Vocal Fold Scar Treatment via Controlled Dexamethasone Delivery With a Light‐Activatable Implant

ABSTRACT

Objective

Intralesional steroid injections for vocal fold (VF) scarring are rapidly cleared, limiting their effectiveness. This study evaluates the efficacy of a light‐activatable dexamethasone implant compared to a single steroid injection for VF scar healing.

Methods

Ten rabbits underwent endoscopic VF injury and were treated with either a dexamethasone injection or a light‐activatable implant; injured‐only and no‐injury VFs served as controls. On days 0 and 21, VF implants were irradiated with a near‐infrared pulsed laser for 1 min. Larynges were harvested after 42 days. High‐speed video (> 10 kHz) of excised larynges captured VF vibrations for kymographic analysis. Normal force, structural stiffness, and displacement were measured. Data were analyzed with appropriate statistical tests.

Results

Compared to injury‐only VFs, implant‐treated VFs demonstrated significantly reduced normal force and stiffness (5.28 ± 0.77 mN vs. 2.34 ± 0.66 mN, p < 0.0001; 17.74 ± 2.45 mN/mm vs. 6.80 ± 1.32 mN/mm, p < 0.0001), and increased displacement at 1.96 mN (0.14 ± 0.02 mm vs. 0.29 ± 0.05 mm, p < 0.0001) along the injury zone. Implant‐treated VFs exhibited larger amplitude ratios and improved tissue architecture than untreated scars. No differences in quantitative measures were observed between implant‐treated and injection‐treated groups.

Conclusion

A light‐activatable dexamethasone implant improved VF biomechanics, vibratory behavior, and histological outcomes in a rabbit injury model comparable to a single steroid injection. This novel approach shows promise for delivering controlled, repeatable therapy to VF scars.

A light‐activatable dexamethasone implant was employed for the treatment of vocal fold (VF) scarring in a rabbit model. The implant enables controlled, repeatable drug release via near‐infrared laser irradiation, improving biomechanics, vibratory function, and histological architecture at the injury site. This approach offers a novel platform for localized, on‐demand therapy for VF scar management.

1. Introduction

Voice disorders affect an estimated 30 million Americans annually, incurring costs of nearly $13 billion, with significant economic and social impacts [1, 2, 3]. Chronic dysphonia diminishes quality of life, with vocal fold (VF) scarring being a common and difficult‐to‐treat etiology [4]. Scars often arise from iatrogenic trauma (e.g., laryngeal surgery or intubation), inflammation, or external injury. Fibrotic changes stiffen the VF, reducing pliability and causing hoarseness or breathiness [5, 6].

Intralesional steroid injections are one strategy to reduce fibrosis. Glucocorticoids are often administered at or soon after VF injury to modulate wound healing [7, 8, 9, 10, 11]. Steroid injections improve function short‐term but often require repeat dosing [12, 13], which is limited by VF access and risks such as atrophy and hematoma [14, 15]. Systemic or inhaled steroids are less effective and are prone to systemic side effects [8, 11, 16]. While surgery is another option, outcomes remain limited [17]. Thus, there remains an unmet need for a long‐lasting, safe, and controllable treatment for VF scarring.

Recent approaches in VF treatment have explored hydrogels and polymers, but these systems provide a fixed dose or release profile [18]. Our previous study demonstrated a light‐activatable implant for controlled drug delivery using light‐sensitive liposomes embedded in a biodegradable polycaprolactone (PCL) capsule [19]. The system is activated by a pulsed near‐infrared (NIR) laser to stimulate dexamethasone release. The implant demonstrated tunable drug release based on irradiation time and released therapeutically desirable doses. In addition, an in vivo study with a sham implant in rabbits confirmed its safety as a delivery vehicle in the VF [19]. This light‐triggered approach enables spatially and temporally controlled therapy.

This study builds on that platform to evaluate its therapeutic efficacy in a rabbit VF injury model. We tested whether our light‐activatable implant would improve healing effectively. We hypothesized that repeated laser‐triggered dexamethasone delivery would lead to improved biomechanical, vibratory, and histological outcomes over a single‐dose steroid injection.

