AModelofIronMetabolismintheHumanBodyTimothyBarry*1,MaryGockenbach*2,JigneshParmar3,PedroMendes3

1UniversityofMaryland,CollegePark,2UniversityofTexasatArlington,3CenterforQuantitativeMedicine,UConnHealth*Theseauthorscontributedequallytothiswork

Introduction Methods AdditionofHepcidin Data

Model

ParameterIdentifiability• Iron-relateddiseasesareprevalentthroughouttheworld.• Anemiaaffectsonequarteroftheworld’spopulation.• Hemochromatosis,adiseaseofironoverload,isthemost

commoninheriteddiseaseofgenemutationinCaucasians.• Understandingthemechanismsofironmetabolismwill

advanceprogresstowardsindividualizedtreatmentstrategies.

• Amathematicalmodelusingordinarydifferentialequations(ODE)isdevelopedtosimulateirondistributioninthemajororgansofthebody.

• Thismodelisbasedonpreviousworkmodelingironmetabolisminmice[6].

• Someparametervalues,includingthevolumeandironconcentrationoftheorgans,wereobtainedfromliterature.

• Theremainderwereestimatedusingexperimentaldatafromsubjectsinjectedwithradioactiveiron[1-3,5,7].

• Parametersetsweregeneratedbyminimizinganobjectivefunction(sumofsquaredresiduals)usingavarietyofoptimizationalgorithms.

• Thechosenparametersetwasrequired(1)toproduceasmallobjectivevalue,and(2)torenderthefollowingquantitiesclosetobiologicallyrealisticranges:

• ParameterestimateswerecarriedoutusingthesoftwareCOPASI,abiochemicalsystemsimulator[4].

ThisworkwascarriedoutatModelingandSimulationinSystemsBiology, anREUattheCenterforQuantitativeMedicine(CQM)atUConnHealth,andfundedbytheNationalScienceFoundation(DMS1460967).WethankDr.Reinhard Laubenbacher,directorofCQM.

[1]Berzuini,C.etal. (1978).ComputersandBiomedicalResearch 11,209-227.[2]Bonnet,John,etal. (1960).Blood 15.1,36-44[3]Ganz,T.(2011).Blood 117.17,4425-4433[4]Hoops,S.etal. (2006).Bioinformatics 22,3067-74[5]Huff,R.,etal.(1951).JournalofClinicalInvestigation 30.12Pt2,1512.[6]Parmar, J.etal (2017).BMCSystemsBiology 11.1,57.[7]Pollycove,M.etal.(1961).JournalofClinicalInvestigation 40.5,753.

• Themodeliscapableoffittingdatafrommultipleindividuals,andaccuratelysimulatesanemiaandtheeffectofmeals.

• Asrelevantdataisadded,parameteridentifiability willimprove.

• Withtheavailabilityofindividualdata,thismodelcouldbepersonalizedforuseinprecisionmedicine.

• Forexample,anindividual’sweight,height,andagecouldbeusedtoestimateorganvolumesandironconcentrations.

• Identifiabilityanalysisisatoolusedtoascertaintheextenttowhichestimatedparametersaredeterminedbydata.

• Toreplicatemealpatterns,ironabsorptionwaschangedfromacontinuousratetoasequenceofdiscreteevents.

• Theresultsofmealsimulationsaligncloselywithtimecoursedataofserumiron[3],andallowestimationofparametersrelatedtohepcidin.

• Oncethemodelwasaligned,thehepcidindatawasincludedinthemodelcalibration.

TotalIron PercentIroninRBC TransferrinSaturation3.5 g 84.1% 35.3%

• Irondeficiencyanemiaistheresultofinsufficientironinthediet.

• Wheninfectedwithapathogen,thebodyincreaseshepcidinproductionsoastodecreaseironconcentrationintheblood.Anemiaofchronicdiseaseresultswhenthisresponseissustainedoveranextendedperiodoftime.

• Simulationsofirondeficiencyanemiaandanemiaofchronicdiseasebroadlyreflectthesepathologies.

• Thevaluesofparametersforwhichdatawasavailable(e.g.,KEPO)aremorecertainthanthoseforwhichdatawasnotavailable(e.g.,kInBM).

Discussion

References

Acknowledgements

SimulationofIronDisorders

ModelCalibration

TotalIron PercentIroninRBC TransferrinSaturation2 – 5g 60 – 75% 24 – 40%

• Themodelconsistsof7compartmentsand23equations,whichdescribetheflowofironbetweencompartments,theirreversiblebindingofirontotransferrinintheplasma,andthesynthesisanddegradationofkeyregulatoryproteinshepcidinanderythropoetin(EPO).

• Ironentersthebodythroughdietaryintakeandleavesthroughintestinalcellshedding(representedaslossfromtheduodenum)andthroughhair,sweat,anddeadskin(representedaslossfromtherestofbody).

• AsanexampleoftheformofanODEinthemodel,theequationthatgovernsironfluxinthespleenis