This is a repository copy of Harvesting the promise of AOPs: An assessment and recommendations. White Rose Research Online URL for this paper: http://eprints.whiterose.ac.uk/128403/ Version: Published Version Article: Carusi, A., Davies, M.R., De Grandis, G. et al. (6 more authors) (2018) Harvesting the promise of AOPs: An assessment and recommendations. Science of the Total Environment, 628-29. pp. 1542-1556. ISSN 0048-9697 https://doi.org/10.1016/j.scitotenv.2018.02.015 Reuse This article is distributed under the terms of the Creative Commons Attribution (CC BY) licence. This licence allows you to distribute, remix, tweak, and build upon the work, even commercially, as long as you credit the authors for the original work. More information and the full terms of the licence here: https://creativecommons.org/licenses/ Takedown If you consider content in White Rose Research Online to be in breach of UK law, please notify us by emailing [email protected] including the URL of the record and the reason for the withdrawal request. [email protected] https://eprints.whiterose.ac.uk/ ScienceoftheTotalEnvironment628–629(2018)1542–1556 ContentslistsavailableatScienceDirect Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv Review Harvesting the promise of AOPs: An assessment and recommendations AnnamariaCarusia,⁎,MarkR.Daviesb,GiovanniDeGrandisc,BeateI.Escherd,e,GeoffHodgesf, KennethM.Y.Leungg,MauriceWhelanh,CatherineWilletti,GeraldT.Ankleyj aMedicalHumanitiesSheffield,UniversityofSheffield,MedicalSchool,BeechHillRoad,SheffieldS102RX,UK bQTInformaticsLimited,MacclesfieldSK105DS,UK cScience,Technology,EngineeringandPublicPolicy(STEaPP),BostonHouse,36-37FitzroySquare,LondonW1T6EY,UK dUFZ–HelmholtzCentreforEnvironmentalResearch,04318Leipzig,Germany eEberhardKarlsUniversityTübingen,EnvironmentalToxicology,CentreforAppliedGeosciences,72074Tübingen,Germany fSafetyandEnvironmentalAssuranceCentre,Unilever,ColworthSciencePark,Sharnbrook,BedfordshireMK441LQ,UK gTheSwireInstituteofMarineScienceandSchoolofBiologicalSciences,TheUniversityofHongKong,Pokfulam,HongKong,China hEuropeanCommission,JointResearchCentre(JRC),Ispra,Italy iTheHumaneSocietyoftheUnitedStates,700ProfessionalDrive,Gaithersburg,MD,20879,USA jUSEnvironmentalProtectionAgency,6201CongdonBlvd,Duluth,MN55804,USA H I G H L I G H T S G R A P H I C A L A B S T R A C T • TheAOPframeworkaimstoincreaseef- ficiencyofchemicalsafetyassessments. • ThestakeholdercommunityforAOPs, however,isbroaderthanchemicalrisk assessors. • Therearescientificandsocialchallenges tosuccessfullyengageallstakeholders. • Multi-facetedcommunicationandgov- ernance strategies will address these challenges. a r t i c l e i n f o a b s t r a c t Articlehistory: TheAdverseOutcomePathway(AOP)conceptisaknowledgeassemblyandcommunicationtooltofacilitatethe Received15December2017 transparenttranslationofmechanisticinformationintooutcomesmeaningfultotheregulatoryassessmentof Receivedinrevisedform2February2018 chemicals.TheAOPframeworkandassociatedknowledgebases(KBs)havereceivedsignificantattentionand Accepted2February2018 useintheregulatorytoxicologycommunity.However,itisincreasinglyapparentthatthepotentialstakeholder Availableonlinexxxx communityfortheAOPconceptandAOPKBsisbroaderthanscientistsandregulatorsdirectlyinvolvedinchem- icalsafetyassessment.Inthispaperweidentifyanddescribethosestakeholderswhocurrently—orinthefuture— Editor:D.Barcelo couldbenefitfromtheapplicationoftheAOPframeworkandknowledgetospecificproblems.Wealsosumma- Keywords: rizethechallengesfacedinimplementingpathway-basedapproachessuchastheAOPframeworkinbiological Toxicology sciences,andprovideaseriesofrecommendationstomeetcriticalneedstoensurefurtherprogressionofthe Adverseoutcomespathways frameworkasauseful,sustainableanddependabletoolsupportingassessmentsofbothhumanhealthandthe Riskassessment environment.AlthoughtheAOPconcepthasthepotentialtosignificantlyimpacttheorganizationandinterpre- Environmentalhealth tationofbiologicalinformationinavarietyofdisciplines/applications,thispromisecanonlybefullyrealized ⁎ Correspondingauthor. E-mailaddresses:a.carusi@sheffield.ac.uk(A.Carusi),[email protected](M.R.Davies),[email protected](G.DeGrandis),[email protected](B.I.Escher), [email protected](G.Hodges),[email protected](K.M.Y.Leung),[email protected](M.Whelan),[email protected](C.Willett),[email protected] (G.T.Ankley). https://doi.org/10.1016/j.scitotenv.2018.02.015 0048-9697/©2018TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/). A.Carusietal./ScienceoftheTotalEnvironment628–629(2018)1542–1556 1543 Chemicalmanagement throughtheactiveengagementof,andinputfrommultiplestakeholders,requiringmulti-prongedsubstantive Social,ethicalandlegalaspects long-termplanningandstrategies. ©2018TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense(http:// creativecommons.org/licenses/by/4.0/). Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1543 2. FeaturesoftheAOPframeworkanditsAOPknowledgebase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1544 3. StakeholdersintheevolutionoftheAOPframework. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1546 3.1. Regulatoryassessmentofchemicalrisks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1546 3.2. Chemicalsafetyassessmentforindustry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1547 3.3. Productdiscoveryanddevelopment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1547 3.4. Medicineandhealth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1547 3.5. Environmentalquality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1548 3.6. Academicapplications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1548 3.7. Non-governmentalorganizations(NGOs). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1549 4. ChallengestoscalinguptheAOPframeworkanditsapplications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1549 4.1. FormalpublicationversusprovidinginputtotheAOP-KB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1549 4.2. PossiblerisksandburdenstostakeholderssupportingtheAOPframework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1549 4.3. Qualityassurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1549 4.4. MaintainingeffectivedialogueabouttheAOPframeworkamongstakeholdergroups. . . . . . . . . . . . . . . . . . . . . . . . . . 1550 4.5. Ethical,legalandsocialaspectsassociatedwithuseoftheAOPframework. