3
METHODVALIDATIONFOR
HPLCANALYSISOFRELATED
SUBSTANCESIN
PHARMACEUTICALDRUG
PRODUCTS
Y.C.LEE,PH.D.
PatheonYM,Inc.
3.1INTRODUCTION
Inthischapterweoutlinethegeneralrequirementsforanalyticalmethodvalida-
tionforHPLCanalysisofrelatedsubstancesinpharmaceuticalproducts.Most
ofthediscussionisbasedonmethodvalidationforpharmaceuticalproductsof
syntheticorigin.Eventhoughmostoftherequirementsaresimilarforothertypes
ofpharmaceuticaldrugproducts(e.g.,biopharmaceuticaldrugproducts),detailed
discussionofmethodvalidationforothertypesofpharmaceuticaldrugproducts
isoutsidethescopeofthischapter.Thediscussionfocusesoncurrentregulatory
requirementsinthepharmaceuticalindustry.Sincetheexpectationsformethod
validationaredifferentatdifferentstagesoftheproductdevelopmentprocess,
theinformationgiveninthischapterismostsuitableforfinalmethodvalida-
tionaccordingtotheICHrequirementstoprepareforregulatorysubmissions
(e.g.,NDA).EventhoughthemethodvalidationisrelatedtoHPLCanalysis,
mostoftheprinciplesarealsoapplicabletootheranalyticaltechniques(e.g.,
TLC,UV).
AnalyticalMethodValidationandInstrumentPerformanceVerification,EditedbyChungChow
Chan,HermanLam,Y.C.Lee,andXue-MingZhang
ISBN0-471-25953-5Copyright?2004JohnWiley&Sons,Inc.
27
28METHODVALIDATIONFORHPLCANALYSISOFRELATEDSUBSTANCES
3.2BACKGROUNDINFORMATION
3.2.1Definitions
Definitionsforsomeofthecommonlyusedtermsinthischapteraregivenbelow.
?Drugsubstance(activepharmaceuticalingredient):apharmaceutical
activeingredient.
?Relatedsubstances:impuritiesderivedfromthedrugsubstanceandthere-
forenotincludingimpuritiesfromexcipients.Relatedsubstancesinclude
degradationproducts,syntheticimpuritiesofdrugsubstance,andmanufac-
turingprocessimpuritiesfromthedrugproduct.
?Authenticsample:apurifiedandcharacterizedsampleofarelatedsubstance.
Unlikereferencestandards,authenticsamplesmaynotbeofhighpurity.
However,thepurityofanauthenticsamplehastobedeterminedbefore
use.Authenticsamplesareusedinmethoddevelopmenttoidentifyrelated
substancesintheanalysis.Inaddition,theyareusedextensivelytoprepare
thespikedsamplesinmethodvalidation.
?Spikedsample:asampleaddedwithaknownamountofrelatedsubstances,
preparedfromauthenticsamplesduringmethoddevelopmentorvalidation.
?Controlsample:arepresentativebatchofdrugsubstance(ordrugproduct).
Typically,controlsamplesaretestedinallanalysestoensureconsistency
inmethodperformanceacrossdifferentruns.Sometimes,theyareusedas
partofthesystemsuitabilitytesttoestablishtherun-to-runprecision(e.g.,
intermediateprecision,reproducibility).
?Responsefactor:theresponseofdrugsubstanceorrelatedsubstancesper
unitweight.Typically,theresponsefactorofdrugsubstance(orrelated
substance)canbecalculatedbythefollowingequation:
Responsefactor=
response(inresponseunits)
concentration(inmg/mL)
?Relativeresponsefactor:theratiooftheresponsefactorofindividualrelated
substancetothatofadrugsubstancetocorrectfordifferencesintheresponse
ofrelatedsubstancesandthatofthedrugsubstance.Itcanbedetermined
usingthefollowingequation:
Relativeresponsefactor=
responsefactorofindividualrelatedsubstance
responsefactorofdrugsubstance
Ifalinearitycurve(Figure3.1)isconstructedforboththerelatedsubstance
andthedrugsubstancebyplottingtheresponseversustheconcentration,
therelativeresponsefactorcanalsobedeterminedby
Relativeresponsefactor=
slope
relatedsubstance
slope
drugsubstance
BACKGROUNDINFORMATION29
x
x
x
x
x
x
P
eakarea
Concentration(mg/mL)
Drugsubstance
Relatedsubstance
slope
relatedsubstance
slope
drugsubstance
Figure3.1.Relativeresponsefactor.