2. Materials and Methods

2.1. Implant Preparation

The light‐activatable dexamethasone implant was fabricated using a biodegradable polymer capsule encapsulating drug‐loaded liposomes as previously described [19]. PCL (65,000–75,000 MW; PolyScitech, West Lafayette, IN) was used to form a nanoporous cylindrical capsule (5‐mm length × 0.46‐mm diameter). Liposomes were composed of 1,2‐distearoyl‐sn‐glycero‐3‐phosphocholine, 1,2‐distearoyl‐sn‐glycero‐3‐phosphoethanolamine‐N‐[methoxy‐(polyethylene glycol)‐5000] (Avanti Polar Lipids Inc., Alabaster, AL), cholesterol (Fisher Chemical, Waltham, MA), and stearylamine (Tokyo Chemical Industry Co. Ltd., Tokyo, Japan) in a 50:5:35:10 M ratio. Dexamethasone sodium phosphate (PCCA, Houston, TX) and sulfo‐Cy5 carboxylic acid (Cy5) (Lumiprobe Co., Hunt Valley, MD) were encapsulated into liposomes with gold nanorods (NanoPartz Inc., Loveland, CO) via a modified reverse evaporation method, and these liposomes were loaded into the capsule. Upon NIR irradiation, localized plasmonic heating from the gold nanorods disrupts liposomal membranes, releasing dexamethasone into the capsule interior that gradually diffuses through the nanoporous membrane in a controlled manner (Figure 1A). Passive drug release is minimal between activations.

FIGURE 1.

(A) Schematic of capsule implantation in rabbit VFs, laser irradiation and mechanism of drug release. Viewpoints of study workflow: (B) VF injury, (C, D) implant treatment, and (E) endoscopic guidance of fiber optic cable prior to irradiation.

2.2. Surgical Procedure

The study was approved by the University of Cincinnati Institutional Animal Care and Use Committee (#23‐05‐31‐02). Ten New Zealand White rabbits (2–3 kg) were anesthetized (35 mg/kg ketamine and 2 mg/kg xylazine for injection, followed by isoflurane for maintenance). VF injuries were created using an established injury model [20]. Under direct endoscopic visualization with a 3‐mm 0° rigid endoscope (Karl Storz, El Segundo, CA, USA) inserted through a Snowden‐Pencer Killian Speculum (V. Mueller, Germany), micro‐cup forceps were used to make precise incisions in the membranous VF epithelium and superficial lamina propria.

2.3. Injection and Implantation and Laser Irradiation

Following endoscopic confirmation of VF injury (Figure 1B), each rabbit received one treatment per VF (one treatment on the left and another on the right) in a randomized fashion. Treatments were (1) a 100 μL injection of dexamethasone (10 mg/mL), (2) a light‐activatable implant, or (3) no treatment—such that there were five VFs per condition. Additionally, five uninjured VFs served as controls. A 25G needle was used to inject dexamethasone into the superficial lamina propria using a cricothyroid approach through a small neck incision adapted from prior work [21]. For implantation, an 18G introducer pre‐loaded with the implant was advanced through the cricothyroid space and advanced beneath the thyroid cartilage into the VF (Figure 1C), with positioning confirmed endoscopically (Figure 1D). To trigger drug release from the implant, a NIR laser (1064 nm, 100 mW) on a fiber optic cable positioned ~1 mm from the VF was applied (Figure 1E). Following prior protocol [19], the laser was applied in three 20‐s cycles, with 40‐s cooling intervals between each cycle to prevent overheating, for a total irradiation time of 1 min. Activation was performed immediately following implantation (day 0) and 21 days post‐injury. Endoscopic images were taken during surgery and at weeks 3 and 6. Topical lidocaine (0.5 mL of 1 mg/mL) was applied to prevent laryngospasm. All rabbits were euthanized at 6 weeks post‐injury to harvest their larynges.

2.4. Ex Vivo Phonation Trial and High‐Speed Imaging

Each larynx was kept in saline at 4°C for phonation trials within 24 h. The vibrating membrane of the VFs measured 9–10 mm in length (vocal process to anterior commissure). Approximately five tracheal rings were preserved below the cricoid to allow mounting onto an airflow inlet. Supraglottic tissues were removed using a surgical blade. The larynx was mounted on an aerodynamic nozzle delivering conditioned airflow at controlled flow rates and pressures. The setup was adapted from previous work to match the nozzle diameter to rabbit trachea size [22]. Medialization was achieved via opposing prongs actuated by a three‐axis micrometric screw system. Airflow through the larynx was gradually increased until stable phonation occurred, typically 2–3 cmH₂O above phonation threshold pressure. Glottal dynamics were recorded using a high‐speed camera (Photron Nova S9; Photron Limited, Tokyo, Japan). For each trial, images were captured at a constant framerate of 10,000 or 12,000 frames per second, and resolution ranged from 25 to 55 pixels/mm depending on the working distance and magnification. These settings provided sufficient resolution to visualize VF vibrations. Only larynges with bilateral VF injuries were used.

