D1 Research and Development

RERC SCI logo

 


Overview of Research and Development Activities


State of the Science on Spinal Cory Injury


Research Projects: Effect of Local Cooling | Effects of Weight Shifting | Handrim Technology


Development Projects: Inflammation Modeling | Low Shear, Cool Cushion | Propulsion Training Tools


 

D1 Mathematical Model of Inflammation and Healing Following Spinal Cord Injury

Task Leader(s): Yoram Vodovotz, Ph.D.

Co-Investigator: Gary An, M.D.; Greg Constantine, Ph.D.; Qi Mi, Ph.D.

Collaborators: Immunetrics, Inc.; IBM; Graduate Student Researcher Alexey Solovyev University of Pittsburgh, Department of Mathematics; Technician Derek Barclay (LuminexTM and ELISA assay technician, University of Pittsburgh, Department of Surgery)

Project Overview

Our long-term goal is to develop mathematical models of inflammation and healing following spinal cord injury for determining risk and monitoring progression of pressure ulcers, urinary tract infections, and musculoskeletal injury. These models will improve rehabilitation technology and approaches by predicting individual needs and the ultimate success of specific interventions. Our specific aims are:

  • Aim 1: To develop a predictive agent-based model of the formation and progression of pressure ulcers following spinal cord injury
  • Aim 2: To develop a preliminary, agent-based model of the formation and progression of urinary tract infection following spinal cord injury
  • Aim 3: To develop a preliminary, agent-based model of the trajectory of inflammation and healing of upper extremity injury following spinal cord injury
  • Aim 4: To carry out in silico (simulated) clinical studies using the agent based models of pressure ulcers, urinary tract infections, and upper extremity injuries

Project Objective(s)

Inflammation is a central, modulating process in many complex diseases (e.g. sepsis, infectious disease, trauma, asthma, allergy, autoimmune disorders, transplant rejection, cancer, neurodegenerative diseases, obesity, and atherosclerosis), and is a central driver of the physiology of people with spinal cord injuries. However, inflammation is not an inherently detrimental process: properly regulated, inflammation is required for successful immune response and wound healing. Inflammation is also a complex biological process that has defied reductionist, linear approaches.

The investigative team on this project has previously created mathematical and agent-based models (ABM) of the integrated inflammatory and damage/healing responses. They have used these models to elucidate basic features of acute inflammation in sepsis, trauma/hemorrhage, and altered healing in diabetic foot ulcers. They have also accomplished translational objectives using these simulations, including both in silico clinical trials and patient-specific, predictive simulations of inflammation.

This track record of success in gaining both basic and translational insights into a highly complex process-inflammation and organ damage/dysfunction in sepsis and trauma-for the first time raises the possibility of deciphering the complexity of the integrated inflammatory/healing response in settings relevant to post-injury rehabilitation. As a first step, based on an analysis of the relevant literature, we propose a systems-based schematic of the interactions among inflammatory elements, anti-inflammatory elements, muscle damage, and neural damage (Figure 1).

Figure 1 - The process of acute inflammation and tissue damage/healing, distilled for mathematical modeling purposes. Solid arrow: induction; dashed line: suppression. ). In our model, an initiating stimulus (e.g. spinal cord injury) stimulates both pro- and anti-inflammatory pathways. Pro-inflammatory agents (e.g. TNF) cause tissue damage/dysfunction, which in turn stimulates further inflammation (e.g. through the release of "danger signals"). Anti-inflammatory agents (e.g. TGF-ß, PGE2) both suppress inflammation and stimulate healing.

We postulate that inflammation and its interrelations with tissue damage/healing are a central cause of pressure ulcers following spinal cord injury, with inflammation detected via circulating cytokines in the blood that exhibits inter-individual variability with regards to risk factors. Therefore, computer simulations encompassing the biology of inflammation and damage / healing that will be calibrated with inflammatory analyte data, as well as non-invasive data on the onset and progression of pressure ulcers, will be capable of predicting the onset, progression, and characteristics of pressure ulcers post-spinal cord injury in individuals. We also postulate that chronic inflammation and its integrated effects on the healing cascade lead to fibrosis, which in turn creates a milieu conducive to urinary tract infections and further fibrosis, with detectable changes in inflammatory mediators in the urine. Therefore, we will be able to create and calibrate a computer simulation of the onset and progression of urinary tract infection following spinal cord injury. Finally, we postulate the inflammatory process set in motion by spinal cord injury, coupled with extensive utilization of a manual wheelchair, will lead to shoulder injuries in persons with spinal cord injuries. Therefore, we will be able to create and calibrate a computer simulation of this process based on analysis of inflammatory cytokines in the blood.

