Standard Practices for Method Validation Forensic Toxicology · PDF fileDarcie Wallace‐Duckworth, Ph.D.; Aegis Sciences Corporation; Nashville, Tennessee Jeff Walterscheid, Ph.D.,

Thisdocumentiscopyrighted©bytheAAFSStandardsBoard,LLC.2017Allrightsarereserved.4200WisconsinAvenue,NW,Suite106‐310,Washington,DC20016‐2143,asb.aafs.org.

ASBStandard036,FirstEdition2017

StandardPracticesforMethodValidationinForensicToxicology

ASBStandard036,1stEd.‐2017


ASBApprovedXxxxx2017

ANSIApprovedXxxxx2017

AcademyStandardsBoard4200WisconsinAvenue,NW

Suite106‐310Washington,DC20016‐2143

Thisdocumentmaybedownloadedforfreeat:http://asb.aafs.org/

ThisdocumentisprovidedbytheAAFSStandardsBoardforfree.Youarepermittedtoprintanddownloadthedocumentandextractsfromthedocumentforyourownuse,providedthat:

youdonotmodifythisdocumentoritsrelatedgraphicsinanyway; youdonotuseanyillustrationsoranygraphicsseparatelyfromanyaccompanyingtext;and, youincludeanacknowledgementalongsidethecopiedmaterialnotingtheAAFSStandardsBoard

asthecopyrightholderandpublisher.

Youexpresslyagreenottoreproduce,duplicate,copy,sell,resell,orexploitforanycommercialpurposes,thisdocumentoranyportionofit.Youmaycreateahyperlinktohttp://asb.aafs.orgtoallowpersonstodownloadtheirindividual,freecopyofthisdocument.YourhyperlinkmustnotportrayAAFS,theAAFSStandardsBoard,thisdocument,ouragents,associatesandaffiliatesinanoffensivemanner,orbemisleadingorfalse.Youmaynotuseourtrademarksaspartofyourlinkwithoutourwrittenagreementforyoutodoso.

TheAAFSStandardsBoardretainsthesolerighttosubmitthisdocumenttoanyotherforumforanypurpose.

Certaincommercialentities,equipmentormaterialsmaybeidentifiedinthisdocumenttodescribeaprocedureorconceptadequately.SuchidentificationisnotintendedtoimplyrecommendationsorendorsementbytheAAFSortheAAFSStandardsBoard,norisitintendedtoimplythattheentities,materials,orequipmentarenecessarilythebestavailableforthepurpose.

Thisdocumentiscopyrighted©bytheAAFSStandardsBoard,LLC.2017Allrightsarereserved.4200WisconsinAvenue,NW,Suite106‐310,Washington,DC20016‐2143,asb.aafs.org.


Foreword

Validationistheprocessofperformingasetofexperimentstoestablishobjectiveevidencethatamethodisfitforpurpose,andtoidentifythemethod'slimitationsundernormaloperatingconditions.Thisstandardwasdevelopedtoprovideguidanceonminimumrequirementsforvalidatinganalyticalmethodsinforensictoxicologylaboratories.

Thisdocumentwasrevised,preparedandfinalizedasastandardbytheToxicologyConsensusBodyoftheAAFSASB.ThedocumentwasoriginallydraftedbytheScientificWorkingGrouponForensicToxicology(SWGTOX).WhenSWGTOXdisbandedin2014,itpassedownershipofallofitsdocumentstotheToxicologySubcommitteeoftheOrganizationofScientificAreaCommittees(OSAC)whointurnupdatedandapprovedthedraftdocument.Allhyperlinksandwebaddressesshowninthisdocumentarecurrentasofthepublicationdateofthisstandard.

Keywords:Validation,ForensicToxicology

Abstract:Thisstandardwasdevelopedtoprovideguidanceonminimumrequirementsforvalidatingmethodsinforensictoxicologylaboratories.Propermethodvalidationensuresthatalaboratoryestablishesobjectiveevidencethatamethodisfitforpurposeandthethemethod’slimitationsundernormaloperatingconditionsareunderstood.


Acknowledgements

BaseDraftDevelopedbytheOrganizationofScientificAreaCommittees(OSAC)ToxicologySubcommittee:

Editor: MarcA.LeBeau,PhDF‐ABFT;FederalBureauofInvestigation;LaboratoryDivision;Quantico,Virginia

WorkingGroup: NicholeBynum,MS;RTIInternational;ResearchTrianglePark,NorthCarolina

ConnieM.Borror,PhD;ArizonaStateUniversity;NewCollegeofInterdisciplinaryArtsandSciences;Phoenix,Arizona

SimonElliott,PhD;ROARForensics;Worcestershire,UnitedKingdom

MarilynA.Huestis,PhD;HuestisandSmithToxicology,SevernaPark,Maryland

MatthewP.Juhascik,PhDF‐ABFT;MontgomeryCountyCoroner’sOffice;Dayton,Ohio

JamesC.Kraner,PhD,F‐ABFT;OfficeoftheChiefMedicalExaminer;Charleston,WestVirginia

LoralieJ.Langman,PhD,F‐ABFT,DABCC;MayoClinic;Rochester,Minnesota

JenniferLimoges,MS,DABC;NewYorkStatePolice‐Forensics;Albany,NewYork

ChristineMoore,PhD,DABCC;ImmunalysisCorporation;Pomona,California

StephenL.Morgan,PhD;UniversityofSouthCarolina;DepartmentofChemistryandBiochemistry;Columbia,SouthCarolina

RobertJ.Osiewicz,PhDF‐ABFT;ErieCountyMedicalExaminer’sOffice(Retired);Buffalo,NewYork

JuliaPearson,PhDF‐ABFT;HillsboroughCountyMedicalExaminer’sOffice;Tampa,Florida

FrankT.Peters,PhD;InstitutfuerRechtsmedizin;UniversitaetsklinkumJena;JenaGermany

SumandeepRana,PhD;RedwoodToxicologyLaboratory;SantaRosa,California

MatthewSlawson,PhD;UtahDepartmentofHealth;Taylorsville,Utah

ElizabethSpratt,MSF‐ABFT;WestchesterCountyDepartmentofLaboratories;Valhalla,NewYork

ScientificWorkingGrouponForensicToxicology(SWGTOX)

ToxicologySubcommitteeoftheOrganizationofScientificAreaCommittees(OSAC)

FinalStandardPreparedandFinalizedbytheASBToxicologyConsensusBody:

Members: WendyAdams,Ph.D.,F‐ABFT;NMSLabs;WillowGrove,Pennsylvania

SabraBotch‐Jones,M.S.M.A.D‐ABFT‐FT;BostonUniversitySchoolofMedicine,BiomedicalForensicSciences;Boston,Massachusetts


RandalClouette,M.S.F.S.,D‐ABFT‐FD;QuestDiagnostics,Inc;Lenexa,Kansas

FionaCouper,Ph.D.,WashingtonStatePatrol;Seattle,Washington

KennethFerslew,Ph.D.,F‐ABFT;EastTennesseeStateUniversity;JohnsonCity,Tennessee

DeanFritch,Ph.D.,F‐ABFT;OraSureTechnologies,Inc;Bethlehem,Pennsylvania

ShannonGeorge,B.S.;IllinoisStatePolice;Springfield,Illinois

MicheleGlinn,Ph.D.,F‐ABFT;EssentialTestingLLC/Avertest;Collinsville,Illinois

JamesHutchison,Jr.,M.S.,D‐ABFT‐FT;MontanaForensicToxicologyServices,LLC;Lolo,Montana

MatthewJuhascik,Ph.D.,F‐ABFT;MontgomeryCountyCoroner’sOffice;Dayton,Ohio

PhilipKemp,Ph.D.,F‐ABFT;CivilAerospaceMedicalInstitute(FAA);OklahomaCity,Oklahoma

MelissaKennedy,M.S.,D‐ABFT‐FA;ANSI‐ASQNationalAccreditationBoard;Garner,NorthCarolina

MarcLeBeau,Ph.D.,F‐ABFT;FederalBureauofInvestigation;Quantico,Virginia

AdamNegrusz,Ph.D.,F‐ABFT;UnitedStatesDrugTestingLaboratories,Inc.;DesPlaines,Illinois

DouglasRohde,M.S.;LakeCountyCrimeLaboratory;Painesville,Ohio

RobertSears,M.S.,F‐ABFT;SouthCarolinaLawEnforcementDivision;Columbia,SouthCarolina

MichaelStypa,M.S.,D‐ABFT‐FT;LasVegasMetropolitanPoliceDepartment;LasVegas,Nevada

TedVosk,J.D.;Kirkland,Washington

FrankWallace,B.A.;AmeritoxLLC;HighPoint,NorthCarolina

DarcieWallace‐Duckworth,Ph.D.;AegisSciencesCorporation;Nashville,Tennessee

JeffWalterscheid,Ph.D.,F‐ABFT;ArmedForcesMedicalExaminerSystem;DoverAFB,Delaware

RuthWinecker,Ph.D.,F‐ABFT;OfficeoftheChiefMedicalExaminer;Raleigh,NorthCarolina

DustinTateYeatman,M.S.F‐ABFT,F‐ABC;PalmBeachCountySheriff’sOfficeCrimeLaboratory;WestPalmBeach,Florida


TableofContents 1. Scope.................................................................................................................................................................................1

2. NormativeReference..................................................................................................................................................1

3. TermsandDefinitions...............................................................................................................................................1

4. WhentoValidateMethods.......................................................................................................................................3

5. MethodDevelopmentandOptimization............................................................................................................3

6. EstablishingaValidationPlan................................................................................................................................4

7. RequiredValidationParametersBasedonScopeoftheMethod............................................................4

8. SpecificRequirementsforConductingMethodValidationExperiments.............................................5

8.1. BiasandPrecision..............................................................................................................................................6

8.2. CalibrationModel...............................................................................................................................................8

8.3. Carryover.............................................................................................................................................................10

8.4. InterferenceStudies........................................................................................................................................10

8.5. IonizationSuppression/Enhancement...................................................................................................11

8.6. LimitofDetection.............................................................................................................................................12

8.7. LowerLimitofQuantitation........................................................................................................................14

9. AdditionalValidationParameters.......................................................................................................................15

9.1 General..................................................................................................................................................................16

9.2 DilutionIntegrityStability............................................................................................................................16

9.3 Stability................................................................................................................................................................16

10. RequiredRevalidationofPreviouslyValidatedMethods.........................................................................17

11. DocumentationRequirementsforMethodValidation...............................................................................17

12. EfficiencywithValidation.......................................................................................................................................18

ANNEXA:QuantitationofDrugXinBloodValidationExample......................................................................19

ANNEXB:ImmunoassayScreenofBenzodiazepinesinUrineValidationExample................................32

ANNEXC:ExampleFlowchartofMethodValidationExperiments................................................................35

ANNEXD:TableofExampleExperimentsforValidationofQualitativeConfirmation/IdentificationMethods...................................................................................................................................................................................36

ANNEXE:TableofExampleExperimentsforValidationofQuantitativeMethods.................................37

ANNEXF:Bibliography.....................................................................................................................................................38


1


1. Scope

Thisdocumentdelineatesminimumstandardsofpracticeforvalidatinganalyticalmethodsinthefieldofforensictoxicology.Thefundamentalreasonforperformingmethodvalidationistoensureconfidenceandreliabilityinforensictoxicologicaltestresultsbydemonstratingthemethodisfitforitsintendeduse.

2. NormativeReferences

Therearenonormativereferences.AnnexG,Bibliography,containsinformativereferences.

3. TermsandDefinitions

Forpurposesofthisdocument,thefollowingdefinitionsapply.

3.1. biasAnestimateofsystematicmeasurementerror,calculatedasthedifferencebetweenthemeanofseveralmeasurementsunderidenticalconditions,toaknown“true”value.Itisoftenreportedasapercentdifference.

3.2. biologicalfluidsAnyliquidbiologicalspecimenthatistypicallypipettedforanalysis(e.g.,blood,urine,bile,serum,vitreoushumor,oralfluid).

3.3. blankmatrixsampleAbiologicalfluidortissue(orsyntheticsubstitute)withouttargetanalyteorinternalstandard.