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1550 4.6. Governanceandsustainability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1550 5. Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1551 5.1. Publishing,depositingandreviewingstrategies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1551 5.2. Educationandtraining:stakeholder-specificapproachesandmaterials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1551 5.3. Stakeholder-specificinteraction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1551 5.4. Translationintoapplication. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1552 5.5. Governanceandsustainability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1553 5.6. Impactindicators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1553 6. Conclusion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1553 Acknowledgments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1554 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1554 1.Introduction (OECD).TheOECDinitiatedaninternationalAOPdevelopmentprogram in2012thathassponsoredandcontributedtoanumberofcriticalactiv- Legislativedictatesandsocietalexpectationsrequireadifferentap- ities,includingthepublicationofstandardizedapproachesforAOPde- proachtochemicalsafetyassessmentthanhasbeenusedinthepast velopmentandreview,andtheformationofanAOP-knowledgebase whenextensivedatawerecollectedforonlyahandfulofchemicalsof (AOP-KB)tosupportAOPdevelopmentanddissemination(Villeneuve concerneithertohumansortheenvironment. Currentlythereis a et al., 2014a, 2014b; Vinken et al., 2017). A critical motivation for needtodeterminethepotentialbiologicaleffectsoftensofthousands OECDinvolvementwastherecognitionbymembercountriesthata ofexistingandnewsubstancesinacost-effectiveandtimelymanner, pathway-basedapproachtochemicalsafetycanpromoteeconomic butresourcesforchemicaltestingcurrentlyareeitherstaticordecreas- growthaswellasthehighestlevelsofhealthandenvironmentalprotec- ing.Hence,thereisincreasingrelianceonapproachestoefficientlygen- tion(Perkinsetal.,2015). erate data concerning the possible biological activity of chemicals AstheAOPframeworkhasevolvedinthecontextofregulatorytox- throughuseofinsilicomodels,invitroassays,andshort-terminvivo icology,ithasbecomerecognizedthatthepotentialstakeholdercom- tests emphasizing molecular and biochemical measures of effects munityisbroaderthanthoseinvolvedinchemicalsafetyassessment, (NationalResearchCouncil,2007).Thiscreatesachallengeforeffec- asillustratedinFig.1.Forexample,theframeworkmaycomplementac- tivelytranslatingresultsfromthesetypesofmechanistic/pathway- tivitiestoapplythetractabilityofmechanism-basedapproachesforad- basedanalysesintoinformationusefultothoseultimatelyresponsible dressingbiomedicalchallenges(Benson,2015;vanHasseltandvander forconductingchemicalsafetyassessments. Graaf,2015;Visseretal.,2014),insafetyevaluationsassociatedwith TheAdverseOutcomePathway(AOP)conceptualframeworkwas drugdevelopment(Miramsetal.,2014),systemstoxicology(Stahlet developedtoserveasaknowledgeassemblyandcommunicationtool al.,2015)andclinicaltrialsimulations(Hanetal.,2016). tofacilitatethetransparenttranslationofmechanisticdataintoout- Ouraimwiththispaperistopresentastrategicvisiontohelpensure comesmeaningfultochemicalsafetyassessment(Ankleyetal.,2010). thattheAOPframeworkcanrealizeitsfullpotentialformultiplestake- Specifically,theAOPframeworkfacilitatestheidentificationandevalu- holdergroups.Thepaperfirstdetailstheuniquefeaturesandqualities ationofcausallinkagesacrossbiologicallevelsoforganization,suchthat oftheAOPapproachandoftheAOPframework,andgoesontoprofile mechanisticresponsescanbereliablyusedindecision-makingtopro- avarietyofstakeholders,describinghowtheystandtobenefitfrom tectfromadverseeffects,suchascancerinhumansorsuppressionofre- thefurtherdevelopmentoftheAOPframeworkanditsapplicationto productionofecologicallyimportantspecies.Manyregulatoryagencies solvingtheirproblems.Ouranalysisincludesconsiderationofthecore acrosstheworldhaverecognizedthepotentialofAOPsinsupporting obstaclestoafull,long-termengagementofthestakeholdercommunity moreefficientassessmentsofchemicalsafety.Onenotableexamplein- withtheframework.Finally,weconcludewithasetofrecommenda- volvestheOrganizationforEconomicCooperationandDevelopment tions regarding the steps that need to be taken to address these 1544 A.Carusietal./ScienceoftheTotalEnvironment628–629(2018)1542–1556 Fig.1.StakeholdersinvolvedinthedifferentareasofresearchanddevelopmentinrelationtoAOPdevelopmentandusage. challenges,andtoensuredevelopmentoftheAOPframeworkasasus- ThegoaloftheAOPframeworkistocompileandsynthesizethis tainabletoolfortheapplicationofourscientificknowledgeoflivingsys- wealthofbiologicalinformationsuchthatitcanbetransparentlyandef- temsinnumerousdecision-makingcontexts. ficientlyemployedfordecision-making.Fig.2providesanoverviewof ThispaperstemsfromaPellstonworkshopheldApril2–6,2017in theAOPframeworkinthecontextofitspotentialapplicationtothe Cornwall,ON,Canada.TheworkshopwascoordinatedbytheSociety translationanduseofdifferenttypesofdatatosupportassessmentof ofEnvironmentalToxicologyandChemistry(SETAC),andfinancially theeffectsofchemicalsonhumanhealthandtheenvironment.Theini- supported by multiple governmental and business organizations. tialinteractionofachemicalwithabiologicalsystemisdepictedasthe Pellstonworkshopstypicallyassemble30–40(invited)expertsona molecularinitiatingevent(MIE),suchasbindingtoaprotein(e.g.,re- giventopicintheenvironmentalsciences,whoarechargedwithdefin- ceptors, enzymes) or DNA, or interactions with membrane lipids. ingandproposingpathsforwardtoaddresschallenge(s)inthetopic TheseMIEscancausesubsequentperturbationsathigherbiological areaofinterest.ThisPellstonworkshopfocusedonanumberofissues levelsoforganization,depictedasintermediatekeyevents(KEs)along associatedwithAOPscienceandimplementation,includingthedevel- anAOP,whichultimatelymayresultinadverseapicalresponsessuch opmentofa“roadmap”topromotesustainableuseoftheconcept.Spe- effectsonsurvival,reproduction,carcinogenesis,etc.Inthecaseofeco- cificneedsaddressedattheworkshopwereidentifiedthroughaglobal logicalassessments,impactsatthepopulationlevelalsooftenarean horizonscanningeffortthatgatheredinputfromscientistsandriskas- endpointofregulatoryconcern.ImplicitinanAOPisthatthedepicted