2.5. Kymograph Generation and Analysis

High‐speed images were loaded into MATLAB using a custom‐made program. A line was drawn across the glottal opening in the mid‐membranous coronal plane, which corresponds to the region of the greatest amplitude during the VF cycle. A kymograph of at least 10 vibratory cycles was then extracted from this axis [23]. Each kymograph was manually analyzed in ImageJ (Version 1.54g, NIH, Bethesda, MD) to determine the amplitude ratio and acoustic metrics. Knowing the acquisition frame rate, the cycle period was measured and converted to fundamental frequency. Variations of frequency and amplitude were computed to obtain jitter and shimmer [24]. To compare the vibratory excursion of each unilateral VF in a manner across different VF sizes and conditions, an amplitude ratio was calculated (Figure 2). The minimal and maximal lateral positions of each VF edge within a single vibratory cycle were recorded, yielding a peak‐to‐peak excursion. The total bilateral glottal width—the distance between the two VFs at maximal glottal opening—was measured. The amplitude ratio was then defined as:

$$\text{amplitude ratio}=\frac{\text{unilateral}\mathrm{VF}\text{peak}-\text{to}-\text{peak excursion}}{\text{total bilateral glottal width}}$$

This produces a dimensionless measure that reflects how much a unilateral VF's excursion contributes to the overall glottal opening, normalizing out differences in VF size or image scaling to directly compare vibration amplitudes across groups. The values were averaged across 10 vibratory cycles for each kymograph.

FIGURE 2.

Kymograph generation and amplitude ratio measurement. Representative high‐speed images (left) at the opening (top) and closing (bottom) phases of the glottis, along with the corresponding kymograph (right). The yellow line indicates the mid‐membranous coronal plane from which pixel data were extracted to generate the kymograph. On the kymograph, the red bracket represents the two measurements used to determine amplitude ratio. The purple arrows denote the peak‐to‐peak excursion of one VF (the distance from its most medial to its most lateral position), while the blue arrows show the maximal glottal opening width between the left and right media edges. The amplitude is calculated by dividing the peak‐to‐peak excursion (purple) by the maximal glottal width (blue), producing a dimensionless measure of unilateral VF displacement.

2.6. Biomechanical Testing and Analysis

Following high‐speed imaging, the larynges were bisected along the mid‐sagittal plane. Custom molds using Ortho‐Jet resin (Lang Dental, Illinois, USA) were made for each sample, which were allowed to harden at −20°C for at least 4 h. Specimens were kept hydrated at room temperature in phosphate‐buffered saline (PBS) until measurement. Each sample was affixed to its respective mold for testing. Biomechanical measurements were acquired using an established VF indentation technique [25]. A 1.0‐mm diameter spherical indenter with a constant velocity of 0.45 mm/s and an indentation depth of 0.7 mm on a Mach‐1 v500css (Biomomentum, Laval, Quebec, Canada) with a 1.5‐N uniaxial load cell was used. Indentations were performed using a grid of at least 40 positions across the glottic and subglottic regions.

Load versus distance plots were acquired at each position during testing to extract structural stiffness, normal force, and normal displacement. Structural stiffness (load per displacement, reported in mN/mm) and normal force (load, reported in mN) were measured at an indentation depth of 0.3 mm, and normal displacement (tissue displacement, reported in mm) was measured under a 1.96 mN load. The mid‐membranous region along the VF edge was identified and designated as the “injury zone” for quantitative comparisons.