3.4. calibrationmodelThemathematicalmodelthatdemonstratestherelationshipbetweentheconcentrationofanalyteandthecorrespondinginstrumentresponse.

3.5. carryoverTheappearanceofunintendedanalytesignalinsamplesaftertheanalysisofapositivesample.

3.6. decisionpointAnadministrativelydefinedcutofforconcentrationthatisatorabovethemethod’slimitofdetectionorlimitofquantitationandisusedtodiscriminatebetweenpositiveandnegativeresults.

3.7. fortifiedmatrixsampleAblankmatrixsamplespikedwithtargetanalyteand/orinternalstandardusingreferencematerials.


2

3.8. interferencesNon‐targetedanalytes(i.e.,matrixcomponents,otherdrugsandmetabolites,internalstandard,impurities)whichmayimpacttheabilitytodetect,identify,orquantitateatargetedanalyte.

3.9. ionizationsuppression/enhancementDirectorindirectalterationorinterferenceintheinstrumentresponseduetothepresenceofco‐elutingcompounds.

3.10. limitofdetectionAnestimateofthelowestconcentrationofananalyteinasamplethatcanbereliablydifferentiatedfromblankmatrixandidentifiedbytheanalyticalmethod.

3.11. lowerlimitofquantitationAnestimateofthelowestconcentrationofananalyteinasamplethatcanbereliablymeasuredwithacceptablebiasandprecision.

3.12. precisionThemeasureoftheclosenessofagreementbetweenaseriesofmeasurementsobtainedfrommultiplesamplingsofthesamehomogenoussample.Itisexpressednumericallyasimprecision.

3.13. referencematerialMaterial,sufficientlyhomogenousandstablewithreferencetospecifiedproperties,whichhavebeenestablishedtobefitforitsintendeduseinameasurementorinexaminationofnominalproperties.

3.14. stabilityAnanalyte’sresistancetochemicalchangeinamatrixunderspecificconditionsforgiventimeintervals.

3.15. tissuesAnysolidbiologicalspecimenthatisgenerallyweighedforanalysis(e.g.,brain,liver,muscle,hair,meconium).

3.16. workingrangeTherangeofconcentrationthatcanbeadequatelydeterminedbyaninstrument,wheretheinstrumentprovidesausefulsignalthatcanberelatedtotheconcentrationoftheanalyte.


3

4. WhentoValidateMethods

Methodsshallbevalidatedtoverifyamethod'sperformanceparametersarefitforuseforaparticularanalysis.Commonexamplesinclude:

a) newanalyticalmethod;

b) modificationsofanestablishedanalyticalmethodtoimproveperformanceorextenditsusebeyondthatforwhichitwasoriginallyvalidated(e.g.,additionofnewcompoundstothemethod’sscope);

c) todemonstrateequivalencebetweenanestablishedmethod/instrumentandanewmethod/instrument;

d) existinganalyticalmethodsthatdonotcurrentlymeettherequirementsofthisdocument.

Theparameterstobeevaluatedforvalidationofmethodswilldependuponthecircumstancesinwhichthemethodistobeused.Likewise,itisrecognizedthataftervalidationhasoccurred,methodsmayberevised.Theextentandfrequencyofrevalidationofpreviouslyvalidatedmethodswilldependuponthenatureoftheintendedchangesorlaboratorypolicy.SeeSection10forfurtherguidanceonrevalidationofpreviouslyvalidatedmethods.

Laboratoriesusingmethodsthatwerevalidatedpriortothepromulgationoftheseminimumstandardsshalldemonstrateanddocumentthatthesemethodsarefitforpurposeunderthesestandards.Thesemethodswilllikelyhavesufficienthistoricalcalibrationandcontroldata,aswellaspreviouslyanalyzedcaseworksampleresults,thatcanbeusedtoaddressanumberoftherequiredvalidationparameters.Whensufficientdataareabsenttofulfilltheseminimumstandards,appropriatestudiesshallbeconductedtoensurecompliancewiththisdocument.

5. MethodDevelopmentandOptimization

5.1. General

Forpurposesofthisdocument,methoddevelopmentwillbeconsideredintwophases:1)instrumentalanddataacquisition/processingparametersand2)samplepreparation.Itisessentialthatvalidationisconductedwiththesameanalyticalconditionsandtechniquesasthefinalmethod.

Theprinciplesofgoodlaboratorypracticeandrecordkeepingshallbeappliedtotheconceptsofthisdocument.Thisincludesdocumentationofparametersthatwereevaluatedduringmethoddevelopment,yetdidnotprovideacceptableresults.

5.2. DevelopmentandOptimizationofInstrumentalandDataProcessingParameters

Instrumentalanddataprocessingparametersaredefinedandoptimizedthroughanalysisofreferencematerialsoftheanalyte(s)ofinteresttoachievetherequiredperformanceoftheinstrument.

5.3. DevelopmentandOptimizationofSamplePreparationTechniques

Thesamplepreparationtechniqueshallbeevaluatedandoptimizedusingreferencematerialsoftheanalyte(s)ofinterest.Theprimarygoalistodemonstratethatthesamplepreparationsteps


4

allowforadequateextraction,detection,identification,and/orquantitationoftheanalyte(s).Samplepreparationshallbeevaluatedwithfortifiedmatrixsamples.

6. EstablishingaValidationPlan

Thelaboratoryisresponsibleforensuringitsmethodsareadequatelyvalidated.Avalidationplanshallbeinplacepriortostartinganyvalidationexperiments.Thevalidationplanisseparatefromalaboratory’sstandardoperatingprocedure(SOP)formethodvalidation.Theplanshallincludetheinstrumentalmethod(s)andsamplepreparationtechnique(s)tobeusedforaspecificmethod.Further,itshalldocumentthevalidationrequirementsofthemethod,aswellasthelimitsofthemethodthatwillallowittobefitforuse.Thevalidationplanprovidesdirectionfortheexperimentsthatwillbeperformedandacceptancecriteriaforeachparameter.AnnexBandAnnexCprovideexamplesofvalidationplans.

7. RequiredValidationParametersBasedonScopeoftheMethod

7.1. Thescopeofforensictoxicologymethodsistypicallycategorizedasscreeningmethods,qualitativeconfirmation/identificationmethods,orquantitativemethods.Assuchthefollowingvalidationparametersshallbeevaluated.

7.2. Screening(Immunoassay‐based):

a) limitofdetection;

b) precision(atthedecisionpoint);

c) dilutionintegrity(ifapplicable);

d) stability(ifapplicable).

7.3. Screening(allothers):

a) interferencestudies;

b) limitofdetection;

c) ionizationsuppression/enhancement(forapplicabletechniques,suchasLC/MS);

d) dilutionintegrity(ifapplicable);

e) stability(ifapplicable).

7.4. Qualitativeconfirmation/identification:

a) carryover;

b) interferencestudies;

c) ionizationsuppression/enhancement[forapplicabletechniques,suchasliquidchromatography/massspectrometry(LC/MS)];


5

d) limitofdetection;

e) dilutionintegrity(ifapplicable);

f) stability(ifapplicable).

7.5. Quantitativeanalysis:

a) bias;

b) calibrationmodel;

c) carryover;

d) interferencestudies;

e) ionizationsuppression/enhancement(forapplicabletechniques,suchasLC/MS);

f) limitofdetection;

g) limitofquantitation;

h) precision;

i) dilutionintegrity(ifapplicable);

j) stability(ifapplicable).

8. SpecificRequirementsforConductingMethodValidationExperiments

8.1. General

Allvalidationexperimentsshallbeconductedusingfortifiedsamplesofthematrixforwhichthemethodisintended,unlessotherwisenoted.Insomeinstances(e.g.,immunoassayscreens),itmaybemoreappropriatetoanalyzepreviouslycharacterizedhumansamplesinsteadoffortifiedsamplesforselectedmethodvalidationstudies.

Validationstudiesshallbeconductedinamannersimilartocasework.Thismayincludeconductingvalidationstudiesondifferentdays,bydifferentanalysts,onallidenticalinstrumentstobeutilizedfortheassay,andensuringthattheinstrumentsmeetthesamedailyperformancerequirementsasforcasework.

Wheneverpossible,fortifiedmatrixsamplesshallbepreparedfromreferencematerialsthatarefromadifferentsource(e.g.,supplierorlotnumber)thanusedtopreparecalibrationsamples.Ininstanceswherethesamesourceshallbeutilized,separateweighingsorsolutionsshallbeusedtopreparethesesamples.

Thefollowingrequirementsaretheminimumforassessingthelistedvalidationparametersinforensictoxicologymethods.Theyarelistedalphabeticallyandnotnecessarilyinproceduralorder.SomeofthevalidationexperimentsaredemonstratedinAnnexBandAnnexC.Section11providesguidanceonhowtoefficientlyperformvalidationexperiments.


6

8.2. BiasandPrecision1

8.2.1. Bias

Biasstudiesshallbecarriedoutforallquantitativemethods.Thesecanbeconductedconcurrentlywithprecisionstudies.Biasshallbemeasuredinpooledfortifiedmatrixsamplesusingaminimumofthreeseparatesamplesperconcentrationatthreedifferentconcentrationpools(low,mediumandhigh2)overfivedifferentruns.Thebiasshallbecalculatedforeachconcentrationusingthefollowingformula:

%

Themaximumacceptablebiasis±20%ateachconcentration.Forsomeanalyseswherelessbiasisrequired(e.g.,ethanol),abiasof±10%orbetterisexpected.Itisrecommendedthatthesamedatausedinbiasstudiesalsobeusedforprecisioncalculations.

8.2.2. Precision

8.2.2.1. General

Precisionstudiesshallbecarriedoutforallquantitativemethods,aswellasatthedecisionpointforimmunoassays.Thesestudiescanbecarriedoutconcurrentlywithbiasstudies,ifrequiredinthevalidationplan.

Precisionisexpressedasthecoefficientofvariation(%CV).Themeanandstandarddeviation(s)oftheresponseiscalculatedforeachconcentrationtodeterminethe%CV.

%

8.2.2.2. PrecisionofImmunoassaysatDecisionPoint

Forimmunoassaysthatcross‐reactwithabroadclassofcompounds(e.g.,benzodiazepines,opiates,amphetamines),ifalaboratorydeclarestotheircustomersthattheyareabletodetectanalyteswithlowcross‐reactivity(lessthanorequaltothetargetanalyte)usingtheimmunoassay,itisessentialtoverifytheirabilitytodetectthesecompounds.Forexample,abenzodiazepineimmunoassaytargetedforoxazepamtypicallyhaslowcross‐reactivitiestootherbenzodiazepinessuchaslorazepam.Ifthelaboratoryusesthisimmunoassaykittoscreenforlorazepam,theyarerequiredtoevaluatetheassay’sabilitytoreliablydetectlorazepam,inadditiontooxazepam.Incontrast,ifalprazolamhasgreatercross‐reactivitythanoxazepam,thereisnorequirementtoevaluatetheabilitytodetectalprazolamprovidedthatthedecisionconcentrationforalprazolamisnotlowerthanthedecisionconcentrationforoxazepam.Thisevaluationmayrequireanadjustment

1Anaccuratemeasurementisonewithacceptablebiasandprecision.2Forpurposesofthisdocument,lowconcentrationsshallbenomorethanapproximately3timesthelowestendoftheworkingrangeofthemethodandhighconcentrationsshallbewithinapproximately80%(ormore)ofthehighestendoftheworkingrangeofthemethod,unlessotherwisenoted.Mediumconcentrationsshallbenearthemidpointofthelowandhighconcentrations.


7

orreevaluationofthedecisionpointorthetargetcompounddependingontheneedsandmissionofthelaboratory.

Ataminimum,precisionatthedecisionpointshallbeassessedusingthreeseparatesamplesfromthreedifferentconcentrationpoolsoverfivedifferentruns:

a) generallynomorethan50%belowdecisionpoint,

b) atdecisionpoint,and

c) generallynomorethan50%forurineand100%forallothermatricesabovedecisionpoint.

Immunoassaysarematrixdependentandtheconcentrationrangearoundthedecisionpointmaybewiderformorecomplexmatricesincomparisontourine.

