View
219
Download
1
Tags:
Embed Size (px)
Citation preview
Agent Based Models of the Acute Agent Based Models of the Acute Inflammatory Response:Inflammatory Response:
Update on Development and Update on Development and Future DirectionsFuture Directions
Swarmfest 2004, Ann Arbor, MISwarmfest 2004, Ann Arbor, MI
May 11, 2004May 11, 2004
Gary An, MDGary An, MD
Department of TraumaDepartment of Trauma
Cook County HospitalCook County Hospital
Acute Inflammatory Acute Inflammatory Response (AIR)Response (AIR)
Initial defense and repair mechanismInitial defense and repair mechanism Specialized cellular/molecular pathwaysSpecialized cellular/molecular pathways Diffusely distributed/Tissue NonspecificDiffusely distributed/Tissue Nonspecific Activation is non-specific to insultActivation is non-specific to insult Precedes Adaptive Immune response Precedes Adaptive Immune response
(self/non-self distinction=>Antibodies)(self/non-self distinction=>Antibodies)
Systemic Inflammatory Systemic Inflammatory Response Syndrome/Multiple Response Syndrome/Multiple
Organ Failure (SIRS/MOF)Organ Failure (SIRS/MOF)
Disease of the ICU => “Unexplored State”Disease of the ICU => “Unexplored State” Pathologic state of Acute Immune Pathologic state of Acute Immune
Response (AIR)Response (AIR) Physiologic manifestations result from Physiologic manifestations result from
endogenous mediatorsendogenous mediators Hyperinflammation vs. Immune-Hyperinflammation vs. Immune-
suppression =>Temporal and Spatialsuppression =>Temporal and Spatial
Challenge of SIRS/MOF
Gap between Pathophysiology and Diagnosis
Gap between Mechanisms and Treatment
Gap between Basic Science and Clinical Implementation
Nonlinear Behavior => Complexity
AIR as a Complex SystemAIR as a Complex System
ComponentsComponents Rules Rules
LocalityLocality Emergent Emergent
PropertiesProperties Unexpected Unexpected
BehaviorBehavior
CellsCells Cellular Cellular
ProgrammingProgramming Membranes/Membranes/
ReceptorsReceptors Organ PhysiologyOrgan Physiology SIRS/MOFSIRS/MOF
Applications of ABM to Applications of ABM to AIR/SIRS/MOF AIR/SIRS/MOF
Base Global ModelBase Global Model– PathophysiologyPathophysiology
– Therapeutic InterventionsTherapeutic Interventions Specific Disease Specific Disease
Processes/Pathogens/MechanismProcesses/Pathogens/Mechanism– Cutaneous and Inhalational AnthraxCutaneous and Inhalational Anthrax
Basic Science Experiment SimulationBasic Science Experiment Simulation– Epithelial Permeability ModelEpithelial Permeability Model
ABM of Global Systemic ABM of Global Systemic InflammationInflammation
Endothelial/Blood interfaceEndothelial/Blood interface Activation/Propagation of InflammationActivation/Propagation of Inflammation Endothelial Cells and White Blood CellsEndothelial Cells and White Blood Cells Dynamics of PathophysiologyDynamics of Pathophysiology Proto-Testing Platform for Systemic Proto-Testing Platform for Systemic
TherapiesTherapies Very Abstract!Very Abstract!