3.2.2DifferentTypesofRelatedSubstanceAnalysis
AreaPercent.Inthisapproach,thelevelofanindividualrelatedsubstanceis
calculatedbythefollowingequation:
%relatedsubstance=
area
relatedsubstance
totalarea
×100%
wherethearea
relatedsubstance
isthepeakareaoftheindividualrelatedsubstance
andthetotalareaisthepeakarea(i.e.,response)ofthedrugsubstanceplus
thepeakareasofallrelatedsubstances.Thisisoneofthesimplestapproaches
forrelatedsubstanceanalysisbecausethereisnoneedforareferencestandard.
Thisisparticularlyimportantduringtheearlyphaseoftheprojectwhenahighly
purifiedreferencestandardisnotavailable.Itisthepreferredapproachaslong
asthemethodperformancemeetsthecriteriadescribedbelow.
LinearityoveraWideRangeofConcentration.Sincetheareasoftherelatedsub-
stances(typically,lessthan1%)anddrugsubstance(typically,morethan95%)
aresummed,itisimportantthatthemethodislinearfromtheconcentrationof
relatedsubstances(e.g.,1%)tothatofthedrugsubstance(e.g.,95%).However,
insomecases,thepeakshapeofthedrugsubstancemaynotbetotallysym-
metricalatsuchahighconcentration.Therefore,theresponsemaynotbelinear
insuchawideconcentrationrange,andtheuseofareapercentagemaynotbe
appropriate.Iftheresponseoftheanalyteisnonlinearathigherconcentrations,
therelatedsubstanceswouldbeoverestimated.Althoughthisisconservativefrom
asafetyperspective,itisinaccurateandthereforeunacceptable.
SampleConcentration(MethodSensitivity).Tomaintainlinearityatthecon-
centrationrangeofthedrugsubstance,scientistsmaytrytolowerthesample
concentrationtoimprovepeakshapeforthedrugsubstance.However,ifthe
sampleconcentrationistoolow,itwillaffectthemethodsensitivity,andthe
abilitytodetectlowlevelsofrelatedsubstancesmaynotbeadequate.
30METHODVALIDATIONFORHPLCANALYSISOFRELATEDSUBSTANCES
ResponseFactor.Theresponsefactorsoftherelatedsubstancesshouldbesim-
ilartothatofthedrugsubstance(i.e.,relativeresponsefactorsclosetounity).
Otherwise,aresponsefactorcorrectionmustbeusedinthecalculation.
High–Low.Thisapproachcanbeusedtoovercomethelimitationoflinear
rangeintheareapercentmethoddiscussedabove.Inthisapproach,samplesare
preparedataconcentration(i.e.,highconcentration)similartothatofthearea
percentmethod(Figure3.2).Inaddition,thehighconcentrationsamplesolutions
aredilutedfurther,tolowconcentrations(Figure3.3).Samplesfrombothhigh-
andlow-concentrationsolutionsareinjectedforanalysis.Intheinjectionsof
thehighconcentration,theresponsesofallrelatedsubstancesaredetermined
asthesesmallpeaksaredetectable.Thehighsampleconcentrationisusedto
allowallrelatedsubstancestobedetectedandquantitated.Intheinjectionof
Peakareafor
relatedsubstances
Time(min)
Absorbance(mA
U)
510152025
0
10
20
30
40
50
60
70
Figure3.2.chromatogramfromhighconcentration.
Peakareafor
drugsubstance
Time(min)
Absorbance(mA
U)
510152025
0
10
20
30
40
50
60
70
Figure3.3.chromatogramfromlowconcentration.
BACKGROUNDINFORMATION31
low-concentrationsample,theresponseofthedrugsubstanceisdetermined.Low
concentrationisusedtoensurethattheresponseofthedrugsubstanceiswithin
thelinearityrange.
