Different Vibratory Conditions Elicit Different Structural and Biological Vocal Fold Changes in an In-Vivo Rabbit Model of Phonation

Abstract

Objective:

Vibration of the vocal folds can disrupt the tissue and induce structural, functional and molecular changes; the presence or absence of contact between the vocal folds during vibration can affect the type and extent of these changes. The purpose of this study was to characterize vocal fold changes following two hours of contact phonation or phonation without vibratory contact.

Methods:

Six New Zealand white breeder rabbits underwent 120 minutes of phonation with or without vibratory contact, and four served as non-phonated controls. The larynx was exposed and current was applied to the cricothyroids bilaterally to achieve vocal fold adduction while humidified airflow was delivered to induce vocal fold vibration. Laryngeal position, airflow, and stimulation levels were adjusted to obtain phonation with or without contact, and phonation was elicited for 120 minutes. Following excision, larynges were stained using Hematoxylin & Eosin, Elastica van Gieson, and Grocott’s Methenamine Silver, or labeled with immunofluorescent markers for E-cadherin, CD31, CD11b, and Vimentin. All images were captured using a Nikon 90i microscope and analyzed using ImageJ.

Results:

Differences between vibratory conditions and control samples were observed. There was more extensive epithelial thinning, reduced epithelial integrity and increased vascularity in the contact phonation group, while both phonatory groups demonstrated a decreased presence of mucous on the luminal surface and a decrease in elastin band thickness and lamina propria depth. Neither condition showed differences in inflammatory cell presence compared to control tissue.

Conclusions:

By showing that these two vibratory conditions result in structural changes of different types and magnitude, we have provided the first empirical evidence that vocal fold tissue is sensitive to differences in forces, and that changes in vibratory pattern can elicit different downstream biological changes within the tissue. The differences described herein are an important step toward understanding the vocal folds’ potential for differential response to phonotraumatic damage following different vibratory behaviors.

1. Introduction

Voice disorders can arise from vocal overuse, and are hypothesized to be caused by chronic damage to the vocal fold tissue. In the case of lesion development, the tissue loses its ability to remodel and recover to its fully functional structure (1-4). Behavioral therapy is commonly recommended for the majority of phonotraumatic lesions (5), and it aims to alter patient phonatory behavior, believed to reduce various mechanical stresses to the vocal folds associated with overuse and misuse, with the ultimate goal of decreasing lesion size and preventing recurrence. Within this dogma is the assumption that there are inherent differences in the types of damage that occur to the vocal fold tissue depending on the nature of the phonatory vibration. Thus, the type and extent of damage to the tissue is at least partially dependent on phonatory behavior and may subsequently change the tissue’s ability to fully recover from such damage. The current study seeks to investigate the structural damage elicited in two vibratory conditions in an in vivo rabbit model of phonation.

To characterize structural changes, and the influence of vibratory pattern on their manifestation, two distinct vibratory conditions were utilized for the current study. One condition was characterized by phonation with consistent contact of the vocal folds in the striking zone during vibration, and the other characterized by a clear absence of this vibratory contact. It is important to note that these conditions were not chosen to represent vibratory behaviors typically elicited in humans, but to demonstrate structural change in two distinct vibratory conditions that can be reliably modeled in an in vivo rabbit phonation paradigm. Should these two conditions elicit different types and severities of damage following equal vibratory durations, it would lend credence to the clinical notion that behavioral modification of the way in which vibration is achieved can lead to structural and physiologic differences within the tissue. While neither type of phonation elicited within the current study are necessarily considered ‘high efficiency’ phonation, the two conditions are easily elicited in a non-overlapping manner (i.e. can be categorized in a binary fashion), and so allow for greater power in detecting significant differences between vibratory conditions in this study.