sessors/managersfromthroughouttheworld(LaLoneetal.,2017a).In KEsarecausally-associatedwithoneanotherviadefinedKErelation- all,41expertsrepresentinggovernment,academiaandbusiness,from ships,anattributethatcanbeassessedusingweight-of-evidenceanaly- ninedifferentcountriesinNorthAmerica,EuropeandAsiaparticipated ses(Beckeretal.,2015).Itisthisdocumented,formalizedlinkageacross intheworkshop. biologicallevelsoforganizationinanAOPthatprovidesthebasisforthe useof“alternative”datastreamstopredictthetypesofapicalresponses 2.FeaturesoftheAOPframeworkanditsAOPknowledgebase deemedcriticaltoregulationsandriskassessments. TheAOPframeworkpotentiallygoesbeyondbiologicalpathways, Overthepastseveraldecadestheglobalscientificacademic,phar- butwillimproveourunderstandingofchemicaltoxicityinrelationto maceutical,chemicalandpersonalcareproductcommunities,aswell theecologicalimpactsofchemicalcontaminantsinair,water,soiland asgovernmentagencies,haveinvestedbillionsof(US)dollarsgenerat- foodaswellasimpactstohumanhealth.Escheretal.(2017)giveade- ingbiologicalinformation.Thegoalhasbeenbothtoextendbasicscien- tailed account of how the mechanistic understanding of the AOP tific knowledge of how biological processes causally unfold and to frameworkcanbeusefulbeyondhumanhealthforbroaderecological informthedevelopmentoflessharmfulchemicalsandmoreeffective issues.TheAOPframeworkhaswideimplicationsforadvancingtheas- diseasediagnosis,preventionandtreatment.Traditionally,thisinfor- sessmentofchemicalhazardsacrossvariousenvironmentalcompart- mationhasbeendepositedinscientificpublications,butthesheernum- mentscoveringatmosphere,hydrosphere,biosphere,lithosphere,and berofpapers,compoundedbythefactthattheinformationisoften anthroposphere. conveyedinunstructuredandinconsistentways,makesitdifficultto TheAOP-KBalsohasspecificallybeendesignedtoaddresstheprob- access,shareandintegrate.Initiativestosummarizeandcompilethis lemofunstructuredandscatteredinformation.TheAOP-KBcurrently informationintocentralizeddatarepositorieshavebeenonlypartially consistsofseveralsoftwarepackages,allofwhichdealwithcollecting, successful,astheyi)donotincludeallpublications,ii)remaincompart- formattingand,tosomeextentevaluating,biologicalinformation.At mentalized,andiii)usedifferentontologies.Thesedisconnecteddata- thetimeofwriting,theAOP-Wikiisthemostprominentandwell-de- baseshaveresultedinthenextlevelof“silo-ed”information.Therise velopedofthepackagesandthereforeitistheillustrativefocusinthis inbioinformatics,includingcollectionsof“omics”informationandasso- paper;however,theneedsandissuesforengagementwiththeotherel- ciatedrelationaldatabases,hasalsoproducedavastamountofpoten- ementsoftheAOP-KBwilllikelybesimilar.TheAOP-Wikiisanopen- tially useful, but still ultimately difficult-to-access information. An sourceplatformforcollectingandorganizingbiologicalinformation. additionalcomplicationisthatlackofcurationmakesitdifficulttoas- AtpresenttheAOP-KBcontainsabout223AOPsapplicabletoassess- sessthequalityofmuchinformation(Tripathietal.,2016;Lægreid mentsfocusedonbothhumanhealthandtheenvironment(http:// andKuiper,2015;Howeetal.,2008). aopkb.org, accessed 20 Sep 2017). Other AOP-KB packages include A.Carusietal./ScienceoftheTotalEnvironment628–629(2018)1542–1556 1545 Fig.2.DepictionoftheroleoftheAdverseOutcomePathway(AOP)frameworkinlinkingvariousdatastreamstooutcomesrelevanttoregulatorydecision-makingforchemicals.MIE– molecularinitiatingevent,KE–keyevent,KER–keyeventrelationship,AO–adverseoutcome. Source:FromAnkleyandEdwards(2018). Effectopedia,AOP Explorer,and theIntermediate Effects Data Base initiators andtheirrelevantchemical properties linkedto theMIE. (Table1).AuniquefeatureoftheAOP-KBisthateachofthemodules ThuswhenonedescribesthetypeofknowledgecoveredbytheAOP isbeingdevelopedandmanagedbydifferentagenciesorentities,but framework,itisimportanttohighlighttheavailabilityofinformation theoverallorganization andsupportingdocumentsarecentralized onmethodsandchemicalsassociatedwithmeasurementoftheKEs. undertheauspicesoftheOECD.Itisnot,however,theonlydatabase Someofthekeyattributesofthewayknowledgeismanagedwithin forAOPs,sincetherewillsoonbeastandardreportingtemplatethat theAOPframeworkaredepictedinFig.3(a).Theyinclude: canbeusedtodepositAOPsinanetworkofinteroperabledatabases Clarity–thedescriptionofanAOPishighlystructuredandfollowsa orknowledgebases,withtheAOP-KBasacentralhub. particular format and well defined conventions and guidance StakeholdersneedtobeawareofalltheuniquefeaturesoftheAOP (Villeneuveetal.,2014a,2014b;OECD,2017,2016a,2016b,2016c). frameworkthatmakeitworthwhileforthemtofullyembraceitand Onceusersbecomefamiliarwiththeformat,itprovidesclarity,ease motivatethemtoactivelyengagewiththebroaderAOPcommunity. andefficiencywhenaccessingAOPknowledgeandultimatelyprovides Atitscore,theAOPframeworkisacomprehensivemeansofgathering, adefactoreportingstandard. integrating,curating,sharing,reviewinganddisseminatingknowledge Synthesis – there is a wealth of existing knowledge relevant to aboutthemechanismsandconsequencesofperturbationofnormalbi- perturbedbiologicalpathwaysspreadacrossmanysources.AnAOPtyp- ologicalfunctionbychemicalornon-chemicalstressorsindifferentor- icallycapturesknowledgefromalargenumberofpeer-reviewedpapers ganisms. Ultimately there are three “selling points” of the AOP andintegratesanddistillsitintoaconciseform,resultinginconsider- framework,namely;i)theAOPknowledgeitself,ii)thewaytheknowl- ablesavingintimeandeffortforusers. edgeisassembled/treatedandiii)thepotentialapplicationsandimpact Sharing–theframeworkisbuiltwithcrowdsourcingandknowl- theknowledgecanhave.Theprimarytypeofknowledgecapturedby edge-sharingverymuchinmind,notonlyfortheinitialdevelopment theframeworkisobviouslyrelatedtotoxicologicalprocessesandmech- ofanAOPbutalsoduringongoingrefinementandreview.Sincean anisms.However,inadescriptionofanAOPthereisalsovaluableinfor- AOPcoversmanylevelsofbiologicalorganization,thisnaturallystimu- mationontestmethodsandmodelsthatcanbeusedtomeasureor latesandreliesuponextensivecollaborationacrossnumerousscientific predicttheKEsthatcompriseanAOP,andinformationonchemical disciplines Table1 ComponentsoftheAOPknowledgebase(KB)(http://aopkb.org/,accessed20Sep2017). Module Description Developing URL entity e.AOP.Portal MainentrypointfortheAOP-knowledgebase OECD http://aopkb.org/ (AOP-KB) AOPWiki Anopen-sourceplatformforcollecting