EnzymeLinkedImmunosorbentAssays(ELISA):Thedifferencebetweentheabsorbanceofanegativesample(Bo)andtheabsorbanceofaspecimen(B)shouldbemeasuredasapercentage:[B/Bo]x100andnotanabsolutevalue.

Liquidreagentassays(e.g.enzymemultipliedimmunoassaytechnique[EMIT],clonedenzymedonorimmunoassay[CEDIA®3],etc.):Theabsorbancevaluecanbeuseddirectly.

a) Runeachconcentration3timesinfiveseparateruns.

b) Calculatethegrandmean(n=15)andrelatedgrandstandarddeviationateachconcentration.

c) The%CVshallnotexceed20%ateachconcentrationusingall15sampleresultsperconcentration.

d) Thegrandmeanplusorminustwostandarddeviations(stddev)foreachconcentrationshallnotoverlapforthedecisionpointtobevalid.

ItshouldbenotedthatthedataobtainedfromthesestudiesarealsousedtoestimatetheLODforimmunoassays.

8.2.2.3. PrecisionofQuantitativeProcedures

8.2.2.3.1. General

Forquantitativeprocedures,twodifferenttypesofprecisionstudiesshallbeassessedduringmethodvalidation:within‐runprecisionandbetween‐runprecision.Ataminimum,precisionshallbeassessedusingthreedifferentsamplesperconcentrationatthreedifferentconcentrationpools(low,mediumandhigh)overfivedifferentruns.Thedifferentrunsusedtoevaluateprecisionmaybeperformedwithinthesameday,providedadifferentcalibrationcurveisusedforeachrun.

The%CVshallnotexceed20%ateachconcentration.Itisnotedthatcertainanalyticalmethods(e.g.,bloodalcoholanalysis)mayrequireamuchlowercoefficientofvariation(≤10%CV).

3Thistermisusedasanexampleonly,anddoesnotconstituteanendorsementofthisproductbytheAAFSStandardsBoard.


8

8.2.2.3.2. Within‐RunPrecisionCalculations

Within‐runprecisionsarecalculatedforeachconcentrationseparatelyforeachofthefiveruns.Within‐runprecisionmaybecalculatedusingthedatafromeachrun’striplicateanalysesateachconcentrationas:

%stddeviation

100

Thelargestcalculatedwithin‐run%CVforeachconcentrationwillbeusedtoassesswithin‐runprecisionacceptability.

8.2.2.3.3. Between‐RunPrecisionCalculations

Between‐runprecisioniscalculatedforeachconcentrationoverthefiveruns.Thismaybedonebyusingthecombineddatafromallreplicatesofeachconcentrationas:

%

100

8.2.2.3.4. One‐WayAnalysisofVariance(ANOVA)ApproachtoCalculateCombinedWithin‐RunandBetween‐RunPrecision

Bothwithin‐runandbetween‐runprecisionsmaybecalculatedusingtheone‐wayANOVAapproachwiththevariedfactor(runnumber)asthegroupingvariable.TheANOVAcalculationscanbeeasilyperformedusingaspreadsheetorastatisticalsoftwareprogram.

Usingthisapproach,within‐runprecisionsarecalculatedforeachconcentrationas:

%

100

whereMSwgisthemeansquarewithingroupsobtainedfromtheANOVAtable.

Likewise,between‐runprecisionsarecalculatedas:

%

1 ∗

100

whereMSbgisthemeansquarebetweengroupsobtainedfromtheANOVAtableandnisthenumberofobservationsineachgroup(e.g.,n=3ifdoingtriplicateanalyses).AnnexBprovidesanexampleofhowtheANOVAapproachmaybeusedtocalculatewithin‐runandbetween‐runprecision.

8.3. CalibrationModel

Thecalibrationmodelshallbedeterminedforallquantitativemethods.Thisisaccomplishedbyfirstdeterminingtherangeofanalyteconcentrationsoverwhichthemethodshallbeused,


9

sometimescalledtheworkingrange.Withinthisrange,therewillbeacorrelationbetweensignalresponse(e.g.,peakarearatioofanalyteandinternalstandard)andanalyteconcentrationinthesample.Thecalibrationmodelisthemathematicalmodelthatdescribesthiscorrelation.Thechoiceofanappropriatemodel(i.e.,linearorquadratic)isnecessaryforaccurateandreliablequantitativeresults.

Calibratorsamplesareanalyzedtoestablishthecalibrationmodel.Theuseofmatrix‐matchedcalibratorsamplesisencouraged,butnotrequired.Regardlessofthematrixusedtopreparecalibratorsamples,alaboratoryshalldemonstrateacceptablebiasandprecisionwithcontrolsamplespreparedinallmatricesintendedtobeanalyzedbythemethod(seeSection7.2).Forexample,bloodalcoholmethodsmaydemonstrateacceptablebiasandprecisioninwholebloodcontrolsusingaqueouscalibratorsamples.Likewise,acceptablebiasandprecisionmaybedemonstratedusingcalibratorsamplespreparedinwholebloodbutusedtoquantitateanalytesindifferentmatrices(e.g.,postmortemtissues,serum,urine).

Thecalibratorsamplesshallspantherangeofconcentrationsexpected.Atleastsixdifferentnon‐zeroconcentrationsshallbeusedtoestablishthecalibrationmodel.Theconcentrationsshallbeappropriatelyspacedacrossthecalibrationrangetoestablishthemostappropriatecalibrationmodel.Aminimumoffivereplicatesperconcentrationisrequired.Thereplicatestoestablishthecalibrationmodelshallbeinseparateruns.Alldatapointsfromthefiverunsshallbeplottedtogether(usingastatisticalsoftwarepackage)toestablishthecalibrationmodel.Theoriginshallnotbeincludedasacalibrationpoint.

Themostoftenusedcalibrationmodelisthesimplelinearregressionmodelusingtheleastsquaresmethod.However,thismodelisonlyapplicablewhenthereisconstantvarianceovertheentireconcentrationrange.Whenthereisanotabledifferencebetweenvariancesatthelowestandhighestconcentrations,aweightedleastsquaresmodelorotherappropriatenon‐linearmodelshallbeapplied.4Thisisgenerallythecasewhentheconcentrationrangeexceedsoneorderofmagnitude.Ultimately,thebestapproachistousethesimplestcalibrationmodelthatbestfitstheconcentration‐responserelationship.

Althoughithasbecomewidespreadpractice,itisemphasizedthatacalibrationmodelcannotbeevaluatedsimplyviaitscorrelationcoefficient(r).Instead,acalibrationmodelshallbevisuallyevaluatedusingstandardizedresidualplots.Theseallowonetocheckforoutliersthatshallbeeliminatediffoundtobestatisticallysignificant(e.g.,outside±3standarddeviations).Further,residualplotsallowonetodetermineifthevariancesappeartobeequalacrossthecalibrationrangewithasimilardegreeofscatterateachconcentration.Theyalsogiveanindicationifthechosenmodeladequatelyfitsthedata.Forexample,randomdistributionofindividualresidualsaroundthezeroline(homoscedasticity)suggeststhatalinearmodelisappropriate.

Finally,thereareotherappropriatealternativestoevaluatecalibrationmodels(i.e.,ANOVAlack‐of‐fittestforunweightedlinearmodels,checkingforsignificanceofthesecondorderterminquadraticmodels,assessmentofcoefficientofdeterminationforlinearmodels).

Ifalinearcalibrationmodelhasbeenestablished,fewercalibrationsamples(i.e.,fewerlevelsorsingle/fewerreplicates)maybeusedforroutineanalysis.However,iffewercalibrationsamplesarechosen,thesamecalibrators(e.g.,number,replicates,andconcentrationlevel)shallbeusedtoconstructthecalibrationcurvesusedforthebiasandprecisionstudies.Furtherthecalibrationdata

4Ingeneral,non‐linearmodelsmayrequireadditionalcalibratorstoaccuratelycharacterizethecurve.


10

shallincludethelowestandhighestcalibrationlevelsusedtoestablishthemodel,aswellasincludenofewerthanfournon‐zerocalibrationpoints.

Additionally,oncethecalibrationmodelisestablishedforavalidatedmethod,itshallnotbearbitrarilychangedtoachieveacceptableresultsduringagivenanalyticalrun.Forexample,oneshallnotswitchfromanunweightedlinearmodeltoaweightedlinearmodelinordertoadjustforchangesininstrumentperformance.

8.4. Carryover

Analytecarryoverintoasubsequentsamplemayleadtoaninaccuratequalitativeorquantitativeresultwhenusinginstrumentalmethods.Carryovershallbeevaluatedduringmethodvalidationintendedforconfirmationand/orquantitation,unlessalaboratoryisconstantlyaddressingcarryoverintheirqualityassurance(QA)/qualitycontrol(QC)practices.

Toevaluatecarryoveraspartofmethodvalidation,blankmatrixsamplesareanalyzedimmediatelyafterahighconcentrationsampleorreferencematerial.Thehighestfortifiedconcentrationatwhichnoanalytecarryoverisobserved(abovethemethod'sLOD)intheblankmatrixsampleisdeterminedtobetheconcentrationatwhichthemethodisfreefromcarryover.Thiscarryoverconcentrationforeachanalyteinthemethodshallbeconfirmedusingtriplicateanalyses.Itisacceptabletolimitthecarryoverstudytothehighestpointofyourcalibrationcurveinquantitativeassays.

Ifpossible,theanalyticalprocedurewillbemodifiedtoremoveanycarryover.Incaseswhenitisnotpossibletoeliminatethecarryover,theSOPshalladdresshowcarryoverwillbemanaged.

8.5. InterferenceStudies

8.5.1. General

Interferingsubstancesfromcommonsourcesshallbeevaluatedinallscreening(exceptimmunoassays),qualitativeidentification,andquantitativemethods.

8.5.2. EvaluatingMatrixInterferences

Wheneverpossible,blankmatrixsamplesfromaminimumoftendifferentsourceswithouttheadditionofaninternalstandard(whenusedinthemethod)shallbeanalyzedtodemonstratetheabsenceofcommoninterferencesfromthematrix.Whilethisapproachmaydetectthemorecommonmatrixinterferences,itisrecognizedthatlesscommoninterferencesmaynotbedetected.

8.5.3. EvaluatingInterferencesfromStable‐IsotopeInternalStandards

Formethodsemployingstableisotopeinternalstandards,theisotopically‐labeledcompoundsmaycontainthenon‐labeledcompoundasanimpurity.Additionally,themassspectraofthelabeledanalogsmaycontainfragmentionswiththesamemass‐to‐chargeratiosasthesignificantionsofthetargetanalyte.Inbothinstances,analyteidentificationorquantitationcouldbeimpacted.

Stable‐isotopeinternalstandardinterferencesshallbeassessedbyanalyzingablankmatrixsamplefortifiedwiththeinternalstandardandmonitoringthesignaloftheanalyte(s)ofinterest.InterferencesbelowtheLODoftheassaymaybeinsignificantdependingonthelaboratory’smission.


11

Likewise,ablankmatrixsamplefortifiedwiththeanalyte(s)attheupperlimitofthecalibrationrangeshallbeanalyzedwithoutinternalstandardtoevaluatewhetherrelevantamountsoftheunlabeledanalyteionsappearasisotopically‐labeledcompoundfragmentswhichcouldimpactquantitation.

8.5.4. EvaluatingInterferencesfromOtherCommonlyEncounteredAnalytes

Forallmethodsotherthanimmunoassays,itisnecessarytoevaluateotheranalyteswhichmaybeexpectedtobepresentincasesamplesfortheirpotentialtointerferewiththemethod’sanalytes.Forexample,amethoddevelopedtoanalyzebloodforcocaineshallevaluatewhetherothercommondrugsofabuse,metabolites,andstructurally‐similarcompoundsinterferewiththeassay.Likewise,aheadspacegaschromatograph‐flameionizationdetection(GC‐FID)methoddevelopedforethanolshallevaluatewhetherothercommonvolatileorganiccompoundsinterferewiththeassay.