Afterdilution,responseofthedrugsubstanceinthelow-concentrationsample
issimilartothatofrelatedsubstanceinthehigh-concentrationsample.Therefore,
onlyasmalllinearityrangeisrequiredforthismethod.Inaddition,sincehigh
sampleconcentrationisusedforthedeterminationofrelatedsubstances,high
methodsensitivitycanbeachieved.Thelimitationofthehigh–lowapproachis
thateachsampleisinjectedatleasttwice(i.e.,highandlowconcentrations)and
thetotalanalysistimewillbedoubled.Inaddition,anadditionalstepisrequired
todilutethehighconcentrationtoalowconcentration,anddilutionerrorcan
occurduringtheseconddilution.
ExternalStandard.Inthisapproach,relatedsubstancelevelsaredetermined
bycalculationusingastandardcurve.Theconcentrationofrelatedsubstanceis
determinedbytheresponse(i.e.,peakareaofindividualrelatedsubstance)and
thecalibrationcurve.Areferencestandardofthedrugsubstanceistypicallyused
inthecalibration.Therefore,aresponsefactorcorrectionmayberequiredifthe
responseofrelatedsubstanceisverydifferentfromthatofthedrugsubstance.A
single-pointstandardcurve(Figure3.4)isappropriatewhenthereisnosignificant
y-intercept.Otherwise,amultipointcalibrationcurve(Figure3.5)hastobeused.
DifferenttypesofcalibrationarediscussedinSection3.2.3.
Theexternalstandardapproachoffersseveraladvantagesovertheareaper-
centmethod,asdiscussedbelow.
ReducedLinearRange.Unliketheareapercentandhigh–lowmethods,which
usetheresponseofthedrugsubstanceinsampleinjectionsforcalculation,an
externalstandardmethodusesastandardcurve.Typically,theconcentrationrange
ofthecalibrationcurveissimilartothatofrelatedsubstancesinthesample(e.g.,
1to5%ofthenominalsampleconcentration).Therefore,thismethodrequiresa
smalllinearrange.
x
x
x
P
eakarea
Concentration(%relatedsubstance)
Referencestandardcalibrationcurve
%relatedsubstancefound
Areafound
Standardconcentration
atthesamelevel
Figure3.4.Single-pointcalibration.
32METHODVALIDATIONFORHPLCANALYSISOFRELATEDSUBSTANCES
x
x
x
Concentration(%relatedsubstance)
Referencestandardcalibrationcurve
%relatedsubstancefound
Standardconcentration
atdifferentlevels
P
eakarea
Areafound
Figure3.5.Multi-pointcalibration.
ImprovedMethodSensitivity.Inthisapproach,onlytheresponsesofindividual
relatedsubstancesareusedinthecalculation.Sincetheareaofdrugsubstance
peakinthesampleinjectionsisnotnecessaryforthecalculation,highsample
concentrationscanbeusedwithoutworryingabouttheoff-scaleresponseofthe
drugsubstance.Thisapproachisparticularlyusefulwhenthescientistswantto
improvethemethodsensitivitybyincreasingthesampleconcentration.
ReferenceStandard.Oneofthelimitationsoftheexternalstandardmethodis
thatawell-characterizedreferencestandardisessential.Inaddition,eachanal-
ysisrequiresaccurateweighingsofsmallquantities(e.g.,10mg)ofreference
standard.Therefore,weighingerrorcanaffectmethodprecisionandaccuracy.
3.2.3SuitabilityofRelatedSubstanceAnalysis
AsdiscussedinSection3.2.2,linearrangeisacriticalfactorfordeterminingthe
suitabletypeofrelatedsubstanceanalysis.Thefollowingaredifferentsituations
toillustratetherationales.Typically,thelowendofalinearitycurveisabout50%
oftheICHreportinglimit(e.g.,50%of0.1%=0.05%).Thisistoensurethatthe
methodwillbeabletocalculateresultsaccuratelybelowtheICHreportinglimit.
Thehighendofthelinearitycurveisthenominalconcentration(i.e.,100%).
Thisisthetargetsampleconcentrationforthedrugsubstance.