Current Model of Global Current Model of Global InflammationInflammation
Cell types Endothelial cells,neutrophils, monocytes,TH0, TH1, TH2, bacteria,white blood cell generativecells
Cell Receptors andFunctions
L-selectin, E/P-selectin,CD-11/18, ICAM, TNFr, IL-1r, adhesion, migration,respiratory burst,phagocytosis, apoptosis
Mediators Endotoxin, PAF, TNF, IL-1,IL-4, IL-8, IL-10, IL-12, IFN-g, sTNFr, IL-1ra, GCSF
Validation StrategiesValidation Strategies
Agent Rules=>Transparency wrt codeAgent Rules=>Transparency wrt code Behavior of Individual wrt global Behavior of Individual wrt global
response to injury=>response to injury=>Individual DynamicsIndividual Dynamics Behavior of Population wrt cytokine Behavior of Population wrt cytokine
patterns=>patterns=>Population DynamicsPopulation Dynamics Behavior of Population wrt outcome to Behavior of Population wrt outcome to
intervention=>intervention=>Population ResponsePopulation Response
Individual Response Individual Response DynamicsDynamics
Four possible dynamics:Four possible dynamics:– Successful healingSuccessful healing
– ““Phase II” or Immune-suppressed Phase II” or Immune-suppressed SIRS/MOFSIRS/MOF
– ““Phase I” or Hyper-inflammatory SIRS/MOFPhase I” or Hyper-inflammatory SIRS/MOF
– Overwhelming insult/infectionOverwhelming insult/infection Function of degree of Initial InsultFunction of degree of Initial Insult
Healing IIN=800
Time
Oxy Deficit Total Infection
Immune-Suppressed MOF IIN=1100
Time
Oxy Deficit Total Infection
Hyper-Inflammatory SIRS IIN=1500
Time
Oxy Deficit Total Infection
Overwhelming Infection IIN=1900
Time
Oxy Deficit Total Infection
Population Dynamics:Population Dynamics:Cytokine ProfilesCytokine Profiles
Patterns of cytokine levels for a Patterns of cytokine levels for a population at a specific IINpopulation at a specific IIN
7 days simulated time7 days simulated time IIN generates 50% mortalityIIN generates 50% mortality N=100N=100 Pattern Oriented/Qualitative (Very Pattern Oriented/Qualitative (Very
Large Range-not shown)Large Range-not shown)
Total Cytokines: TNF and IL-10
0
20
40
60
80
100
120
0 50 100 150 200
Hours
Total TNF Total IL-10
Population Response:Population Response:Simulating Anti-inflammatory Simulating Anti-inflammatory
InterventionsInterventions
Any mediator represented as a variable Any mediator represented as a variable can be manipulatedcan be manipulated
Modified based on published effectsModified based on published effects No other modifications of the ABM other No other modifications of the ABM other
than simulated interventionthan simulated intervention Results all generated prospectivelyResults all generated prospectively
List of In-Silico ExperimentsList of In-Silico Experiments
Phase III ClinicalTrials
3 day anti-TNF (Reinhart)3 day rhIL-1ra (Opal)7 day GCSF (Root)
Smaller Clinical Trials 1 dose anti-CD18 (Rhee)
Animal Studies 3 day combination anti-TNF and IL-1ra (Remick)
Hypothetical Multi-modal Regimes
anti-CD-18/anti-TNF/IL-1raGCSF/anti-TNF/IL-1ra
ABM of Anthrax InfectionABM of Anthrax Infection
Modification of Base Global ModelModification of Base Global Model Specific Characteristics of B. anthracisSpecific Characteristics of B. anthracis
– Both Cutaneous and Inhalational FormsBoth Cutaneous and Inhalational Forms
– Reproduce effects of Toxin-Component Reproduce effects of Toxin-Component (Lethal Factor, Edema Factor and (Lethal Factor, Edema Factor and Protective Antigen) knockout species of B. Protective Antigen) knockout species of B. anthracisanthracis
0
5
10
15
20
25
30
