The development of personalized medicine is a primary objective of the medical community and increasingly also of funding and registration agencies. Modeling is generally perceived as a key enabling tool to target this goal. Agent-Based Models (ABMs) have previously been used to simulate inflammation at various scales up to the whole-organism level. We extended this approach to the case of a novel, patient-specific ABM that we generated for vocal fold inflammation, with the ultimate goal of identifying individually optimized treatments. ABM simulations reproduced trajectories of inflammatory mediators in laryngeal secretions of individuals subjected to experimental phonotrauma up to 4 hrs post-injury, and predicted the levels of inflammatory mediators 24 hrs post-injury. Subject-specific simulations also predicted different outcomes from behavioral treatment regimens to which subjects had not been exposed. We propose that this translational application of computational modeling could be used to design patient-specific therapies for the larynx, and will serve as a paradigm for future extension to other clinical domains.
Citation: Li NYK, Verdolini K, Clermont G, Mi Q, Rubinstein EN, Hebda PA, et al. (2008) A Patient-Specific in silico Model of Inflammation and Healing Tested in Acute Vocal Fold Injury. PLoS ONE3(7): e2789. https://doi.org/10.1371/journal.pone.0002789
Editor: Marie Csete, California Institute for Regenerative Medicine, United States of America
Received: February 28, 2008; Accepted: May 12, 2008; Published: July 30, 2008
Copyright: © 2008 Li et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The study was supported in part by the National Institutes of Health grants R01-DC-005643 and R01-DC-008290 to K. Verdolini from the National Institute on Deafness and Other Communication Disorders; grant P50-GM-53789-09 (Project V) from the National Institute of General Medical Sciences to Y. Vodovotz and G. Clermont, and grant H133E070024 (Developmental Project 1) from the National Institute for Disability Rehabilitation Research to Y. Vodovotz and Q. Mi. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: Ms. Li, Dr. Verdolini, Dr. Clermont, Dr. Mi, Dr. Hebda and Dr. Vodovotz are the co-inventors of this agent-based model of vocal fold inflammation, which is covered under a pending patent application entitled “Modeling Wound Healing” (Serial Nos. 60/850,690 and 60/850,896; PCT/US2007/080893).
The vocal folds are exposed to nearly continuous biomechanical stress during phonation. Increased intrafold contact stresses associated with certain voicing patterns can result in structural damage to the vocal fold mucosa , . Specifically, phonotrauma can (1) alter the tissue's physical properties by disrupting intracellular adhesion , and (2) modulate the tissue's cellular/molecular responses by altering gene expression . Persistent stress can further lead to tissue disorganization , , , stimulation of extracellular matrix synthesis , , and ultimately, the deposition of frank phonotraumatic lesions, dysphonia, and quality-of-life changes –.
Our long-range goal is to generate a technology that will allow clinicians to prescribe a personalized vocal exercise (or rest) program that should optimize tissue healing in cases of both acute and chronic phonotrauma –. The first-line approach to the treatment of phonotrauma is usually behavioral –. Traditionally, behavioral voice treatment involves complete or partial voice rest with the hope that the ensuing inflammatory response will subside on its own . However, our recent in vitro data from vocal fold fibroblast cultures and human data derived from concentrations of inflammatory mediators in laryngeal secretions suggest that contrary to clinical wisdom, some forms of vocal fold tissue mobilization—specifically mobilization in so-called “resonant voice” (roughly, classical singing) exercises may be able to modulate the inflammatory and healing process by blunting the cells' pro-inflammatory responses as well as enhancing their anti-inflammatory responses. From the wound healing perspective, this modulation may optimize the healing outcomes of the stressed/traumatized tissue in acute phonotrauma , . Equally important, both in vitro and human clinical data suggest that the benefits of these tissue mobilization exercises for acute vocal fold inflammation are dose-dependent . These observations suggest a commonality across tissue types in the response to injury, given that active rehabilitation is now used to treat many types of injuries –. However, details are lacking about mobilization dose that may optimize healing in laryngeal tissues, and how optimized doses may interact with the specific initial inflammatory status of the tissue. Purely empirical approaches to addressing this question are unattractive because of the relatively invasive and expensive nature of the research protocols . The cumbersome nature of data collection also complicates the potential for biologically oriented clinical trials on the value of therapeutic interventions for phonotrauma in humans.
Inflammation and Healing in Phonotrauma
Phonotrauma, like all other forms of trauma, is a highly complex process induced by a variety of stimuli, modulated by numerous cells and their products, and affecting different tissues in diverse ways. At the heart of the response to phonotrauma is the intertwined process of inflammation and wound healing –. Inflammation is the earliest and necessary response for the subsequent phase of wound healing . The whole inflammatory process can be understood as an information system that processes and controls the biochemical signals induced by injury and/or infection. These signals include: (1) “Go” signals to initiate the inflammatory reactions, (2) “Stop” signals to temper the inflammation, and (3) “Switch” signals to convert a tissue-damaging mode to a healing mode . Also, these signals are actively monitored and regulated by a number of checkpoints within the system. If any of the aforementioned signals is missing or one of the checkpoints malfunctions, the normal inflammatory process is altered possibly resulting in persistent inflammation, abnormal healing and tissue distortion. A prompt transition from the inflammatory phase to the healing phase is a key determinant of good wound healing, that is the replacement of traumatized tissue by healthy tissue .
Wound healing outcomes depend on the original insult to a certain degree, and are also likely influenced by individual genetics. Under ideal conditions, the inflammatory response would be limited by the antagonistic interactions among the various pro- and anti-inflammatory agents, followed by a transition to the later healing process in an orderly fashion. However, if repeated injuries occur over a period of time, the normal healing process would be disrupted. Two possible healing processes can occur in that case, namely, reparative regeneration and constructive repair. Reparative regeneration is scarless wound healing, in which traumatized tissue is completely restored in normal architecture and function. On the other hand, constructive repair is more common in humans. The lost tissue is replaced by granulation tissue which matures into a scar. Depending on the amount of scarring, scarred tissue usually does not properly restore the architecture and perform the pre-morbid function of the injured tissue. The vocal folds are generally capable of withstanding phonatory stresses and have the reparative capability of resolving microscopically phonotraumatic damage incurred during daily voice use. However, when injury exceeds a critical threshold, inflammation is usually clinically evident and constructive repair is involved. As a result, macroscopic vocal fold lesions may develop, with prolonged dysfunction in vocal fold vibratory function and voice output characteristics , .
Our research team has published four studies on the development of a novel method for obtaining quantitative information about the inflammatory status of the larynx from laryngeal secretions. Our prior studies ,  demonstrate the presence of various mediators (interleukin-1 beta [IL-1β], tumor necrosis factor alpha [TNF-α]), prostaglandin E2 [PGE-2], and matrix metalloproteinase 8 [MMP-8] ) in vocal fold surface secretions of human subjects in response to phonotrauma. Our first study showed that pre- to post-vocal loading shifts in mediator concentrations were clearly evident at 10 and 20 min post-loading for IL-β, TNF-α, and MMP-8, reflecting the presence of acute phonotrauma. In contrast, concentration shifts were not shown for TGF-β1 or PGE-2 . Another, intraoperative human study confirmed that IL-1β was an indicator of acute inflammation, whereas PGE-2 characterized chronic wounds . A third study showed that IL-1β was an early indicator of inflammation, and PGE-2 was a later indicator of wound healing in rabbits subjected to surgical trauma, with IL-1β returning to baseline by Day 7 post-injury and PGE2 remaining elevated until the final time-point of three weeks post-injury . Finally, a fourth study in rabbits subjected to surgical trauma assessed the degree to which assays of laryngeal secretions may reflect wound healing processes deep to the epithelium . That study showed that the time point associated with spikes in IL-1β (24 hours) corresponded to the presence of fibrinous clot. The time point associated with maximum PGE-2 levels (7 days) was associated with the presence of mature collagen. Massive cellular infiltration and complete epithelial coverage were found at intermediate time points. Taken together, these studies provide robust evidence that secretions from the laryngeal surfaces can provide a quantitative reflection of the current inflammatory and wound healing state of vocal fold tissue. The attractiveness of this marginally invasive technology is that it can be readily used in human subjects—although not without some difficulties on the part of both subjects and examiners–and thus the data gain external validity over data obtained from more invasive technologies involving animal subjects. Our research team has also examined the potential of mathematical modeling to elucidate apparently contradictory and unpredictable behavior emerging from the plethora of interactions among biologic pathways involved in the acute inflammatory response –. We have recently coined the term “translational systems biology” to refer to the process of creating, calibrating, and validating computational simulations in settings of complex diseases, simulations that are designed a priori for the purpose of modifying clinical treatment, carrying out in silico clinical trials, generating novel therapies, and refining diagnoses , .
The objective of the present study was to use empirical data from human subjects to develop a mechanistic computational simulation for the biological dynamics of vocal fold inflammation and wound healing following acute phonotrauma. The modeling approach employed in the present study involves Agent-Based Models (ABMs). In such models, individual components of a given complex system interact based on rules whose outcomes are partially determined by stochastic processes . More specifically, ABM involves discrete event simulation to study the behavior of complex systems. ABM is the most direct initial approach to simulate the temporal evolution of a complex system and to encode complicated time-dependent cellular and molecular events that occur during inflammation and wound healing –. Our results suggest that patient-specific, mechanistic, and predictive computational simulations of inflammation can be generated, raising the possibility that similar methods could be utilized in other complex diseases.