Case1.Linearitydemonstratedfrom50%oftheICHreportinglimittoanominal
concentrationofdrugsubstanceinthesamplesolution.Inaddition,nosignif-
icanty-interceptisobserved(Figure3.6).Inthiscase,areapercentcalculation
issuitablebecausethelinearityrangecoverstheresponsesofrelatedsubstances
andthatofthedrugsubstanceinthesamplesolution.Therefore,theseresponses
canbeuseddirectlytocalculatetheareapercentageofeachrelatedsubstance.
Case2.Linearitydemonstratedfrom50%oftheICHreportinglimitto150%
oftheshelflifespecificationofrelatedsubstance.Nosignificanty-intercept
isobserved(Figure3.7).Inthiscase,ahigh–lowcalculationismoresuitable,as
BACKGROUNDINFORMATION33
x
x
x
x
x
x
P
eakarea
Concentration(%relatedsubstance)
Drugsubstance
100%(nominalsampleconcentration)
50%of
ICHreportinglimit
Figure3.6.Linearity:case1.
x
x
x
x
x
x
100%
x
150%of
shelflifespecification
Nosignificant
y-intercept
Concentration(%relatedsubstance)
Drugsubstance
50%ofICH
reportinglimit
P
eakarea
Figure3.7.Linearity:case2.
theresponseislinearonlyuptotheshelflifespecificationlevel.Drugsubstance
concentrationinsamplesolution(highconcentration)shouldbedilutedtothe
linearrangetoobtainthelow-concentrationsolution.Therefore,theresponse
ofdrugsubstanceinlowconcentrationwillbewithinthelinearityrangeand
suitableforcalculation.Alternatively,asingle-pointexternalstandardcalibration
ofconcentrationwithinthelinearityrangecanalsobeused.
Case3.Linearitydemonstratedfrom50%oftheICHreportinglimitto150%
oftheshelflifespecificationofarelatedsubstance,andasignificanty-intercept
isobserved(Figure3.8).Duetothesignificanty-intercept,asingle-pointcali-
bration(e.g.,high–loworone-pointexternalstandardcalibration)isnotsuitable.
Inthiscase,multiple-pointexternalstandardcalibrationisthemostappropriate.
SeeSection3.3.3formorediscussionofthesignificanty-intercept.
3.2.4PreparationbeforeMethodValidation
CriticalRelatedSubstances.Criticalrelatedsubstancesarethosethatmayexist
atsignificantlevelsinthedrugproduct.Authenticsamplesofthesecriticalrelated
34METHODVALIDATIONFORHPLCANALYSISOFRELATEDSUBSTANCES
x
x
x
x
x
x
Drugsubstance
x
150%of
shelflifespecification
Significanty-intercept
Concentration(%relatedsubstance)
50%of
ICHreportinglimit
P
eakarea
Figure3.8.Linearity:case3.
substancesshouldbeavailableformethodvalidation.AccordingtotheICH
guidelines,allrelatedsubstancesatalevelexceedingtheidentificationthreshold
havetobeidentified.Theserelatedsubstancesshouldbeconsideredcriticaland
includedinthemethodvalidation.
Todeterminethecriticalrelatedsubstances,onecanreviewtherelatedsub-
stanceprofilewhenthedrugsubstance(ordrugproduct)issubjecttostress
testing.Themostsignificantrelatedsubstancesinstresstestingshouldbecon-
sideredcritical.Inaddition,significantrelatedsubstances(i.e.,greaterthanICH
identificationthreshold)observedinstabilitystudiesduringproductdevelopment
shouldalsobeincludedinthemethodvalidation.Therelatedsubstancemethod
hastobevalidatedwithrespecttoeachcriticalrelatedsubstance;therefore,
theworkloadassociatedwithmethodvalidationwillincreasedrasticallyifthe
numberofcriticalrelatedsubstancesislarge.