35
0 200 400 600 800
Initial Infection Number
Inhalation Inhalation Abs Cutaneous Cutaneous Abs
Inhalation EFKO Inhalation LFKO Inhalation PAKO
Basic Science ABMsBasic Science ABMs
Basic Science Paradigm = Linear Basic Science Paradigm = Linear analysisanalysis
Examine Component Sub-SystemsExamine Component Sub-Systems Improve efficiency of Basic Science Improve efficiency of Basic Science
experimentsexperiments Guide further investigationGuide further investigation Modular Components of System-wide Modular Components of System-wide
ModelModel
ABM of Epithelial Cell ABM of Epithelial Cell Permeability: StructurePermeability: Structure
Based on model of DeludeBased on model of Delude Epithelial cell culture => Grid of Epi Cell Epithelial cell culture => Grid of Epi Cell
AgentsAgents Agent rules => Tight Junction (TJ) Agent rules => Tight Junction (TJ)
FormationFormation TJ status determines permeabilityTJ status determines permeability
ABM of Epithelial Cell ABM of Epithelial Cell Permeability: ResultsPermeability: Results
Increased Permeability to NO/Pro-Increased Permeability to NO/Pro-inflammatory cytokine mixinflammatory cytokine mix
Blocked with NO scavenger/iNOS Blocked with NO scavenger/iNOS inhibitorinhibitor
Matches Basic Science resultsMatches Basic Science results Potential Modular ModelPotential Modular Model
Uses of ABM of the AIRUses of ABM of the AIR
Formalize Mental ModelsFormalize Mental Models– Functional Repository of Basic Science Functional Repository of Basic Science
InformationInformation
– ModularModular
– Community-dependentCommunity-dependent Drug EngineeringDrug Engineering
– Identify targets for manipulationIdentify targets for manipulation
– Use to pre-test a planned treatment regimes => Use to pre-test a planned treatment regimes => Multi-Modal regimesMulti-Modal regimes
Uses of ABM of the AIR cont.Uses of ABM of the AIR cont.
Clinical Therapeutics DesignClinical Therapeutics Design– Patient Population Sub-stratificationPatient Population Sub-stratification
– Generate Cytokine Profiles => “Finer Grained” Generate Cytokine Profiles => “Finer Grained” Theoretical ToolTheoretical Tool
– Mathematical characterization of system to Mathematical characterization of system to guide future therapiesguide future therapies
– ““Cross Platform” ValidationCross Platform” Validation
Future DevelopmentFuture Development
Multi-Tissue ModelMulti-Tissue Model– Directional FlowDirectional Flow– CoagulationCoagulation– Multiple Organ Failure/SupportMultiple Organ Failure/Support
Modular ModelModular Model– Basic Science ModelsBasic Science Models– Community/Web-basedCommunity/Web-based– ““Functional Data-bank”Functional Data-bank”
Complex SystemsComplex Systems
Rules drive Local interactions between individual components
Feedback loops =>non-linearity Interaction dynamics result in meta-
stable structures=> Emergence Hierarchies of Emergent properties Non-intuitive, paradoxical behavior
Agent Based Modeling (ABM)Agent Based Modeling (ABM)
System of Components=>System of Components=>AgentsAgents Agent Rule systems=>Basic ScienceAgent Rule systems=>Basic Science PopulationsPopulations of agents in virtual world of agents in virtual world Runs = agent actions/interactions=> Runs = agent actions/interactions=>
LocalityLocality Multiple runs=Random Number Multiple runs=Random Number
Generators=> basic science experimentsGenerators=> basic science experiments Stochastic and DeterministicStochastic and Deterministic
Why use ABM to model Why use ABM to model AIR/SIRS/MOF?AIR/SIRS/MOF?