USEnvironmentalProtectionAgency http://www.oecd. andorganizingbiologicalinformation OfficeofResearchandDevelopment org/chemicalsafety/testing/adverse-outcome- pathways-molecular-screening-and-toxicogenomics. htm Effectopedia Anopen-knowledgeandstructured OECD https://www.effectopedia.org/ platformabletodisplayquantitative informationonAdverseOutcome Pathways(AOPs) AOPExplorer AOPnetworkvisualizationandanalysistool USArmyCorpsofEngineers https://github.com/DataSciBurgoon/aopxiv IntermediateEffectsData Repositoryforkeyeventinformation EuropeanCommissionJointResearch aopkb.org Base Centre 1546 A.Carusietal./ScienceoftheTotalEnvironment628–629(2018)1542–1556 Fig.3.Anillustrationof(a)theattributesoftheAOPframeworkand(b)thepotentialapplicationsofAOPs. Knowledgebase–themostimportantaspectoftheframeworkispro- analysis,itisevenpossibletorelatetheeffectsobservedinthecomplex vidingreadyaccesstoAOPcontent.Further,withtheanticipatedadop- environmentalsamplestoknownchemicalsandtocharacterisetheef- tionofanOECDharmonisedtemplateforreportingAOPsandtherecent fectsofmixturesofunknownchemicals(Bracketal.,2016;Nealeetal., launchofthee.AOP.Portal(Table1),itispossibletoenvisioninteroper- 2015). ableAOPplatformsmaintainedbydifferentpartiesacrosstheworld. TheAOPframeworkcanalsobedescribedrelativetocurrentregula- Transparency–AOPdevelopmentandreviewprocessesareiterative torychallengesandpoliticalconcerns(Fig.3b).Manyagenciesworld- innatureandfulltransparencyhelpstoengagepotentialcontributors wideareunderpressuretoidentifyandregulateendocrinedisruptors directly,eitherasco-developersorreviewers,orindirectlythrough, arelookingtoAOP-basedapproachesforhazardandriskassessment forexample, thepostingof comments on theAOP-Wiki discussion thatcanhelpfilldatagapswithinareasonabletimeframe,atanaccept- pages.Theidentityofdevelopers,contributorsandreviewsarealso ablecost,andkeepinganimaltestingtoaminimum(Coadyetal.,2017; displayedwithintheAOP-KB. Browneetal.,2017).Theeffectsof“real-life”chemicalmixturesrepre- Quality–thequalityofanAOPisassessedandassuredthroughsev- sentanotherconsiderablechallengeforriskassessors.Inthiscontext, eralsteps:theavailabilityofdetailedguidancefordevelopers(OECD, AOPnetworkscouldprovetobeavitaltooltoexploremixtureeffects 2017,2016a,2016b,2016c),continuouscrowdsourcedpeerreview, inordertofocusattentiononchemicalsandAOPsthatreallymatter andcurationbyscientistsinvolvedinmaintainingtheAOP-KB.More- (Villeneuveetal.,2014a,2014b).Finally,thesafesubstitutionofhazard- over,AOPsdevelopedwithinthecontextofOECDprojectsaresubject ouschemicalsinproductswithlesshazardousalternativesisanareaof totworoundsofrigorousexpertreview,onebyanOECDexpertadvi- growingimportance,buttheuseoftraditionaltoxicitytestingmethods sory group, and the other by a panel of independent international toinformtheprocessareoftentooslow,ineffectiveandexpensive. experts. Thus,AOP-basedhazardprofilingandrankingpotentialsubstitutes Endorsement–onceanAOPhassuccessfullypassedthroughexpert basedontheirrelativetoxicitywouldbeofgreatvaluetomanycompa- reviewwithintheOECDprocess,itcanbesubmittedtotheWorking niesoperatinginavarietyofsectors. PartyofNationalCoordinatorsfortheTestGuidelinesProgrammeand theWorkingPartyforHazardAssessmentfortheircollectiveendorse- 3.StakeholdersintheevolutionoftheAOPframework ment,leadingultimatelytopublicationoftheAOPintherecentlycre- atedseriesdedicatedtoAOPs(OECD,2016c). TheAOPapproachcannotdeliveronitsfullpromisewithoutacriti- Quantification–thedescriptionofanAOPcanbeenhancedbythein- calmassofstakeholders,representingdifferentdisciplinesandsectors, clusionofquantitativedata,suchasthedynamicsunderpinningKEre- engagingitindifferentways.Thereisawiderangeofexistingandpo- lationships (Villeneuve et al., 2014a; Wittwehr et al., 2017). In tentialstakeholdersintheframework(Fig.1).Inthissection,weprofile addition, more complex processes can be represented through the sevenareasandfourgroupsofstakeholderswhoarealreadyengaged weightedcombinationofasetofinterconnectedAOPswithinanetwork (or could easily become more engaged) with the AOP framework. (e.g.,LaLoneetal.,2017b). Eachprofileidentifiesatypicalcommunityofstakeholders,outlining TheAOPframeworkcanbeviewedbystakeholdersfromtheper- whytheyare(orwouldbe)interestedinAOPsandhowtheycouldfur- spectiveofitsdifferentpotentialapplications(Fig.3b).Someofthese therdeveloptheframework.Thelistisnotexhaustiveordefinitivebut arequitecloselyrelated,suchasreducinganimaltesting,predictivetox- ismeanttoserveasaninitialanalysisofpotentialstakeholdersandtheir icologyandsafetyassessment.Essentially,AOPsprovideamechanistic rolesfortheadvancementoftheAOPframework. blueprinttodesignintegratedassessmentapproachesbasedoncombi- nationsofinvitroandinsilicomethodsthatpredicttoxicologicaleffects 3.1.Regulatoryassessmentofchemicalrisks ofinterest,thusreducingtheneedforanimaltesting(Tollefsenetal., 2014).Moreover,includingspecies-specificknowledgewithinAOPs AprincipalstakeholdergroupfortheAOPframeworkandassociated helpstoaddresstheincreasingdemandforspecies-relevantassess- knowledgeconsistsofregulatorytoxicologists,riskassessorsandman- ments.Mechanisticprofilingofmolecularlibrariesduringthedrugde- agersdirectlyinvolvedinevaluatingchemicalsafetyandimplementing velopmentprocesscanbeeffectivelyinformedbyAOPknowledgeto riskmitigationmeasures.Theresponsibilitiesandneedswithinthis guidetheselectionofappropriatebatteriesofinvitrohigh-throughput groupcanbeverydiverse,dependinguponjurisdiction,policycontext screening(HTS)assaysforfingerprintingtheeffectsofchemicalstosup- andlegislativemandate.Regulatorytoxicologistsandriskassessors porttheirsafetyassessment.HTSbioassayscanbeemployedtocharac- havetodealwithvariedprotectiontargets(e.g.,varioushumangroups terise not only single chemicals but also be used to test complex orfacetsoftheenvironment),differenthazard/riskassessmentscenar- environmentalsamplesinenvironmentalmonitoring(Schroederetal., ios,varyingamountsofavailabledataand/orlevelsofdatacollectionca- 2016). By using concepts of mixture toxicity and effect-directed pabilities(rangingfromverylittleformostindustrialchemicals,tolarge A.Carusietal./ScienceoftheTotalEnvironment628–629(2018)1542–1556 1547 amountsforpesticides)anddifferingnumbersofchemicalsforwhich thepurposeofassessingtheriskofallergicreactioninhumanswithout theyareresponsible(afewranginguptothousands).Riskmanagers