2.7. Histological Preparation and Analysis

VF samples were sectioned along the anterior–posterior axis and trimmed using two razor blades attached to a 3D‐printed 3‐mm spacer, with cuts spaced approximately 1.5 mm apart from the ventral edge. The trimmed samples were embedded in paraffin, and 5‐μm sections were cut and stained with hematoxylin and eosin (H&E) to visualize tissue morphology and measure epithelial thickness using ImageJ, and Masson's trichrome to assess collagen distribution within the lamina propria as an indicator of fibrosis. Slides were imaged at 10× magnification on a Nikon Ti2 inverted microscope (Nikon Corporation, Tokyo, Japan), and five epithelial thickness measurements were averaged to yield a mean thickness value per sample. For immunohistochemical (IHC) analysis, collagen type I (COL1A1) and fibronectin were evaluated to visualize tissue remodeling. Specifically, COL1A1 was used to detect collagen type I expression, while fibronectin served as a marker of ongoing wound healing.

Immunohistochemistry (IHC) staining was performed in part using the Cincinnati Children's Integrated Pathology Research Facility (RRID: SCR_022637) according to standard protocol. Paraffin‐embedded sections (4–5 μm) were deparaffinized and processed on the Ventana BenchMark ULTRA using the ULTRAView Universal DAB Detection Kit. For collagen I, antigen retrieval was performed with Cell Conditioner 2 (CC2) at 91°C for 68 min, followed by manual application of Collagen I antibody (MA1‐26771, ThermoFisher) at 1:200 for 32 min. For fibronectin, Cell Conditioner 1 (CC1) was used at 95°C for 56 min, with the antibody applied at 1:20 for 32 min. Slides were counterstained with hematoxylin and bluing reagent. PBS‐only controls on positive tissues confirmed staining specificity.

2.8. Statistical Analysis

Multivariate analysis (MANOVA) with Pillai's trace was performed to assess differences across experimental groups on the biomechanical data. When MANOVA results were significant, follow‐up one‐way analysis of variance (ANOVA) and post hoc Tukey's tests were conducted for each individual biomechanical variable. The kymographic amplitude ratio and epithelial thickness were analyzed by ANOVA with post hoc Tukey tests. Differences in the other acoustic metrics were assessed using the Kruskal–Wallis test. A p‐value < 0.05 was considered statistically significant. Statistical analyses were conducted in R (4.4.1) and RStudio (2025.05.1 + 513). Plots were drawn with GraphPad Prism (Version 10.5, GraphPad Software LLC., San Diego, CA). All values are presented as mean ± standard deviation, unless otherwise noted.

3. Results

3.1. Light‐Activated Implant Improves Injured VF Pliability

Stiffness heatmaps display changes in VF stiffness after injury (Figure 3A). The structural stiffness values in the injury‐only group were elevated relative to both treatment groups and the no‐injury VFs. The injury‐only VFs had a structural stiffness of 17.74 ± 2.45 mN/mm, which was greater than no‐injury VFs (7.26 ± 0.63 mN/mm, p < 0.0001) and injured VFs treated with either a single injection (6.65 ± 1.40 mN/mm, p < 0.0001) or light‐activated implant (6.80 ± 1.32 mN/mm) (Figure 3B). The injury group also displayed the greatest normal force (0.3 mm indentation depth) at 5.28 ± 0.77 mN, higher than the no‐injury VFs (2.36 ± 0.18 mN, p < 0.0001), injection‐treated VFs (2.31 ± 0.50 mN, p < 0.0001), and implant‐treated VFs (2.34 ± 0.66 mN, p < 0.0001) (Figure 3C). Normal displacement under a 1.96 mN force in the injury‐only VFs at 0.14 ± 0.02 mm was lower than the no‐injury group (0.27 ± 0.02 mm, p = 0.0002), injection‐treated group (0.28 ± 0.05 mm, p = 0.0001), and implant‐treated group (0.29 ± 0.05 mm, p < 0.0001) (Figure 3D). No differences were observed between the implant group, single injection group, or no‐injury group across these metrics (p > 0.05). Treatment with the light‐activated implant yields comparable biomechanical improvements to a single injection.

FIGURE 3.

Biomechanical changes of injured VFs after treatment. (A) Representative stiffness heatmaps of VFs of each experimental groups, with the red boxes on the heatmaps indicating the injury zone (mid‐membranous zone along VF edge). Comparing (B) structural stiffness, (C) force at 0.3 mm indentation, and (D) displacement at 1.96 mN load within the injury zone across groups. Force and displacement were measured along the axis normal to the VF surface. The heatmap color gradient range (in mN/mm) was standardized across all samples. The large circles on the boxplots represent the mean value within the injury zone of each sample; the small circles on the boxplots indicate individual measurements per sample. Significance: ***p < 0.001, ****p < 0.0001.