The premise of mathematical models involves experimental validation and feedback between the models and experiments. The details of our model calibration and validation are described in Materials and Methods. In brief, we first calibrated the ABM using data from a base cohort (Subjects 1, 2 and 3), using their mediator levels in laryngeal fluid at baseline, immediately after phonotrauma induction, and following a 4-hr treatment that involved either voice rest, “resonant voice” exercises, or spontaneous speech) (Figures 1–2, dark circles) . The calibrated ABM was run 10 times for the full cohort of 7 subjects, up to 5 simulated days post baseline under the condition of (1) each subject's actual treatment group and (2) hypothetical randomization to either of the other two treatment groups, i.e., hypothetical treatments, using each subject's baseline mediator profile. Thus, a large simulation data set of subject-specific mediator trajectories (3 treatment conditions×10 runs×7 subjects = 210 runs) was generated for evaluating benefits of behavioral treatments for acute phonotrauma.
Panels A, D and G display empirical and predicted trajectories of IL-1β. Panels B, E and H show empirical and predicted trajectories of TNF-α. Panels C, F and I show empirical and predicted trajectories of IL-10. Inflammatory marker concentrations are in pg/ml. The grey bars represent the mean of the simulated data, and the error bars represent standard deviations in the simulated data. The dark circles represent the input data for the first three time-points (baseline, post-loading, 4-hr post treatment onset), obtained from human laryngeal secretion data. The empty circles represent the validation data at the 24-hr time point from the human laryngeal secretion data. B: baseline; PL: post vocal loading; 4hrPRx: following a 4-hr treatment. Note that human validation data for Days 2–5 have not yet been generated.
Panels A, D and G display empirical and predicted trajectories of IL-1β. Panels B, E and H show empirical and predicted trajectories of TNF-α. Panels C, F and I show empirical and predicted trajectories of IL-10. Inflammatory marker concentrations are in pg/ml. The grey bars represent the means from the simulated data, and the error bars represent the standard deviation from the simulated data. The dark circles represent the input data for the first three time-points (baseline, post-loading, 4-hr post treatment onset) from the human laryngeal secretion data. The empty circles represent the validation data at the 24-hr time point from the human laryngeal secretion data. B: baseline; PL: post vocal loading; 4hrPRx: following a 4-hr treatment. Note that human validation data for Days 2–5 have not yet been generated.
Quantitative validation of the ABM was carried out by comparing the predicted mediator levels with the empirical mediator levels at 24 hr (1) for the base cohort (Subjects 1–3), as well as (2) for Subjects 4–7 whose data had never been used for the model calibration, under the actual treatment condition. The validation results showed that predicted mediator levels generally matched empirical mediator levels for the ABM at 24 hr post baseline. Also, results from the simulation data showed that the predicted mediator trajectories varied with a function of treatment and initial mediator profile.
Predicted Trajectories of Inflammatory Mediators
The ABM reproduced and predicted subject-specific mediator trajectories (Figures 1 and 2). In the base cohort (Subjects 1–3), the ABM predicted 24-hr mediator values in 80% (12/15; p<0.05) of instances (+9 mediator measurements from 3 between-group subjects, each with one of the 3 different treatments; +9 mediator measurements from one within-group subject—Subject #3– who received each of the treatments; -3 mediator measurements to account for double-counting the “spontaneous speech” treatment for Subject #3 in both between- and within-group measurements; hence 15 different mediator measurements in total). For Subjects 4–7, z-tests (model data vs. empirical data) indicated that the ABM predicted empirically obtained 24-hr mediator values 67% (2/3; p<0.05) of the times for markers that were considered “valid,” as described shortly, and 44% (4/9; p<0.05) of the time for markers that were considered “pre-inflamed and/or non-responsive” (see Materials and Methods).
For the single within-group subject (Subject 3; Figure 1), both empirical data and simulation results showed that the concentrations of pro-inflammatory mediators (IL-1β and TNF-α) spiked immediately following 1 hr of vocal loading, whereas the anti-inflammatory mediator (interleukin-10 [IL-10]) showed a more protracted course. The model predicted that following the 4-hr treatment, inflammatory mediators would have distinctive temporal and quantitative expression patterns across treatment assignments. For the spontaneous speech condition, the ABM predicted that the inflammatory response would be further escalated, i.e., would involve markedly increased secretion of both pro- and anti-inflammatory mediators following the 4-hr treatment. The concentrations of pro-inflammatory mediators (IL-1β and TNF-α) reached their peaks at Day 1 post-injury and resolved to baseline concentrations around Day 2–3 post-injury. The model also predicted that the anti-inflammatory mediator, IL-10, would be secreted in great quantities by wound macrophages during the first 5 days post-injury. On the other hand, under conditions of voice rest and “resonant voice” exercises, the predicted concentrations of the pro-inflammatory mediators dropped rapidly after the 4-hour treatment and then remained low at the end of simulation, i.e., Day 5. The anti-inflammatory mediator IL-10 was predicted to be secreted rapidly after the 4-hr treatment and remain elevated up to Day 3 post-injury. Similar mediator patterns were found in the larger dataset from the between-group subjects (Figure 2).
Predicted Outcomes of In Silico Therapies
In order to demonstrate the application of ABM for running in silico (“virtual”) clinical studies, we ran a series of simulations for the full cohort of subjects under the conditions of (1) actual treatment received and (2) hypothetical treatments. The simulation results showed theoretical individual-specific trajectories of mediator levels across the treatments. For between-group comparisons, linear mixed models showed significant treatment effects on the predicted levels of mediators (IL-1β: F (2, 1260) = 223.17, p<0.001; TNF-α: F (2, 1260) = 12.27, p<0.001; IL-10: F (2, 1260) = 215.65, p<0.001). Post-hoc pairwise comparisons with Bonferroni adjustments showed that among the three treatment groups, the predicted level of the pro-inflammatory marker IL-1β ensuing from spontaneous speech was the highest (p<0.001), whereas concentrations of the same marker in the resonant voice condition was the lowest (p<0.001). At the same time, among the three treatment groups, the predicted level of the anti-inflammatory marker IL-10 ensuing from voice rest was the lowest (p<0.001), whereas the level for the resonant voice condition was the highest (p<0.001).
We asked several experimental questions in the current study. First, we sought to determine if an ABM could offer an individualized prediction of biological readouts presumed to be closely related to tissue status. Second, we wished to know if the simulated wound healing response would vary specifically as a function of the initial settings of mediator level and treatment prescribed. Positive answers to these experimental questions would strengthen confidence in the eventual utility of mathematical modeling in general, and of ABM in particular, for understanding the complex vocal fold wound healing system and for predicting the healing outcome after phonotrauma and varied treatment approaches.
The current study describes the development of an ABM that reproduced diverse trajectories of inflammatory mediators in the laryngeal secretions of different human subjects at early time-points up to 4 hrs post-phonotrauma, and furthermore was capable of predicting the levels of these mediators at 24 hrs. The subject-specific ABMs were further used to explore the effects of treatment regimens to which subjects had not been exposed and the predicted levels of mediators under each condition were compared. The models predicted that the wound healing outcomes as informed by the mediator trajectories would be dramatically different given variations in the initial mediator profile, presumably due to some combination of pre-existing vocal fold conditions and the treatment prescribed following phonotrauma.
Contemporary therapeutic interventions in phonotrauma are oriented towards modulating the inflammatory and healing processes to promote reparative healing of the traumatized vocal folds. A plausible approach is to both blunt the inflammatory response and activate the healing program. Upon mechanical challenge, an acute inflammatory cascade is immediately activated in damaged or stressed tissues , . The inflammatory cells infiltrate the area of injury to remove damaged and dead cells and tissue debris. This inflammatory reaction contributes to additional tissue damage and cell death, which exacerbate the initial tissue damage and amplify the signals for scarring. Theoretically, inflammation-blocking interventions may reduce the “secondary” tissue damage and the possibility for fibrosis and scarring. At the same time, this approach may potentially reduce the supply of growth factors and cytokines from the inflammatory cells that facilitate tissue repair. Thus, a therapeutic balance between the need to limit inflammation causing tissue damage and the need for inflammation to initiate tissue repair is important to optimize the quality of healing outcomes and the recovery of physiological functions –. In addition to in vitro and human data from our laboratory, the current in silico study also suggests that modified tissue mobilization exercise in the form of “resonant voice production” may have the effect of blunting the cells' pro-inflammatory responses (e.g. IL-1β and TNF-α) but enhancing their anti-inflammatory responses (e.g. IL-10), which may ultimately promote tissue regeneration , . At this point, only the ABMs' molecular outputs were calibrated with experimental data, whereas the ABMs' tissue-level outputs are yet to be experimentally validated. We have embarked on the work of evaluating the “net effect” of the complex interactions among these inflammatory mediators at the tissue level. The success of this work will lead us to predict actual vocal fold tissue status following injury, which ultimately may advance understanding of inflammation and healing in a clinically useful way.
Our ABM predicted that the secretion of pro-inflammatory mediators would be both prolonged and elevated subsequent to spontaneous speech following an episode of phonotraumatic injury. This prolongation would be due to a positive feedback loop involving “Pro-inflammation → Damage → Pro-inflammation,”  thereby delaying the transition to the subsequent healing process. Clinically, visible vocal fold inflammation would be expected. On the other hand, under conditions of either voice rest or resonant voice, the ABM predicted that the pro-inflammatory response would be attenuated. It also predicted that the anti-inflammatory response may be escalated in response to the tissue mobilization involved in resonant voice—and the pro-inflammatory response attenuated by the limitation of impact stress—thereby further prompting rapid healing . These predictions suggest that the repair process would bypass the “Pro-inflammation → Damage → Pro-inflammation” positive feedback loop and avoid a full-scale inflammatory and repair response. As a result, a reparative regeneration (as opposed to constructive repair) of mucosal structure and function would be observed clinically and the probability of scar formation would be minimized in the long run.