LowerandUpperConcentrationRangeforMethodValidation.Theconcentra-
tionrangeofrelatedsubstancesistypicallyrelatedtothetargetedquantitation
limit(QL)atthelowendandtheproposedshelflifespecificationatthehigh
end.Therefore,itisimportanttohaveagoodestimateoftheselimits;otherwise,
inappropriateconcentrationsmaybeusedinmethodvalidation.Eventhough
ICHproposesamethodvalidationrangefromtheICHreportinglimitto120%
ofspecification,onewouldwanttoextendtherangeto50%oftheICHreport-
inglimitto150%ofspecificationtoensurethatthemethodissuitableformost
intendeduses.TheICHreportinglimitisgiveninTable3.1.Ingeneral,the
quantitationlimitshouldbelowerthanthecorrespondingICHreportinglimit.
Thisistoensurethatthemethodisaccurateandpreciseenoughtoreportresults
attheleveloftheICHreportinglimit.
MethodProcedure.Sincethemethodprocedureisundergoingconstantmodifi-
cationsduringmethoddevelopment,itisveryimportanttodefinetheprocedure
beforemethodvalidation.Thiswillensurethatthesamemethodprocedurewill
beusedinallmethodvalidationexperiments.
METHODVALIDATIONEXPERIMENTS35
Table3.1.VariousICHThresholdsRegardingDegra-
dationProductsinNewDrugProductsasStatedin
theCurrentICHGuidelinesQ3B(R)
MaximumDailyDose
a
Threshold
b
ThresholdsforReporting
≤1g0.1%
>1g0.05%
ThresholdsforIdentification
<1mg1.0%or5μgTDI
c
whichever
islower
1–10mg0.5%or20μgTDI,whichever
islower
>10mg–2g0.2%or2mgTDI,whichever
islower
>2g0.1%
ThresholdsforQualification
<10mg1.0%or50μgTDI,whichever
islower
10–100mg0.5%or200μgTDI,
whicheverislower
>100mg–2g0.2%or2mgTDI,whichever
islower
>2g0.1%
a
Theamountofdrugsubstanceadministeredperday.
b
Thresholdisbasedonpercentofthedrugsubstance.
c
Totaldailyintake.
CriticalExperimentalParametersforRobustness.Criticalexperimentalparam-
etersshouldbeidentifiedduringmethoddevelopment,andtheywillbeinvesti-
gatedintherobustnessexperiments.
SystemSuitabilityTests.Theappropriatesystemsuitabilitytestsshouldbedefi-
nedbeforemethodvalidation(e.g.,precision,resolutionofcriticalrelatedsub-
stances,tailing,detectorsensitivity).Thesesystemsuitabilitytestsshouldbe
performedineachmethodvalidationexperiments.Systemsuitabilityresultsfrom
themethodvalidationexperimentcanbeusedtodeterminetheappropriatesystem
suitabilityacceptancecriteria.
3.3METHODVALIDATIONEXPERIMENTS
Inthissectionweoutlinetherequirementsformethodvalidationaccordingto
currentICHguidelines.
36METHODVALIDATIONFORHPLCANALYSISOFRELATEDSUBSTANCES
3.3.1Specificity
ICHdefinition:Specificityistheabilitytoassessunequivocallytheanalyteinthe
presenceofcomponentsthatmaybeexpectedtobepresent.
Mostrelatedsubstancemethodswillbeusedinastabilitystudy,andtherefore
theyhavetobestabilityindicating.Stabilityindicatingmeansthatthemethod
hassufficientspecificitytoresolveallrelatedsubstancesandthedrugsubstance
fromeachother.Typically,fortherelatedsubstancemethodforadrugproduct,
degradationproductsarethemostcriticalrelatedsubstances.Therefore,asa
minimumrequirement,themethodshouldhavesufficientspecificitytoresolve
thedegradationproductsandthedrugsubstance.Inaddition,alldegradation
productsshouldberesolvedfrompotentialinterferencewiththeexcipients.
SamplesforSpecificity
?BlanksolutiontoshownointerferencewithanyHPLCsystemartifactpeak.
?Placebotodemonstratethelackofinterferencefromexcipients.
?Drugsubstancetoshowthatallsignificantrelatedsubstancesareresolved
fromthedrugsubstance.
?Authenticsamplesofcriticalrelatedsubstancestoshowthatallknown
relatedsubstancesareresolvedfromeachother.