Vibration with vocal fold contact implies a large physical burden on the superficial structures of the epithelium as they collide with the contralateral vocal fold. In addition to the shearing and stretching forces associated with any vibratory movement, the tissues undergo significant deceleration forces as one vocal fold collides with the other (6,7). This is in contrast to the forces associated with vibration without contact, where collision forces are absent and deceleration rate is reduced, but stretching and shearing forces may be greater. Regardless of the type of forces (shearing, stretching, collision, etc.), repeated exposure to the various mechanical forces associated with vibration is the major contributor to alterations of the tissue’s native microarchitecture, i.e. phonotrauma, and the formation of benign vocal fold lesions (1). We hypothesized that there would be measurable differences with distinct damage profiles elicited in these two vibratory conditions, such that both groups exhibit changes to structures within the lamina propria compared to the control tissue, but the nature and severity of these changes would be different between the groups. We speculated that the contact group would exhibit greater damage to superficial structures while the non-contact group might exhibit more change in the deeper structures of the lamina propria. Areas of interest include characteristics associated with the biomechanical and structural integrity of the tissue, as well as markers of inflammation.

Structural characteristics assessed in this study include epithelial thickness and integrity, basement membrane and lamina propria thickness, and mucosal blanket integrity. Differences between conditions in any of these structural measures will indicate differences in biomechanical tissue tolerance to different types of vibration. Because an intact vocal fold epithelial layer is the first defense to physical trauma (8), as well as being critical to normal vibratory properties (9), the epithelium was an important target to assess in this study. The integrity and thickness of the epithelium will be assessed using an immunofluorescence label to E-cadherin, an epithelial barrier protein. The overall integrity and thickness of the epithelial barrier will provide insight to the overall structural damage endured by the vocal fold. Additionally, the presence of a thick mucosal layer paired with a thick and well hydrated basement membrane and lamina propria are fundamentally related to the vibratory function of the vocal folds (10) and were therefore also an important target of investigation in this study. Both the basement membrane and mucosal blanket are easily identified using specialized histochemical stains: Grocott’s Methenamine Silver (GMS) for the mucosal blanket; Elastica van Gieson (EVG) for the basement membrane. Lamina propria depth was identified by gross tissue structure analysis with Hematoxlin and Eosin (H&E) staining.

Further, greater structural changes may elicit a more significant inflammatory tissue response. The increased need for remodeling may also point to a higher risk of lasting and additive injury, as more substantial initial phonatory structural changes may not have yet had an opportunity to fully resolve when the next injury occurs. It is therefore important to understand how different vibratory patterns may contribute to the inflammatory process and wound healing cascade. Markers of inflammation include an increased presence of inflammatory cells which help to break down damaged tissue and clear away debris, and mesenchymal cells which aid in the structural remodeling of fibers within the damaged tissue (11). An increase in tissue vascularity is also expected, as dilation of the vessels within the tissue allows for greater access of nutrients and cells to the area of injury (12). All three of these markers of inflammation were identified and assessed using immunofluorescence labeling. The anti-CD11b antibodies target inflammatory cells, the anti-vimentin antibodies target mesenchymal cells, and the anti-CD31 antibodies targed the endothelial lining of vessels.

In order to better understand the contribution of vibratory patterns on vocal fold tissue changes, it is necessary to investigate the cellular responses associated with these behaviors. The current study investigated the structural, molecular, and inflammatory response to extended phonation exposure in two distinct vibratory conditions: 120 minutes of vibration with consistent contact at the mid-membranous portion of the vocal folds (contact condition), or 120 minutes of vibration without contact between the vocal folds (non-contact condition). The structural changes elicited in these two distinct vibratory conditions are compared to non-phonated healthy vocal fold tissue (control condition) and quantified using tissue sectioning and staining procedures. This investigation will build a foundation for understanding cellular changes associated with behavioral patterns observed in phonation.

2. Materials and Methods

2.1. Surgical Procedure:

Ten New Zealand white breeder rabbits (Oryctolagus cuniculus) were used for this study. Four animals served as controls, three animals underwent 120 minutes of experimentally-induced phonatory vibration characterized by vocal fold contact (contact condition), and three animals underwent 120 minutes of experimentally-induced phonatory vibration characterized by no vocal fold contact (non-contact condition). All procedures were approved by the Vanderbilt University Medical Center Institutional Animal Care and Use Committee (protocol M1600081).