Methods

This project will create, calibrate, and validate computer simulations of the interlinked inflammatory and damage/healing responses following spinal cord injury. Three complications of spinal cord injury linked to inflammation and healing are pressure ulcers, urinary tract infections, and musculoskeletal injuries. Extrapolating from our work on diabetic foot ulcers, we will model and predict, on a patient-specific basis, the pressure ulcer response post-spinal cord injury. We have carried out preliminary studies on the formation of urinary tract infections in an inflammatory milieu, and this preliminary work will be used to build a computer simulation of this process following spinal cord injury. Though we have not carried out preliminary work on how inflammation leads to musculoskeletal injury, we have done work in trauma/hemorrhagic shock and phonotraumatic injury. Moreover, we have created a schema of how nerve and muscle injury interact to promote inflammation (Figure 1). This preliminary work will guide our effort to create a computer simulation of the inflammation and musculoskeletal tissue damage/healing following spinal cord injury.

Based on existing literature, we understand that spinal cord injury results in an elevation in baseline, circulating inflammatory cytokines and other inflammatory mediators such as nitric oxide (NO). The characteristics of the inflammatory response in each patient are likely to be partially dictated by the nature of the original spinal cord trauma and partially by genetic variations (single-nucleotide polymorphisms) in genes coding for cytokines. In the case of Specific Aim 1, we believe that elevated inflammatory milieu interacts non-linearly with local, pressure-driven ischemia-reperfusion and secondary oxygen radical formation to cause pressure ulcers. In the case of Specific Aim 2, we believe that the initial elevation in pro-inflammatory cytokines drives the increase in anti-inflammatory cytokines, which predispose the patient to urinary tract infections. The ongoing presence of pro-inflammatory stimuli (including infections) maintains and propagates the anti-inflammatory response, which in turn results in fibrosis, tissue damage, and further inflammation. Accordingly, we will develop ABM capable of reproducing the dynamics (time courses) of inflammatory cytokines in the blood and urine of patients sampled frequently at early time points post-injury. Based on these data, we will create patient-specific ABM. In subsequent work, these calibrated agent-based models will be used to predict the rates of formation of pressure ulcers and urinary tract infections in patients at intermediate and late time points post-injury. Finally, we will also create similar models of musculoskeletal injury following spinal cord injury. These models will include interactions between damaged nerves and muscles. We believe that the inflammatory milieu characteristic of the spinal cord injury patient, in conjunction with the biomechanical stress associated with continuous use of a manual wheelchair, will result in increased incidence of musculoskeletal injury. We will create a computer simulation of this response that, in future studies, will be used to predict patient-specific predisposition to musculoskeletal injury.

Dr. Constantine will analyze the data obtained in the proposed studies using standard statistical methods to determine significant changes in biofluids and other markers, along with associations with specific outcome measures related to pressure ulcers, urinary tract infections, and musculoskeletal injuries. In addition, these data will be used for data-driven modeling, an approach parallel to our ABM. We have made much progress in the use of mathematical and agent-based models to gain insights into the acute inflammatory response. The work we propose herein would greatly extend our progress. However, it is prudent to take a parallel approach in our modeling efforts. As described above, we will obtain repeated samples from the same patient over time. This rich dataset will allow us to use statistical modeling approaches sufficiently powered to augment our ABM. Our goal will be to compare the predictions from this statistical model with the predictions of the ABM, both of which will be based on the data obtained.

The Center for Inflammation and Regenerative Modeling (CIRM) and the Society for Complexity in Acute Illness

Mathematical modeling of a complex biological process such as inflammation benefits tremendously from interdisciplinary expertise. Dr. Vodovotz established the first institution dedicated solely to studying inflammation through an interdisciplinary process consisting of biomarker analysis, modeling/simulation, and experiments (Figure: the Center for Inflammation and Regenerative Modeling (CIRM, www.mirm.pitt.edu/cirm). The CIRM includes investigators both at the University of Pittsburgh and at collaborating institutions, both academic and industrial. The CIRM was created under the auspices of the University of Pittsburgh's McGowan Institute for Regenerative Medicine (MIRM; www.mirm.pitt.edu), the first national center for regenerative medicine. At the CIRM, the two main types of simulations being carried out involve either differential equations or ABM. The former are a class of models that has been time-tested and that forms much of the underlying knowledge base of classical physiology. The latter involves the simulation of discreet "agents" that move and interact with other agents, leading to emergent phenomena for relatively simple underlying rules.

Drs. Vodovotz and An, as well as Steve Chang of Immunetrics and collaborators at IBM, were among the founders of the Society for Complexity in Acute Illness (SCAI; www.scai-med.org), with over 160 members worldwide. Dr. Vodovotz is currently SCAI's Secretary, and Dr. An and Mr. Chang serve on SCAI's Board of Directors. The Society's International Conference on Complexity in Acute Illness (www.iccai.org) serves as the only dedicated venue for work on basic and translational applications of mathematical modeling in inflammatory diseases.

Figure 2

Figure 2 - Evidence-based modeling. Model building is an iterative process involving calibration from existing or new data, and validation from prediction of data. This process identifies both areas where a model is correct and where it is deficient relative to data and therefore must be corrected

Expected Findings and Deliverables

We will create various ABM’s related to the process of spinal cord injury, the subsequent inflammatory response, and the impact of inflammation on pressure ulcers, urinary tract infections, and musculoskeletal injuries. We will obtain serial time course data on plasma and urine inflammation biomarkers from SCI patients, and utilize them to create patient-specific, predictive ABM’s for these secondary complications of SCI.