?Typically,astressedsampleofabout10to20%degradationisusedto
demonstratetheresolutionamongdegradationproducts.A10to20%de-
gradedsampleisusedbecauseithasasufficientlyhighconcentrationlevelof
criticalrelatedsubstance.Therefore,theserelatedsubstancescanbedetected
easily.Inaddition,10to20%degradationisnottooexcessive,andthe
relatedsubstanceprofileshouldbeclosetothatofatypicalstabilitysample.
?Stressedplacebotoshowthatthedegradationproductsfromtheexcipients
willnotinterferewiththedegradationproductsofthedrugsubstance.
DifferentApproaches
1.Whenauthenticsamplesofrelatedsubstanceareavailable.Analyze
stresseddrugproduct,placebo,drugsubstance,stressedplacebo,and
solutionsspikedwithauthenticsamplesofrelatedsubstances.TheHPLC
chromatogramsareusedtoshowtheresolutionamongrelatedsubstances,
drugsubstance,andotherpotentialinterferences.Inaddition,check
thepeakhomogeneityofthesignificantdegradationproductsanddrug
substancebyaphotodiodearraydetector(PDA)ormassspectrometer.
Thisverifiesthatnosignificantrelatedsubstancecoelutewitheachother.
2.Whenauthenticsamplesofimpuritiesarenotavailable.Astresseddrug
productcanbeanalyzedtoshowseparationofthemostsignificantrelated
substances.Inaddition,thepeakhomogeneityofthestressedsampleshould
beinvestigatedbyPDAormassspectrometry.Alternatively,onemayuse
anorthogonalproceduretoverifythemethodspecificity.Theorthogonal
METHODVALIDATIONEXPERIMENTS37
methodcanbeadifferenttechnique(e.g.,capillaryelectrophoresis,thin-
layerchromatography)ordifferenttypeofHPLCanalysis(e.g.,reversed
phaseversusnormalphase).Forexample,comparetherelatedsubstance
profileintheoriginalHPLCmethodandthatoftheorthogonalmethod.To
demonstratemethodspecificity,thesignificantrelatedsubstancesshouldbe
consistentinthesemethods.
3.3.2QuantitationLimit(and/orDetectionLimit)
ICHdefinition:Thequantitationlimitofanindividualanalyticalprocedureisthe
lowestamountofanalyteinasamplethatcanbedeterminedquantitativelywith
suitableprecisionandaccuracy.Thedetectionlimitofanindividualanalytical
procedureisthelowestamountofanalyteinasamplethatcanbedetectedbut
notnecessarilyquantitatedasanexactvalue.
Twotypesofapproachescanbeusedtodeterminethequantitationlimitor
detectionlimit,asdescribedbelow.
Signal-to-NoiseApproach.Quantitationlimit(QL;Figure3.9)isdefinedasthe
concentrationofrelatedsubstanceinthesamplethatwillgiveasignal-to-noise
(S/N)ratioof10:1.Detectionlimit(DL)correspondstotheconcentrationthat
willgiveasignal-to-noiseratioof3:1.Thequantitationlimitofamethodis
affectedbyboththedetectorsensitivityandtheaccuracyofsamplepreparation
atsuchalowconcentration.Inpractice,thequantitationlimitshouldbelower
thanthecorrespondingICHreportinglimit(Table3.1).
Toinvestigatetheeffectofbothfactors(i.e.,samplepreparationanddetector
sensitivity),solutionsofdifferentconcentrationsneartheICHreportinglimits
arepreparedbyspikingknownamountsofrelatedsubstancesintoexcipients.
Eachsolutionispreparedaccordingtotheprocedureandanalyzedrepeatedlyto
determinetheS/Nratio.TheaverageS/Nratiofromallanalysesateachconcen-
trationlevelisusedtocalculatetheQLorDL.Thefollowingequationcanbe
usedtoestimatetheQLateachconcentrationlevel.Sincedifferentconcentration
levelsgivedifferentQLs,typicallytheworst-caseQLwillbereportedastheQL
ofthemethod.
QLateachconcentration=10×
concentration(in%relatedsubstance)
S/N(averageateachconcentration)
Signal(S)
Noise(N)
Figure3.9.Quantitationlimit.