The phonation procedure described in this study was based on previously described methods (13). Following induction of anesthesia, the rabbits were shaved from submentum to chest, exposing the neck. The anterior neck was dissected at midline from the hyoid bone down to the sternal notch to expose the larynx, and the trachea was transected proximal to the sternum. An un-cuffed endotracheal tube was placed in the lower portion of the trachea to provide a stable airway for spontaneous respiration, and a cuffed endotracheal tube was inserted into the upper portion of the trachea terminating approximately 1.5cm inferior to the opening of the glottis. The endotracheal tube was oriented to direct humidified airflow through the anterior portion of the glottis, and the cuff of this tube was inflated to seal the trachea. To produce vocal fold adduction, two pairs of custom stainless-steel hooked wire electrodes delivered electrical stimulation to the laryngeal musculature in 10 second trains (3 seconds on, 7 seconds off). One electrode was inserted into the belly of each cricothyroid muscle, and a second pair of electrodes was inserted into the cricothyroid membrane bilaterally. To induce electrical stimulation to the laryngeal apparatus, electrodes were connected to a Grass S-88 stimulator and constant current isolation unit. Humidified airflow at 37°C was delivered to the glottis using a Gilmont Instruments flowmeter and ConchaTherm Neptune humidifier to induce vocal fold vibration. Laryngeal position, airflow, and stimulation levels were adjusted to obtain consistent and audible phonation in both the contact and non-contact vibratory conditions, and experimental conditions were visually confirmed via high speed videoendoscopy every 15-30 minutes throughout the 120-minute phonation procedure (representative digital kymography images in Fig. 1). Digital kymography images were extracted from each high speed videoendoscopy sample and used to estimate fundamental frequency of vibration. A precision flowmeter (Alicat Scientific, Tucson, AZ) was positioned in series with the humidified airflow source upstream of the glottis, such that direct subglottic pressure measurement was possible. Direct, real-time subglottic pressure was collected every 30 minutes during the procedure.

Figure 1:

Digital Kymography. Representative digital kymography of rabbit vocal fold phonation in (a) the non-contact condition or in (b) the contact condition. A line scan of high-speed video at 8,000fps was selected through the mid-third portion of the vocal fold and is displayed over time from left to right.

Following 120 minutes of phonation, the animals were euthanized and the larynges were excised and bisected. The right half of the larynx was fixed with formalin, embedded in paraffin, sectioned, and processed for histochemical staining analysis. The left half of the larynx was fresh-frozen in Optimal Cutting Temperature compound (OCT), cryosectioned, and processed for immunofluorescence labeling.

2.2. Histochemical Staining and Imaging:

Paraffin embedded samples were sectioned using a Histocore Biocut microtome (Leica) at 5μm, mounted on glass slides, and stained with Hematoxylin and Eosin (H&E), Elastica van Gieson (EVG), and Grocott’s Methenamine Silver (GMS). Staining was performed in triplicate for each subject.

Images were captured along the apical surface of the superficial vocal fold, including the full depth of the lamina propria. Imaging was performed using a Nikon Eclipse 90i widefield microscope with Nikon Plan Fluor 20x/0.50 or 40x/0.75 coverslip corrected objectives and captured using a coupled Nikon DS-Fi2 color camera.

2.3. Immunofluorescence Labeling and Imaging:

OCT blocks were sectioned at 12μm using a CM3050S cryostat (Leica) and mounted on charged glass slides. Sections were immunolabeled with antibodies against e-cadherin (BD Laboratory, 610181, 1:100), CD31 (Novus Bio, NB600–562, 1:100), CD11b (Kingfisher Biotech, WS0780U, 1:200), and Vimentin (Abcam, Ab28028, 1:100), tagged with a fluorescent secondary antibody (Invitrogen, A-11005, 1:500), and counter stained with 4’,6-diamidino-2-phenylindole (DAPI) and mounted using Vectashield (Vector Laboratories).

Once the antibodies target their respective antigens, fluorescent labels are attached to allow for microscopic visualization and quantification of the cells/structures of interest. Immunofluorescence imaging was performed using a Nikon Eclipse 90i widefield microscope with Nikon Plan Fluor 20x/0.50 or 40x/0.75 coverslip-corrected objectives. Images were captured using a Hamamatsu Cl0600 Orca-R² digital CCD camera.