Project Updates

2012

Recruitment, data collection and sample processing:
To date, a total of 325 subjects with spinal cord injury have been screened for enrollment in our study. We have successfully recruited 102 subjects. We have assayed over 1600 plasma and urine samples for 25 inflammatory cytokines (using the CIRM’s Luminex™ platform) along with NO2-/NO3- (using the nitrate reductase method).

We have created a new web based SQL database system, which will allow for direct entry, eliminating the need for duplicate entries to confirm data, and will allow direct access to all members of the research team, improving communication and capacity of the data dissemination.

Computational Model Development:
In this final year of funding, Dr. Qi Mi, Alexey Solovyev, and Gary An, along with a Computational Biology graduate student (Cordelia Ziraldo) have continued to work on all of the simulations described in previous reports, and to develop our agent-based modeling (ABM) platform. We continue to develop our novel ABM framework, SPARK (Simple Platform for Agent-based Representation of Knowledge).

We have fully merged the shear stress mechanisms with the Ischemia/Reperfusion Injury model and now we are in the iterative process of calibration and refinement of the model, using images of pressure ulcers from subjects in the study. We are also continuing to incorporate the contributions of neutrophils to both inflammation and reperfusion injury. Our initial data suggest that we can match aspects of the pattern of ulcer formation with our simulations. We continue to explore the capabilities of this model, focusing on how certain parameters (such as oxygen availability) affect the differential morphology of the developing ulcer. We have developed a method to track the evolving outlines of the pressure ulcers, and are using it to calculate metrics that can be directly compared to images of real ulcers. We will also apply these metrics to the real images, and then compare the metric results of the model to the metric calculations on real images. With this goal in mind, we have refined the rendering to produce more tissue-realistic simulations.

Using an IBM 32-node supercomputer, it is possible to run several long simulations in parallel very quickly. This allows us to generate sensitivity analyses much more efficiently, and we are focusing on this now. Sensitivity analysis is a group of methods that seeks to identify the parameters of the model whose values most influence the outcome of the model. This analysis leads us to another set of iterations where we refine the mechanisms of the model to truly reflect the biology we are modeling. We are currently focusing our sensitivity analysis on the inflammatory mechanism in the model and we recently completed analyses of parameters affecting oxygen availability and pressure intensity and duration. We are able to use the visual output of the model as an additional gauge of parameters in the model. We have completed several iterations of sensitivity analysis and model updating. Through these analyses, we are discovering control points in the model that suggest possible avenues for therapeutic interventions. Equally important, we are finding other avenues that seemed promising are not as sensitive as we had hoped. We are continuing to explore the model through sensitivity analyses and hope to understand these dynamics better. We have added IL-1alpha as a second pro-inflammatory mediator and are calibrating the dynamics of this well- studied system to data from the literature, based on published studies that suggest a link between sepsis/pneumonia and the formation of pressure ulcers. By including these long-range effects in the model, we will be more able to study how secondary illnesses such as fever, infection, etc can affect ulcer development. We are in the process of doing the sensitivity analysis for the model parameters and also writing the manuscript. Finally, we have carried out a simulated clinical study of steroid administration to the simulated PU at different times following SCI, finding that this treatment reduces inflammatory damage, but does not prevent ulceration.

A separate manuscript describes a hybrid (equation- and agent-based) computational model of reactive hyperemia and pressure ulcers in people with and without spinal cord injury. The relationship between pressure and the course of ulcer formation is demonstrated in this model. This efficient model also captures several other important characteristic patterns of pressure ulcer formation. Based on data on blood flow in non-injured human subjects vs. SCI patients, the parameters gained from this hybrid model predicted the higher likelihood of pressure ulcer development in SCI patients. We suggest that this platform may at some point have diagnostic utility.

Additional analyses of plasma and urine:
We are continuing to analyze the plasma and urine of SCI patients; the initial plasma and urine cytokine data are now being analyzed. We have collected data on circulating monocyte profiles from the whole blood samples, and completed a comparison with circulating cytokine profiles. The results have demonstrated interesting novel phenotypes of pro-and anti-inflammatory phenotypes in patients with traumatic spinal cord injury. The results have been presented at the ISCoS meeting in June, 2011, and AAP meeting in March 2012. A manuscript is currently in preparation. We have also begun to characterize plasma cytokine levels in a population of blunt trauma patients without SCI, in order to compare the inflammatory characteristics of these two populations.

New projects:

  1. A new project has been initiated with the working hypothesis: Salzburg scale will be more sensitive in predicting PU in SCI population than the Braden scale. We will be collecting additional data from the clinical record, including albumin and hematocrit to facilitate calculation of the Salzburg scale.
  2. Comparison of cytokine profiles, Injury severity scores, and possibly PHQ-9 between SCI and TBI populations has been initiated.