3.2. Light‐Activated Implant Treatment Increases Mid‐Membranous Excursion in Injured VFs

The acoustic metrics (fundamental frequency, jitter, and shimmer) of the injured VFs that were untreated or treated with either the single injection or implant are described in Table 1. Videokymographic analysis of injured larynges revealed that both treatments mitigated the vibratory deficits caused by scarring (Figure 4). Injury‐only VFs exhibited reduced mucosal wave motion. The amplitude ratio (unilateral peak‐to‐peak excursion normalized to maximal glottis width) in the injury‐only group was 0.42 ± 0.05 (n = 3), lower than the injection‐treated group (0.61 ± 0.03, p = 0.0020, n = 4) and the implant‐treated group (0.61 ± 0.03, p = 0.0023, n = 3). There was no difference between treatment groups (p = 0.919). Two larynges were excluded from analysis due to an inability to phonate during acquisition.

TABLE 1.

Unilateral acoustic metrics of the injured VFs.

Group	Sample size (n)	Fundamental frequency (Hz)			Jitter (%)		Shimmer (%)
		Mean	SD	p	Mean	p	Mean	p
Injury‐only	3	251.97	28.26	0.215	8.17	0.705	15.43	0.567
Injection	4	285.57	18.92		4.58		6.41
Implant	3	233.30	35.36		8.47		12.06

FIGURE 4.

(A) Representative unilateral kymographs for injured VFs at the mid‐membranous plane, and (B) amplitude ratios of injured VFs across different treatment groups. Kymographs display 10 VF vibratory cycles. Sample size: N = 3, 4, 3 for Injury Only, Dex Injection, and Light Implant, respectively. **p < 0.01.

3.3. Local Tissue Changes on Injured VFs Following Light‐Activated Implant Treatment

Histological VF tissue evaluations demonstrate differences in epithelial structure and extracellular matrix remodeling across groups (Figure 5). H&E staining showed disrupted epithelium and disorganized tissue architecture in the injury‐only group, while both treatment groups improved epithelial continuity and reduced inflammatory infiltration. Trichrome staining revealed dense, disorganized collagen in untreated scars, while collagen distribution appeared more organized and less extensive in the injection and implant groups. To further assess fibrosis‐related changes, immunohistochemical staining for COL1A1 and fibronectin was performed. COL1A1 was used to highlight collagen production during fibrotic remodeling, and fibronectin served as a marker of ongoing wound healing. Compared to the injury‐only group, both treatments showed lower expression levels and more localized distribution of these markers, suggesting reduced fibrotic activity. The implant group showed preserved tissue architecture with no evidence of a local inflammatory reaction.

FIGURE 5.

Representative histological sections from each treatment group 6 weeks after injury. (A–D) No‐injury control, (E–H) injury‐only, (I–L) dexamethasone injection, and (M–P) light‐activated implant. From left to right: H&E to assess epithelial and overall tissue morphology; Masson's Trichrome to evaluate collagen deposition within the lamina propria; immunohistochemical staining for COL1A1 and fibronectin as markers of fibrotic remodeling and wound healing. Black dashed lines (A, E, I, M) outline the epithelial layer, and LP indicates the lamina propria, the primary site of injury and matrix remodeling. Red dashed circles (M–P) indicate the location of the implant. Scale bar = 250 μm. Created in BioRender. Dion, G. (2025) https://BioRender.com/vgof0ka.

Epithelial thickness was greatest in the injury‐only group at 154.8 ± 18.1 μm, higher than the no‐injury group (98.4 ± 12.4 μm, p = 0.0001), injection‐treated group (103.1 ± 13.0 μm, p = 0.0003), and implant‐treated group (109.5 ± 9.1 μm, p = 0.0012) (Figure 6). No differences were observed between the no‐injury, injection‐treated, and implant‐treated groups (p > 0.05).

FIGURE 6.