2.4. Image Quantification and Assessment

Mucosal blanket and basement membrane were qualitatively assessed in GMS stained samples. Epithelial thickness was quantified from H&E stained slides imaged at 40x magnification using the ‘Grid’ plugin (ImageJ) to manually measure the depth of the epithelium every 22.4μm (√500μm) along the length of the vocal fold edge, and averaged across the image and subject (Fig. 2a)(14). The number of breaks in the epithelial layer were quantified in 20x images of E-cadherin labeled samples using the “count” plugin (ImageJ). Similar methods to measure epithelial thickness were employed in EVG stained slides to quantify the thickness of the subepithelial elastin band.

Figure 2:

ImageJ Quantification. (a) demonstrates the grid added as an overlay (green lines) to each H&E image, such that the area of each square was equal to 500μm². The thickness of the epithelium at the center of each vertical grid segment was manually selected (yellow line), and the length of the line for each segment was populated in a separate window. For each analysed image, the compiled length measurements were exported for further statistical analysis. (b) represents an immunofluorescence image of CD31 labeling (red) with region of interest (yellow line) selected without the “Otsu” intensity threshold applied and then (c) with the intensity threshold (red) applied.

H&E stained samples were also imaged using a FITC (fluorescein isothiocyanate) fluorescent filter set and a Hamamatsu C10600 Orca-R² digital charge-coupled device (CCD) camera on the same microscope. These fluorescence images were used to visualize the autofluorescence of mature elastin fibers within the lamina propria. Images were captured at 20x magnification and the ‘Grid’ plugin was used to manually measure the depth of the lamina propria (perpendicular distance from basement membrane to muscle fibers) every 63.2μm (√4000μm) along the length of the vocal fold edge and averaged across the image and subject. Endothelial, inflammatory, and mesenchymal cell types were quantified in the lamina propria from CD31, CD11b, and Vimentin labeling respectively. The lamina propria was manually selected as the region of interest, and percent area of positive labeling was quantified using the “color threshold” function (ImageJ). For CD31 and CD11b “Triangle” threshold presets were used, while the “Otsu” thresholding algorithm was applied for Vimentin labeling (Fig. 2b-c). The “analyze particles” function was then used to quantify the percent area within the selected region of interest.

2.5. Statistical Analysis:

Statistical comparisons were performed for quantitative measures in Graphpad Prism V7. Statistical analysis of subglottic pressure and frequency of vibration was completed using a two-tailed unpaired Student’s t-test. Main effects of condition on structural changes were tested using a Kruskal-Wallis test of variance, followed by post-hoc Dunn’s test with correction for multiple comparisons. Data reported are mean +/− standard deviation, with significance determined at p<0.05.

3. Results

All quantitative results are provided in summary in Table 1.

Table 1:

Summary of quantitative findings and statistical analysis. Values are reported as averages across subjects and replicates between conditions with their respective standard deviations (SD). Subglottic pressure and frequency of vibration averages were calculated from values collected every 30 minutes during the 120-minute phonation period.

	Control (±SD)	Non-Contact (±SD)	Contact (±SD)	p-value
Subglottic Pressure (cmH2O)	N/A	11.2±0.36	10.8±0.17	<0.0001
Frequency of Vibration (Hz)	N/A	839±148	713±139	0.0023
Epithelial Breaks	0.0±0.0	0.1±0.3	3.3±1.8	<0.0001
Epithelial Thickness (μm)	9.9±2.2	9.5±1.1	8.2±1.7	0.023
Elastin Band Thickness (μm)	5.4±1.0	4.1±1.2	4.3±0.5	0.016
Lamina Propria Depth (μm)	233.9±68.7	170.8±32.8	170.3±10.8	0.006
CD11b %Area	0.4±0.3	0.3±0.3	0.7±0.5	0.270
Vimentin %Area	7.5±2.1	6.7±1.5	8.7±1.9	0.117
CD31 %Area	2.1±0.8	2.1±0.5	3.6±0.7	0.002

3.1. Subglottic Pressure and Frequency of Vibration

Due to the nature of vibration in an open glottal configuration, where more airflow and pressure is required to create vibration (15,16), the non-contact condition had significantly higher measured subglottic pressure than the contact condition (p<0.0001, Table 1). Complementary to this finding, and due to the glottal configuration necessary to ensure lack of midline contact, the non-contact condition also had a higher average fundamental frequency (p=0.0023, Table 1).