As additional data are acquired, the current ABM is under continuous revision and augmentation. Our ultimate, long-term goal is to generate in silico models that can be queried to identify biomechanical treatments that will optimize the wound healing process in the vocal folds, as a function of patient-specific inflammatory profiles. Although these results are encouraging in terms of the potential translational utility of ABM in the setting of vocal fold inflammation, at least five limitations can be noted in our study.
First, the current ABM mainly simulated (1) inflammation, (2) proliferation and (3) collagen formation. The model did not account for a final phase of the wound healing process, which involves extracellular matrix (ECM) reorganization. According to the literature on dermal wound healing, ECM reorganization is initiated once a neo-matrix involving materials such as collagen is deposited at the wound site –. Collagen is indeed a core component of the ECM, and undergoes remodeling that is dependent on both continued collagen synthesis and compensatory collagen degradation. The degradation of wound collagen is controlled by collagenases and other proteolytic enzymes, and the net increase in wound collagen is determined by the balance of these opposing mechanisms. Compared to the large body of literature on dermal wound healing, research on ECM reorganization in vocal fold wound healing is sparse. No in vivo measurement of collagen remodeling in human vocal folds is currently available. Thus, in its use of empirical data obtained from non-destructive methodologies only, the current model did not incorporate aspects of collagen remodeling that might prove to be important.
A second limitation in the current study is that healing outcomes in this ABM were primarily informed by interactions among inflammatory mediators and cells. However, a growing literature supports the idea that several ECM components such as fibronectin, hyaluronic acid, and decorin could also be involved in regulating the wound healing process. Studies to date have shown that aberrant scarring/fibrosis is at least partly due to the response of fibroblasts in the wound to both inflammatory mediators and extracellular matrix components , –, some of which are known to constitute alarm/danger signals (e.g. fragments of hyaluronic acid);  and therefore may be currently abstracted under the “damage” variable in our model. Future iterations of the ABM could be augmented with rules developed around the interactions among inflammatory mediators, cells and ECM components to yield more precise predictions.
Third, our model assumes that biomechanical stresses during phonation cause mucosal damage. However, the current biologically-based ABM lacks the ability to receive input from physical models of phonation, because data are lacking regarding the link between the output of physical models—i.e. distributed tissue stresses – and the biological consequences of those physical stresses. Although biochemical networks may be reasonably modeled by using stochastic simulations, many cell biology phenomena relating to wound healing require the calculation of biophysical processes, such as tissue deformation and disruption of intracellular adhesion . Ideally, our synthesized biochemical networks should be coupled with these biophysical processes to yield a more complete picture of vocal fold wound healing in response to biomechanical stresses during phonation.
A fourth limitation relates to the inclusion of the multi-functional anti-inflammatory mediator TGF-β1 in our model, although our earlier work on assaying biochemical markers of vocal fold wound healing failed to detect this mediator in laryngeal secretions pre- or post-traumatically . TGF-β1 is known to be involved in the regulation of cell proliferation, cell differentiation and extracellular matrix formation in all phases of inflammation and wound healing , . This mediator exerts both anti-inflammatory and pro-fibrotic effects that could convert an active site of inflammation into a site dominated by subsequent tissue repair . We suspect that TGF-β1 might be highly cell-associated –, and that this property might have led to our inability to detect this growth factor in vocal fold secretions . To determine if TGF-β1 is necessary for a correct simulation of inflammation and wound healing in the vocal folds, a qualitative validation procedure was carried out to determine what the simulated data would show in the presence or absence of TGF-β1. In the latter case, the ABM predicted that no cellular or molecular events would be triggered for any ranges of initial damage (data not shown). In the presence of TGF-β1, the ABM predicted different inflammatory and wound healing curves that vary with initial mediator profile and treatment modalities following phonotrauma (Figures 1 and 2). These results indicated that TGF-β1 (or a qualitatively similar cytokine) is essential for the wound healing process and should be included in the ABM structure.
The fifth, and perhaps most important limitation concerns the validation of the ABM predictions with regard to the different treatment modalities. We have in this study shown data from seven humans and the capacity of our ABM to predict mid-term (24-hr) inflammation based on earlier short-term assays. The subject population in this study was relatively homogeneous in terms of age. The subjects were all relatively young (21–46 years) and it is known that tissue regenerative processes usually decline with aging. We have carried out preliminary mathematical modeling of the effects of aging on the acute inflammatory response . We have not, however, verified in a large cohort of patients the validity of ABM predictions with regard to treatment outcome. It should be noted that such studies in humans are complex from regulatory, practical, and ethical points of view. We have embarked on a larger clinical study to validate the ABM described herein. Indeed, clinical management of phonotrauma remains a challenge to clinicians. Large clinical trials are needed to establish optimized patient-specific treatment interventions. In addition to validating the ABM with further experimental studies, a parallel development of ordinary differential equation-based models (EBMs) has been pursued in the interest of cross-platform comparison of results . This “model docking” is a well-vetted validation strategy based on a comparison of predictions of different models across an array of user input data. The finding of similar predictions in ABMs and EBMs would increase confidence in the underlying assumptions in the current ABMs. Also, the EBMs alone allow formal mathematical analysis of the simulation results, such as, bifurcation and stability analysis, that will facilitate capture of the systems dynamics of the inflammatory and healing responses .
Finally, for the approach ultimately to be useful in a real-life clinical setting, the time lag introduced by ELISAs will have to be addressed, or an alternate methodology will have to be identified that can yield biological profiles of materials in a shorter period of time. In the meantime, there is encouragement that models such as ABMs and EBMs have potential to illuminate important information about individualized wound healing processes in the larynx.
More globally, we suggest that a systems biology approach that involves modeling is integral to sorting through the perplexing array of factors that dictate success or failure of clinical trials for complex diseases . Ultimately, this process would be augmented by the inclusion of genetic variability in inflammatory and wound healing components, typically mediated via single-nucleotide gene polymorphisms in relevant genes .
In summary, this study suggests for the first time that patient-specific, individualized models of inflammation and healing are possible. This demonstration extends the power of translational simulations of acute inflammation beyond the responses of idealized organisms –, quantitative prediction of inflammation occurring in experimental animals , and simulations of populations (clinical trials) , , . It is hoped that this work will point the way to addressing other complex disease processes.
Materials and Methods
Experimental Protocol for Acute Phonotrauma in Different Treatment Modalities
The study was approved by the Institutional Review Board at the University of Pittsburgh. A total of nine subjects participated in the study; six females (21–46 years) and three males (21–29 years). Eight of nine subjects participated in a between-subjects study design, which involved exposure to one “treatment” condition (voice rest, “resonant voice” exercises or spontaneous speech) following a vocal loading task. One female subject (Subject 3) was involved in a within-subjects design, and received all three treatments, randomly ordered without replacement on different pairs of days separated by intervals ranging from 1–6 months. Prior to subjects' participation in the experimental part of the protocol, written informed consent was obtained by an investigator or research coordinator and then subjects received a screening in the clinic for gag response and nasal patency. Exclusion criteria for the study included gag reflex with tooth-brushing or history of exaggerated gag reflex, deviated septum (based on the otolaryngology exam), current or recent voice problems (within 1 year), current or any history of speech or language deficits, current use of drugs that may influence the voice (e.g., diuretics, decongestants), and allergy to local anesthetics (especially lidocaine).
In this experimental protocol, a vocal loading task aimed to induce an acute phonotrauma or acute laryngeal inflammation. The protocol for vocal loading entailed three consecutive cycles, each involving 15 minutes of loud phonation (∼75–90 dB @ 15 cm microphone-to-mouth distance) followed by 5 minutes of silence, for a total 60 minutes. Then, the subjects were randomly assigned to one of three treatment groups: voice rest, “resonant voice” exercises or spontaneous speech for 4 hours in the clinic, under the careful supervision of a voice trainer, who-for the majority of subjects-was blinded to the experimental hypotheses. These three treatment modalities can be considered on a continuum of tissue mobilization magnitude and as important, intra-vocal fold impact stress magnitude: no mobilization or impact stress (voice rest), normal-to-large magnitude but relatively low impact stress (“resonant voice” exercise; , ) and normal to larger magnitude mobilization but potentially greater impact stress, depending on the mode of phonation (spontaneous speech; , ). The “resonant voice” exercise condition involved cycles of 4 minutes of voice exercise using “resonant voice,” defined as “easy” voice associated with perceptible anterior oral vibrations followed by 16 minutes of rest, whereas the spontaneous speech condition involved repeating cycles of 16 minutes of conversational speech followed by 4 minutes of silence. After a 4-hour treatment period monitored in the clinic, participants were discharged to home with instructions to continue their corresponding treatments (in somewhat less intense cycles for resonant voice and spontaneous speech conditions). The following morning, participants were required to observe complete voice rest until their arrival at the clinic between 7:30 and 8:30 a.m.