 

2011

To date, a total of 212 subjects with spinal cord injury have been screened for enrollment in our study and 79 successfully consented. From these patients, we have assayed over 1300 plasma and urine samples for 25 inflammatory cytokines (using the CIRM's Luminex™ platform) along with NO2-/NO3- (using the nitrate reductase method). Due to difficulty with merging our clinical data with cytokine data in addition to errors in data entry in our Access database, plans are in place to enter data into a new web based SQL database system, which will allow for direct entry, eliminating the need for duplicate entries to confirm data, and will allow direct access to all members of the research team, improving communication. In addition, modifications to the database have included improvements in collection of pressure ulcer data to ensure greater accuracy and partnered with wound care team to facilitate timely assessments, and collection of data on non-pressure ulcer skin conditions, which could potentially affect the inflammatory profile. IRB modification and a secondary analysis were initiated, with measurement of cell surface markers of circulating monocytes from whole blood. Results from the cell phenotype study pilot of 22 subjects were accepted to be presented at the ISCoS meeting in June, 2011.

In this fourth year of funding, Dr. Qi Mi, Alexey Solovyev, and Gary An, along with a Computational Biology graduate student (Cordelia Ziraldo) have continued to develop an efficient, open-source agent-based modeling (ABM) platform using the Java programming language (Simple Platform for Agent-based Representation of Knowledge [SPARK]; Refs 18-19; freely downloadable at http://www.pitt.edu/~cirm/spark/). Using SPARK, we have begun to model the post-SCI insult(s) that lead(s) pressure ulcer formation. In this simulation, we hypothesize that constriction of blood vessels due to pressure initiates the activation of inflammatory cells that in turn lead to further reduced tissue health. The relationship between pressure and the course of ulcer formation is demonstrated in this model. The model suggests that a high turning frequency can lead to reduced incidence of pressure ulcers. This efficient model also captures several other important characteristic patterns of pressure ulcer formation.

We are continuing to develop our SCI ABM platform. We fully merged the shear stress mechanisms with the Ischemia/Reperfusion Injury model and now we are in the iterative process of calibration and refinement of the model, using images of pressure ulcers from subjects in the study. We are also continuing to incorporate the contributions of neutrophils to both inflammation and reperfusion injury. Our initial data suggest that we can match aspects of the pattern of ulcer formation with our simulations. We continue to explore the capabilities of the model, focusing on how certain parameters (such as oxygen availability) affect the differential morphology of the developing ulcer. We have developed a method to track the evolving outlines of the pressure ulcers, and are using it to calculate metrics that can be directly compared to images of real ulcers. We will also apply these metrics to the real images, and then compare the metric results of the model to the metric calculations on real images. With this goal in mind, we have refined the rendering to produce more tissue-realistic simulations, and have secured access to the subject database, which gives us a larger number of photos and patients to compare against our model. We have culled through the clinical database of photographs and identified patients with relevant time courses of ulcer development (due to both pressure and shear) that we can compare to our model output. We now also have access to an IBM 32-node supercomputer, making it possible to run several long simulations in parallel very quickly. This allows us to generate sensitivity analyses much more efficiently, and we are focusing on this now. Sensitivity analysis is a group of methods that seeks to identify the parameters of the model whose values most influence the outcome of the model. This analysis leads us to another set of iterations where we refine the mechanisms of the model to truly reflect the biology we are modeling. We are currently focusing our sensitivity analysis on the inflammatory mechanism in the model. We have completed two iterations of sensitivity analysis and model updating. Through these analyses, we are discovering control points in the model that suggest possible avenues for therapeutic interventions. Equally important, we are finding other avenues that seemed promising are not as sensitive as we had hoped. We are continuing to explore the model through sensitivity analyses and hope to understand these dynamics better. We have added IL-1beta as a second pro-inflammatory mediator and are calibrating the dynamics of this well-studied system to data from the literature. We also would like to incorporate the effects of systemic inflammation on the developing ulcer, based on published studies that suggest a link between sepsis/pneumonia and the formation of pressure ulcers. By including these long-range effects in the model, we will be more able to study how secondary illnesses such as fever, infection, etc can affect ulcer development. We are continuing to work on developing the hybrid model (combining both ordinary differential equations and ABM approaches) for the simulation of blood flow during the ischemia/reperfusion process. We are in the process of estimating the model parameters that characterize the control and SCI patients. Finally, we are continuing to analyze the levels of inflammatory mediators in the plasma and urine of SCI patients; these data are now being analyzed by Dr. Sowa's group as well as by Dr. Constantine.

2010

In total, we have screened 128 patients with spinal cord injury and successfully consented 54. From these patients, we have assayed a total of 225 samples for 25 inflammatory cytokines (using the CIRM's Luminex™ platform) along with NO2-/NO3- (using the nitrate reductase method). 46 patients were not eligible; the exclusions included Diabetes Mellitus (4), Rheumatoid arthritis (2), Age <18 (3), autoimmune disease (4), prior spinal cord injury (7), non-traumatic (4), multiple sclerosis (1), stroke (1) and lower extremity motor score >30 (20, which occurred prior to our removal of this exclusion criteria).