3.2. Mucosal Blanket

To assess mucosal blanket integrity and thickness, GMS stained vocal folds were observed for qualitative differences. Characterization of the mucosal blanket identified a thick band of black staining with no visible breaks or interruption in the control tissue, indicating that the mucin glycoprotein layer remains intact; however, disruption and thinning of the mucosal blanket was observed in the phonated tissue, with the mucosal blanket being the most disrupted in samples in the contact phonation condition (Fig. 3).

Figure 3:

Mucosal Blanket. Representative images of tissue stained with Grocott’s Methenamine Silver (GMS). GMS staining highlights mucous with a dark black stain, shown here along the surface of the epithelial layer. Tissue shown is in each of the three experimental conditions: (a) control, (b) non-contact phonation, and (c) contact phonation. Scale bars represent 20μm.

3.3. Epithelial Integrity

3.3.1. Epithelial Discontinuity:

E-cadherin is an integral epithelial tight junction protein present throughout the full length and depth of the vocal fold epithelium. To evaluate the effect of collision forces on epithelial damage, images of tissue sections labeled for E-cadherin were analyzed for “gaps” or “breaks” in the epithelial barrier (i.e. where discontinuity of labeling through the epithelium was observed). We identified a main effect of experimental group on the number of epithelial breaks (p<0.0001; Table 1, Fig. 4a-d). Post-hoc analyses demonstrated differences between contact phonation and both non-contact phonation (p<0.001) and the control condition (p<0.0001), with no differences observed between the non-contact and control conditions (p>0.99).

Figure 4:

Epithelial Changes. (a)-(c) are representative images of E-cadherin fluorescence labeling of the vocal fold epithelium, where E-cadherin is represented in green, and cell nuclei are represented in blue; (a) control, (b) non-contact phonation, and (c) contact phonation, and scale bars represent 50μm. (d) shows summative data and statistics for epithelial discontinuity, and error bars represent standard error of the mean, (e)-(g) are representative images of H&E stained sections of the vocal fold, demonstrating differences in epithelial thickness; (e) control, (f) non-contact phonation, and (g) contact phonation, and scale bars represent 20μm. (h) Shows summative data of epithelial thickness, and error bars represent standard error of the mean. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

3.3.2. Epithelial Thickness:

To assess the contribution of vocal fold phonation with vibratory contact on epithelial morphology, epithelial thickness was evaluated from mean measurements across the full length of the epithelium in H&E stained samples from the mid-membranous portion of the vocal fold. Analyses confirmed a main effect of condition on epithelial thickness (p=0.023; Table 1), with a significant decrease in epithelial thickness in the contact phonation group vs. the control group (p=0.03, Fig. 4e-h). Epithelial thickness appeared reduced in the non-contact vibratory condition compared to control, however this change was non-significant (p=0.12).

3.4. Extracellular Matrix Components

3.4.1. Elastin Band Thickness:

To assess the effects of vibratory contact on organization of extracellular matrix proteins within the superficial lamina propria, quantitative analysis of the subepithelial elastin band, a unique structure within the rabbit vocal fold characterized in a study by Pitman et al., was performed (17). There was a main effect of condition on elastin band thickness (p=0.016; Table 1), and the sub-epithelial elastin band was compressed in tissues from both the contact and non-contact phonation conditions compared to control tissue (p=0.04 and p=0.035, respectively; Fig. 5a-c). There was no significant difference in elastin band thickness between the two phonation conditions (p>0.99; Fig. 5a).

Figure 5:

Extracellular Matrix Components. (a) Summative data and statistics for elastin band thickness and (d) lamina propria depth across experimental conditions. Error bars represent standard error of the mean. * p<0.05. (b) and (c) are representative images of EVG-stained tissue sections in (b) control and (c) phonated tissue with the subepithelial elastin band stained in dark blue, identified with red brackets. Scale bars represent 20μm.

