an improvement on day 14, which is the same time point that demonstrated decreases in mechanical properties. While this is counter intuitive, there are no negative changes in mechanics in the early group where there are more pronounced improvements in tissue organization. Therefore, these organizational changes may be countering other mechanisms induced by the decrease in vascularity contributing to the reduction in mechanical properties in the late group.

While the VEGF treatment caused multiple changes, they were in fewer parameters and at less time points than the B20 treatment. Specifically, in the late group, the largest vascular increases manifested at later time points (day 21) in ultrasound evaluations. Additionally, VEGF caused a reduction in the speed of perfusion of the tissue at early time points. However, in the early group, there was an increase in amplitude-based contrast-enhanced ultrasound parameters early, but a decrease in blood oxygenation. Histologically, VEGF caused an increase in vessel size but not an increase in vessel density. These findings evaluated together indicate that VEGF is causing more structural changes to the forming vessels than increasing the density of new vessels. During angiogenesis, VEGF expression causes vessels to become more unstable and leakier to allow for branching and vessel growth. Therefore, larger but possibly more leaky or disorganized vessels could cause a slower perfusion of the tissue with a decreased blood oxygenation. However, the larger vessels could explain the increases in area under the curve and peak enhancement at early time points. This finding is also supported by a study evaluating how the dosage of VEGF effects vessel morphology. Specifically, lower dosages of VEGF induce increased vessel density, whereas higher doses increase vessel size.³³

Our previous work demonstrated more increases in vascularity with VEGF delivery after injury than in this work.¹¹ These smaller changes could be due to the use of younger animals with a more robust vascular response to injury. Interestingly, the only mechanical change observed with this treatment was a decrease in percent relaxation (less viscous and more similar to healthy tendon values). Since there were few alterations to mechanical properties, it is not surprising that there were no changes in histology. When comparing to previous literature, there is support for increased vascular response with VEGF delivery during tissue healing,^{6; 7; 31; 34–36} including tendon healing However, most studies evaluate vascular changes using only histology, where we performed a comprehensive vascular analysis using in vivo ultrasound imaging. Contrary to our results, many of the tendon or ligament studies evaluating VEGF delivery found improvements in mechanical properties.^{3; 5; 31} However, none of the tendon or ligament studies with mechanical evaluation directly measured what vascular changes occurred with their treatment, so it is difficult to compare to these models. Skin healing models have also reported improved healing outcomes and increased vascularity with VEGF treatment. ^34–37 It is possible that our VEGF delivery model did not cause a large enough vascular change to induce the expected improvements in healing.

In addition to mechanical outcomes, we evaluated in vivo measures of joint function. While there were few transient changes observed, most parameters were not affected, suggesting insignificant alterations to animal ambulation and ankle function with these treatments. This information is beneficial because though we did not find improved healing, we know that this level of vascular modulation does not create detriments to joint function, which could be promising for future studies.

In this work, we successfully implemented color Doppler, photoacoustics, and non-linear contrast-enhanced ultrasound imaging to provide multiple metrics for evaluating in vivo vascular alterations, including vessel density, blood flow velocity, blood oxygenation, and time- and amplitude-based perfusion parameters. This is the first time that all of these in vivo vascular properties were evaluated in the context of tendon healing. Interestingly, we found that Doppler ultrasound and photoacoustics are just as sensitive to changes in vascularity as contrast-enhanced ultrasound. This is surprising given that contrast can detect perfusion through much smaller vessels. It is possible that the larger vessels are undergoing more of the changes due to these treatments, particularly in the case of VEGF delivery, allowing these less sensitive ultrasound measures to also detect alterations due to treatment and injury. This hypothesis is supported by the late delivery histological results, where the B20 group generally caused a reduction in vessel density and the VEGF group generally caused an increase in vessel size as compared to the saline group. Another potential explanation is that since microvascular flow is closely correlated to the upstream blood flow through larger blood vessels, the changes in small blood vessels are also detectable in larger blood vessels. Finally, we found that these imaging parameters were more consistent at detecting vascular changes than the more commonly used histological methods.

This work is not without limitations. We chose to perform this analysis using a partial laceration model instead of a full laceration model, which does not exactly parallel the human condition. However, we believe that the partial model was the best choice for this particular study design because it did not require suture repair or limb immobilization for proper end-to-end healing to occur, both of which would greatly interfere with ultrasound imaging and functional analysis at early time points. Additionally, we were able to perform bilateral surgeries using the partial model, which decreased the number of animals required for this study design. Another limitation is that we only evaluated one dosage for each factor. While we previously tested multiple dosages of B20,¹¹ there was ample literature support for the VEGF dosage. However, we found only mild vascular changes with VEGF and no major differences in healing outcome, suggesting a higher VEGF dosage may be necessary to induce more vascular changes for this model. Another limitation is that we did not evaluate other angiogenic factors, such as bFGF and PDGF-BB,^{8; 38; 39} or other anti-angiogenic therapies^{26; 27; 29} that may induce vascular changes through different mechanisms of action. However, the factors evaluated in this work were chosen due to significant literature support to induce the desired vascular changes. Further, in addition to vascular changes, VEGF has also been shown to promote collagen production, which we hypothesized may increase tendon mechanical properties.⁴⁰ Additionally, while we did perform repetitive dosing over the course of 3 days for each treatment group due to the concern of quick clearance of these drugs, a sustained delivery model may have allowed for the drug to remain in the injury area longer and provide more significant changes. This could be a consideration for future work in this space. Another limitation is that we only evaluated mechanical outcomes at days 14 and 28. It is possible that mechanical changes existed in the early delivery group at earlier time points when B20 induced a reduced vascular response, but this would be transient since there were no changes later. While we did not include any analysis of uninjured tendon properties in this study, general comparisons can be made to previous work.^{41; 42} The mechanical properties of the tendons for all 3 groups are similar to uninjured properties at 28 days post-injury, with max stress around 20 MPa and tendon modulus around 300 MPa, with the late B20 injury site modulus closest to uninjured properties. Additionally, circular standard deviation at the injury site generally remains slightly elevated compared to uninjured tendon data, indicating the tendon structure is not fully back to normal at 28 days post injury. However, the B20 groups do exhibit properties closest to uninjured tendons. While these general comparisons can provide some additional context, direct comparison with these studies is not possible due to the use of different rat ages and strains in the previous work. We also did not perform other biological assays besides immunohistochemistry. These methods were chosen to evaluate the localized deposition of proteins, but the sensitivity for quantifying immunohistochemistry is limited. Further analysis specifically looking at collagen production on a molecular level could prove to be beneficial as VEGF has been shown to increase collagen production in tendons.⁴⁰

This study was the first to evaluate vascular changes using both in vivo imaging methods and ex vivo histological methods, as well as functional and mechanical outcomes associated with these vascular changes. This work is also the first to substantially use new imaging methods, such as photoacoustics, to evaluate blood oxygenation alterations during tendon healing. We demonstrated that reducing the vascular response following injury impairs healing potential only at early time points, but may improve healing potential at later time points when vascularity is increased. While these results do not provide a straightforward answer regarding the benefits or deficits of angiogenic treatments, they clearly demonstrate that vascular perturbations impact healing capacity uniquely depending on dosage and timing of delivery, and further work could help to titrate these effects to produce a beneficial outcome.

Clinical Significance:

This work demonstrates that the alteration of vascular response through the delivery of angiogenic growth factors has the ability to alter tendon healing properties.