Thisevaluationisaccomplishedbyanalyzingfortifiedmatrixsamples,previouslyanalyzedcasesamples,orneatreferencematerialsofthepotentialinterference(s)athightherapeuticorlethalconcentrations,dependingontheanalyte,thematrix,andthelaboratory’smission.Themostcommondrugs/metabolitesencounteredinthelaboratoryshallbeincludedintheevaluationtogetherwithothercommondrugswithintheclassification,whereappropriate.

8.6. IonizationSuppression/Enhancement

8.6.1. General

Theenhancementorsuppressionofanalyteionizationresultingfromthepresenceofco‐elutingcompoundsisaphenomenoncommonlyencounteredinliquidchromatography/massspectrometry(LC‐MS)applications.

Whenaveragesuppressionorenhancementoftheanalyte’stargetionoriontransitionandqualifyingions,ifapplicable,exceeds±25%orthe%CVofthesuppressionorenhancementexceeds20%,alaboratoryshalldemonstratethatthereisnoimpactonothercriticalvalidationparameters.Forexample,suppressionorenhancementofionizationismostlikelytoimpactthelimitofdetectionofaqualitativemethod.Likewise,thelimitofdetection,thelimitofquantitation,andbiasmaybeaffectedbyionizationsuppressionorenhancementinquantitativemethods.Theinfluenceontheaboveparametersshallbeassessedbyincreasingthenumberofdifferentsourcesofblankmatricesusedintheirevaluation.Forexample,iftheaveragesuppressionorenhancementexceeds±25%,theLODdeterminationshallbeperformedinmorethanthreeblankmatrices.

Laboratoriesshallalsoassesstheimpactofionizationsuppressionorenhancementonthemethod’sinternalstandards.

Ionizationsuppression/enhancementshallbeevaluatedusingeitheroftheapproachesthatfollow.

8.6.2. Post‐columnInfusiontoAssessIonizationSuppression/Enhancement

Thisapproachprovidesinformationonretentiontimeswhereionizationsuppression/enhancementoccurs.Itisusefulformethoddevelopment,aswellastoassesstheamountofionizationsuppressionorenhancementforLC‐MSbasedconfirmationmethods.Solutionsatbothlowandhighconcentrationsoftheanalyteareindividuallyinfusedwithasyringepumpintotheeluentfromthecolumnviaapost‐column“T”‐connectorandaconstantbaseline


12

signalfortheanalyteofinterestismonitored.Wheneverpossible,aminimumoftendifferentprocessedblankmatrixsamplesthatarerepresentativeofthequalityofsamplestypicallyencounteredincaseworkareinjectedintotheLC‐MSduringinfusionofthesolutions.5Ifthereisanyconsiderablesuppressionorenhancement(>25%)oftheinfusedanalytesignalattheretentiontimeoftheanalyte,thenmodificationofthechromatographicsystemorthesamplepreparationmayberequiredtominimizetheeffectofionizationsuppressionorenhancement.

8.6.3. Post‐ExtractionAdditionApproachtoAssessIonizationSuppression/Enhancement

Thisapproachyieldsaquantitativeestimationofionizationsuppression/enhancement.ItisusefulforassessingtheamountofionizationsuppressionorenhancementforLC‐MSbasedquantitativemethods.Twodifferentsetsofsamplesarepreparedandtheanalytepeakareasofneatstandardsarecomparedtomatrixsamplesfortifiedwithneatstandardsafterextractionorprocessing.

Setoneconsistsofneatstandardspreparedattwoconcentrations–onelowandonehigh.Eachoftheseneatstandardsisinjectedaminimumofsixtimestoestablishameanpeakareaforeachconcentration.

Settwoconsistsofaminimumoftendifferentmatrixsources,wheneverpossible.6Eachmatrixsourceisextractedinduplicate.Aftertheextractioniscomplete,eachmatrixsampleisthenreconstituted/fortifiedwitheithertheloworhighconcentrationneatstandard.

Theaverageareaofeachset(X )isusedtoestimatethesuppression/enhancementeffectateachconcentrationasfollows:

%X 2

X 11 100

Twoionizationsuppressionorenhancementpercentageswillbeestablished–oneatthelowconcentrationandoneatthehighconcentration.

8.7. LimitofDetection

8.7.1. General

Limitofdetection(LOD)studiesshallbecarriedoutforallmethods.ThereareanumberofdifferentapproachesfordeterminingtheLOD.Selecttheapproachthatprovidesthemostreasonableestimationofthedetectionlimitgiventheanalyticalinstrumentation(orlackthereof)utilizedinthemethod.

Amethod'sLODincorporatesinstrumentalperformance,aswellasthesamplematrixandinherentprocedurallimitations.Therefore,theLODshallbeassessedovermultiplerunsusingfortifiedmatrixsamplesfromatleastthree(3)differentsourcesofblankmatrix,unlessotherwiseindicatedbelow.Further,whenpossible,itisnecessarytoensurethedefinedLODstillsatisfiesthenecessary

5Additionalmatrixsamplesmayberequiredinpostmortemtoxicologygiventhevarietyofsampleconditionstypicallyencounteredinthiswork.6Additionalmatrixsamplesmayberequiredinpostmortemtoxicologygiventhevarietyofsampleconditionstypicallyencounteredinthiswork.


13

parametersforidentification.Forexample,matchingofamassspectrumtoareferencespectrumwithinanacceptablematchfactorcanonlybeachievedbyexperimentaldeterminationofLODratherthantheoreticalcalculation.

TheLODshallbedeterminedbyoneofthefollowingapproaches.

8.7.2. EstimatingLODforaNon‐InstrumentalMethod

Thisapproachismostoftenusedwhenscreeningforthepresenceorabsenceofaspecifiedanalyteorclassofanalytes(e.g.,colortests).ToestimatetheLODforavisual,non‐instrumentalmethod,samplesfortifiedwithdecreasingconcentrationsofanalyteareanalyzedoveraminimumofthreeruns.Whenpossible,multipleanalystsshouldbeinvolvedinestimatingtheLODusingthisapproach.ThelowestconcentrationofanalytethatyieldsapositiveresultonallrunsisconsideredtheLOD.

8.7.3. UsingtheLowestNon‐ZeroCalibratorastheLOD

Thistechniqueisusefulforquantitativemethods.Insomeinstances,itmaybesufficienttodefinetheLODasthevalueofthelowestnon‐zerocalibrator.Aminimumofthreesamplesperrunofthelowestcalibratorshalleachbeanalyzedoverthreerunstodemonstratethatalldetectionandidentificationcriteriaaremet.Ifdesired,itisacceptabletousethesamecalibratorreplicatesusedtoestablishthecalibrationmodel(Section8.3)forsomeofthesamplesusedforthismethod,butadditionalsamples/replicateswillbeneededtomeettheminimumofninedatapoints.

8.7.4. UsingtheDecisionPointConcentrationastheLOD

Thistechniqueisusefulforqualitativeandquantitativemethods.Insomeinstances,itmaybesufficienttodefinetheLODasthevalueofanadministratively‐defineddecisionpoint.Forexample,alaboratorymaychoosetodefineamethod’sLODforethanolas0.02g/dLforbloodbasedonthelaboratory’sadministrativelydefineddecisionpointforreportingthisanalyte,eventhoughalowerLODisanalyticallyachievable.Likewiseforanimmunoassay,alaboratorymaychoosetousethedecisionpointconcentration[thathasdemonstratedappropriateprecision(Section8.2.2.2)]astheassay’sLOD.Aminimumofthreesamplesperrunofafortifiedmatrixsampleattheconcentrationofthedecisionpointshallbeanalyzedoverthreerunstodemonstratethatalldetectionandidentificationcriteriaaremet.ThedatageneratedfortheprecisionatthedecisionpointconcentrationinimmunoassayswillsufficeforthisLODrequirement.

8.7.5. EstimatingLODUsingBackgroundNoise

8.7.5.1. General

TheseapproachesfordeterminingLODareonlyusefulforinstrumentalmethodsthatdemonstratebackgroundnoise.Aminimumofthreedifferentblanksourcematricesshallbeused.Forexample,iftheassayistobeusedforpostmortembloodsamples,threeindependentrepresentativepostmortembloodsourcesareneeded.

8.7.5.2. EstimatingLODUsingReferenceMaterials

Three(ormore)sourcesofblankmatrixsamplesfortifiedatdecreasingconcentrationsareanalyzedinduplicate(twoseparatesamples)foratleastthreeruns.TheLODisconsideredthelowestconcentrationthat1)yieldsareproducibleinstrumentresponsegreaterthanorequalto3.3


14

times7thenoiselevelofthebackgroundsignalfromthenegativesamples,and2)achievesacceptablepredefineddetectionandidentificationcriteria(e.g.,retentiontime,peakshape,massspectralionratios).

Whileitmaybepossibletovisuallyassessthesignaltonoiseratio,suchanapproachissubjective.Thereforecalculatethesignal‐to‐noiseratiooruseinstrumentationsoftwaretodeterminetheratio.Ifmanuallycalculated,thesignalisdefinedastheheightresponseoftheanalytepeakandthenoiseisdefinedastheamplitudebetweenthehighestandlowestpointofthebaselineinanareaaroundtheanalytepeak.Eachreplicateshallbeindependentlyevaluated.

‐ ‐

8.7.5.3. EstimatingLODUsingStatisticalAnalysisofBackground

TodeterminetheLODusingthisapproach,aminimumofthreesourcesofblankmatrixareanalyzedinduplicate(twoseparatesamples)overatleastthreeruns.Theaverageandstandarddeviation(sblank)ofthesignal(e.g.,integratedareaofsignalattheanalyte’sretentiontime)fromallblankmatrixsamplesiscalculated.Likewise,fortifiedmatrixsamplesofdecreasingconcentrationareanalyzedinduplicateoverthecourseofatleastthreeruns.Thelowestconcentrationofafortifiedmatrixsamplethatconsistentlyyieldsasignalgreaterthantheaveragesignaloftheblank

matrixsamples(X blank)plus3.3timesthestandarddeviationisidentifiedastheLOD:

LOD=X blank+3.3sblank

8.7.6. EstimatingLODUsingaLinearCalibrationCurve

Thistechniqueisusefulforanyquantitativemethodthatfollowsalinearcalibrationmodel.Aminimumofthreeindependentcalibrationcurvesareconstructedacrosstheworkingrangeoftheanalyticalmethodoverdifferentruns.TheLODcanbeestimatedfromthestandarddeviationoftheyintercept(sy)andtheaverageslope(Avgm)as:

LOD=(3.3sy)/Avgm

8.8. LowerLimitofQuantitation

8.8.1. General

Lowerlimitofquantitation(LLOQ)studiesshallbecarriedoutforallquantitativemethods.Thereareanumberofdifferentapproachesfordeterminingamethod'sLLOQ.Selecttheapproachthatprovidesthemostreasonableestimationofthequantitationlimitgiventheanalyticalinstrumentationutilizedinthemethod.Amethod'sLLOQincorporatesinstrumentalperformance,aswellasthesamplematrixandinherentprocedurallimitations.TheLLOQshallbeassessedovermultiplerunsusingfortified,blankmatrixsamplesfromatleastthreedifferentsourcesofblankmatrix,unlessotherwiseindicatedbelow.

7Useof3.3intheLODcalculationprovidesafalsepositiveerrorrateof0.0005%.(seeBoyd,RobertK.,CeciliaBasic,andRobertA.Bethem.TraceQuantitativeAnalysisbyMassSpectrometry.Hoboken,N.J.:JohnWiley(2008).


15

8.8.2. UsingtheLowestNon‐ZeroCalibratorastheLLOQ

Insomeinstances,itmaybesufficienttodefinetheLLOQasthevalueofthelowestnon‐zerocalibrator.Aminimumofthreesamplesperrunofthelowestcalibratorshallbeanalyzedoverthreerunstodemonstratethatalldetection,identification,bias,andprecisioncriteriaaremet.Ifdesired,itisacceptabletousethesamecalibratorreplicatesusedtoestablishthecalibrationmodel(Section8.3)forsomeofthesamplesusedforthismethod,butadditionalsamples/replicateswillbeneededtomeettheminimumofninedatapoints.