We have expanded recruitment to include UPMC Mercy inpatient units. We have expanded the capacity of the database to include new more sensitive measures of pressure ulcers, information regarding ambulation status after discharge, and information about injury severity to facilitate future comparisons to other traumatic conditions (non-spinal cord). We have begun collecting data on circulating monocyte cell surface markers to use as potential prognostic indicators, and have begun regular data management meetings, which serve to formulate new research hypotheses that can be tested using the existing database.

In the third year of funding for D1, Dr. Qi Mi, Alexey Solovyev, and Gary An, along with a new Computational Biology graduate student (Cordelia Ziraldo) have developed an efficient agent-based models using the Java programming language in order to support all of the modeling efforts within D1. This work has been accepted for publication (Solovyev, A.; Mikheev, M.; Zhou, L.; Dutta-Moscato, J.; Ziraldo, C.; An, G.; Vodovotz, Y.; Mi, Q., SPARK: A framework for multi-scale agent-based biomedical modeling. Int. J. Agent Technologies and Systems. 2010, In Press).

Using SPARK, we have begun to model the post-SCI insult(s) that lead(s) pressure ulcer formation. In this simulation, we hypothesize that constriction of blood vessels due to pressure initiates the activation of inflammatory cells that in turn lead to further reduced tissue health. The relationship between pressure and the course of ulcer formation is demonstrated in this model. The model suggests that a high turning frequency can lead to reduced incidence of pressure ulcers. This efficient model also captures several other important characteristic patterns of pressure ulcer formation. This work was presented at the 2009 RESNA meeting.

We are continuing to develop the pressure ulcer ABM to incorporate the effects of both shear and ischemia and reperfusion injury on the development of pressure ulcers. The simulation currently incorporates epithelial cells, inflammatory cells, blood vessels, inflammatory cytokines, and danger signals ("damage-associated molecular pattern" molecules), and incorporates mechanistic rules for pressure-induced ischemia/reperfusion and shear stress-mediated injury. We have partially calibrated the simulation with blood flow data obtained in collaboration with R1. We simulated two types of initial injury: 1) a circumscribed area of pressure whose intensity decreases radially, simulating the effect of a bony protrusion against soft tissue, followed by ischemia/reperfusion injury, activation of an inflammatory cascade, and production of oxygen free radicals; or 2) shear force injury, in which epithelial cells are damaged by tissue stretching. In both cases, tissue damage stimulates inflammatory cells to produce cytokines, and further damage caused by cytokines leads to ulceration. The shapes of the simulated ulcers qualitatively match patterns seen clinically, including position and size relative to the protrusion and the development of secondary satellite ulcers that eventually coalesce with the primary ulcer. Simulated ulcers induced by ischemia/reperfusion injury are predicted to be circular, whereas ulcers induced by shear stress are predicted to be elliptical. Photos of ulcers taken at early timepoints often show an elliptical shape. This work was presented in abstract form at the 2010 Wound Healing Society annual meeting. We are continuing to work to analyze images of pressure ulcers from subjects in the study that will serve as further validation of our ABM. In the coming year, we will explore in silico the predicted impact of local cooling on pressure ulcer formation. We suggest that this evolving model can serve as a platform for personalized clinical decision support for the management of SCI patients. We are preparing a manuscript on this study.

As mentioned above, we have obtained extensive time course data on numerous inflammation biomarkers. We are performing several novel analyses on these data. One of these analyses utilizes a novel method developed by Drs. Constantine and Mi, to which we refer as Dynamic Profiling. In this method, a recursive process of clustering based on parameters of injury, demographics, and inflammation biomarkers will be used to predict the likelihood of secondary complications post-SCI (pressure ulcers, urinary tract infections, and musculoskeletal injuries). We are also pursuing an alternative clustering approach based the cytokine data alone, as well as Principal Component Analysis on the inflammation biomarkers as a method for predicting the likelihood of secondary complications of SCI. We are currently preparing a manuscript on these data-driven approaches to predicting outcomes post-SCI.

2009

In total, we have screened 54 patients with spinal cord injury and successfully consented 19. Of the 29 not eligible, 19 of these were due to having a lower extremity motor score of greater than 30. As a result of this finding, we have recently expanded our inclusion criteria to include patients with a lower extremity motor score of greater than 30 to allow for testing of novel pilot hypotheses as well as examine differences among ambulators and non-ambulators. Of the 19 enrolled, 7 were lost to outside rehabilitation facilities. Therefore, we have also added a modification and additional consent form to allow us to recapture these patients after discharge home. We have also expanded our capacity for recruitment through partnering with critical care medicine and accessing the regularly updated UPMC trauma list. We have also obtained IRB approval to obtain saliva samples to facilitate genotyping in addition to the originally proposed serum and urine measurements.

In the second year of funding for D1, Dr. Qi Mi, Alexey Solovyev, and Gary An have developed an efficient agent-based models using the Java programming language in order to support all of the modeling efforts within D1. Though the NetLogo software we have used in the past is a good framework for relatively simple and small models, it is important for us to be able to work with a large number of agents (several thousand). This is computationally intensive, and therefore we carried out initial efforts at scaling us using the Repast Symphony framework. However, for this large-scale work, even this software was inefficient. This framework is now fully implemented, including a high-level programming language as well as support for parallel processing. This work has been presented in abstract form (SPARK: A systems biology framework for agentbased biomedical modeling; Computational Cell Biology; Cold Spring Harbor Laboratory, NY; March 24-27, 2009).