Laryngeal Secretion Procedure and Assessment of Inflammatory Analytes
A total of 4 secretion specimens were collected from each subject at 4 different times per treatment condition: at baseline, immediately after vocal loading, following the 4-hr treatment and 24 hours post-baseline. For baseline secretion collection, an otolaryngologist first examined the subject's oral cavity, oropharynx, and nasal cavity and placed a cotton pledget (a flat absorbent pad) soaked with lidocaine and decongestant into the subject's more patent nasal cavity. Cetacaine® was sprayed into the oropharynx. Rigid laryngeal stroboscopy was performed to obtain a baseline stroboscopic evaluation of the patient. Then 4% lidocaine was dripped onto the endolarynx through the working channel of a chip-tip flexible laryngoscope. After approximately 5 minutes, subsequent to verification of anesthesia to light touch, a one millimeter plastic cannula was passed through the working channel of the scope and guided down to the free edge and superior surface of the vocal folds while suction was applied to the catheter. That procedure allowed for the collection of a small amount of vocal fold secretions (about 100 µl), while minimizing contact of the scope with the vocal folds bilaterally. Secretions were captured in a modified sinus trap and then transferred into a 0.2 ml microfuge tube using a 1cc syringe. The tubes were labeled using codes that could not be traced to the subject or the subject's condition—except by way of a secret list retained by one investigator who was not involved with secretion data analysis–and the tubes were placed on dry ice. Tubes were then stored at −80°C until analysis.
All secretion analyses were carried out by an investigator who was blinded to subjects' conditions (time point and treatment condition). For the analyses, a known volume was aliquoted for analysis and served as the dilution factor. The appropriate volume of sterile saline was added to the tube to bring the total volume up to 2.0 ml. Standard enzyme-linked immunosorbent assays (ELISAs) were performed for IL-1β, IL-6, IL-8, TNF-α, matrix metalloproteinase (MMP)-8, and IL-10 utilizing the manufacturer's recommended protocol (R&D Systems, Minneapolis, MN). All samples were run on the same kit to avoid inter-kit variability.
Inspection of the laryngeal secretion data was carried out in order to identify the cleanest data for the development of ABM. To do so, the secretion data were sorted into three main categories for each subject and inflammatory marker: (1) data showing high baseline concentrations of pro-inflammatory markers (≥1 standard deviation from the sample mean; designated as “pre-inflamed” data); (2) data showing normal baseline concentrations of markers (<1 standard deviation from the sample mean), but paradoxically decreasing post loading (“non-responsive” data); and (3) data showing normal baseline concentrations of markers (<1 standard deviation from the sample mean) and increase after loading (“responsive” data) . In addition, data from two subjects (Subject 8 and Subject 9) were considered as invalid due to thick secretions that compromised interpretation of ELISA results. Therefore, their data were not used in the development of the ABM.
As a result of this process, a base data set was identified using data from three subjects (Subject 1—voice rest, Subject 2—resonant voice, and Subject 3—spontaneous speech), whose mediator data were considered “responsive” and “not pre-inflamed”. Of note, Subject 3 was the one who participated in both the between- and within-groups design. The first three time points of the base data set were used for ABM calibration. Also, mediator data at 24 hr for (1) Subjects 1, 2 and 3 as well as (2) Subjects 4–7, whose inflammatory mediator patterns were anomalous in some fashion, was used for model validation.
The ABM of inflammation and tissue damage/healing was a modification of one we developed previously to address the prototypical wound healing scenario, namely skin healing . The freeware Netlogo (Center for Connected Learning and Computer-Based Modeling, Northwestern University, Evanston, IL) was used as the platform for model building and simulation. This first-generation ABM aimed to reproduce the basic and generally-accepted mechanisms of wound healing. Thus, detailed literature on inflammation and wound healing was reviewed to identify the essential components and rules for this ABM , –, . Then, experimental measures of inflammatory mediators in human laryngeal secretions  were used to adapt the model to the setting of vocal fold injury.
Simply put, our ABM of phonotrauma represents processes thought to occur in the vocal fold mucosal tissue and to simulate the mucosal repair response to biomechanical damage during phonation. The model consists of platelets, inflammatory cells (neutrophils, macrophages, and fibroblasts), mediators that involve in inflammation and wound healing (IL-1β, TNF-α, IL-10, and TGF-β1), a representative component of the extracellular matrix (collagen), and, perhaps most important, a tissue damage function functionally analogous to alarm/danger signals  that produces positive feedback to induce further inflammation  (Figure 3; Table 1).
The model assumes that biomechanical stress during phonation causes mucosal damage and activates platelets, neutrophils and macrophages. Platelets produce TGF-β1, which chemoattracts both neutrophils and macrophages. Activated neutrophils and macrophages secrete pro-inflammatory mediators, which in turn induce anti-inflammatory mediator release. Pro-inflammatory mediators also induce neutrophils and macrophages to produce free radicals that damage tissue. In our model, the activity of free radicals was subsumed in the actions of TNF-α. Anti-inflammatory mediators contribute to fibroblast activation. Activated fibroblasts secrete collagen that mediates tissue repair. In the model, collagen accumulation is considered as the surrogate for healing outcome following phonotrauma. Collagen is an important ECM protein involving both structural and biomechanical functions in the vocal folds (Gray & Titze, 1988; Gray et al., 2000).
Regions, patches and agents of the vocal fold ABM
A typical ABM is composed of three elements: region, patch and agent. The region is composed of small patches. The patches are immobile components that characterize the physical-spatial environment, where the agents operate. Agents are the active objects that move and interact within the region. In our ABM, the “world” is a square grid of 120×120 patches, with the origin in the center of the grid. Two regions were created to simulate (1) blood and (2) mucosal tissue itself. These two regions do not intersect. Specifically, the tissue region is a circle with a diameter of 55, centered at the origin and bounded by the blood region. The blood region is the source of the inflammatory cells that infiltrate the wounded tissue. At the same time, the region of mucosal tissue is the source of some resident cells and is also the site in which phonotraumatic injury occurs and is subsequently repaired by fibroblasts .
Patch variables were used to represent (1) tissue status (healthy, damaged, and healed); (2) platelets; (3) collagen; and (4) inflammatory mediators. Platelets are important to initiate the inflammatory process following tissue damage. The initial number of platelets is spatially distributed based on the rules in our model. Collagen is the major structural protein in the vocal folds and its content and organization are prone to be disturbed following repetitive phonotrauma –. In our model, the amount of collagen was required not to exceed the existing amount of damage in the same patch. Two pro-inflammatory mediators (IL-1β and TNF-α) and two anti-inflammatory mediators (IL-10 and TGF-β1) were also selected as patches because they are generally believed to play an important role in wound healing environment and because we had prior data on their expressions in the vocal folds –. The concentrations of the inflammatory mediators on each patch are controlled by formulae for mediator synthesis, mediator degradation and mediator diffusion.
Agent variables were used to represent (1) tissue damage and (2) cells. Tissue damage is induced by the initial injury and the subsequent inflammatory response of the pro-inflammatory mediators (IL-1β and TNF-α). Tissue damage also acts as a stimulus for further inflammation. Another class of agent is cells, namely, neutrophils, macrophages and fibroblasts. In our model, cells have three states: resting, activated or dead. Cells are represented as agents because they can be organized based on common behavioral rules, and because the response of a particular cell type to various mediators is readily characterized in the literature . Cell behavior was governed by rules based on existing wound healing literature , –, . Depending on the cell type, the cellular responses included activation, migration, proliferation, cell death, secretion of inflammatory mediators, tissue debridement, and collagen generation. The complete rules of the ABM with explanations are attached in Supporting Information (Table S1).
Simulation of acute phonotrauma
For each simulation, the user can define the initial levels of IL-1β, TNF-α and IL-10, add a phonotraumatic event, and then a 4-hr treatment event (voice rest, “resonant voice” exercises or spontaneous speech). We assumed in the model that one step of simulated time represents 0.1 days or approximately 2.4 hours. The changes in temporal concentration of inflammatory cells, mediators, tissue damage and collagen were plotted and refolded into the model at each time step.
Initially, some resting neutrophils are in the blood region, whereas some macrophages and fibroblasts are present with a random distribution in both blood and tissue regions. Simulated phonatory stresses traumatized the mucosal tissue in the middle of region, triggering platelet degranulation. Shortly afterwards, a chemoattractant gradient is created that stimulates the infiltration and activation of neutrophils and macrophages. Later on, fibroblasts are activated by tissue damage and TGF-β1. Fibroblasts secrete collagen to repair both the initial and the inflammation-induced damage. Last, additional mechanical stresses are applied to the traumatized tissue based on the treatment selected (voice rest: no additional mechanical stress; resonant voice: low mechanical stress; spontaneous speech: high mechanical stress).
Model Calibration and Validation
Standard procedures to evaluate the fit of ABM to empirical data have not been established in the literature. In the present study, pattern-oriented analysis  was used to estimate the conformity of simulation-generated data curves with the inflammatory and wound healing patterns reported in the literature as well as the empirical data sets around acute phonotrauma (Table 2).
Using this approach, the user-defined initial magnitude of mucosal injury (range 0–40 in arbitrary units of damage) was set at a value of 20 as a “comparison condition”, because that setting resulted in realistic predictions of mucosal damage and healing when compared with the general consensus around laryngeal wound healing documented in the literature –, . When the qualitative behavior of the simulation appeared satisfactory, we proceeded to calibrate the model by adjusting parameter values not found in the literature to fit the quantity and time-course of measured vocal fold mediators. We calibrated the ABM using data from the base cohort (Subjects 1, 2 and 3) using their baseline mediator levels in laryngeal fluid, immediately after phonotrauma induction, and following a 4-hr treatment (voice rest, “resonant voice” exercises, or spontaneous speech) (Figures 1–2, dark circles) . Subsequently, we validated the ABM quantitatively by comparing the predicted mediator levels with the empirical mediator levels at 24 hr (1) for Subjects 1, 2 and 3, as well as (2) Subjects 4–7 whose data had never been used for the model.
