GAMPGoodPracticeGuideforValidationofLaboratWord文档格式.docx

GAMPGoodPracticeGuideforValidationofLaboratWord文档格式.docx

《GAMPGoodPracticeGuideforValidationofLaboratWord文档格式.docx》由会员分享，可在线阅读，更多相关《GAMPGoodPracticeGuideforValidationofLaboratWord文档格式.docx（15页珍藏版）》请在冰豆网上搜索。

GuidancetohelpvalidationofcomputerizedsystemsusedwithinregulatedlaboratoriesisalwayswelcomebutisitalwayshelpfulInthefirstpart,IpresentanoverviewoftheGuide,differentapproachtolifecyclevalidationandsystemclassification.

OverthepastyearsIhavenotspokeninanygreatdetailaboutguidancedocumentsoncomputervalidationforchromatographicsystemsandchromatographydatasystems（CDS）butconcentratedonaspecifictopicfromtheregulationsthemselves.Thisisbecausemostguidancehasconcentratedlargelyoncomputerizedmanufacturingandcorporatesystems,ratherthanlaboratorysystems.

ThishaschangednowwiththepublicationoftheGoodAutomatedManufacturingPractice（GAMP）Forum'

sGoodPracticeGuide（GPG）onValidationofLaboratoryComputerizedSystems.1However,thispublicationneedstobecomparedandcontrastedwiththeAAPSpublicationonQualificationofAnalyticalInstruments（AIQ）.2Bothpublicationshavebeenwrittenbyacombinationofrepresentativesfromthepharmaceuticalindustry,regulators,equipmentvendorsandconsultants.

Thiswillbeatwo-partdiscussionoftheguideandwhereweshouldgotocoveradequatelybothequipmentqualificationandvalidationofchromatography-basedlaboratorysystems.

OverviewoftheGuide

Publishedin2005,thestatedaimoftheGPGistodeveloparationalapproachforcomputerizedsystemvalidationinthelaboratoryandprovideguidanceforstrategicandtacticalissuesinthearea.Section5oftheGPGalsonotesthat"

...thefocusshouldbeontherisktodataintegrityandtherisktobusinesscontinuity.TheGuideassumesthatthesetwofactorsareofequalimportance."

1

TheGPGnotesthatcompaniesneedtoestablishtheirownpoliciesandproceduresbasedontheirownriskmanagementapproaches.Ofinterest,theinsidepageoftheGPGstatesthatifcompaniesmanagetheirlaboratorysystemswiththeprinciplesintheguidethereisnoguaranteethattheywillpassaninspection—thereforecaveatemptor!

Table1:

ContentsoftheGAMPGPGonvalidationoflaboratorysystems.

TheguideconsistsofanumberofchaptersandappendicesasshowninTable1.Asyoucansee,theorderofsomeofthechaptersisalittlestrange.Forexample,whyisthevalidationplanwrittensolateinalifecycleorwhyisthechapterontrainingofpersonnelpositionedafterthevalidationreporthasbeenwrittenHowever,atleastthemaincomputervalidationsubjectsarecoveredinthewholelifecycleincludingsystemretirement.TheGPGalsocrossreferencesthemainGAMPversion4publicationforanumberoftopicareasforfurtherinformationwhereappropriate.3

OnemajorcriticismisthattheninereferencescitedinAppendix5areveryselectiveand,therefore,theGPGignoressomekeypublicationsinthisareasuchas:

∙Furmanetal.onthedebateofholistic（orsystem）versusmodularvalidationorqualificationofcomputerizedequipment.4ThispaperwaswrittenbythreeFDApersonnelaboutthevalidationofcomputerizedchromatographicequipment;

ignoringitisnotanoptionasitprovidesascientificrationaleforthistwo-levelapproach.

∙PDATechnicalReport18onvalidationofcomputer-relatedsystemsthatcontainsamorespecificcomputervalidationdefinitionthantheFDAprocessvalidationdefinitionquotedinSection3.1oftheGPG.5,6

∙AAPSAnalyticalInstrumentQualificationwhitepaperpublishedin2004,whichwastheoutcomeofajointFDA-AAPSconferencefrom2003.2

Ignoringthesepapersbiasestheapproachthatthisguidehastakenandisafatalflawasweshalldiscusslaterinthisarticle.

Overall,theproblemwiththisGPGisthatyouhavetocherrypickthegoodbitsfromthebad.Aswithanyperformanceappraisalsystem,let'

sstartwiththegoodnewsfirstandworkourwaydownhillafterwards.

TheGoodNews

Figure1:

Systemdevelopmentandsystemimplementationlifecycles（adaptedfromtheGAMPlaboratoryGPG）.

ThebestpartsoftheGAMPlaboratorysystemGPGarethelifecyclemodelsforbothdevelopmentandimplementationofcomputerizedlaboratorysystems.1ThewritersoftheguidearetobecongratulatedforproducinglifecyclemodelsfordevelopmentandimplementationthatreflectcomputerizedsystemsratherthanmanufacturingandprocessequipmentpresentedintheoriginalGAMPguide.3ThelatterVmodellifecycleistotallyinappropriateforcomputerizedandlaboratorysystemsasitbearslittlecomparisonwithreality.TheproblemwiththeGAMPVmodelwasthatimmediatelyafterprogrammingthesystemitundergoesinstallationqualification（IQ）.Unit,moduleandintegrationorsystemtestingisconvenientlyforgotten,ignoredtotallyorimpliedratherthanexplicitlystated.ThetwomodelsillustratedinthelaboratoryGPGareshowninFigure1.Theleft-handsideshowsthesystemdevelopmentlifecycle（SDLC）thatisintendedformorecomplexsystemsandtheright-handsidewhichshowsthesystemimplementationlifecycle（SILC）forsimplersystems.