8.8.3. UsingDecisionPointConcentrationastheLLOQ

Insomeinstances,itmaybesufficienttodefinetheLLOQasthevalueofanadministratively‐defineddecisionpoint.Forexample,alaboratorymaychoosetodefineamethod’sLLOQforGHBas5mg/Lforantemortembloodbasedonthelaboratory’sadministrativelydefineddecisionpointforreportingthisanalyte,eventhoughalowerLLOQisanalyticallyachievable.Theconcentrationsusedforthisapproachshallremainwithinthepreviouslyestablishedcalibrationcurve.Aminimumofthreesamplesperrunofafortifiedmatrixsampleattheconcentrationofthedecisionpointshallbeanalyzedoverthreerunstodemonstratethatalldetection,identification,bias,andprecisioncriteriaaremet.

8.8.4. EstimatingLLOQUsingReferenceMaterials

Three(ormore)sourcesofblankmatrixsamplesfortifiedatdecreasingconcentrationsareanalyzedinduplicate(twoseparatesamples)overaminimumofthreeruns.Theconcentrationsusedforthisapproachmaybelowerthanthepreviouslyestablishedcalibrationcurve,ifyoudesiretoreportquantitativevaluesunderthelowestpointofyourcalibrationcurve.Inthiscase,biasandprecisionbelowthelowestcalibratorshallbeestablishedwithqualitycontrolsampleswithineachanalyticalbatch.Thelowestconcentrationthatiscapableofachievingacceptabledetection,identification,bias,andprecisioncriteriainallthreefortifiedsamplesisconsideredtheestimatedLLOQ.

9. AdditionalValidationParameters

9.1. General

Incertaininstances,itisimportanttoevaluateadditionalvalidationparameters,ifapplicable.Theseincludeanalytestabilitywhenthematrixisfrozenandthawed,processedsamplestabilityandtheeffectofsampledilutiononbiasandprecision.Alaboratoryshallincludetheseparametersintheirvalidationplan,anddetermineiftheyareapplicabletotheanalyticalmethodoriftheyarealreadyaddressedthroughothermeans(i.e.,qualityassurancepractices,publishedreferences).Thelaboratoryvalidationplanshallincludedocumentationofthisevaluation.

9.2. DilutionIntegrity

Theeffectofsampledilutionshallbedeterminedduringvalidationofquantitativemethodsifthisisaroutinepracticewithinthelaboratory.Attimes,thismaybeduetolowspecimenvolumerequiringthesampleorassaytobeadjustedappropriately.Inotherinstances,excessivelyhighconcentrationsmaybeencounteredthatareabovetheestablishedcalibrationrange.Tobringtheanalyteconcentrationwithinthevalidatedconcentrationrange,thelaboratoryproceduremayallowforreanalysisafterdilutionofthesample.


16

Ifdilutionofasampleisallowedbecauseofhighanalyteconcentrationorlowsamplevolume,thenthelaboratoryshallevaluatetheeffectofdilutiononthemethod'sbiasandprecision.Thisisaccomplishedbyrepeatingbiasandwithin‐runprecisionstudiesatcommondilutionratios(e.g.,1:2,1:10,1:50)utilizedbythelaboratoryanddeterminingifperformancecriteriaarestillmet.

9.3. Stability

9.3.1. General

Analytestabilitymaybeaffectedbyanumberofvariablesincludingstorageconditionsandsampleprocessing.Stabilityexperimentsshallbedesignedandcarriedouttoaddresssituationsnormallyencounteredinlaboratoryoperations,unlessanalytestabilityisalreadyaddressedthroughothermeans(i.e.,qualityassurancepractices,publishedreferences).Allstabilitydeterminationsshallincludeasetofsamplespreparedfromreferencematerials.Thereferencematerialsareusedtopreparefortifiedsamplesoftheanalyte(s)atbothlowandhighconcentrationsineachmatrixthatwillbeanalyzedinthemethod.Itisimportantthatalargeenoughvolumeofeachofthesefortifiedsamplesispreparedinordertocompletethestudiesusedinthesectionsbelow.Thesefortifiedsamplesshallinitiallybeanalyzedintriplicatetoestablishtimezeroresponses.Theaveragetimezeroresponseforeachsetofsamplesiscomparedtotheaveragesignalsfromeachofthefollowingstabilitystudies.Linearregressionoftheaveragesignal(e.g.,peakareas8orratiosofpeakareaofanalytetointernalstandard)versustimewillallowforanassessmentoftrends.Alternatively,concentrationsateachtimeintervalcouldbemonitoredprovidedtheconcentrationdeterminationisbasedonthetimezerocalibrationcurve.

9.3.2. Stability–Freeze/Thaw9

Ifitispartofalaboratory’sstandardpracticetofreezesamplespriortoanalysisandtherearenopublisheddatatorelyupon,analytestabilityshallbedeterminedafterthreefreezeandthawcycles.Theabovefortifiedsamples(Section8.3)arealiquotedintoaminimumofthreeseparatestoragecontainersperconcentrationandthenfrozenattheintendedstoragetemperaturefor24hours.Thisisfollowedbyanunassistedthawatroomtemperature.Whencompletelythawed,thefirstsetofsamplesshallbeanalyzedintriplicate,whiletheothersarerefrozenfor12to24hoursunderthesameconditions.Thefreeze/thawcycleandanalysisshallberepeatedtwomoretimes.Theanalyteshallbeconsideredasstableuntiltheaveragesignal(e.g.,peakareaorratiosofpeakareaofanalytetointernalstandard)comparedtothetimezeroaveragesignalfallsoutsideofthemethod’sacceptablebias.Forexample,ifthemethodbiasis±10%andthetimezeroaveragesignalis100,000,theanalyteisconsideredstableuntiltheaveragesignalfallsoutsideofthe90,000–110,000range.

9.3.3. Stability–ProcessedSample

Circumstancesmayariseinwhichsamplesthathaveundergoneroutinepreparationforinstrumentalanalysiscannotbeimmediatelyanalyzed.Itmaybenecessarytorunthesamplethefollowingdayorlater.Intheseinstances,itisimportanttoevaluatethelengthoftimeaprocessed

8Whenmonitoringpeakareas,theinstrument’sresponseshouldbeconstantoverseveraldaysforreliableinterpretationofthedata.9Itisrecognizedthatfreeze/thawandstoragestabilitystudiesinsolidsamples(e.g.,hair,tissues,foodproducts)maynotbepossiblebyfortificationduetothenatureofthesematrices.Cautionshouldbeemployedininterpretingresultsofsolidsampleswhenstabilityinformationisnotavailable.


17

samplecanbemaintainedbeforeitundergoesunacceptablechanges,preventingreliableanalytedetection,identification,orquantitation.

TypicallyprocessedsamplesfortifiedperSection8.4arecombinedperconcentrationandthendividedintodifferentautosamplervials.Asindicatedabove,thefirstvialsofeachconcentrationareimmediatelyanalyzedintriplicatetoestablishthetimezeroresponses.Allremainingvialsarestoredinamannerthattheywouldtypicallybestoredduringroutineanalysis(e.g.,refrigerated,atroomtemperatureonautosampler).Theremainingvialsarethenanalyzedintriplicateatdifferenttimeintervals.Averageresponsesateachtimeintervalarecomparedtothetimezeroresponses.Theanalytewillbeconsideredstableuntiltheaveragesignal(e.g.,peakareaorratiosofpeakareaofanalytetointernalstandard)comparedtothetimezeroaveragesignalfallsoutsideofthemethod’sacceptablebias.Forexample,amethod’sbiaslimitis±15%andthetimezeroaveragesignalis100,000.Processedsamplesindifferentautosamplervialsareanalyzedrepeatedlyupto72hours.Theprocessedsample’sanalyteisconsideredstableuntiltheaveragesignalfallsoutsideofthe85,000–115,000range.

10. RequiredRevalidationofPreviouslyValidatedMethods

Modificationstoavalidatedmethodrequireevaluationtoconfirmthatthechangesdonothaveanadverseeffectonthemethod’sperformance.Thedecisionregardingwhichperformancecharacteristicsrequireadditionalvalidationisbasedonconsiderationofthespecificparameterslikelytobeaffectedbythechange(s).Thesechangesmayinclude,butarenotlimitedto:

a) analyticalconditions,

b) instrumentation,

c) sampleprocessing,

d) datasoftware.

Forexample,changesofextractionsolventorbuffermayaffectlinearity,interferences,LLOQ,precision,andbias.Achangeoftheanalyticalcolumnstationaryphaseorachangeinmobilephasecompositionmayaffectlinearityandinterferences.Further,considerationshouldbegiventoconductingparallelstudieswithknownorproficiencysamplesutilizingbothapreviouslyvalidatedmethodandthemodifiedmethodtoevaluatetheeffectsofthechanges.Thegoalistodemonstratetheimpactthechangeshaveontheperformanceofthepreviouslyvalidatedprocedure.

11. DocumentationRequirementsforMethodValidation

Recordkeepingisanessentialpartoflaboratoryoperatingproceduresandisakeycomponentofmethodvalidation.Thedatageneratedduringmethodvalidationstudiesshallbemaintainedandavailableforaudits,reviews,orinspections.Theserecordsshallbeorganizedforeasyretrievalandreview.

Methodvalidationrecordsshallincludeasummaryofthevalidationstudiesconductedandtheirresults.Theformatofthissummaryreportmaybeabriefbulletedreportortablesummaryformattofacilitateaswiftreviewofvalidationstudies.Thesummaryshallminimallyincludethefollowing:

a) scope;


18

b) validationplan;

c) descriptionofalltheparametersevaluated,ifanyoftheparameterswerenotevaluated,thenthereasonshallbestatedorjustified;

d) samplepreparationstepstoincludeconcentrationsandmatrices;

e) rawdataorreferencetowheretherawdataarestored;

f) resultsandcalculations;

g) conclusions;

h) references;

i) documentationofmanagementreviewandapproval.

Itisimportantthatthevalidationrecordscontainspecificdetailsregardingthestudiesconducted,including:

a) individualsinvolvedinthemethodvalidation,

b) specificinstrumentation,

c) dates.

Methodvalidationdocumentationshallalsoincludeacopyofthenewlydevelopedanalyticalmethodorareferencetoitslocation.Further,validationdocumentationshouldberetainedforaminimumof10yearsafterthemethodisretired.

12. EfficiencywithValidation

Keepinmindthatsomevalidationexperimentsmaybeconductedconcurrentlywiththesamefortifiedsamples.AnnexD,E,andFpresentexampleapproachestoassistinstreamliningvalidationexperiments.


19

AnnexA(informative)

FoundationalPrinciples

Customersofforensicscienceserviceprovidersrequirethatthemethodsusedtoanalyzeevidenceisfit forpurpose. Further,customersneedtoknowthatthemethod'slimitationsareunderstoodwhen used under normal operating conditions. This is one ofmany steps toward ensuring thatqualityresultsareprovidedinlegalmatters.


20

AnnexB(informative)

QuantitationofDrugXinBloodValidationExample

Thefollowingisanexampleofsomeofthevalidationstepsoutlinedinthisdocument.Itisnotintendedtoprovidespecificguidanceforanyparticularmethod.

Inthisexample,assumealaboratoryvalidatedaLC/MS/MSmethodforanewopiate,DrugX,inwholeblood.

CreateValidationPlan(Section6)

Beforestartingthevalidationexperiments,thelaboratorypreparedthevalidationplan.Intheplan,theyspecifiedthatanexistingSPEprocedure,alreadyusedfortheextractionofotheropiates,wouldberelieduponforextractingDrugX(Section5).Further,instrumentconditionswerepreviouslyoptimized(Section5),sothoseconditionswerealsolistedintheplan(notshown).Asthisisaquantitativeprocedure,thevalidationparameterslistedinTableB.1—ValidationParameterstobeAssessedwereassessedagainstthelaboratory’spre‐definedacceptancecriteria.

InterferenceStudies(Section8.5)

Ten(10)independentsourcesofblankwholebloodweresecuredfrompreviouslyanalyzedcasestoevaluatematrixinterferences(Section8.5.2).Theblankmatrixsampleswereextractedwithouttheadditionofinternalstandard(d3‐DrugX)andanalyzedusingthenewlydevelopedmethod.NointerferencesattheretentiontimeforDrugXwerenotedafteranalysisoftheblankwholebloodsamples.