Due to the inherent stochasticity of the ABM framework, we performed ten runs of the calibrated ABM for each subject up to five simulated days under the condition of (1) each subject's actual treatment group and (2) hypothetical randomization to either of the other two treatment groups for all subjects, i.e., hypothetical treatments. The means and standard deviations of model variables (concentrations of inflammatory cells, mediators, and tissue damage) were computed at each time point for subsequent analysis. Hypothetical treatments were included for simulation to demonstrate the application of ABM for running individual-specific in silico clinical trials.
Z-tests using the p-value approach to testing statistical significance were carried out to compare the human empirical data and the data predicted by the model for IL-1β, TNF-α and IL-10 at the 24-hr time point for (1) the base cohort (Subjects 1–3) and (2) Subjects 4–7. Also, linear mixed models with fixed and random effects were used to compare the predicted levels of IL-1β, TNF-α, and IL-10. Fixed factors included: (1) patient, (2) treatment condition (spontaneous speech, voice rest and “resonant voice” exercises) and (3) time point (immediately after phonotrauma induction, following a 4-hr treatment and 24-hr post baseline). The random factor is the number of simulation runs and is nested within the patient factor.
The authors would like to acknowledge the following clinicians and investigators: Dr. Clark Rosen, Dr. Kim Steinhauer, Dr. Ryan Branski and Dr. Elizabeth U. Grillo, for their contributions to this work. We would also like to thank Dr. Gary An for his valuable inputs to the computer model.
Conceived and designed the experiments: NYKL KV. Performed the experiments: NYKL KV. Analyzed the data: NYKL KV GC PAH YV. Contributed reagents/materials/analysis tools: NYKL KV GC QM PAH YV. Wrote the paper: NYKL KV GC QM PAH YV. Statistical consultation: ENR. Agent-based model construction: NYL GC QM YV.
- 1. Gray SD, Titze IR (1988) Histologic investigation of hyperphonated canine vocal cords. Ann Otol Rhinol Laryngol 97: 381–388.SD GrayIR Titze1988Histologic investigation of hyperphonated canine vocal cords.Ann Otol Rhinol Laryngol97381388
- 2. Gray SD, Titze IR, Alipour F, Hammond TH (2000) Biomechanical and histologic observations of vocal fold fibrous proteins. Ann Otol Rhinol Laryngol 109: 77–85.SD GrayIR TitzeF. AlipourTH Hammond2000Biomechanical and histologic observations of vocal fold fibrous proteins.Ann Otol Rhinol Laryngol1097785
- 3. Titze IR, Hitchcock RW, Broadhead K, Webb K, Li W, et al. (2004) Design and validation of a bioreactor for engineering vocal fold tissues under combined tensile and vibrational stresses. J Biomech 37: 1521–1529.IR TitzeRW HitchcockK. BroadheadK. WebbW. Li2004Design and validation of a bioreactor for engineering vocal fold tissues under combined tensile and vibrational stresses.J Biomech3715211529
- 4. Gray SD (1991) Basement membrane zone injury in vocal nodules;. In: Gauffin J, Hammarberg B, editors. San Diego: Singular Press. pp. 1–35.SD Gray1991Basement membrane zone injury in vocal nodules;J. GauffinB. HammarbergSan DiegoSingular Press135
- 5. Gray SD, Titze IR, Lusk RP (1987) Electron microscopy of hyperphonated canine vocal cords. Journal of Voice 1: 109–115.SD GrayIR TitzeRP Lusk1987Electron microscopy of hyperphonated canine vocal cords.Journal of Voice1109115
- 6. Branski RC, Perera P, Verdolini K, Rosen CA, Hebda PA, et al. (2006) Dynamic Biomechanical Strain Inhibits IL-1beta-induced Inflammation in Vocal Fold Fibroblasts. J Voice. RC BranskiP. PereraK. VerdoliniCA RosenPA Hebda2006Dynamic Biomechanical Strain Inhibits IL-1beta-induced Inflammation in Vocal Fold Fibroblasts.J Voice
- 7. Jacobson BH, Johnson A, Grywalski C, Silbergleit A, Jacobson G, Benninger MS, Newman CW (1997) The voice handicap index (VHI): Development and validation. American Journal of Speech-Language Pathology 6: 66–70.BH JacobsonA. JohnsonC. GrywalskiA. SilbergleitG. JacobsonMS BenningerCW Newman1997The voice handicap index (VHI): Development and validation.American Journal of Speech-Language Pathology66670
- 8. Ma EP, Yiu EM (2001) Voice activity and participation profile: assessing the impact of voice disorders on daily activities. J Speech Lang Hear Res 44: 511–524.EP MaEM Yiu2001Voice activity and participation profile: assessing the impact of voice disorders on daily activities.J Speech Lang Hear Res44511524
- 9. Raaijmakers MF, Dekker J, Dejonckere PH (1998) Diagnostic assessment and treatment goals in logopedics: impairments, disabilities and handicaps Folia Phoniatr Logop 50: 71–79.MF RaaijmakersJ. DekkerPH Dejonckere1998Diagnostic assessment and treatment goals in logopedics: impairments, disabilities and handicaps Folia Phoniatr Logop507179
- 10. Smith E, Verdolini K, Gray SD, Nichols S, Lemke J (1996) Effect of voice disorders on quality of life. Journal of Medical SpeechLanguage Pathology 4: 223–244.E. SmithK. VerdoliniSD GrayS. NicholsJ. Lemke1996Effect of voice disorders on quality of life.Journal of Medical SpeechLanguage Pathology4223244
- 11. Goldring SR (2003) Inflammatory mediators as essential elements in bone remodeling. Calcified Tissue International 73: 97–100.SR Goldring2003Inflammatory mediators as essential elements in bone remodeling.Calcified Tissue International7397100
- 12. Hart J (2002) Inflammation. 1: Its role in the healing of acute wounds. J Wound Care 11: 205–209.J. Hart2002Inflammation. 1: Its role in the healing of acute wounds.J Wound Care11205209
- 13. Hart J (2002) Inflammation. 2: Its role in the healing of chronic wounds. J Wound Care 11: 245–249.J. Hart2002Inflammation. 2: Its role in the healing of chronic wounds.J Wound Care11245249
- 14. Redd MJ, Cooper L, Wood W, Stramer B, Martin P (2004) Wound healing and inflammation: embryos reveal the way to perfect repair. Philos Trans R Soc Lond B Biol Sci 359: 777–784.MJ ReddL. CooperW. WoodB. StramerP. Martin2004Wound healing and inflammation: embryos reveal the way to perfect repair.Philos Trans R Soc Lond B Biol Sci359777784
- 15. Morrison M, Rammage L (1994) The Management of Voice Disorders. San Diego: Singular Publishing Group Inc. M. MorrisonL. Rammage1994The Management of Voice Disorders.San DiegoSingular Publishing Group Inc
- 16. Stemple JC, Lee L, D'Amico B, Pickup B (1994) Efficacy of vocal function exercises as a method of improving voice production. Journal of Voice 8: 271–278.JC StempleL. LeeB. D'AmicoB. Pickup1994Efficacy of vocal function exercises as a method of improving voice production.Journal of Voice8271278
- 17. Verdolini K (2000) Case Study: Resonant Voice Therapy;. In: Stemple J, editor. San Diego: Singular Publishing Group Inc. pp. 46–62.K. Verdolini2000Case Study: Resonant Voice Therapy;J. StempleSan DiegoSingular Publishing Group Inc4662
- 18. Sataloff RT (1997) Voice rest;. In: Sataloff RT, editor. San Diego: Singular Publishing Group, Inc. pp. 453–456.RT Sataloff1997Voice rest;RT SataloffSan DiegoSingular Publishing Group, Inc453456
- 19. Branski RC (2005) Vocal fold fibroblast response to mechanical stress [doctoral dissertation]. Pittsburgh: University of Pittsburgh. pp. 1–123.RC Branski2005Vocal fold fibroblast response to mechanical stress [doctoral dissertation].PittsburghUniversity of Pittsburgh1123
- 20. Verdolini K, Li NYK, Branski RC, Rosen CA, Urban EG, et al. (in preparation) The effect of targeted vocal exercise on recovery from acute inflammation. K. VerdoliniNYK LiRC BranskiCA RosenEG Urbanin preparationThe effect of targeted vocal exercise on recovery from acute inflammation.