thenecessity of animal testing (Maxwell et al.,2014; OECD,2012; thenhavetousetheoutputofariskassessmenttomakedecisionson MacKayetal.,2013). whethermeasuresshouldbetakenasrequiredbyrelevantlawsandas- sociatedprotectiongoals.Decisionsmightincluderestrictingtheuseof 3.3.Productdiscoveryanddevelopment ahazardouschemicalorrequiringitspriorauthorisation,andoftenin- cludeconsiderationofsocio-economicconsequencesofintervention, Thepharmaceutical(bothhumanandveterinary)andagrochemical whichcanbeconsiderable.Accordingly,theutilityandperceptionof industriesareprimarilyfocussedondevelopingnovelchemicalentities valueoftheAOPframeworkwillvaryfordifferentmembersofthis forthepurposeofmedicinalandagriculturalusage.Herearequirement stakeholdercommunity.Forexample,assessorsresponsibleforrapid istoefficientlyidentifycompoundsthatmaybeharmfultohuman,an- processingoflargenumbersofchemicalswithlimiteddatamayrely imalhealthortheenvironment.Themotivationsaresignificant,asa upontheAOPframeworkfortheutilizationofinsilicoand/orinvitro numberofchemicalsused,forexample,incommonlyprescribedmedi- dataforhazardprofiling,whichwouldserveasthebasisforidentifying cationsmaybehazardoustotheenvironment(seereviewbyDaughton, chemicalsofpotentialconcernandprioritizingthemforfurthertesting 2003).Acomplementaryneedisthenecessitytoretainchemicalsof orassessment.Thisscenarioisexemplifiedbyregulatorsinvolvedinthe proventherapeuticeffectiveness.Thereareanumberofopportunities US Environmental Protection Agency's (EPA)'s endocrine disruptor tousetheAOPframeworktosupporttheseactivitiessuchasenriching screeningprogram,whousetheAOPframeworkasabasisforprioritiz- thetriagingprocessidentifyingundesirablechemical/pharmacological ingchemicalsforpossibleinvivotestingbasedontheirendocrineactiv- propertiesappliedearlyindrugdiscoverytodeprioritizechemicals ity(Coadyetal.,2017;Browneetal.,2017).Anotherscenario–suchas likelytocauseharm.Secondly,AOPsfromtheAOP-KBcouldactasapre- inpesticideriskassessment–mayinvolvemembersofastakeholder cursorforonward,morequantitativeapproachesbasedonsystemstox- grouprequestingthegenerationofnewdata,targetingtests/endpoints icology(Sturlaetal.,2014)forsimulatingthepotentialconsequences mostlikelytobesensitivetoagivenchemical.Inthiscase,AOPknowl- on adverse outcomes that are still poorly characterised. For some edgecanserveasabasisforguidingandoptimizingtestselection. types of adverse effects such as hepatotoxicity (Woodhead et al., 2017)andcardiactoxicity(Daviesetal.,2016),therehavealready 3.2.Chemicalsafetyassessmentforindustry beendemonstrationsofusinginsilicoapproachesinthepharmaceutical industry.However,thereisscopeforextendingconsiderationofthese Thechemicalindustryisrequiredtoprovideinformationforspecific toxicitiesinsupportofpreclinicalsafetyassessmentofdrugcandidates apicalendpointsrelevanttohumanhealthandtheenvironmenttosup- onsensorialsystems,suchasdrug-induceddisturbancesontasteand portsafetyassessments,aslaidoutundervariousregulationssuchas smell or sleep disorders (Cavero and Holzgrefe, 2015; Cavero and TSCAintheUS(https://www.epa.gov/tsca-inventory,accessed20Sep Holzgrefe,2017).WideremploymentoftheAOPframeworktocomple- 2017),REACHRegulationintheEU(EuropeanParliament,2006),and mentnascentapproachesofferstangiblebenefitsinbeingabletoenrich ChinaREACH(Lauetal.,2012;Wangetal.,2012;CRS,2014).Thesein theavailablechemistryforcompoundsthatwouldbesubsequentlyless turnaresupportedbyguidancedocumentsoutliningtheacceptability likelytofailduetounforeseentoxicitiesorenvironmentalharm. ofmethodsconcerningdatarequirements.Theseregulatoryrequire- mentsareparticularlyrelevantforcompaniesinvolvedinthedevelop- 3.4.Medicineandhealth mentofnewchemicals/products,butcanvaryaccordingtoindustry sector(e.g.,personalandhomecareproducts,solvents,etc.).However, TheAOPframeworkwouldbeenrichedbytheknowledgeandex- therearealsoanumberofothersafetyrequirementswithinindustry pertiseofclinicians,whocouldcontributeunderstandingofthebiolog- thataredrivenbystewardship.Theseincludeidentifyingchemicalal- icalandphysiologicalcontextofAOPs,andthereforeoftheirplausibility ternativesthathavereducedenvironmentalorhumanhealthimpacts orpotentialgaps.Inturn,theAOPframeworkisapplicabletounder- (NationalAcademyofSciences,2014),enhancingapproachesforchem- standingdiseasepathwaysacrossmultiplebiologicallevelsandisa icalread-acrossandcategorisation(chemicalgroupingandextrapola- highly useful addition to the “toolbox” of approaches increasingly tionbetweenchemicals),performingstrategic/targetedtesting(data- beingusedintheclinicforprevention,diagnosisandtreatment,and gapfilling),conductingbiologicalread-across(interspeciesextrapola- inbiomedicalandclinicalresearch fordrugdiscovery,efficacyand tion),identifyingnon-animalapproachestosafetyassessment,and safety testing (Langley et al., 2015; EFPIA MID3 Workgroup et al., conductingpost-productionenvironmentalmonitoring. 2016).Cliniciansareimportantforfullydevelopingthebiomedicalpo- PerhapsthegreatestcontributionthatanAOP-basedapproachcan tentialoftheAOPframework,astheycouldincreaseaccesstohuman- providetosupportregulatorytoxicologyandriskassessmentisareduc- baseddata(Rodriguezetal.,2015),andgiveamorecontextualisedun- tionoftheuncertaintyinunderstandinghazardouseffectsforsafetyas- derstandingofinvitroassays.Theframeworkcouldalsoofferunique sessment(andhenceincreasedconfidenceinthesafetyassessment) formsofsupportforcliniciansusingtoolstoensurethatprescribingis throughabetterunderstandingofpathwayplausibilityandbiological optimisedforthepatient.Tothisend,anumberofdecisionsupportsys- read-acrossfortargeteddata-gapfilling,etc.