3.4.2. Lamina Propria Depth:

To further understand the effects of vibration on the extracellular matrix within the vocal fold lamina propria, lamina propria depth was assessed across conditions. Indeed, there was a main effect of condition on lamina propria depth (p=0.006; Table 1). Similar to the elastin band changes described in 3.4.1, images of tissue from the contact and non-contact phonation conditions showed a decreased lamina propria depth compared to controls (p=0.019 and p=0.025, respectively; Fig. 5d). There was no significant difference in lamina propria depth between the two vibratory conditions (p>0.99; Fig. 5d).

3.5. Vascularity

To evaluate vascularity of the vocal folds, the abundance of platelet endothelial adhesion molecule (PECAM-1, or CD31) immunolabeling was quantified. There was a significant main effect of vibratory condition on this vascularity marker (p=0.002; Table 1). CD31-positive labeling was increased in tissues that had undergone contact phonation compared to both the non-contact phonation condition (p=0.0045) and control tissue (p=0.0137, Fig. 6).

Figure 6:

Vascularity. (a)-(c) are representative images of CD31 fluorescence labeling of the vocal fold epithelium, where E-cadherin is represented in red, and cell nuclei are represented in blue; (a) control, (b) non-contact phonation, and (c) contact phonation, and scale bars represent 50μm. (d) shows summative data and statistics for CD31+ labeling, representing vascularity. Error bars represent standard error of the mean. * p<0.05, ** p<0.01.

3.6. Immune Cell Infiltration

To evaluate the effect of vibratory contact in phonotrauma on cellular infiltration of the lamina propria, we evaluated two cell types, CD11b-positive immune cells (e.g. macrophages, monocytes, natural killer cells, and granulocytes) and Vimentin-positive mesenchymal cells. Quantification of CD11b-positive immune cells in the lamina propria did not detect any significant differences across control or vibratory conditions (p=0.2696; Table 1, Fig. 7a). Similarly, Vimentin-positive labeling within the lamina propria was not found to differ between the control or vibratory conditions (p=0.1168; Table 1, Fig. 7b).

Figure 7:

Immune Cell Infiltration. Summative data for (a) CD11b+ cells and (b) Vimentin+ labeling across experimental conditions. No statistically significant differences were identified, error bars represent standard error of the mean.

4. Discussion

The changes in epithelial thickness and integrity are consistent with previous studies of phonotraumatic injury, which have demonstrated that either higher intensity or longer duration of phonation creates more damage to the epithelial surface than shorter, lower intensity phonatory events (18-22). The findings of the current study complement previous work, as contact phonation created greater overall damage to the superficial structures of the vocal fold (the mucosal blanket and epithelial layer) compared to non-contact phonation. These changes were deteceted even though the group with greater measured change had lower fundamental frequency. These findings support the conclusion that the changes detected and reported in this study are in fact a direct consequence of the vibratory condition, and not related to vocal dose.

An intact subepithelial elastin band in rabbit vocal folds has been previously associated with better vibratory function (17). In the current study, when this sub-epithelial elastin band was assessed, its integrity was maintained across all samples. However, its thickness was decreased in both phonation groups. It appears that this sub-epithelial elastin band becomes compressed following vibration, but the physiological implications of this change are yet to be determined. It may be that this structure is absorbing some of the shock associated with both compression and shearing forces of vibration, and thus protecting the underlying tissues. Even in samples where the epithelium was broken, the subepithelial elastin structure was maintained, and thus the overall gross structure of the vocal fold lamina propria was also maintained. This structure (and the basement membrane more generally) likely plays an important role in the protection of the lamina propria – even in the absence of an intact epithelial barrier. Disruption of this subepithelial layer was not achieved in the current study, suggesting that more significant trauma is required to disrupt this thick fibrous capsule.