- 21. Burroughs P, Dahners LE (1990) The effect of enforced exercise on the healing of ligament injuries. Am J Sports Med 18: 376–378.P. BurroughsLE Dahners1990The effect of enforced exercise on the healing of ligament injuries.Am J Sports Med18376378
- 22. Eiff MP, Smith AT, Smith GE (1994) Early mobilization versus immobilization in the treatment of lateral ankle sprains. Am J Sports Med 22: 83–88.MP EiffAT SmithGE Smith1994Early mobilization versus immobilization in the treatment of lateral ankle sprains.Am J Sports Med228388
- 23. Kerkhoffs GM, Struijs PA, Marti RK, Assendelft WJ, Blankevoort L, et al. (2002) Different functional treatment strategies for acute lateral ankle ligament injuries in adults. Cochrane Database Syst Rev CD002938.GM KerkhoffsPA StruijsRK MartiWJ AssendelftL. Blankevoort2002Different functional treatment strategies for acute lateral ankle ligament injuries in adults.Cochrane Database Syst RevCD002938
- 24. Kerkhoffs GM, Struijs PA, Marti RK, Blankevoort L, Assendelft WJ, et al. (2003) Functional treatments for acute ruptures of the lateral ankle ligament: a systematic review. Acta Orthop Scand 74: 69–77.GM KerkhoffsPA StruijsRK MartiL. BlankevoortWJ Assendelft2003Functional treatments for acute ruptures of the lateral ankle ligament: a systematic review.Acta Orthop Scand746977
- 25. Kvist J (2004) Rehabilitation following anterior cruciate ligament injury: current recommendations for sports participation. Sports Med 34: 269–280.J. Kvist2004Rehabilitation following anterior cruciate ligament injury: current recommendations for sports participation.Sports Med34269280
- 26. Pijnenburg AC, Van Dijk CN, Bossuyt PM, Marti RK (2000) Treatment of ruptures of the lateral ankle ligaments: a meta-analysis. J Bone Joint Surg Am 82: 761–773.AC PijnenburgCN Van DijkPM BossuytRK Marti2000Treatment of ruptures of the lateral ankle ligaments: a meta-analysis.J Bone Joint Surg Am82761773
- 27. Mulligan B (1995) Manual Therapy “NAGS,” “SNAGS,” “MWMSs” etc. Wellington, New Zealand: Plane View Service. pp. 78–88.B. Mulligan1995Manual Therapy “NAGS,” “SNAGS,” “MWMSs” etc.Wellington, New ZealandPlane View Service7888
- 28. Paungmali A, O'Leary S, Sowvlis T, Vicenzino B (2003) Hypoalgesic and sympathoexcitatory effects of mobilization with movement for lateral epicodylalgia. Physical Therapy 83: 374–383.A. PaungmaliS. O'LearyT. SowvlisB. Vicenzino2003Hypoalgesic and sympathoexcitatory effects of mobilization with movement for lateral epicodylalgia.Physical Therapy83374383
- 29. Salter RB (1994) The physiologic basis of continuous passive motion for articular cartilage healing and regeneration. Hand Clinics 10: 211.RB Salter1994The physiologic basis of continuous passive motion for articular cartilage healing and regeneration.Hand Clinics10211
- 30. Salter RB (1996) History of rest and motion and the scientific basis for early continuous passive motion. Hand Clinics 12: 1–11.RB Salter1996History of rest and motion and the scientific basis for early continuous passive motion.Hand Clinics12111
- 31. Thornton GM, Shrive NG, Frank CB (2003) Healing ligaments have decreased cyclic modulus compared to normal ligaments and immobilization further compromises healing ligament response to cyclic loading. Journal of Orthopaedic Research 21: 716–722.GM ThorntonNG ShriveCB Frank2003Healing ligaments have decreased cyclic modulus compared to normal ligaments and immobilization further compromises healing ligament response to cyclic loading.Journal of Orthopaedic Research21716722
- 32. Threlkeld AJ (1992) Manual Therapy: An American Physical Therapy Association Monograph. Physical Therapy 59–68.AJ Threlkeld1992Manual Therapy: An American Physical Therapy Association Monograph.Physical Therapy5968
- 33. Visser NA, de Koning MH, Lammi MJ, Hakkinen T, Tammi M, et al. (1998) Increase of decorin content in articular cartilage following running. Connect Tissue Res 37: 295–302.NA VisserMH de KoningMJ LammiT. HakkinenM. Tammi1998Increase of decorin content in articular cartilage following running.Connect Tissue Res37295302
- 34. Williams JM, Moran M, Thonar EJ, Salter RB (1994) Continuous passive motion stimulates repair of rabbit knee articular cartilage after matrix proteoglycan loss. Clin Orthop 304: 252–262.JM WilliamsM. MoranEJ ThonarRB Salter1994Continuous passive motion stimulates repair of rabbit knee articular cartilage after matrix proteoglycan loss.Clin Orthop304252262
- 35. Verdolini K, Li NYK, Rosen CA, Grillo E, Branski RC, et al. (in preparation [b]) Effect of resonant voice excercises on the biological recovery from acute phonotrauma.K. VerdoliniNYK LiCA RosenE. GrilloRC Branskiin preparation [b]Effect of resonant voice excercises on the biological recovery from acute phonotrauma.
- 36. Clark RAF (1998) Wound Repair. Overview and General Considerations. In: Clark RAF, editor. The Molecular and Cellular Biology of Wound Repair. New York: Plenum Press. RAF Clark1998Wound Repair. Overview and General Considerations.RAF ClarkThe Molecular and Cellular Biology of Wound RepairNew YorkPlenum Press
- 37. Clark RAF (1988) Cutaneous wound repair: Molecular and cellular controls. Prog Dermatol 22: 1–12.RAF Clark1988Cutaneous wound repair: Molecular and cellular controls.Prog Dermatol22112
- 38. Clark RAF, editor. (1998) The Molecular and Cellular Biology of Wound Repair. New York: Plenum Press. RAF Clark1998The Molecular and Cellular Biology of Wound Repair.New YorkPlenum Press
- 39. Calvin M (1998) Cutaneous wound repair. Wounds 10: 12–32.M. Calvin1998Cutaneous wound repair.Wounds101232
- 40. Mast BA, Schultz GS (1996) Interactions of cytokines, growth factors, and proteases in acute and chronic wounds. Wound Repair Regen 4: 411–420.BA MastGS Schultz1996Interactions of cytokines, growth factors, and proteases in acute and chronic wounds.Wound Repair Regen4411420
- 41. Werner S, Grose R (2003) Regulation of wound healing by growth factors and cytokines. Physiol Rev 83: 835–870.S. WernerR. Grose2003Regulation of wound healing by growth factors and cytokines.Physiol Rev83835870
- 42. Gillitzer R, Goebeler M (2001) Chemokines in cutaneous wound healing. J Leukoc Biol 69: 513–521.R. GillitzerM. Goebeler2001Chemokines in cutaneous wound healing.J Leukoc Biol69513521
- 43. Hardy MA (1989) The biology of scar formation. Phys Ther 69: 1014–1024.MA Hardy1989The biology of scar formation.Phys Ther6910141024
- 44. Nathan C (2002) Points of control in inflammation. Nature 420: 846–852.C. Nathan2002Points of control in inflammation.Nature420846852
- 45. Walter JB, Israel MS, editors. (1987) General Pathology. Churchill: Livingstone. JB WalterMS Israel1987General Pathology.ChurchillLivingstone
- 46. Catten M, Gray SD, Hammond TH, Zhou R, Hammond EH (1998) Analysis of cellular location and concentration in vocal fold lamina propria. Otolaryngology Head and Neck Surgery 118: 663–667.M. CattenSD GrayTH HammondR. ZhouEH Hammond1998Analysis of cellular location and concentration in vocal fold lamina propria.Otolaryngology Head and Neck Surgery118663667
- 47. Gray SD (2000) Cellular Physiology of the Vocal Folds. In: Rosen CA, Murry T, editors. The Otolaryngologic Clinics of North America. Philadelphia: W.B. Saunders Company. SD Gray2000Cellular Physiology of the Vocal Folds.CA RosenT. MurryThe Otolaryngologic Clinics of North AmericaPhiladelphiaW.B. Saunders Company
- 48. Branski RC, Verdolini K, Rosen CA, Hebda PA (2004) Markers of wound healing in vocal fold secretions from patients with laryngeal pathology. Annals of Otology, Rhinology, and Laryngology 113: 23–29.RC BranskiK. VerdoliniCA RosenPA Hebda2004Markers of wound healing in vocal fold secretions from patients with laryngeal pathology.Annals of Otology, Rhinology, and Laryngology1132329
- 49. Verdolini K, Rosen CA, Branski RC, Hebda PA (2003) Shifts in biochemical markers associated with wound healing in laryngeal secretions following phonotrauma: A preliminary study. Annals of Otology, Rhinology, & Laryngology 112: 1021–1025.K. VerdoliniCA RosenRC BranskiPA Hebda2003Shifts in biochemical markers associated with wound healing in laryngeal secretions following phonotrauma: A preliminary study.Annals of Otology, Rhinology, & Laryngology11210211025
- 50. Branski RC, Rosen CA, Verdolini K, Hebda PA (2005) Biochemical markers associated with acute vocal fold wound healing: a rabbit model. J Voice 19: 283–289.RC BranskiCA RosenK. VerdoliniPA Hebda2005Biochemical markers associated with acute vocal fold wound healing: a rabbit model.J Voice19283289