(Perkinsetal.,2015).In temsexisttohelptheclinicianbettertailorprescribedmedicinesthat theshorttermtheuniqueAOPconstructallowsinformationonKEs donotgiverisetounintendedadverseeffects,suchasco-medication and KE relationship-derived causal links from widely-varying data error.Longer-termambitionsmayincludedecisionsupportsystems sources(rangingfrominsilicotoinvivo)toenablesupportingweight thataretailoredtoindividualpatientcharacteristicsbeyondco-medica- ofevidenceapproachestobeusedin,forexample,biologicalread- tions,suchasdiet,weight,sexandlifestyle.Onesuchexampleisthe across,withouttheneedforapicalendpointtesting(e.g.,Burdenetal., CredibleMedsdatabase(SchwartzandWoosley,2016),whichprovides 2015). easy-to-useguidancetotheclinicianonthepropensityforaparticular Inthemidtolongerterm,theAOPapproachhasthepotentialtobe medicationtocauseproarrhythmia,particularlywhentherearepredis- usedtohelpidentifyrelevantpathwaysinthederivationofbiological posinggeneticfactors.Whereasdecisionsupportsystemsinclinical pathwayalteringdoses(Judsonetal.,2011)orforhelpingtodefinebi- contextscurrentlyarelargelybasedonstatisticalandepidemiological ologicalpointsofdeparture(Bercuetal.,2016),aspartoftheriskas- correlations,theAOPframeworkandAOP-KBwouldaugmentdecision sessmentparadigmshiftoutlinedbytheNationalAcademyofSciences supportbyaddingthelayerofmechanisticdata,makingthebiological intheUS(NAS,2014).Onehigh-profileexampleoftheuseofAOPsin basisforoutcomesmoretransparent.Thisdevelopmentwouldfitcon- chemicalsafetyassessmentinthisforward-lookingmannerhasbeen sistentlywiththecurrentaspirationtowardsPersonalisedMedicine incollatingscientificdatasupportinganAOPforskinsensitizationfor (e.g., pharmacogenomics), but would have to take heed of the 1548 A.Carusietal./ScienceoftheTotalEnvironment628–629(2018)1542–1556 challengesthatsuchattemptsencounteratthelevelofclinicalimple- molecularandcellbiologists,pharmacologistsandbioinformaticians. mentation(Hedgecoe,2004;Dayetal.,2017). AnexamplegroupofacademicswhohaveadoptedtheAOPconcept TheAOPframeworkalsocomplementsandsupportstherepresenta- areenvironmentaltoxicologists,whohavepursueddifferentfacetsof tionofdiseaseprogression,oftenreferredtoinmedicineasthe“natural mechanisticresearchformanydecades(Hermensetal.,1985;McKim historyofdisease”.Thismightbeparticularlyrelevanttotheidentifica- et al., 1987; van Welie et al., 1992; van Wenzel and Opperhuizen, tionofdiagnosticmarkersofdiseaseonset/progression,whichcouldes- 1995;EscherandHermens,2002),butwiththeadventoftheAOPcon- sentiallycorrespondtodiscreteKEswithinanAOP.Futuredevelopment ceptwereabletoassemblethispastworkinthecontextofaunifying wouldbeparticularlyrelevantforrarediseases,whereprogressionis approachandcommonlanguage. oftenpoorlyunderstood(FDA,2015).GiventhematurityoftheAOP- Academicresearchershavedualincentivesasstakeholdersforthe basedassessmentforthelikelihoodofchemicalstocauseskinsensitisa- AOPframework:i)theyhavevocationalmotivationsasscientistswho tion,thismaysubstantiatethehopeforfurtherimplementablesystem aspiretoadvancingknowledgeintheirfieldandteachingstudents; forcliniciandecisionsupport. andii)theyhavecareermotivations.Forthelatter,publicationsarea mainindicatorofscientificproductivityandperformance,andneces- 3.5.Environmentalquality saryforcareerprogressionintermsofpositions,resources,andpromo- tions. While there is some variability depending on country and Similarstakeholdergroupsasinchemicalriskassessmentarere- institutions,ingeneralthereisincreasingpressureonacademicsto sponsibleforassessingtheeffectsofchemicalmixturesandcomplex demonstratetheimpactoftheirresearchbeyondacademia.Iftheim- environmentalmixturesintheenvironment-inwaterandinfood. portanceofimpactonpolicy-makingandindustrialinnovationcon- Complex environmental mixtures poseachallenge thatcan be ad- tinuetogrow,itcouldprovideanimportantincentivetocontributeto dressedmoresystematicallythroughtheapplicationofAOPnetwork AOPdevelopment. concepts(Villeneuveetal.,2014a,2014b).Invitrobioassaysincreas- Academicscientistsareinacriticalpositiontomakeasignificant inglyarebeingusedforenvironmentalmonitoringtoprocessmany contributiontotheuptakeandestablishmentoftheAOPframework samplesinatime-andcost-effectivemanner.Recentlyithasbeensug- throughtrainingnewgenerationsofresearchersandgraduateswho gestedthatinvitroHTSassayscouldprovidewater-monitoringpro- willbeemployedinindustry,policyandregulation.Forexample,aca- grams focused either on human health (e.g., drinking water) or demicsareabletopromoteAOPsviacoursecurriculaandincorporating ecological(e.g.,effluentdischarge)effects(Schroederetal.,2016).The theAOP-KBandapplicationsintoxicologyandpharmacologycourses. maincriteriaforaninvitroassaytobeapplicableformonitoringcom- AsexemplifiedinateachingmonographbyEscherandLeusch(2012) plexenvironmentalmixturesisthatitsensitivelydetectschemicalsof thatheavilyreliesontheAOPframework,textbookscangreatlycontrib- environmentalconcern,andthatitresultsinameasurementthatcan utetopropagationoftheAOPapproach. beclearlyrelatedtorelevantapicalresponsesinorganismspotentially AnexampleofengagementofacademicstakeholderswiththeAOP exposedtothecontaminantmixtures(Fig.4).TheAOPconceptthere- frameworkisthelargeEuropeanresearchconsortiumprojectSOLU- forecanserveasjustificationforthechoiceofinvitrobioanalytical TIONS(Bracketal.,2015),whichhasfullyadoptedAOPsasaunifying tools(Escheretal.,2014;Nealeetal.,2017).Byanchoringinvitro concepttoaddressthechallengesrelatedtoidentifyingandassessing assayresults(whichtypicallyreflectMIEsbutinsomecasesalsointer- waterpollution.SOLUTIONSisanEU-fundedresearchprogramwith mediateKEs)torelevantAOPs,thisinformationisprovidedefficiently thegoalofdevelopingtoolsfortheidentification,prioritisationandas- and,inturn,itispossibletodevelopnewinvitromethodsfromknowl- sessmentofwatercontaminantsthatmayposerisksbothtoecosystems edgeofenvironmentallyrelevantAOPs(Fig.4). and human health, to support the EU Water Framework Directive (Bracketal.,2017;EuropeanParliamentandEuropeanCouncil,2000). 3.6.Academicapplications SOLUTIONSreliesheavilyontheapplicationofinvitrowhole-organism andcell-basedbioassayswithinanAOPframeworkforassessingtheef- AcademicscientistscurrentlyinvolvedwiththeAOPframeworkare fectsofcomplexmixturesofaquaticcontaminants(Altenburgeretal., mostlyhumanhealthandenvironmentaltoxicologists,biochemists, 2015).Anotherprominentexampleofamajorresearchconsortium Fig.4.ApplicationofinvitrobioassaysforenvironmentalqualitysurveillancemonitoringthatareanchoredintheAOPframework. A.Carusietal./ScienceoftheTotalEnvironment628–629(2018)1542–1556 1549 adoptingtheAOPframework,thistimeinthehumanhealtharena,is voluntaryeffort,andtheformatisnotnecessarilycompatiblewiththe theEUHorizon2020fundedproject,EU-ToxRisk(www.eu-toxrisk.eu). contentofpeer-reviewedarticlesthatarethebasisfordevelopingim- Theprojectaimstoemploynon-animalmethodstodevelopnewap- pactmetricsnecessaryforcareeradvancement.Prior-publicationrules proachestoassessingthesafetyofchemicalsinavarietyofsectors,ad- imposedbymostscientificjournalswouldprecludepublicationifthe dressing in particular their potential chronic, developmental and