Thelaboratoryrandomlyselectedoneoftheblankmatrixsamples,addedd3‐DrugXtothesample(250ng/mL),extractedthesample,andanalyzedit.ThiswastodemonstratethattheinternalstandardwouldnotinterferewiththesignalforDrugX(Section8.5.3).Likewise,anotherrandomblankmatrixsamplewasfortifiedwithDrugXat2000ng/mLandanalyzedwithoutinternalstandard.Thiswastoevaluatewhethertheunlabeledanalyteionsinterferewiththesignalford3‐DrugX.Theresultsdemonstratednointerferencesbetweentheanalyteandinternalstandard.


21

TableB.1—ValidationParameterstobeAssessed

Parameter: AcceptanceCriteria:

Bias Shallnotexceed±20%

CalibrationModel 10–1000ng/mL(linearmodeldesired)

Carryover Carryoverafterhighestcalibratordoesnotexceed10%ofsignaloflowestcalibrator

InterferenceStudies Nointerferingsignalfrommatrix,internalstandard,commondrugsofabuse(includingothercommonopiates/metabolites),OTCdrugs,andprescriptionmedications

IonizationSuppression/Enhancement

<25%suppressionorenhancementand<15%CVduetomatrix(ifnot,evaluateimpactonLOD,LLOQ,andBias)

LimitofDetection Shallbe10ng/mLorlower

LowerLimitofQuantitation Shallbe10ng/mLorlower

Precision %CVshallnotexceed20%

DilutionIntegrity Biasandprecisioncriteriashallbemetwithdilutionofsamples.Dilutionratiosevaluatedwilldependonlinearrangeoffinalcalibrationcurve.

ProcessedSampleStability Evaluatelengthoftimethatanalyteinextractedsamplesstoredatroomtemperatureonautosamplerremainsstable

Lastly,toevaluateinterferencesfromothercommonlyencounteredanalytes(Section8.5.4),thelaboratoryinjectedneatsolutionsdilutedinmobilephasetoaconcentrationof5000ng/mL(orhigher)ofallcommonopiatesandmetabolitesobservedintheircasework,othercommonrecreationaldrugsofabuseandtheirmetabolites,othercommonprescriptionmedicationsandtheirmetabolites,andcommonover‐the‐countermedicationsandtheirmetabolites.TableB.2showshowthelaboratoryefficientlypreparedthesesolutionsintofourinjectionstandards.ThelaboratoryobservednointerferenceforthesignalofDrugXord3‐DrugXfromanyofthesecompounds.


22

TableB.2—ExampleDrugs/MetabolitesUsedinInterferenceStudy

InjectionMix IncludedDrugs/Metabolites(5000ng/mLunlessnotedotherwise)

OpiatesandRelated codeine,morphine,heroin,6‐acetylmorphine,hydrocodone,hydromorphone,oxycodone,oxymorphone,levorphanol,meperidine,methadone,tramadol,fentanyl

DrugsofAbuse amphetamine,cocaine,benzoylecgonine,ecgoninemethylester,methamphetamine,PCP,MDA,MDMA,THC,THC‐COOH

PrescriptionDrugs antidepressants(amitriptyline,imipramine,doxepin,amoxapine,trazodone,bupropion,fluoxetine,sertraline,citalopram),benzodiazepines(alprazolam,chlordiazepoxide,clonazepam,clorazepate,diazepam),antiarrhythmics(verapamil,diltiazem,lidocaine),barbituratesat500,000ng/mL(amobarbital,butalbital,pentobarbital,phenobarbital),otherCNSdepressants(zopiclone,buspirone,zolpidem)

OTCDrugs antihistamines(diphenhydramine,doxylamine,chlorpheniramine),analgesicsat500,000ng/mL(acetaminophen,ibuprofen),antitussive(dextromethorphan)

CalibrationModel(Section8.3)andCarryover(Section8.4)

Thelaboratoryindicatedadesireforthemethod’scalibrationmodeltobelinearandincludetherangeof10–1000ng/mL.However,toevaluateifthemethodcouldexceedthisrange,thecalibrationsampleswerepreparedinblankbloodattheconcentrationsof10,20,50,100,250,500,1000,1500,and2000ng/mL.Eachcalibratorwasanalyzedonceperruninfiveseparateruns(TableB.3).Anextractedmatrixblankwasanalyzedaftereachcalibratortoevaluatecarryoverateachconcentration.Thedataofallrunswerecombinedintoasinglecalibrationcurve.ItwasnotedthatcarryoverwasnotpresentforDrugXortheinternalstandardinanyoftheextractedblankmatrixsamplesthatfollowedthecalibratorsintherangeof10–1500ng/mL;however,asmallamountofcarryoverforDrugXwasobservedintwoofthefiveblankmatrixsamplesthatfollowedthe2000ng/mLcalibrator.TheintegratedareasofDrugXinthesetwosampleswerelessthan10%ofthesmallestareaofthelowest(10ng/mL)calibrator,sothecarryoverfromthe2000ng/mLcalibratorwasdeemedacceptable.

Thefirstevaluationofthesedatasuggestedthatlinearitymaybreakoffabove1000ng/mL(TableB.3andFigureB.1).Aresidualplotwasusedtofurtherevaluatethesedata(FigureB.2).

TableB.3—CalibrationCurveData


23

FigureB.1—CombinedCalibrationCurveDemonstratingLossofLinearityAbove1000G/Ml

TheresidualplotshowedaninvertedU‐shapeddistributionsuggestinganon‐linearmodelwouldbethebestcalibrationmodelforthesedata(FigureB.2).FigureB.2—StandardResidualPlotofCalibrationCurveDatawithanInvertedU‐Shaped

Distribution

Becausethelaboratory’spreferencewastousealinearcalibrationmodel,theyre‐evaluatedthesedataafterdroppingthe1500and2000ng/mLcalibrators.Doingsoallowedfortheiroriginalvalidationplanrequirements(10ng/mL–1000ng/mL)tostillbemet.Therevisedcalibrationcurveappearedtoprovideabetterfitofthesedatausinganunweightedlinearmodel(FigureB.3).Thiswasconfirmedbytheresidualplotthatshowedarandomdistributionaroundthezerolinesuggestingalinearmodelwasthemostappropriateforthesedata(FigureB.4).

R² = 0.9906

0

1

2

3

4

5

6

7

8

0 500 1000 1500 2000

Ratio

Concentration (ng/mL)

‐3

‐2

‐1

0

1

2

3

0 500 1000 1500 2000 2500

Stan

dard Residual



24

FigureB.3—RevisedCalibrationCurve

FigureB.4—ResidualPlotofCalibrationCurvedDatawithaRandomDistribution

Forfuturevalidationexperiments,thelaboratoryusedcalibratorspreparedat10,50,100,250,500,and1000ng/mL.

Sinceaccuratequantitativeresultscannotbeassumedabove1000ng/mL,thelaboratoryknewtheywouldhavetore‐extract(withdilution)anysamplesthatexceed1000ng/mL.Therefore,theyplannedtoevaluatedilutionintegrityinratiosupto1:5whenconductingthebiasandprecisionexperiments.

R² = 0.9989

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

0 200 400 600 800 1000

Ratio

Concentation (ng/mL)

‐3

‐2

‐1

0

1

2

3

0 200 400 600 800 1000

Stan

dard Residual



25

Recallthatnocarryoverwasobservedupto1500ng/mLinthelaboratory’sstudy.However,sincetheworkingcalibrationrangewillendat1000ng/mL,thelaboratoryrecognizedthataccuratequantitativeresultscannotbeachievedabovetheworkingrange.Socarryoverwillneedtobeevaluatedinsamplesthatfollowthosethatexceed1000ng/mL.IftheamountofDrugXinsamplesfollowingthosewithconcentrationsgreaterthan1000ng/mLisabovethemethod’sLOD,thesampleswithpotentialcarryoverwillbere‐extractedandanalyzed.

LimitofDetection(Section8.7)

ToestimatetheLOD,thelaboratorychosetoutilizetheresultsfromtheirpreviouslygeneratedcalibrationcurvedata(Section8.7.6).Boththeslopeandy‐interceptoftheindividualcalibrationcurves(10–1000ng/mL)weredeterminedinordertocalculatetheaverageslopeandstandarddeviationofthey‐intercepts(TableB.4).

TableB.4—SlopeandY‐InterceptDatafromCalibrationCurves

Slope y‐interceptRun1 0.003980 ‐0.00050Run2 0.003828 ‐0.01543Run3 0.004009 ‐0.01247Run4 0.003934 0.00695

Run5 0.003995 ‐0.00318Average 0.003949 0.00125StdDev 0.000073 0.01054

TheLODwascalculatedusingtheformula:3.3×0.01054/0.003949=8.8ng/mL

LowerLimitofQuantitation(Section8.8)

ThelaboratorychosetoanalyzereferencematerialstoestablishtheirLLOQ(Section8.8.4).Threesourcesofwholebloodwereeachfortifiedat20,15,and10ng/mL.Theywereextractedandanalyzedinduplicateagainstafreshlypreparedcalibrationcurveonthreedifferentdays.Thelowestconcentrationthatwascapableofreproduciblyprovidingsymmetricalpeaksandtheminimummassspectralidentificationratios,whilemaintainingabiasof±20%anda%CVof<20%wasthe10ng/mLsample.Thisconcentrationwasdeemedasthemethod’sLLOQandreaffirmedacceptableresultsatthelowestcalibrationpoint.

BiasandPrecision(Section8.2)

Toestablishthemethod’sbiasandprecision,thelaboratorypreparedthreepoolsoffortifiedmatrixsamplesatthefollowingconcentrations:low(30ng/mL);medium(400ng/mL);andhigh(800ng/mL).Eachconcentrationpooloffortifiedsampleswasanalyzedintriplicateonfiveseparatedaysalongwithafreshlypreparedcalibrationcurve(TableB.5).

Thelaboratorycalculatedthebias(Section8.2.1)byfirstdeterminingthemeanforeachconcentration.ThisresultedinthevalueslistedinTableB.6.

Fromthesevalues,thebiaswascalculatedateachconcentration.Forexample,forthelowconcentrationsample,thebiaswasdeterminedas:BiasLow=((28–30)/30)×100)=(‐6.7%)


26

Likewise,thebiasforthemediumandhighconcentrationswascalculatedas9.3%and–2.4%,respectively.

TableB.5—QuantitativeResults(ng/mL)ofBiasandPrecisionRuns

Low(30ng/mL)

Run1 Run2 Run3 Run4 Run5

Rep1 32 26 29 26 28Rep2 28 24 31 35 25Rep3 27 28 27 30 29Med

(400ng/mL) Run1 Run2 Run3 Run4 Run5Rep1 412 435 427 455 444Rep2 444 410 419 438 442Rep3 422 450 479 452 423High

(800ng/mL) Run1 Run2 Run3 Run4 Run5Rep1 892 793 761 742 820Rep2 827 741 729 734 749Rep3 850 769 803 720 791

TableB.6—MeanConcentrations(ng/mL)forBiasCalculationsConc(ng/mL) CalculatedMean BiasLow(30) 28 ‐6.7%

Med(400) 437 9.3%

High(800) 781 ‐2.4%

Within‐runandbetween‐runprecisionswerecalculatedusingtheone‐wayANOVAapproach(Section8.2.2.3.4).UsingtheANOVA:SingleFactoranalysisinpopularspreadsheetorstatisticsprograms(seeTableB.7forLowConcentration),thelaboratorywasabletoobtainvaluesforthemeansquarewithingroupsforthelowconcentrationandintroducedintotheappropriateformulaasfollows:

%

100

%√7.93328

100

%2.81728

100

% 10.1%


27

TableB.7—ANOVACalculationsfor30ng/mLSample

ANOVA:SingleFactor

SUMMARY

Groups Count Sum Average Variance

Column1 3 87 29 7

Column2 3 78 26 4

Column3 3 87 29 4

Column4 3 91 30.33333 20.33333

Column5 3 82 27.33333 4.333333

ANOVA

SourceofVariation

SS df MS F P‐value Fcrit

BetweenGroups

34 4 8.5 1.071429 0.420175 3.47805

WithinGroups 79.33333 10 7.933333

Total 113.3333 14

Thebetween‐runprecisionforthelowconcentrationwascalculatedusingtheformulaandthemeansquarebetweengroupsfromtheANOVAtable:

%

1 ∗

100

%

8.5 3 1 ∗ 7.9333

28100

%2.84528

100

% 10.2%

Usingthedataforthemediumandhighconcentrations,theANOVA:SingleFactoranalysiswasconductedontheselevels(datanotshown)andappropriatevaluesintroducedintotheformulastoobtainthewithin‐runandbetween‐runprecisions.TableB.8liststhecalculatedresultsforallconcentrations.