- 51. Branski RC, Rosen CA, Verdolini K, Hebda PA (2005) Acute vocal fold wound healing in a rabbit model. Ann Otol Rhinol Laryngol 114: 19–24.RC BranskiCA RosenK. VerdoliniPA Hebda2005Acute vocal fold wound healing in a rabbit model.Ann Otol Rhinol Laryngol1141924
- 52. Chow CC, Clermont G, Kumar R, Lagoa C, Tawadrous Z, et al. (2005) The acute inflammatory response in diverse shock states. Shock 24: 74–84.CC ChowG. ClermontR. KumarC. LagoaZ. Tawadrous2005The acute inflammatory response in diverse shock states.Shock247484
- 53. Day J, Rubin J, Vodovotz Y, Chow CC, Reynolds A, et al. (2006) A reduced mathematical model of the acute inflammatory response II. Capturing scenarios of repeated endotoxin administration. J Theor Biol. J. DayJ. RubinY. VodovotzCC ChowA. Reynolds2006A reduced mathematical model of the acute inflammatory response II. Capturing scenarios of repeated endotoxin administration.J Theor Biol
- 54. Prince JM, Levy RM, Bartels J, Baratt A, Kane JM 3rd, et al. (2006) In silico and in vivo approach to elucidate the inflammatory complexity of CD14-deficient mice. Mol Med 12: 88–96.JM PrinceRM LevyJ. BartelsA. BarattJM Kane 3rd2006In silico and in vivo approach to elucidate the inflammatory complexity of CD14-deficient mice.Mol Med128896
- 55. An G, Faeder J, Vodovotz Y (2008) Translational systems biology: introduction of an engineering approach to the pathophysiology of the burn patient. J Burn Care Res 29: 277–285.G. AnJ. FaederY. Vodovotz2008Translational systems biology: introduction of an engineering approach to the pathophysiology of the burn patient.J Burn Care Res29277285
- 56. Vodovotz Y, Csete M, Bartels J, Chang S, An G (2008) Translational systems biology of inflammation. PLoS Comput Biol 4: e1000014.Y. VodovotzM. CseteJ. BartelsS. ChangG. An2008Translational systems biology of inflammation.PLoS Comput Biol4e1000014
- 57. Ermentrout GB, Edelstein-Keshet L (1993) Cellular automata approaches to biological modeling. J Theor Biol 160: 97–133.GB ErmentroutL. Edelstein-Keshet1993Cellular automata approaches to biological modeling.J Theor Biol16097133
- 58. An G (2005) Mathematical modeling in medicine: a means, not an end. Crit Care Med 33: 253–254.G. An2005Mathematical modeling in medicine: a means, not an end.Crit Care Med33253254
- 59. Smallwood RH, Holcombe WML, Walker DC (2004) Development and validation of computational models of cellular interaction. Journal of Molecular Histology 35: 659–665.RH SmallwoodWML HolcombeDC Walker2004Development and validation of computational models of cellular interaction.Journal of Molecular Histology35659665
- 60. Vodovotz Y, Clermont G, Chow C, An G (2004) Mathematical models of the acute inflammatory response. Current Opinion in Critical Care 10: 383–390.Y. VodovotzG. ClermontC. ChowG. An2004Mathematical models of the acute inflammatory response.Current Opinion in Critical Care10383390
- 61. Walker DC, Hill G, Wood SM, Smallwood RH, Southgate J (2004) Agent-based computational modeling of wounded epiethelial cell monolayers. IEEE Transactions on Nonobioscience 3: 153–163.DC WalkerG. HillSM WoodRH SmallwoodJ. Southgate2004Agent-based computational modeling of wounded epiethelial cell monolayers.IEEE Transactions on Nonobioscience3153163
- 62. Butterfield TA, Best TM, Merrick MA (2006) The dual roles of neutrophils and macrophages in inflammation: a critical balance between tissue damage and repair. J Athl Train 41: 457–465.TA ButterfieldTM BestMA Merrick2006The dual roles of neutrophils and macrophages in inflammation: a critical balance between tissue damage and repair.J Athl Train41457465
- 63. Toumi H, F'Guyer S, Best TM (2006) The role of neutrophils in injury and repair following muscle stretch. J Anat 208: 459–470.H. ToumiS. F'GuyerTM Best2006The role of neutrophils in injury and repair following muscle stretch.J Anat208459470
- 64. Tidball JG (1995) Inflammatory cell response to acute muscle injury. Med Sci Sports Exerc 27: 1022–1032.JG Tidball1995Inflammatory cell response to acute muscle injury.Med Sci Sports Exerc2710221032
- 65. Eming SA, Krieg T, Davidson JM (2007) Inflammation in wound repair: molecular and cellular mechanisms. J Invest Dermatol 127: 514–525.SA EmingT. KriegJM Davidson2007Inflammation in wound repair: molecular and cellular mechanisms.J Invest Dermatol127514525
- 66. Stramer BM, Mori R, Martin P (2007) The inflammation-fibrosis link? A Jekyll and Hyde role for blood cells during wound repair. J Invest Dermatol 127: 1009–1017.BM StramerR. MoriP. Martin2007The inflammation-fibrosis link? A Jekyll and Hyde role for blood cells during wound repair.J Invest Dermatol12710091017
- 67. Vodovotz Y (2006) Deciphering the complexity of acute inflammation using mathematical models. Immunol Res 36: 237–245.Y. Vodovotz2006Deciphering the complexity of acute inflammation using mathematical models.Immunol Res36237245
- 68. Cockbill S (2002) Wound: the healing process. Hospital Pharmacist 9: 255–260.S. Cockbill2002Wound: the healing process.Hospital Pharmacist9255260
- 69. Martin P (1997) Wound Healing-aiming for perfect skin regeneration. Science 276: 75–81.P. Martin1997Wound Healing-aiming for perfect skin regeneration.Science2767581
- 70. Witte MB, Barbul A (1997) General principles of wound healing. Surgical Clinics of North America 77: 509–528.MB WitteA. Barbul1997General principles of wound healing.Surgical Clinics of North America77509528
- 71. Hirschi SD, Gray SD, Thibeault SL (2002) Fibronectin:an interesting vocal fold protein. J Voice 16: 310–316.SD HirschiSD GraySL Thibeault2002Fibronectin:an interesting vocal fold protein.J Voice16310316
- 72. Thibeault SL, Bless DM, Gray SD (2003) Interstitial protein alterations in rabbit vocal fold with scar. J Voice 17: 377–383.SL ThibeaultDM BlessSD Gray2003Interstitial protein alterations in rabbit vocal fold with scar.J Voice17377383
- 73. Thibeault SL, Gray SD, Bless DM, Chan RW, Ford CN (2002) Histologic and Rheologic characterization of vocal fold scarring. Journal of Voice 16: 96–104.SL ThibeaultSD GrayDM BlessRW ChanCN Ford2002Histologic and Rheologic characterization of vocal fold scarring.Journal of Voice1696104
- 74. Thibeault SL, Gray SD, Li W, Ford CN, Smith ME, et al. (2002) Genotypic and phenotypic expression of vocal fold polyps and Reinke's edema: a preliminary study. Ann Otol Rhinol Laryngol 111: 302–309.SL ThibeaultSD GrayW. LiCN FordME Smith2002Genotypic and phenotypic expression of vocal fold polyps and Reinke's edema: a preliminary study.Ann Otol Rhinol Laryngol111302309
- 75. Taylor KR, Trowbridge JM, Rudisill JA, Termeer CC, Simon JC, et al. (2004) Hyaluronan fragments stimulate endothelial recognition of injury through TLR4. J Biol Chem 279: 17079–17084.KR TaylorJM TrowbridgeJA RudisillCC TermeerJC Simon2004Hyaluronan fragments stimulate endothelial recognition of injury through TLR4.J Biol Chem2791707917084
- 76. Gunter HE (2003) A mechanical model of vocal-fold collision with high spatial and temporal resolution. J Acoust Soc Am 113: 994–1000.HE Gunter2003A mechanical model of vocal-fold collision with high spatial and temporal resolution.J Acoust Soc Am1139941000
- 77. Gunter HE (2004) Modeling mechanical stresses as a factor in the etiology of benign vocal fold lesions. Journal of Biomechanics 37: 1119–1124.HE Gunter2004Modeling mechanical stresses as a factor in the etiology of benign vocal fold lesions.Journal of Biomechanics3711191124
- 78. Jiang JJ, Diaz CE, Hanson DG (1998) Finite modeling of vocal fold vibratioin in normal phonation and hyperfunctional dysphonia: implications for the pathogenesis of vocal nodules. Ann Otol Rhinol Laryngol 107: 603–610.JJ JiangCE DiazDG Hanson1998Finite modeling of vocal fold vibratioin in normal phonation and hyperfunctional dysphonia: implications for the pathogenesis of vocal nodules.Ann Otol Rhinol Laryngol107603610
- 79. Thomson SL, Mongeau L, Frankel SH (2005) Aerodynamic transfer of energy to the vocal folds. J Acoust Soc Am 118: 1689–1700.SL ThomsonL. MongeauSH Frankel2005Aerodynamic transfer of energy to the vocal folds.J Acoust Soc Am11816891700
- 80. Roberts AB, Sporn MB (1996) Transforming growth factor-b. In: Clark RAF, editor. The Molecular and Cellular Biology of Wound Repair. New York: Plenum Press. pp. 275–308.AB RobertsMB Sporn1996Transforming growth factor-b.RAF ClarkThe Molecular and Cellular Biology of Wound RepairNew YorkPlenum Press275308
- 81. O'Kane S, Ferfuson MW (1997) Transforming growth factor betas and wound healing. International Journal of Biochemistry & Cell Biology 29: 63–78.S. O'KaneMW Ferfuson1997Transforming growth factor betas and wound healing.International Journal of Biochemistry & Cell Biology296378