informationwerealreadymadeavailableintheAOP-KBand,oncepub- reproductiveeffects. lishedinapeer-reviewedjournal,researchersmayhavelittleincentive todeposittheirdataintheAOP-KB.Potentialcopyrightinfringements 3.7.Non-governmentalorganizations(NGOs) alsomayoccurthroughuploadingpreviouslypublishedmaterialonto theAOP-KB(suchastheWiki)(seeSection4.5below). NGOsgivevoicetoconcernsemergingfromcivilsocietyandtothe interestsofthosewhoareaffectedbychemicalsafetyassessments 4.2. Possible risks and burdens to stakeholders supporting the AOP and resulting decisions. Apart from this common goal, individual framework NGOstypicallyhavediffering,sometimescontrastinginterests,and thereforehavedifferentincentivesforengagement.Despitethis,they WhilewetaketheviewthattheAOPframeworkisthemostprom- arepotentialusersofAOPknowledgeandparticipantsindeveloping isingandmostrealisticanswertothechallengesofchemicalsafetytest- andusingtheAOPframework. ing,itisnecessarytohighlightpossiblerisksandburdens,soasto EngagementoftheanimalwelfareNGOsstemsfromthedesireto addressthem. moveawayfromtheuseofanimalsforchemicalevaluationandbiolog- Policy-orientedorganizations(likeOECD)playacriticalroleinsus- icalresearch(Tayloretal.,2008).Acompanionconcernisforimproved tainingthecredibilityoftheAOPframeworkandinprovidingassurance humanandenvironmentalhealth,butthemainfocusisontheanimals thatAOPscanimprovedecisionmaking.However,therearesignificant usedinlaboratories.SeveralanimalwelfareNGOshaverecognizedthe reputationalrisks,shouldtherebenon-optimaldecision-makingbased promiseofAOPs,andareparticipatinginandsupportingtheirdevelop- onAOPinformationorknowledge.Thisriskincreasesasthepopularity ment(see,forexample,Sullivanetal.,2017)andpromotionthrough oftheframeworkandthenumberofAOPsgrowandthequalitycontrol education,training,anddissemination(seeTable3).TheseNGOsare processbecomesitselfincreasinglycrowdsourcedandnotlimitedtoa alsoinvolvedinpromotingtheAOPframeworktothebiomedicalcom- tightcommunityofhighlydedicatedscientists,ascurrentlyisthecase. munity(Langleyetal.,2015,2016). Governmentalandindustrialorganizationsinvolvedinsafetyas- AsenvironmentalNGOs'coremissionistofosterenvironmental sessmentarekeyplayersinsecuringtheactualusefulnessandhence protection,theyareprimarilyconcernedabouttheeffectsofchemicals theultimatesuccessoftheAOPframework.Buttheyarealsothosefac- on humans and the environment. There is a strong public concern ingthemostseriousrisksinadoptingthem,sincetheycouldbeheldac- aboutthelargenumbersofchemicalsforwhichlittlesafetyinformation countable for harm to consumers, the environment or business isavailable.Theyarealsoconcernedaboutperceivedandpotential interests.ThereneedstobeconfidencethatdecisionsbasedonAOP harmthatis,orcouldbeoccurringviachemicalssuchasendocrine-ac- considerationsareasgoodasconventionalapproaches. tive substances, andpotential developmentally-toxic or neurotoxic Seriousobstaclesmaycomefromlackofadequateincentivesforde- chemicals(e.g.,TheEndocrineDisruptionExchange[TEDX],http:// velopingandadoptingnewtestingmethods.Forthoseinindustry,the endocrinedisruption.org/,lastaccessed20Sep2017).Thusfar,there pressurestodeliverassessmentstoanexistingregulatorystandard has been only a somewhat limited engagement of environmental meanthattherearefewresourcestoconsideralternativeandnewap- NGOs, with a few notable exceptions. Specifically, since pathway- proaches.Withsignificantinitialdevelopmentcostsrequiredtodemon- basedinformationisusedinresearchintotheeffectsofpesticideson straterobustnessofanewapproach,bothintermsofintellectualand theenvironment(Kongsbaketal.,2014;Zhuetal.,2018)itcanbeex- financialinput,thismeansthatoftencorporationsarenotinaposition pectedthatthiswillincreasinglybeusedintheanalysesofenvironmen- tosupportsignificantchangesinanapproachuntilithasalreadybeen talNGOs,andpotentiallytheycouldbothparticipateinbuildingAOPs, sufficientlyvettedbyotherorganizations.Regulatoryagenciesarealso andbenefitfromthedevelopmentanduseofAOPs.Atthesametime, resource-andtime-strapped,suchthataddinganothersourceofinfor- however,someenvironmentalNGOgroupshaveexpressedconcernre- mationtobeweighedagainstothertypesofscientificevidencecan gardingthescientificbasistotheuseofAOPstosupportchemicalsafety makeassessmentsmorecomplicated. decisionsandhavequestionedtheindependenceandtransparencyof theprogramattheOECD(PesticideActionNetwork[PAN]EU,2016). 4.3.Qualityassurance IncreasedparticipationofenvironmentalNGOscouldhelpconfront theseissuesandfacilitatetheirinvolvementinfuturedevelopments. AOPscanserveanumberoffunctionsrangingfrom,attheirmost basic,assemblingavailabledataasabasisforunderstandingwhatis 4.ChallengestoscalinguptheAOPframeworkanditsapplications andwhatisnotknownaboutagivenpathway,torelativelycomplete, quantitativedepictionsofapathwaythatcouldbeappliedtoriskas- BroadutilizationoftheAOPframeworkisdependentoninvolvinga sessment.Accordingly,Villeneuveetal.(2014a)describeaspectrum criticalmassofstakeholders,whoseparticipationwouldensurethatthe ofAOPdevelopmentfrom“putative”to“quantitative”.Beckeretal. frameworkreflectsandco-evolveswiththeircurrentandfutureneeds. (2015)notethatAOPsatallstagesofthisspectrumcouldbeemployed Thereareanumberofchallengesinensuringthattheframeworkade- fordifferentapplications,butusersneedtoconsiderthedegreeoftheir quatelyengagesandcontinuestomeettheimmediateandlong-term developmentandlevelofconfidenceinagivenAOPinthecontextof needsofthedifferentstakeholdersprofiledabove;hereweidentify whetheritwouldbe“fitforpurpose”foragivenapplication.Basically, someofthesechallengesandinthesubsequentsectionrecommend thegreatertheimpactoftheapplication,thegreatertheconfidencere- cross-cuttingstrategiesthatcanaddressdifferentaspectsofthesechal- quiredinasupportingAOP.Thisrequiresthattherebeaformalprocess lenges(seeTable2). toassessAOPquality. Fromasciencepointofview,theissueofqualitycontrolistightlyin- 4.1.FormalpublicationversusprovidinginputtotheAOP-KB terconnectedwiththatofpublicationandpeerreview:the“goldstan- dard”forqualityassuranceofresearchresults.Accordingly,theAOP Themostsignificanthurdleforresearchersforwhomjournalpapers communityhasestablishedamulti-levelsystemofreviewingwell-de- areanimportantoutputisthestatusoftheAOP-KBrelativetoformal velopedAOPs,internallyandexternally.Forexample,OECD-endorsed publication.PopulatingtheAOP-KBcurrentlyrequiresconsiderable