28

TableB.8—PrecisionResults

Low Medium HighWithin‐Run 10.1%CV 4.5%CV 3.9%CV

Between‐Run 10.2%CV 4.2%CV 2.2%CVIonizationSuppression/Enhancement(Section8.6)

AstheinstrumentalportionofthemethodinvolvesLC/MS/MS,thelaboratorywasrequiredtoconductionizationsuppression/enhancementexperiments.Thepost‐columnextractionapproachwaschosen(Section8.6.3).

Twosetsofsampleswerepreparedfortheexperiment.Setoneconsistedofstandardspreparedinmobilephaseat30and800ng/mL.Theywerenotextracted,butinsteadsimplyinjectedsixtimeseach.

Settwowaspreparedintenblankmatrixsamples.Eachblankmatrixwasfromanindependentsourceofblankwholebloodfrompreviouslyanalyzedcases.Thesewerethesametenblankmatrixsamplesusedintheinterferencestudies.Theblankmatrixsampleswereextractedinduplicateandthenfortifiedto30and800ng/mLwithDrugXand250ng/mLwithd3‐DrugX.Eachconcentrationsetsamplewasinjectedonetimeeach.

AveragepeakareasforboththeDrugXandthed3‐DrugXarefoundinTableB.9.

TableB.9—AveragePeakAreasFromSuppression/EnhancementExperiments

AveragePeakAreas30ng/mL 800ng/mL

DrugX d3‐DrugX DrugX d3‐DrugXSet1 13890 110381 330822 112827Set2 11812 102444 303992 105923

Usingtheabovedatasets,thelaboratorycalculatedthe%ionizationsuppression/enhancementforthetargetiontransitionsateachconcentrationusingtheformula:

%Ionizationsuppression/enhancementDrugX(Low)=((11812/13890)‐1)×100=(‐15.0%)

Thenegativevaluesuggestedsomesuppressionwasoccurring,butitwaslessthan25%.

Similarly,thelaboratorycalculatedthe%suppression/enhancementfor800ng/mLandfortheinternalstandardinbothsets.Theresultssuggestedsuppressionof‐8.1%forDrugXatthe800ng/mLconcentration.Althoughatthesameconcentrationinboththelowandhighsamples,thed3‐DrugXdemonstratedionizationsuppressionsof7.2%and6.1%,respectively.

Thedatawerealsousedtocalculatethe%CVateachconcentration.All%CVswere<14%(datanotshown).

Sincetheaveragesuppressionforallanalytesdidnotexceed±25%andthecalculated%CVvaluewas<15%,thevariationwasconsideredinsignificant.Nofurtherworkwasrequiredforothervalidationparameters.


29

DilutionIntegrity(Section9.2)

Whilethelaboratoryindicatedthataminimumworkingrangeforthecalibrationcurvewasbetween10–1000ng/mL,theyanticipatedoccasionalsamplesthatcontainDrugXatconcentrationsabove1000ng/mL.Theirinitialattempttoextendthecalibrationrangeto2000ng/mLwasabandonedwhentheyrealizedthatanon‐linearcalibrationmodelwouldbeneeded.Therefore,theyconducteddilutionintegrityexperimentstodemonstrateacceptablebiasandprecisionresultswhensamplesaredilutedindeionizedwater.Theyevaluatedtwodilutionsratios:1:2and1:5.

Thelaboratorypreparedtwofortifiedmatrixsamplesatconcentrationsof1600ng/mLand3000ng/mL.The1600ng/mLsamplewasdiluted1:2beforeextractionandanalysis.Likewise,the3000ng/mLsamplewasdiluted1:5.Bothdilutionsampleswereanalyzedintriplicateoverfivedifferentruns;eachwithafreshlypreparedcalibrationcurve.Biasandprecisioncalculationswereperformedandresults(TableB.10)demonstratedcomparablevaluescomparedtotheresultsobtainedwithoutdilution.Thisprovidedproofofnodetrimentalimpactwhendilutingthesamplesbeforeextraction.

TableB.10—EffectofDilutiononBiasandPrecision

1600ng/mL(1:2dilution)

3000ng/mL(1:5dilution)

Bias 8.2% 9.9%Within‐RunPrecision 4.0% 2.9%Between‐RunPrecision 4.4% 3.7%ProcessedSampleStability(Section9.3)

Thelaboratoryrecognizedthatsamplesarenotalwaysanalyzedimmediatelyafterextractionduetolargebatchesorunforeseendelays.Forexample,theinstrumentmaylosecommunicationwithitscontroller,inadvertentlyshuttingdownabatchrun.Therefore,toevaluatetheimpactofroomtemperaturestorageofprocessedsamplessittingontheautosamplerbeforeanalysis,thelaboratoryconductedastabilitystudyonextractedsamples.Thiswasachievedbypreparingfortifiedmatrixsamplesattwoconcentrations,30ng/mLand800ng/mL.Twelvealiquotsofeachconcentrationwereextracted.Reconstitutedextractsforeachconcentrationwerecombinedandvortexedtoensureadequatemixing.Theconcentrationpoolwasthendividedinto12autosamplervialsandplacedontheautosampler.Thefirstvialofeachlevelwasinjectedthreetimestorepresentthetimezero(t0)sample.Theremainingvialsforeachconcentrationwereanalyzedintriplicateeverysixhoursupto66hours.Analytesignalsfromthetriplicateanalyseswereaveragedandcomparedtothet0signals(TableB.11).


30

TableB.11—AveragePeakAreasforProcessedSampleStabilityStudy

Time(hr)AveragePeakArea

30ng/mL 800ng/mL

DrugX d3‐DrugX DrugX d3‐DrugX0 12490 101832 332554 1004236 12289 100382 331820 10032812 12198 100432 330779 10110118 11732 100733 330246 10098724 10983 100992 329787 10083230 10101 101789 326048 10082136 10328 100904 327238 10023442 10281 100086 326838 10032348 10271 100183 315009 9972754 10612 100309 315772 9942160 10402 100233 316231 9638166 10183 100872 315499 94832

ByplottingtheaveragepeakareasforbothDrugXandtheinternalstandard,thelaboratorycouldevaluatetheprocessedsampleswhiletheywerestoredontheautosampler.Astheirrequiredbiasis±20%,theyconsideredthecompoundsstableuntiltheysawadecrease(orincrease)insignalofmorethan20%fromthet0averagepeakarea.Theplotforthe30ng/mLconcentrationofDrugXisshowntodemonstratethisconcept(FigureB.5).

FigureB.5—ChangeinDrugXPeakAreaOver66Hours

5000

6000

7000

8000

9000

10000

11000

12000

13000

14000

15000

0 10 20 30 40 50 60 70

Peak Area

Hours

Drug X ‐ 30 ng/mL


31

ThesedataappeartosuggestDrugXremainedstablewithinthepre‐definedlimitsfortheentire66‐hourperiodofthestudy.However,thetrendlineshowsthat66hoursmaybethemaximumperiodoftimebeforethesamplesmayneedtobere‐extracted.Itwasnotedthatatthe30‐hourmark,stabilityseemedtohavedroppedveryclosetothe“instability”point.Sincethepreviouslydeterminedbiaswasactuallymuchbetterthanthe±20%requiredintheirvalidationplan,thelaboratorymadeadecisiontore‐extractanysamplesthatremainontheautosamplermorethan24hours.

DocumentationofResults(Section11)

AlongwithalloftheotherrequireddocumentationlistedinSection11,thelaboratorycomparedtheresultsfromthevalidationstudiesconductedtotheoriginallydefinedrequirements,asdemonstratedinTableB.12.

TableB.12—SummaryofValidationResults

Parameter: AcceptanceCriteria: Result:

Bias Shallnotexceed±20% ‐6.7to9.3%

CalibrationModel 10–1000ng/mL(linearmodeldesired) 10–1000ng/mL(linearmodel)

Carryover Carryoverafterhighestcalibratordoesnotexceed10%ofsignaloflowestcalibrator.

Nosignificantcarryoverat2000ng/mL.Re‐extractandanalyzesamplescontainingDrugXabovetheLODifthatsamplefollowsonethatexceeds1000ng/mLofDrugX.

InterferenceStudies Nointerferingsignalfrommatrix,internalstandard,commondrugsofabuse(includingothercommonopiates/metabolites),OTCdrugs,andprescriptionmedications.

Noobservedinterferencesfrommatrixorfromcommondrugs/metabolites.

IonizationSuppression/Enhancement

<25%suppressionorenhancementand<15%CVduetomatrix(ifnot,evaluateimpactonLOD,LLOQ,andBias).

‐8.1to‐15.0%;<14%CV

LimitofDetection Shallbe10ng/mLorlower 8.8ng/mL

LowerLimitofQuantitation

Shallbe10ng/mLorlower 10ng/mL

Precision %CVshallnotexceed20% Within‐run(3.9to10.1%)

Between‐run(2.2to10.2%)

DilutionIntegrity Biasandprecisioncriteriashallbemetwithdilutionofsamples.

Using1:2and1:5aqueousdilutions,bias(8.2to9.9%)andprecision(within‐run(2.9to4.0%);between‐run(3.7to4.4%).Comparabletoresultsobtainedwithoutdilution.

ProcessedSampleStability

Evaluatelengthoftimethatanalyteinextractedsamplesstoredatroomtemperatureremainsstable

24hours


32

AnnexC(informative)

ImmunoassayScreenofBenzodiazepinesinUrineValidationExample

Thefollowingisanexampleoftheimmunoassayvalidationstepsoutlinedinthisdocument.Itisnotintendedtoprovidespecificguidanceforanyparticularmethod.

Inthisexample,assumealaboratoryvalidatedanimmunoassaykitforitsabilitytoscreenurineforbenzodiazepines.

CreateValidationPlan(Section6)

Beforestartingthevalidationexperiments,thelaboratorypreparedthevalidationplan.Intheplan,theyspecifiedthattheywilluseCompanyABC’sELISAImmunoassayKitforBenzodiazepines(Oxazepam)designedwitha“cutoff”of300ng/mL.Thelaboratoryplannedtoselecttheirowncutoffconcentration(decisionpoint)of50ng/mLforthetargetcompoundofoxazepam.Thesamplepreparationsteps,aswellasinstrumentalsettingswerelistedintheplan.Thevalidationparameterswereassessedagainstthepre‐definedrequirementslistedinTableC.1.

TableC.1—ValidationParameterstobeAssessed

Parameter: DesiredLimit:LimitofDetection Sameasdecisionpoint(50ng/mLforoxazepam,lorazepamandalpha‐

hydroxyalprazolamand25ng/mLforalprazolam)

Precision %CVshallnotexceed20%;grandmeans±2StdDevcannotoverlapPrecisionattheDecisionPoint(Section8.2.2.2)

Theproductbrochurelistedthecross‐reactivitiesforoxazepam,otherbenzodiazepines,andtheirmetabolites.Anabbreviatedlistofthesecross‐reactivitiesisasfollows:oxazepam(100%);nordiazepam(425%);lorazepam(175%);alprazolam(450%);andalpha‐hydroxyalprazolam(340%).