Upon quantification of the depth of the lamina propria, depth was decreased in both phonation conditions compared to control. This suggests that the shearing forces associated with vibration elicit changes to the structural integrity of the lamina propria irrespective of the presence or absence of vocal fold contact. The results described here, wherein the depth of the lamina propria decreases with increased perturbation, are contrary to the known physiologic response to phonotraumatic injury, where edema increases the overall volume of the vocal fold tissue (12,23,24). The method of fixation may lend a possible explanation for this finding, as formalin fixation and paraffin infiltration (steps in the process of obtaining tissue sections for histological staining) is known to induce tissue shrinkage during processing (25-27). In the case of the tissues presented in this study, it may be that the tissue with greatest structural fiber integrity (control tissue) shrinks the least in response to paraffin infiltration, and tissues with greater structural perturbations have less integrity with which to resist the shrinking associated with tissue processing. With this perspective, the results presented in this study are better understood, as measured lamina propria depth may best be correlated to lamina propria fiber integrity, and not with pre-fixation water content. Thus, we conclude that phonated tissue (regardless of presence or absence of contact) has decreased structural fiber integrity within the lamina propria compared to control tissue.

Based upon the current findings, changes in vascularity of the lamina propria are also sensitive to differences in vibratory characteristics. Vasodilation is a well understood component of the inflammation and tissue repair process, and it is no surprise to see increased CD31 expression in phonated vocal fold tissue (28). Interestingly, it is only in the contact condition that there was an observed increase in the percent positive labeling of CD31. It appears that the forces associated with vibratory contact are necessary to elicit these vascular changes. It is possible that damaged epithelial cells drive this change in vascularity, which would align with described evidence of significant epithelial damage in the contact condition. While there is currently a paucity of research related to acute changes to vocal fold tissue vascularity following phonotrauma, there is a large body of literature that reports the increased vascularity of benign lesions (29). It is also well reported that varices and ectasias (visible capillaries in the superficial vocal fold tissue) place patients at higher risk of vocal fold hemorrhage (30). As such, increased vascularity in response to acute injury, while important in the wound healing cascade, may also place patients at higher risk of hemorrhagic lesions.

Vasodilation is an indication that the pro-inflammatory wound-healing cascade has been initiated, although there were no detected differences between conditions in the amount of inflammatory cell infiltration or in the density of vimentin-positive mesenchymal cells in the lamina propria at the time of tissue fixation/freezing. Previous literature has shown changes to pro-inflammatory cytokine expression following phonotrauma, but these expression levels tended to peak between 4 and 8 hours after the damage was inflicted (20). Despite our intention to detect early changes in the inflammatory wound healing cascade, there was likely not enough time allowed in this study for later repair processes such as the infiltration of inflammatory cells to begin. Due to the limitations of the tissue collection and processing methods utilized in this study, gene expression analyses were not performed. Future studies should assess gene expression of inflammatory cytokines as well as delay the tissue harvest past the two-hour phonation to confirm the infiltration of immune cells at a later time point following acute injury.

5. Conclusions

While the phonation conditions presented in this study do not directly translate to the behavioral modifications we aim to elicit in patients undergoing behavioral voice therapy, the differences described herein are an important step toward understanding the vocal folds’ potential for differential response to phonotraumatic damage. Further, this study serves as a proof of concept, showing that evoked phonation in a rabbit model can be effectively modulated to produce distinct vibratory patterns. The current study confirms that structural and biological changes to vocal fold tissue following phonation are dependent upon the vibratory characteristics of phonation. The vibratory conditions in this study, while not necessarily representative of vibratory characteristics in human phonation, serve to provide the first empirical demonstration that differences in vibratory patterns result in differences in structural and biological tissue changes. Continued work is required to further elucidate the cellular and molecular pathways that drive tissue repair following different types of vibratory trauma in humans, but this line of questioning has proven informative not only from a mechanistic and physiologic perspective, but also from a clinical one. Future work will continue to examine the ways in which behavioral voice modifications can be used to prevent vocal fold damage and ultimately reduce the severity of phonotraumatic damage and the incidence of benign phonotraumatic lesions.

Abbreviations:

CCD

Charge-coupled Device

CD11b

Cluster of Differentiation 11b

CD31

Cluster of Differentiation 31

COX-2

Cyclooxygenase 2

DAPI

4’,6-Diamidino-2-Phenylindole

EVG

Elastica van Gieson

FITC

Fluorescein Isothiocyanate

GMS

Grocott’s Methenamine Silver

H&E

Hematoxylin and Eosin

PECAM-1

Platelet Endothelial Adhesion Molecule