Using SPARK, we have begun to model the post-SCI insult(s) that lead(s) pressure ulcer formation. In this simulation, we hypothesize that constriction of blood vessels due to pressure initiates the activation of inflammatory cells that in turn lead to further reduced tissue health. The relationship between pressure and the course of ulcer formation is demonstrated in this model. The model suggests that a high turning frequency can lead to reduced incidence of pressure ulcers. This efficient model also captures several other important characteristic patterns of pressure ulcer formation. This work was recently accepted for presentation at the 2009 RESNA meeting. We are currently in the process of obtaining calibration and validation data in both normal subjects and SCI patients, in collaboration with R1. We will explore in silico the predicted impact of local cooling on pressure ulcer formation. We suggest that this evolving model can serve as a platform for personalized clinical decision support for the management of SCI patients, as we continue calibrate and validate it. We are preparing a manuscript on this study. We are also continuing to model pressure ulcer formation using a variant of our published diabetic foot ulcer model, in order to cross-validate model predictions.

We have obtained extensive time course data on 25 cytokines (using the Luminex™ platform) as well as NO2-/NO3- (using the nitrate reductase method) in both plasma and urine from 11 SCI patients during the acute post-injury (in-hospital) period. Ten of these patients continue to be followed, with one dropping out due to transfer to another facility. We are performing several novel analyses on these data. One of these analyses utilizes a novel method developed by Drs. Constantine and Mi, to which we refer as Dynamic Profiling. In this method, a recursive process of clustering based on parameters of injury, demographics, and inflammation biomarkers will be used to predict the likelihood of secondary complications post-SCI (pressure ulcers, urinary tract infections, and musculoskeletal injuries). We are also pursuing an alternative clustering approach based the cytokine data alone, as well as Principal Component Analysis on the inflammation biomarkers as a method for predicting the likelihood of secondary complications of SCI. We are currently preparing a manuscript on these data-driven approaches to predicting outcomes post-SCI.

During this funding period, we published two primary articles related to our modeling and systems biology work as well as to inflammation in the setting of urinary tract infection/inflammation (Smaldone et al., 2008; Wognum et al., 2009). Several primary articles on urinary tract inflammation are currently under review (Tyagi, P.; Barclay, D.; Zamora, R.; Yoshimura, N.; Vodovotz, Y.; Chancellor, M. Urine cytokines suggest an inflammatory response in the overactive bladder-A pilot study. Tyagi, V.; Tyagi, P.; Yoshimura, N.; Witteemer, E.; Barclay, D.; Loughran, P.; Zamora, R.; Vodovotz, Y. Gender-based reciprocal expression of transforming growth factor-?1 and the inducible nitric oxide synthase in a rat model of cyclophosphamide-induced cystitis) and in preparation (Solovyev, A.; Zhou, L.; Mikheev, M.; Dutta-Moscato, J.; Ziraldo, C.; An, G.; Vodovotz, Y.; Mi, Q. SPARK: A Framework for Agentbased Biomedical Modeling. Mi, Q.; Solovyev, A.; Zhou, L.; Mikheev, M.; Dutta-Moscato, J.; An, G.; Vodovotz, Y. An agent-based mathematical model of pressure ulcer following spinal cord injury: An application of a novel agent-based biomedical modeling framework). In addition, we have published several review manuscripts and book chapters (Vodovotz et al., 2009; An et al., 2009; Vodovotz and An, 2009). On limited existing SCI data, we have a paper in progress entitled “SCI Dynamic Patient Profiling based on plasma and urine cytokine readings” with Greg Constantine as coauthor.
In collaboration with Nicole Li (graduate student researcher [mentor: Dr. Verdolini], Department of Communication Science and Disorders) and Silvia Wognum (graduate student researcher [mentor: Michael Sacks], Department of Bioengineering), we have created a urinary tract infection/fibrosis agent-based model, on which work is ongoing. Dr. An is writing two other manuscripts, one on qualitative dynamic knowledge representation of intracellular processes and the other on data-parallel techniques for agent-based tissue modeling on graphics processing units.

2008

We have created a high-level agent-based model (ABM) of pressure ulcer formation following spinal cord injury, which demonstrates a plausible relationship between pressure and the course of ulcer formation. This ABM also captures several other important characteristic patterns of pressure ulcer formation, and may provide insights into the pathogenesis and effective treatment of such ulcers. The simulation is based on the ABM framework (SPARK; Simple Platform for Agent-based Representation of Knowledge) that we recently developed (www.pitt.edu/~cirm/spark). It has also been successfully tested on Immunetrics's high-performance computing cluster, and its parallelized implementation is being ported to that cluster. We envision this ABM as a starting point for rational prediction of pressure ulcer therapy, such as optimal patient turning frequency and localized cooling. In parallel, we are investigating the immune response in SCI patients, who are being recruited following Institutional Review Board approval. These patients exhibit typical secondary complications of SCI, including pneumonia, sacral and mid buttock pressure ulcers, urinary tract infections, and pleural effusions.