Sincetheassaywastobeusedtodeterminetheuseofthebroadclassofbenzodiazepines,thelaboratorywasrequiredtoverifyprecisionforoxazepamandanyotheranalytesthattheychosetoscreenforusingthisassaywithcross‐reactivitieslessthan100%orwithadecisionconcentrationlessthanthatofoxazepam(50ng/mL).Forexample,thislaboratorydecidedtousetheassaytoscreenforlorazepam(decisionconcentration50ng/mL),alprazolam(decisionconcentration25ng/mL)andalpha‐hydroxyalprazolam(decisionconcentration50ng/mL).Bothoxazepamandalprazolamhadtobeevaluatedforprecisionattheirdecisionconcentration.However,sincelorazepamandalpha‐hydroxyalprazolamhadcross‐reactivitiesgreaterthan100%andthesamedecisionpointasoxazepam(50ng/mL),precisionstudieswerenotrequiredfortheseassays.

Thelaboratorypreparedthreepoolsofoxazepam‐fortifiedmatrixsamplesatthefollowingconcentrations:25ng/mL(50%below);50ng/mL(decisionpoint);and75ng/mL(50%above).Eachofthefortifiedsamplesetsisanalyzedintriplicateonfiveseparatedays.TheresultsareshowninTableC.2.


33

TableC.2—Results(B/BoforELISA;AbsorbanceValuesForLiquidReagentAssays)ofPrecisionRunsfortheOxazepamSampleSets.

25ng/mL Run1 Run2 Run3 Run4 Run5Rep1 44.396 39.911 39.171 41.213 41.929Rep2 43.896 40.632 37.556 39.111 41.293Rep3 45.323 44.814 39.789 41.454 44.843

GrandMean 41.689StdDev 2.443%CV 5.9%





Theresultforthe50ng/mLdecisionpointconcentrationwhenconsideringthegrandmeanofthemeasurementplusorminustwostandarddeviations(30.444±(2×1.557))wasbetween27.330‐33.558.Thisrangedidnotoverlapwiththerangescalculatedforthe25ng/mLor75ng/mLsamples.

The%CVforeachconcentrationwas5.9%,5.1%,and10.6%,respectively;wellbelowtherequirementtonotexceed20%.

Similarexperimentswereconductedforalprazolamatthe25ng/mLdecisionpoint,aswellasconcentrations‐50%and+50%ofthedecisionpoint(datanotshown).

LimitofDetection(Section8.7.4)

Thelaboratoryusedthedecisionpointconcentrationsastheassayslimitofdetectionforeachofthebenzodiazepinesandmetabolites.

DocumentationofResults(Section11)

AlongwithalloftheotherrequireddocumentationlistedinSection11,thelaboratorycomparedtheresultsfromthevalidationstudiesconductedtotheoriginallydefinedrequirements,asdemonstratedinTableC.3.


34

TableC.3—SummaryofValidationResults

Parameter: DesiredLimit: Result:LimitofDetection Sameasdecisionpoint:

Oxazepam(50ng/mL)Lorazepam(50ng/mL)Alprazolam(25ng/mL)Alpha‐hydroxyalprazolam(50ng/mL)

50ng/mL50ng/mL25ng/mL50ng/mL

Precision %CVshallnotexceed20%Grandmeans±2StdDevcannotoverlap

5.1to10.6%25ng/mL:36.803–46.57550ng/mL:27.330–33.55875ng/mL:15.952–24.560


35

AnnexD(informative)

ExampleFlowchartofMethodValidationExperiments


36

AnnexE(informative)

TableofExampleExperimentsforValidationofQualitativeConfirmation/IdentificationMethods

Interference(Section8.5)

10differentsourcesofeachmatrix,noIS 1blanksample+isotopically‐labeledIS 1fortifiedsamplewithhighanalyteconcentrations,noIS Neat,fortified,orauthenticsamplescontainingpotentiallyinterfering

compounds/metabolitesbutnoanalyte

Carryover(Section8.4)

AddressedinroutineQCpracticesbyanalyzingextractedblankmatrixsamplesbetweencasesamples

LimitofDetection(Section8.7.5.2)

Fortifiedmatrixsamplesfortifiedatincreasinglylowerconcentrationsandanalyzedinduplicateover3days.Lowestconcentrationthatreproduciblyyieldssignalgreaterthanorequalto3.3timesthenoiseofbackgroundsignalistheLOD.


37

AnnexF(informative)

TableofExampleExperimentsforValidationofQuantitativeMethods

InterferencesIonization

Suppression/Enhancementa CalibrationModel

10differentsourcesofeachmatrix,withoutIS

1blanksamplewithIS 1fortifiedsamplewithhighanalyteconcentrationsandwithoutIS

Neat,fortified,orauthenticsamplescontainingpotentiallyinterferingcompounds/metabolitesbutnoanalyte

Post‐columninfusion: 10blankextractsfortifiedafterextractionatlowconcentration

10blankextractsfortifiedafterextractionathighconcentration

Analytesolutionsforinfusion(lowandhighconcentrations)eachinjected6times

6concentrationlevels,5replicateseach(maybeaccomplishedwithcalibrationcurvesgeneratedforstudiesbelow)

Mainvalidationphase

Bias&Precision DilutionIntegrity

Run Calibration Low Medium High LODb LLOQbBias&Precision

1 6 3 3 3 3 3 3

2 6 3 3 3 3 3 3

3 6 3 3 3 3 3 3

4 6 3 3 3 ‐ ‐ 3

5 6 3 3 3 ‐ ‐ 3aLC‐MS(/MS)methodsonlybForthisexample,thereferencematerialapproachisusedtoestimatetheLOD(Section8.7.5.2)andLLOQ(Section8.8.4)


38

AnnexG(informative)

Bibliography

1] AraujoP,KeyAspectsofAnalyticalMethodValidationandLinearityEvaluation,JChromatogrB,877,(2009),2224‐2234.

2] Boyd,RobertK.,CeciliaBasic,andRobertA.Bethem.TraceQuantitativeAnalysisbyMassSpectrometry.Hoboken,N.J.:JohnWiley(2008)

3] BressolleF,Bromet‐PetitM,Audran,M,ValidationofLiquidChromatographicandGasChromatographicMethods,ApplicationstoPharmacokinetics,JChromatogrB,686,(1996),3‐10.

4] BruceP,Minkkinen,RiekkolaM,PracticalMethodValidation:ValidationSufficientforanAnalyticalMethod,MikrochimActa,128(1998),93‐106.

5] ClinicalandLaboratoryStandardsInstitute,C50‐AMassSpectrometryintheClinicalLaboratory:GeneralPrinciplesandGuidance;ApprovedGuideline,Vol27,No24(2007).

6] CorleyJ,BestPracticesinEstablishingDetectionandQuantificationLimitsforPesticidesResiduesinFoodsinHandbookofResidueAnalyticalMethodsforAgrochemicals,JohnWiley&SonsLtd,Chichester(2003).

7] DrummerO,RequirementsforBioanalyticalProceduresinPostmortemToxicology,AnalBioanalChem,388(2007),1495‐1503.

8] EurachemGuide,TheFitnessforPurposesofAnalyticalMethods,(1998).

9] HealthSciencesAuthority,GuidanceNotesonAnalyticalMethodValidation:Methodology,(Sept.6,2004).

10] HubertPh,Nguyen‐HuuJ‐J,BoulangerBetal.,HarmonizationofStrategiesfortheValidationofQuantitativeAnalyticalProceduresASFSTPProposal–PartI,JPharmBiomedAnal,36(2004),579‐586.

11] HubertPh,Nguyen‐HuuJ‐J,BoulangerBetal.,HarmonizationofStrategiesfortheValidationofQuantitativeAnalyticalProceduresASFSTPProposal–PartII,JPharmBiomedAnal,45(2007),70‐81.

12] HubertPh,Nguyen‐HuuJ‐J,BoulangerBetal.,HarmonizationofStrategiesfortheValidationofQuantitativeAnalyticalProceduresASFSTPProposal–PartIII,JPharmBiomedAnal,45(2007),82‐96.

13] InternationalVocabularyofMetrology–BasicandGeneralConceptsandAssociatedTerms(VIM),ISO/IECGuide99:2007,OrganizationforStandardization(ISO)/InternationalElectrotechnicalCommission(IEC),Geneva,2007.

14] KushnirMM,RockwoodAL,NelsonGJ,etal.,AssessingAnalyticalSpecificityinQuantitativeAnalysisUsingTandemMassSpectrometry,ClinBiochem,12:003,(2004),319‐327.


39

15] LiangHR,FoltzRL,MengM,BennettP,Ionizationenhancementinatmosphericpressurechemicalionizationandsuppressioninelectrosprayionizationbetweentargetdrugsandstable‐isotope‐labeledinternalstandardsinquantitativeliquidchromatography/tandemmassspectrometry,RapidCommMassSpec,17(2003),2815‐2821.

16] MatuszewskiBK,ConstanzerML,Chavez‐EngCM,StrategiesfortheAssessmentofMatrixEffectsinQuantitativeBioanalyticalMethodsBasedonHPLC‐MS/MS,AnalChem,75:13,(2003),3019‐3030.

17] NiedbalaRSandGonzalezJM,ImmunoassaysinClarke’sAnalysisofDrugsandPoisons,4thedition,ACMoffat,MDOsselton,BWiddop,andJWattsEds.PharmaceuticalPress:London(2011).

18] PetersF,MaurerHH,BioanalyticalMethodValidationandItsImplicationsforForensicandClinicalToxicology–AReview,AccredQualAssur,7,(2002),441‐449.

19] PetersFT,MethodValidationinApplicationsofLC‐MSinToxicology,PharmaceuticalPress,London(2006).

20] PetersF,DrummerO,MusshoffF,ValidationofNewMethods,ForensicSciInt,165(2007),216‐224.

21] RemaneD,MeyerMR,WissenbachDK,MauerHH,Ionsuppressionandenhancementeffectsofco‐elutinganalytesinmulti‐analyteapproaches:systematicinvestigationusingultra‐highperformanceliquidchromatography/massspectrometrywithatmosphericpressurechemicalionizationorelectrosprayionization,RapidCommMassSpec,24(2010),3103‐3108.

22] ShahV,MidhaK,DigheS,etal,AnalyticalMethodsValidation:Bioavailability,BioequivalenceandPharmacokineticsStudies,PharmRes,9:4,(1992),588‐592.

23] ShahV,MidhaK,FindlayJ,etal.,BioanalyticalMethodValidation–ARevisitwithaDecadeofProgress,PharmRes,17:12(2000),1551‐1557.

24] ShultzEK,AnalyticalGoalsandClinicalInterpretationofLaboratoryProceduresinTietzTextbookofClinicalChemistry,2ndedition,CABurtisandERAshwood,Eds.W.B.SaundersCompany:Pennsylvania(1994).

25] StocklD,D’HondtH,ThienpontLM,MethodValidationAcrosstheDisciplines‐CriticalInvestigationofMajorValidationCriteriaandAssociatedExperimentalProtocols,JChromatogrB,877,(2009),2180‐2190.

26] ThompsonM,EllisonSLR,WoodR.HarmonizedGuidelinesforSingle‐LaboratoryValidationofMethodsofAnalysis,IUPACTechnicalReport,PureApplChem,74:5,(2002),835‐855.

27] U.S.DepartmentofHealthandHumanServices,FoodandDrugAdministration,GuidanceforIndustry,BioanalyticalMethodValidation,(2001).

28] VanEeckhartA,LanckmasK,SarreS,etal.,ValidationofBioanalyticalLC‐MS/MSAssays:EvaluationofMatrixEffects,JChromatogrB,877,(2009),2198‐2207.

29] ViswanathanCT,BansalS,BoothB,etal.,QuantitativeBioanalyticalMethodsValidationandImplementation:BestPracticesforChromatographicandLigandBindingAssays,PharmRes,24:10,(2007),1962‐1973.


40

30] ViswanathanCT,BansalS,BoothB,etal.,Workshop/ConferenceReport‐QuantitativeBioanalyticalMethodsValidationandImplementation:BestPracticesforChromatographicandLigandBindingAssays,AAPSJ,9:1,(2007),E30‐E42.

AcademyStandardsBoard4200WisconsinAvenue,NW

Suite106‐310Washington,DC20016‐2143

http://asb.aafs.org/

Documents

Standard Practices for Method Validation Forensic Toxicology · PDF fileDarcie Wallace‐Duckworth, Ph.D.; Aegis Sciences Corporation; Nashville, Tennessee Jeff Walterscheid, Ph.D.,