Thisreflectsthefactthatwecanpurchaseasystem,installitandthenoperateitasshownontheright-handofFigure1.Thevastmajorityofequipmentandsystemsinourlaboratoriesaresimilartothis,butconsiderthequestion:

hastheSILCoversimplifiedtheimplementationprocessforallspectroscopyandotherlaboratorycomputersystems

TheargumentfortheSILC:

FormostcomputerizedchromatographsandCDSinapost-Part11world,youwillneedtoaddusertypesanduserstothesystemthatwillneedtobedocumentedforregulatoryreasons,forexample,authorizedusersandaccesslevelsrequiredbybothpredicaterulesand21CFR11.However,afterdoingthisforsimplersystems,suchasanintegrator,wecangointotheperformancequalificationstage;

thisisnotmentionedspecificallyintheSILC.

TheargumentagainsttheSILC:

Thesoftwareusedinsomelaboratorycomputerizedsystemsmayneedtobeconfigured—thisisatermforeitherselectinganoptioninthesoftwaretoalteritsfunctionwithinlimitssetbythevendor.Thiscanbeassimpleasselectingwhichfunctionwillbeusedfromtwoorthreeoptionsor,inmorecomplexsystemssuchasalaboratoryinformationmanagementsystem（LIMS）,usingalanguagetosetupalaboratoryprocedureormethod.Thisfactorisnotaccommodatedspecificallyineitherofthelifecyclemodels.However,regardlessoftheapproachtaken,thesystemconfigurationmustbedocumentedpartlyforyourbusinesstoallowreconfigurationofthesystemincaseofdisasterbutalsoforperformingthevalidation.

Figure2:

Modifiedsystemimplementationlifecycleoptionsandmacrodevelopment.

Inmyview,theconceptoftheSILCisgoodbutthescopeandextentofitcanbetakenfurtherthantheLaboratoryGPGsuggests.Furthermore,itcanalsobealignedwiththeexistingsoftwarecategoriescontainedinAppendixM4ofthecurrentGAMPGuide.3Therationaleisthatmanylaboratorysystemsareconfiguredratherthancustomized;

thereforeweneedmoreflexibilitythanthesimpleSILCpresentedintheLaboratoryGPG.ThisiswheretheGAMPsoftwarecategories,outlinedinVersion4AppendixM4,comeinandwheretheLaboratoryGPGmakescomputervalidationmorecomplexthanitneedstobe.

Figure2ismyattempttotaketheSILCprinciplesfurtherthantheGPGandalignthemwiththeexistingGAMPsoftwarecategories.Auserrequirements'

specificationistheentrypointforallvariationsillustratedinFigure2,theexitfromthisfigureistheroutetothequalificationandconfigurationofthesystem.ThedefinitionsofthedifferenttypesofGAMPsoftwareare

∙GAMP3software:

Thisiscommercialoff-the-shelfsoftware（COTS）.TheSILCforthistypeofsoftwareisshownontheright-handsideofFigure2Inessence,thisisamodificationoftheGPGimplementationcyclewherethedocumentationofsecurity,accesscontrolandanyothersmallsoftwareconfigurationsforruntimeoperationaresubstitutedforthedesignqualification.

∙GAMP4software:

Thisisconfigurablecommercialoff-the-shelfsoftware（configurableCOTS）.Oncethesoftwarefunctionshavebeenunderstood,anapplicationconfigurationspecificationcanbewrittenthatwillstatewhatfunctionsinthesoftwarewillbeused,turnedon,turnedofformodified.AfterthesoftwarehasbeeninstalledandundergonetheIQandoperationalqualification（OQ）hasbeenperformed,thenthesoftwarecanbeconfiguredaccordingtotheconfigurationspecificationdocuments.

∙GAMP5software:

Althoughthisisusuallyauniqueandcustomapplication.However,inthecontextofCDSsoftware,thisistypicallyacustomcalculation,amacroorcustomprogramthatiswrittentoperformaspecificfunction.ThereforebothGAMP4andGAMP5softwarecanexistinthesamesystemandtheGAMP5aspectsaretypicallyanadditiontothenormalfunctionalityofthesoftwareratherthanasubstitute.

Therefore,thelifecyclesforaCDSinthiscategorywillbetheGAMP4software（centralflowinFigure2）plustheadditionalstepsforGAMP5macrosorcalculations（picturedontheleft-handsideofthesamediagram）.Here,thereneedstobeaspecificationforthemacro（nameplusversionnumber）,thecalculationortheprogrammingorrecordingofthemacro.Againstthiswillbeformaltestingtoensurethatthefunctionalityworksasspecified.Oncethishasbeenperformed,thenthemacroisinstalledwiththeapplicationanditistestedundertheperformancequalification（PQ）phaseofvalidationasanintegralpartoftheoverall