Blood and urine samples are routinely assayed for various inflammatory analytes. Based on the above studies, we are beginning to gain insight into the inflammatory dynamics of SCI patients, and to create clinically relevant simulations relating to the formation of pressure ulcers post-SCI.

Publications

Constantine, G.; Bartels, J.; Buliga, M.; Clermont, G.; Vodovotz, Y. A linear code parameter search algorithm with applications to immunology. Computational Optimization and Applications. 2007. doi:10.1007/s10589-007-9118-9. EISSN: 15732894 ISSN: 09266003.

Smaldone, M.C.; Vodovotz, Y.; Philips, B.J.; Barclay, D.; Prantil, R.; Yoshimura, N.; Chancellor, M.B.; Tyagi, P. Multiplex analysis of urinary cytokine levels in a rat model of cyclophosphamide-induced cystitis. J. Urology. 73:421-426.

Daun, S.; Rubin, J.; Vodovotz, Y.; Roy, A.; Parker, R.; Clermont, G. An ensemble of models of the acute inflammatory response to bacterial lipopolysaccharide in rats: Results from parameter reduction. J. Theoretical Biol. 2008. 253:843-853.

Li, N.Y.K.; Verdolini, K.; Clermont, G.; Mi, Q.; Hebda, P.A.; Vodovotz, Y. A patient-specific in silico model of inflammation and healing tested in acute vocal fold injury. PLoS ONE. 2008. 3:e2789.

Barclay, D.; Zamora, R.; Torres, A.; Namas, R.; Steed, D.; Vodovotz, Y. A simple, rapid, and convenient Luminex™-compatible method of tissue isolation. J. Clin. Lab. Analysis. 2008. 22:278-281.

Vodovotz Y, Csete M, Bartels J, Chang S, An G. Translational systems biology of inflammation. PLoS Comput. Biol. 2008. 4:1-6.

Daun S, Rubin J, Vodovotz Y, Clermont G. Equation-based models of dynamic biological systems. J. Crit. Care. 2008. 23:585-594.

Wognum S, Lagoa C, Nagatomi J, Sacks M.S, Vodovotz Y. An exploratory temporal analysis of the rat bladder wall transcriptome after spinal cord injury: Insights on remodeling, inflammation and infection using pathways analysis. PLoS ONE. 2009. 4:e582.

Tyagi V, Tyagi P, Yoshimura N, Witteemer E, Barclay D, Loughran P, Zamora R, Vodovotz Y. Gender-based reciprocal expression of transforming growth factor-β1 and the inducible nitric oxide synthase in a rat model of cyclophosphamide-induced cystitis. J. Inflammation. 2009. 6:23.

Vodovotz Y, Constantine G, Rubin J, Csete M, Voit E, An G. Mechanistic simulations of inflammation: Current state and future prospects. Mathematical Biosciences. 2009. 217:1-10.

Namas R, Ghuma A, Hermus L, Zamora R, Okonkwo D.O, Billiar T.R, Vodovotz Y. The acute inflammatory response in trauma / hemorrhage and traumatic brain injury: Current state and emerging prospects. Libyan Journal of Medicine. Vol 4, No 3. 2009.

An G, Mi Q, Dutta-Moscato J, Solovyev A, Mikheev M, Vodovotz Y. Agent-based models in translational systems biology. Wiley Interdisciplinary Reviews – Systems Biology. 2009. 1:159-171.

Rajaie Namas, Alexey Solovyev, Maxim Mikheev, Derek Barclay, Ruben Zamora, Qi Mi and Yoram Vodovotz: Post-traumatic Spinal Cord Injury: Inflammatory Response and Modeling (abstract), McGowan Institute for Regenerative Medicine 2009 retreat

Alexey Solovyev, Leming Zhou, Maxim Mikheev, Joyetta Dutta-Moscato, Gary An, Yoram Vodovotz and Qi Mi: SPARK: A Systems Biology Framework for Agent-based Biomedical Modeling (abstract), Cold Spring Harbor 2009 Computational Cell Biology meeting

Rivière, B.; Epshteyn, Y.; Swigon, D.; Vodovotz, Y. A mathematical model of signaling resulting from the binding of lipopolysaccharide with Toll-like receptors demonstrates inherent preconditioning behavior. Math. Biosci Jan 2009. 217:19-26.

Solovyev A, Mikheev M, Zhou L, Dutta-Moscato J, Ziraldo C, An G, Vodovotz Y, Mi Q. SPARK: A framework for multi-scale agent-based biomedical modeling. Int. J. Agent Technologies and Systems. 2010. 2:18-30

Mi Q, Li N.Y.K, Ziraldo C, Ghuma A, Mikheev M, Squires R, Okonkwo D.O, Verdolini-Abbott K, Constantine G, An G, Vodovotz Y. Translational systems biology of inflammation: Applications to personalized medicine. Personalized Medicine. 2010. 7:549-559

An G.; Bartels J.; Vodovotz Y. In silico augmentation of the drug development pipeline: Examples from the study of acute inflammation. Drug Dev. Res. 2010. 72:1-14

Tyagi P, Barclay D, Zamora R, Yoshimura N, Peters K, Vodovotz Y, Chancellor M. Urine cytokines suggest an inflammatory response in the overactive bladder-A pilot study. International Urology and Nephrology. 2010. 42:629-635.