AOPshavealevelofpeerreviewequivalentorsuperiortothatofa 1550 A.Carusietal./ScienceoftheTotalEnvironment628–629(2018)1542–1556 Table2 Multiplemeansforaddressingthedifferentchallenges. Publishingandreview Educationand Stakeholderspecific Translationinto Governanceandfunding strategy training interaction application structures Tensionsbetweenpublicationandknowledgebase x x Risks:financial,reputational,etc. x x x Qualitycontrol x x Cross-disciplinary/sectoralunderstanding x x x Ethical,legalandsocialissues x x x Governanceandsustainability x publishedpaper.OECDendorsementisrecognizedasbringingwithitan plagiarismrules.Inaddition,AOPspublishedbytheOECDdonotspecify importantreputationaladdedvalue,especiallybytheregulatorycom- whatkindofcopyright,ifany,protectstheintellectualworkandwhich munity.Careerwisefortheacademiccommunity,thereisstillaprefer- usesareauthorised.Thisisstrikingwhenintellectualworkfreelyacces- enceforpublishinginknown,andhighimpactjournals,eventhough siblethroughtheInternetoftencomeswithclearlyspecifiedlicences. theymayrecognisethatthelevelofpeerreviewisequivalent.There- ThesameappliestowhatisavailablethroughtheAOPWiki,whichby viewprocessisalsoanimportantwayofbuildingtheAOPcommunity, itsnatureasawikiplatform,isoftenassociatedwithopenaccessand asitdrawsinmorecontributorsthanthosewhodevelopandupload freesharingorreuseofcontent.Authorsneedtohaveaclearperception AOPs.However,thereisaquestionwhethersuchawork-intensivepro- ofthekindofIPRprotectionavailableforworksharedthroughtheAOP- cesscanbesustainedinthelongterm,especiallyinconsiderationofthe KB. lackofincentivesforsomeofthereviewers.Anincreasingnumberof AOPsarenotonlyatoolforsystematicallyorganizingandcurating AOPsuploadedontotheAOP-KBwouldalsomakegreaterdemands scientificknowledgeandindicatingknowledgegapsthatneedtobe onthepoolofqualified,oftenover-committedreviewers. filled,buthavebeenpromotedandareperceivedasanimportantin- strumentforregulatorsdealingwithprotectingtheenvironmentand 4.4.MaintainingeffectivedialogueabouttheAOPframeworkamongstake- humanhealth.AssuchAOPsmaybeusedinregulatorycontextswhere holdergroups thestakesanduncertaintyarehigh,fiercecontroversiesmayariseand stakeholdersmaybecomepolarized,asexemplifiedbyrecentactivism TheAOPframeworkoriginallyarosefromthedisciplineofecotoxi- aroundendocrine-disruptingchemicalsintheEU(LeMonde,2016). cologytosupportregulatorydecision-making(Ankleyetal.,2010). AnobviousethicalissuethatAOPusagebringstotheforeisthewide- Thismeansthatcollaboratorsinotherfieldsmaynotnecessarilybein- spreaddesiretoreduceorultimatelyreplaceanimaltestingforchemical terestedinallitsfeatures,andmaymisunderstandaspectsofthecon- safetyassessment.Asnotedabove,thisisastrongmotivatorforanimalwel- cept,orfindtheframeworktooconstrainingorsimplifiedfortheir fareNGOstakeholders,butitmayalsoresultintensionswithotherethical ownneeds.ArecentsurveydescribedbyLaLoneetal.(2017a)identified priorities,forexample,associatedwithperceptionsregardingtheroleofan- avarietyofconcernsaboutAOPs,andattemptedtoaddresstheseusing imaltestsintheeffectiveprotectionofhumanhealthandtheenvironment. a“frequentlyaskedquestions”approach.Forexample,anissueoften Theinterplayandpotentialconflictsbetweentheseethicalprioritiesneeds raisedbyresearchersinbiology,fromacademiaandelsewhere,isthe furtheranalysis. perceivedlinearityoftheAOP,whichisamisconception. ThesocialaspectsofAOPsarerelatedtothefactthattheycanbe Specifically, AOPs can be assembled into ‘non-linear’ networks usedbymanydifferentpartieswithdifferentagendasandpurposes, whichattimescanbeabetterreflectionofrealityandthusbeafocus andthiscouldleadtomisusesprejudicialforhumanorenvironmental ofapplication(Villeneuveetal.,2014a,2014b;LaLoneetal.,2017b).It safety,oreconomicallyinefficientandburdensome.Recenttrendsin mustbenotedhowever,thatalthoughtheseprovisionsinAOPdevelop- sciencepolicyandresearchfundingemphasisethatscientistsneedto mentallowfornon-linaritesandmodulatingfactorstobeincludedas makeanefforttoforeseeandpreventthemisuseoftheirknowledge narrativeinthedescriptionofanAOP,theAOPconstructitselfwas andresults(Ziman,2001;GustonandSarewitz,2002;vonSchomberg, neverintendedtoprovideameansofrepresentingatoxicologicalpro- 2007,2013;Stilgoeetal.,2013),asisalsoevidentintheEuropean cessasaformalized(mathematical)dynamicalsystemsbiologytype Commission-Horizon2020(n.d.)-SciencewithandforSocietypro- model. gramme(https://ec.europa.eu/programmes/horizon2020/en/h2020- Animportantchallengeinstakeholdersdialoguearisesbecause section/science-and-society,accessed20Sep2017).Arobustconsensus of conflicting perceptions of the utility associated with different abouttheappropriateconditionsanddomainsofapplicationofAOPsis levelsofdetailincludedintheAOP-KB.Thetensionbetweendevel- thereforeneeded.Inotherwords,differentstakeholdersneedtoagree opmentandcommunicationintheAOP-KBisparticularlyprominent onwhen(i.e.,atwhatlevelofdevelopment)AOPsarefitforwhatpur- fordifferentstakeholdercommunitiesreflecting“users”asopposed pose(e.g.,prioritisationofchemicalsfortesting,tomakearegulatory to“developers”,withtheformeroftenrequiringlesscomplexity decision,orforuseasevidenceincourtsetc.;Beckeretal.,2015).In thanthelatter. practice,thisdemandsattentionbothtotheproductionofknowledge (makingsurethatAOPsmeethighstandardsofepistemicrobustness 4.5. Ethical, legal and social aspects associated with use of the AOP andarenotinfluencedbyresearchsusceptibletoconflictsofinterests framework orbiases)andtothetranslationandintegrationofthatknowledgeintoso- cialprocesses.Suchintegrationalsoraisesthequestionofhowtodealwith Therearelegalissuesaroundintellectualpropertyrights(IPR)inthe incompletenessofscientificknowledgeandwhetherandhowtocomple- productionanddisseminationofAOPsthatdeserveattentionandac- mentitwithnon-scientificknowledgedispersedinsociety(Funtowicz tion.Sincemanyscientistsneedfirsttopublishtheresultsoftheirre- andRavetz,1993;Nowotnyetal.,2001;DeGrandis,2016). searchinacademicjournals,stepsshouldbetakeninordertoenable scientiststoreusetheirworkforcontributingtoorauthoringAOPs 4.6.Governanceandsustainability without infringing on copyright ownership. Conversely, there is a needtoenableresearcherstoreusepartsoftheirresearchthathave AstheAOPframeworkcontinuestoevolve,governanceandcoordi- beenuploadedontheAOP-KBinpublicationswithoutviolatinganti- nationoftheeffortinthefuturewouldideallyinvolveallstakeholders