- 82. Barcellos-Hoff MH, Ehrhart EJ, Kalia M, Jirtle R, Flanders K, et al. (1995) Immunohistochemical detection of active transforming growth factor-b in situ using engineered tissue. American Journal of Pathology 147: 1228–1237.MH Barcellos-HoffEJ EhrhartM. KaliaR. JirtleK. Flanders1995Immunohistochemical detection of active transforming growth factor-b in situ using engineered tissue.American Journal of Pathology14712281237
- 83. Chong H, Vodovotz Y, Cox GW, Barcellos-Hoff MH (1999) Immunocytochemical detection of latent TGF-b activation in cultured macrophages. Journal of Cellular Physiology 178: 275–283.H. ChongY. VodovotzGW CoxMH Barcellos-Hoff1999Immunocytochemical detection of latent TGF-b activation in cultured macrophages.Journal of Cellular Physiology178275283
- 84. Vodovotz Y, Chesler L, Chong H, Kim SJ, Simpson JT, et al. (1999) Regulation of transforming growth factor beta1 by nitric oxide. Cancer Res 59: 2142–2149.Y. VodovotzL. CheslerH. ChongSJ KimJT Simpson1999Regulation of transforming growth factor beta1 by nitric oxide.Cancer Res5921422149
- 85. Vodovotz Y, Hoffman R, Baratt A, Wei J, Lagoa C, et al. (2006) In silico and in vivo studies modeling the aged acute inflammatory response. Shock 25: 39–40 (Abstract).Y. VodovotzR. HoffmanA. BarattJ. WeiC. Lagoa2006In silico and in vivo studies modeling the aged acute inflammatory response.Shock253940 (Abstract)
- 86. Li NYK, Verdolini K, Clermont G, Mi Q, Hebda PA, et al. Simulating acute phonotrauma: a comparison of agent-based and equation-based models; 2006; Tokyo, Japan. NYK LiK. VerdoliniG. ClermontQ. MiPA HebdaSimulating acute phonotrauma: a comparison of agent-based and equation-based models; 2006; Tokyo, Japan
- 87. Whitcomb DC, Aoun E, Vodovotz Y, Clermont G, Barmada MM (2005) Evaluating disorders with a complex genetics basis. the future roles of meta-analysis and systems biology. Dig Dis Sci 50: 2195–2202.DC WhitcombE. AounY. VodovotzG. ClermontMM Barmada2005Evaluating disorders with a complex genetics basis. the future roles of meta-analysis and systems biology.Dig Dis Sci5021952202
- 88. Clermont G, Bartels J, Kumar R, Constantine G, Vodovotz Y, et al. (2004) In silico design of clinical trials: a method coming of age. Crit Care Med 32: 2061–2070.G. ClermontJ. BartelsR. KumarG. ConstantineY. Vodovotz2004In silico design of clinical trials: a method coming of age.Crit Care Med3220612070
- 89. An G (2001) Agent-based computer simulation and sirs: building a bridge between basic science and clinical trials. Shock 16: 266–273.G. An2001Agent-based computer simulation and sirs: building a bridge between basic science and clinical trials.Shock16266273
- 90. Kumar R, Clermont G, Vodovotz Y, Chow CC (2004) The dynamics of acute inflammation. J Theor Biol 230: 145–155.R. KumarG. ClermontY. VodovotzCC Chow2004The dynamics of acute inflammation.J Theor Biol230145155
- 91. Reynolds A, Rubin J, Clermont G, Day J, Vodovotz Y, et al. (2006) A reduced mathematical model of the acute inflammatory response: I. Derivation of model and analysis of anti-inflammation. J Theor Biol. A. ReynoldsJ. RubinG. ClermontJ. DayY. Vodovotz2006A reduced mathematical model of the acute inflammatory response: I. Derivation of model and analysis of anti-inflammation.J Theor Biol
- 92. An G (2004) In silico experiments of existing and hypothetical cytokine-directed clinical trials using agent-based modeling. Critical Care Medicine 32: 2050–2060.G. An2004In silico experiments of existing and hypothetical cytokine-directed clinical trials using agent-based modeling.Critical Care Medicine3220502060
- 93. Kumar R, Chow CC, Bartels J, Clermont G, Vodovotz Y (In Press) A mathematical simulation of the inflammatory response to anthraz infection. Shock. R. KumarCC ChowJ. BartelsG. ClermontY. VodovotzIn PressA mathematical simulation of the inflammatory response to anthraz infection.Shock
- 94. Peterson KL, Verdolini-Marston K, Barkmeir JM, Hoffman HT (1994) Comparison of aerodynamic and electroglottographic parameters in evaluating clinically relevant voicing patterns. Ann Otol Rhinol Laryngol 103: 335–346.KL PetersonK. Verdolini-MarstonJM BarkmeirHT Hoffman1994Comparison of aerodynamic and electroglottographic parameters in evaluating clinically relevant voicing patterns.Ann Otol Rhinol Laryngol103335346
- 95. Verdolini K, Druker DG, Palmer PM, Samawi H (1998) Laryngeal adduction in resonant voice. J Voice 12: 315–327.K. VerdoliniDG DrukerPM PalmerH. Samawi1998Laryngeal adduction in resonant voice.J Voice12315327
- 96. Verdolini-Marston K, Burke MK, Lessac A, Glaze L, Caldwell E (1995) Preliminary study of two methods of treatment for laryngeal nodules. Journal of Voice 9: 74–85.K. Verdolini-MarstonMK BurkeA. LessacL. GlazeE. Caldwell1995Preliminary study of two methods of treatment for laryngeal nodules.Journal of Voice97485
- 97. Mi Q, Riviere B, Clermont G, Steed DL, Vodovotz Y (2007) Agent-based modeling of inflammation and wound healing: insights into diabetic foot ulcer pathology and the role of transforming growth factor-beta 1. Wound Repair and Regeneration 15: 671–682.Q. MiB. RiviereG. ClermontDL SteedY. Vodovotz2007Agent-based modeling of inflammation and wound healing: insights into diabetic foot ulcer pathology and the role of transforming growth factor-beta 1.Wound Repair and Regeneration15671682
- 98. Robson MC, Steed DL, Franz MG (2001) Wound healing: biologic features and approaches to maximize healing trajectories. Curr Probl Surg 38: 72–140.MC RobsonDL SteedMG Franz2001Wound healing: biologic features and approaches to maximize healing trajectories.Curr Probl Surg3872140
- 99. Matzinger P (2002) The danger model: a renewed sense of self. Science 296: 301–305.P. Matzinger2002The danger model: a renewed sense of self.Science296301305
- 100. Mi Q, Riviere B, Clermont G, Steed DL, Vodovotz Y (in press) Agent-based modeling of inflammation and wound healing: insights into diabetic foot ulcer pathology and the role of transforming growth factor-beta 1. Wound Repair and Regeneration. Q. MiB. RiviereG. ClermontDL SteedY. Vodovotzin pressAgent-based modeling of inflammation and wound healing: insights into diabetic foot ulcer pathology and the role of transforming growth factor-beta 1.Wound Repair and Regeneration
- 101. Hahn MS, Kobler JB, Zeitels SM, Langer R (2006) Quantitative and comparative studies of the vocal fold extracellular matrix II: collagen. Ann Otol Rhinol Laryngol 115: 225–232.MS HahnJB KoblerSM ZeitelsR. Langer2006Quantitative and comparative studies of the vocal fold extracellular matrix II: collagen.Ann Otol Rhinol Laryngol115225232
- 102. Madruga de Melo EC, Lemos M, Aragao Ximenes Filho J, Sennes LU, Nascimento Saldiva PH, et al. (2003) Distribution of collagen in the lamina propria of the human vocal fold. Laryngoscope 113: 2187–2191.EC Madruga de MeloM. LemosJ. Aragao Ximenes FilhoLU SennesPH Nascimento Saldiva2003Distribution of collagen in the lamina propria of the human vocal fold.Laryngoscope11321872191
- 103. Tateya T, Tateya I, Bless DM (2006) Collagen subtypes in human vocal folds. Ann Otol Rhinol Laryngol 115: 469–476.T. TateyaI. TateyaDM Bless2006Collagen subtypes in human vocal folds.Ann Otol Rhinol Laryngol115469476
- 104. Courey MS, Shohet JA, Scott MA, Ossoff RH (1996) Immunohistochemical characterization of benign laryngeal lesions. Ann Otol Rhinol Laryngol 195: 525–531.MS CoureyJA ShohetMA ScottRH Ossoff1996Immunohistochemical characterization of benign laryngeal lesions.Ann Otol Rhinol Laryngol195525531
- 105. Gray SD, Hammond EH, Hanson D (1995) Benign pathologic response of the larynx. Ann Otol Rhinol Laryngol 104: 8–13.SD GrayEH HammondD. Hanson1995Benign pathologic response of the larynx.Ann Otol Rhinol Laryngol104813
- 106. Neves BMJ, Neto JG, Pontes P (2004) Histopathological and immunohistochemical differentiation of epithelial alterations in vocal nodule comparing to polyps and to laryngeal edema. Brazillian Journal of Otolaryngology 70: 439–448.BMJ NevesJG NetoP. Pontes2004Histopathological and immunohistochemical differentiation of epithelial alterations in vocal nodule comparing to polyps and to laryngeal edema.Brazillian Journal of Otolaryngology70439448
- 107. Railback SF (2001) Getting “results”: the pattern-oriented approach to analyze natural systems with individual-based models. Natural Resource Modeling 14: 465–474.SF Railback2001Getting “results”: the pattern-oriented approach to analyze natural systems with individual-based models.Natural Resource Modeling14465474