Vodovotz Y. Translational systems biology of inflammation and healing. Wound Repair and Regeneration. 2010. 18:3-7.

An G. Translational systems biology using an agent-based approach for dynamic knowledge representation: An evolutionary paradigm for biomedical research. Wound Repair and Regeneration. 2010. 18:8-12

Martins-Green M, Vodovotz Y, Liu P. Systems biology applied to wound healing. Wound. Repair and Regeneration. 2010. 18:1-2.

Solovyev A, Mikheev M, Zhou L, Dutta-Moscato J, Ziraldo C, An G, Vodovotz Y, Mi Q. SPARK: A Framework for Multi-scale Agent-based Biomedical Modeling. Proceedings of Agent-Directed Simulation Symposium ADS’10. Apr 2010. (In Press)

Vodovotz, Y.; Constantine, G.; Faeder, J.; Mi, Q.; Rubin, J.; Bartels, J.; Sarkar, J.; Squires, R.; Okonkwo, D.O.; Gerlach, J.; Zamora, R.; Luckhart, S.; Ermentrout, B.; An, G. Translational systems approaches to the biology of inflammation and healing. Immunopharmacol and Immunotoxicol. 2010. 32:181-195.

Mi Q, Constantine G, Ziraldo C, Solovyev A, Torres A, Namas R, Bentley T, Billiar T, Zamora R, Puyana J.C, Vodovotz Y. A Dynamic View of Trauma/Hemorrhage-Induced Inflammation in Mice: Principal Drivers and Networks, PLoS One. 2011 (In Press).*

Mi, Q.; Constantine, G.; Ziraldo, C.; Solovyev, A.; Torres, A.; Namas, R.; Bentley, T.; Billiar, T.; Zamora, R.; Puyana, JC.; Vodovotz, Y. (2011). A Dynamic View of Trauma/Hemorrhage-Induced Inflammation in Mice: Principal Drivers and Networks. PLoS ONE, 6(5), 19424.  
Vodovotz, Y. (2012). At the Interface between Acute and Chronic Inflammation: Insights from Computational Modeling. In Roy, S.; Bagchi, D.; Raychaudhuri, S. (Ed.), Chronic Inflammation - Molecular Pathophysiology, Nutritional and Therapeutic Interventions (pp. Section 1). Florence, KY: Taylor & Francis.  

An, G. ; Nieman, G.; Vodovotz, Y. (2012). Toward Computational Identification of Multiscale ‘‘Tipping Points’’ in Acute Inflammation and Multiple Organ Failure. Annals of Biomedical Engineering, DOI, 1-11.  

An, G.; Nieman, G.; Vodovotz, Y. (2012). Computational and systems biology in trauma and sepsis: Current State and Future Perspectives. Int J Burn Trauma, 2(1), 1-10.  

Huang, W; Vodovotz, Y; Kusturiss M; Barcley, D; Greenwald, K; Boninger, M; Coen, P; Brienza, D; Sowa, G (2012). Correlation between Circulating Cytokine Profiles and Monocyte Phenotypes in Acute Response to Spinal Cord Injury (SCI). Proceedings of the Association of Academic Physiatrists. Las Vegas, NV: AAP.

 

Presentations

Rajaie Namas, Alexey Solovyev, Maxim Mikheev, Derek Barclay, Ruben Zamora, Qi Mi and Yoram Vodovotz: Post-traumatic Spinal Cord Injury: Inflammatory Response and Modeling (abstract), McGowan Institute for Regenerative Medicine retreat. Mar 2009.

Alexey Solovyev, Leming Zhou, Maxim Mikheev, Joyetta Dutta-Moscato, Gary An, Yoram Vodovotz and Qi Mi: SPARK: A Systems Biology Framework for Agent-based Biomedical Modeling (abstract), Cold Spring Harbor Computational Cell Biology meeting. Mar 2009.

Ziraldo C, Solovyev A, Henzel M, Sowa G.A, Brienza D, An G, Mi Q, Vodovotz Y. Inferring mechanism from morphology: Understanding pressure ulcer formation in spinal cord injury patients using agent-based mechanistic simulations. 20th Wound Healing Society Annual Meeting (Abstract accepted) Apr 2010.

Return Home

This work is funded by the National Institute on Disability and Rehabilitation Research (NIDRR),
Rehabilitation Engineering Research Center (RERC) on Spinal Cord Injury, Grant #H133E070024
The ideas and opinions expressed herein are those of the authors and not necessarily reflective of the NIDRR.

Contact Webmaster | Accessibility Statement


Last Updated: 07.10.2012 | 16:40

Smart Code