Lots of information about potential Lots of information about potential agents (cells and molecules)agents (cells and molecules)
Process is driven by Process is driven by locallocal interactions interactions Dynamics Dynamics maymay be too complex for top- be too complex for top-
down modelingdown modeling Multiple possible levels of model Multiple possible levels of model
validationvalidation Integration of Models => Total SystemIntegration of Models => Total System
Doing Science with ABMDoing Science with ABM
In-Silico ExperimentsIn-Silico Experiments => Virtual => Virtual control and experimental populationscontrol and experimental populations– Apply standard statistical toolsApply standard statistical tools– UseUse Pattern Oriented Analysis Pattern Oriented Analysis
Formalize mental model Formalize mental model building/testing hypothesesbuilding/testing hypotheses
Develop TheoriesDevelop Theories
Population RunsPopulation Runs
Random number generators are Random number generators are active=> Heterogeneityactive=> Heterogeneity
Multiple runs at a specific IIN generates Multiple runs at a specific IIN generates a study “population”a study “population”
Generates a “mortality rate” for a Generates a “mortality rate” for a particular IIN (Mortality at >80% Total particular IIN (Mortality at >80% Total Damage)=>“Control Population”Damage)=>“Control Population”
Results of In-Silico Experiments Results of In-Silico Experiments in Sterile Mode (n=100)in Sterile Mode (n=100)
Model Run Mortality Chi SquareBase 86%
Antibiotics only 37% Significant,p=.01 vs. No Abs
Abs/anti-TNF 39% NS
Abs/rh-IL-1ra 41% NS
Abs/anti-CD18 42% NS
Abs/rhIL-1ra/anti-TNF
38% NS
Abs/anti-CD18/rhIL-1ra/anti-TNF
45% NS
Results of In-Silico Experiments Results of In-Silico Experiments in Infectious Mode (n=100)in Infectious Mode (n=100)
Model Run Mortality Chi-squareBase 100%Antibiotics only 40% Significant, p=.01
vs No AbsAbs/anti-TNF 42% NSAbs/rhIL-1ra 39% NSAbs/GCSF 36% NSAbs/rhIL-1ra/anti-TNF
37% NS
Abs/GCSF/rhIL-1ra/anti-TNF
38% NS
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
0 200 400 600 800
Intial Infection Number
Inhalation Inhalation Abs Cutaneous Cutaneous Abs
Inhalation EFKO Inhalation LFKO Inhalation PAkO
What ABM is not!What ABM is not!
NOT a replacement of current NOT a replacement of current techniques of scientific investigation. techniques of scientific investigation. => “Software vs. Hardware”=> “Software vs. Hardware”
NOT a clinical tool to provide a NOT a clinical tool to provide a prognosis or determine a treatment prognosis or determine a treatment course for an individual patient*course for an individual patient*
Translation, Synthesis and Translation, Synthesis and ABMsABMs
Requires data from Basic ScienceRequires data from Basic Science– ““What we look for and find out.”What we look for and find out.”
Places it into Synthetic frameworkPlaces it into Synthetic framework– ““How do the pieces fit together.”How do the pieces fit together.”
Uses Multiple HierarchiesUses Multiple Hierarchies– ““Little pieces make big pieces.”Little pieces make big pieces.”
We Do This Already!We Do This Already!– Mental Models => Software EngineeringMental Models => Software Engineering
““Theories of SIRS/MOF”Theories of SIRS/MOF”
““Dynamic Equilibrium”=>Response is Dynamic Equilibrium”=>Response is appropriate, degree is notappropriate, degree is not
Concept of “anatomic containment” and Concept of “anatomic containment” and “physiologic containment” “physiologic containment”
Identify “Amplifiers” of responseIdentify “Amplifiers” of response Importance of all aspects of the response => “If Importance of all aspects of the response => “If
a mediator does a lot of different stuff don’t a mediator does a lot of different stuff don’t mess with it.”mess with it.”
Supplementation, not Blockade (WBCs smarter Supplementation, not Blockade (WBCs smarter than ICU MDs)than ICU MDs)
Summary of Key PointsSummary of Key Points
The Acute Inflammatory Response is a The Acute Inflammatory Response is a complex system that cannot be fully complex system that cannot be fully characterized using existing characterized using existing techniques.techniques.
Agent Based Modeling is well suited to Agent Based Modeling is well suited to modeling the Inflammatory Response.modeling the Inflammatory Response.
ABM would be an useful adjunct to ABM would be an useful adjunct to existing techniques of investigation. existing techniques of investigation.