(A) H&E‐stained section of the vocal fold showing anatomical landmarks including the epithelium (outlined with dashed line), lamina propria (LP), and thyroarytenoid muscle (TA). (B) High‐magnification image from the epithelial surface with yellow lines indicating five independent measurements of epithelial thickness. (C) Changes in epithelial layer thickness between groups. **p < 0.01, ***p < 0.001, n = 5 for each group. Created in BioRender. Dion, G. (2025) https://BioRender.com/r7en0fn.

4. Discussion

This study demonstrates that a light‐activated, controlled‐release dexamethasone implant can modulate VF wound healing, yielding outcomes comparable to an intralesional steroid injection. Our implant improved pliability, vibratory excursion, and local histological architecture. These changes were like those achieved by an intralesional injection, supporting the therapeutic potential of this novel approach. However, instead of relying on repeated, invasive intralesional injections, which are associated with side effects from bolus doses of dexamethasone, this light‐activated approach provides a non‐invasive alternative. It enables local, sustained drug delivery at lower doses, effectively minimizing side effects while optimizing therapeutic outcomes. To our knowledge, this is the first evidence that an on‐demand, light‐triggered drug delivery system that can attenuate scar formation and restore tissue properties.

Our light‐activated implant, as described in our previous study [19], continuously released dexamethasone over 6 weeks with controlled delivery upon laser irradiation confirmed by dye visualization endoscopically and after harvesting the larynx. In contrast, the sustained‐release implant showed no visible dye retention by week two, as most of the dye/drug was released and cleared within that timeframe. Fluorescence intensity and area analysis estimate that the drug released by the light‐activated implant in the rabbit VF at 6 weeks was approximately 102 μg with 1‐min irradiation, matching our in vitro observations [19]. Dexamethasone has a duration of action of 36–72 h [11], and its high potency has been shown on dose–response curves and the pharmacokinetics of local injections in preclinical studies [26, 27, 28, 29, 30]. Therefore, the amount of dexamethasone released by the light‐activatable implant at 6 weeks should continue to elicit anti‐inflammatory and antifibrotic effects in the VFs.

Consistent with previous reports of steroid efficacy in laryngeal scarring [7, 13], localized dexamethasone treatment led to a more supple VF with improved mucosal wave propagation than no treatment. Both treatments reduced stiffness and improved relative amplitude in the mid‐section of each VF, which is critical in phonation. The finding that the two spaced doses of light activation in the implant did not surpass a single injection suggests that additional tuning for optimized dosing is needed to maximize the antifibrotic benefit. Our previous work demonstrates that dexamethasone release from the implant extends past 42 days [19], suggesting the presence of unreleased dexamethasone at the study's endpoint. The implant may achieve comparable outcomes at a lower delivered dose, though it is possible the total released was equivalent to the bolus injection. The light activation also provides the capacity to control dose release timed to critical periods of healing. Future studies can explore additional or differently timed laser activations and higher drug loading to determine if outcomes can be further improved beyond what a single injection achieves.

One study limitation is the modest sample size in high‐speed data. While we found improvements in the mid‐membranous excursion with treatment, the low sample sizes may affect overall statistical power. Future studies with larger cohorts are essential to validate these results. Other limitations include using only a single set of laser parameters (e.g., irradiation time and laser power) and a single endpoint (precluding observation of longer‐term or functional voice outcomes). Future studies with multiple endpoints will be important to establish long‐term outcomes. Despite these limitations, the findings suggest that a light‐activatable implant can achieve therapeutic effects in scarred VFs without the need for repeated injections. Additionally, comparative evaluations with other emerging therapeutic strategies, such as stem cell and growth factor hydrogel injections [31, 32, 33], would provide a comprehensive assessment of the translational potential of this controllable delivery system as another clinical option for VF treatment. Clinically, this technology could reduce the number of injections. The implant remained well‐tolerated in the tissue, with no significant foreign body reaction observed, which is encouraging for biocompatibility.

5. Conclusion

This study shows that the treatment efficacy with the light‐activated implant is comparable to a single intralesional steroid injection. These results highlight the implant's potential to modulate VF wound healing and warrant further investigation into optimal dosing and activation timing.

Conflicts of Interest

The authors declare no conflicts of interest.

Cruz D. R. D., Awad N., Zheng A., et al., “Vocal Fold Scar Treatment via Controlled Dexamethasone Delivery With a Light‐Activatable Implant,” The Laryngoscope 136, no. 3 (2026): 1311–1319, 10.1002/lary.70149.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.