38METHODVALIDATIONFORHPLCANALYSISOFRELATEDSUBSTANCES
Alternatively,thespikesolutioncanbedilutedseriallytolowerconcentrations.
TheS/Nratioateachconcentrationlevelisdetermined.Theconcentrationlevel
(inpercentrelatedsubstance)thatgivesanS/Nvalueofabout10willbereported
astheQL.
StandardDeviationApproach.Thefollowingequationscanbeusedtodeter-
minequantitationlimitanddetectionlimitbystandarddeviationoftheresponse
atlowconcentrations:
QL=10×
SD
S
DL=3.3×
SD
S
whereSDisthestandarddeviationoftheresponsenearQLandSistheslope
ofthelinearitycurvenearQL.
TherearetwowaystodetermineSD:
1.Usingexperimentssimilartothosegivenforthesignal-to-noiseapproach,
determinethestandarddeviationoftheresponsesbyrepeatanalysisofa
solutionnearthetargetedQL.
2.ConstructacalibrationcurvenearthetargetedQL:
a.Determinetheresidualstandarddeviationoftheregressionlineofcal-
ibration,or
b.Determinethestandarddeviationofthey-intercept.
OtherConsiderationsforQL.Toaccountforinstrument-to-instrumentvariation,
onemayneedtoverifytheQLinmultiplerunsusingdifferentinstruments.
ThedesiredQLshouldbelessthantheICHreportinglimit(e.g.,50%ofICH
reportinglimit).QLshouldbeappropriate;toohighindicatesthatthemethodis
notsensitiveenoughtoreportresultsattheICHreportinglimit.Toolowindicates
thatinsignificantdegradationproducts,eventhoughmuchlowerthantheICH
reportinglimit,mayneedtobereported.
ToensurethattheHPLCsystemineachanalysisissufficientlysensitive
toreportresultsattheICHreportinglimit,onemayuseadetectorsensitivity
solutionaspartofthesystemsuitabilitytest.SincetheICHreportinglimit
correspondstoQL(i.e.,S/N=10),one-thirdoftheICHreportinglimitshould
correspondtoDL(i.e.,S/N=3).Therefore,aspartofthesystemsuitabilitytest,
adetectorsensitivitysolutionofaconcentrationofaboutone-thirdoftheICH
reportinglimitlevelwouldbeinjected.Theresponseofthedetectorsensitivity
solutionshouldmeetthedetectionlimitandshouldbevisuallydistinguishable
frombaseline.
Alternatively,onemayevaluatetheS/Nratioofthestandardsolutionduring
methoddevelopmentorvalidation.Partoftheroutinesystemsuitabilitytestisto
METHODVALIDATIONEXPERIMENTS39
determinetheS/Nofthestandardsolutionbeforeeachanalysis.Therefore,the
S/Nofeachanalysisneedstobegreaterthantheestablishedlimit.
3.3.3Linearity
ICHdefinition:Thelinearityofananalyticalprocedureisitsability(withina
givenrange)toobtaintestresultsthataredirectlyproportionaltotheconcentra-
tion(amount)ofanalyteinthesample.
GeneralRequirements
Range.Ideally,linearityshouldbeestablishedfrom50%oftheICHreporting
limittothenominalconcentrationofdrugsubstanceinthesamplesolution(for
areapercentmethod).Ifthelinearitydoesnotsupportsuchawiderangeof
concentration,determinethelinearityfrom50%oftheICHreportinglevelto
150%oftheproposedshelflifespecificationsoftherelatedsubstance(forthe
high–lowandexternalstandardmethods)asaminimum.Thiswillensurea
linearresponseforrelatedsubstancesatallconcentrationlevelstobedetected
duringstability.
ExperimentalRequirements.Solutionsofknownconcentrationsareusedtodeter-
minethelinearity.Aplotofpeakareaversusconcentration(inpercentrelated
substance)isusedtodemonstratethelinearity.Authenticsamplesofrelatedsub-
stanceswithknownpurityareusedtopreparethesesolutions.Inmostcases,for
thelinearityofadrugproduct,spikingtherelatedsubstanceauthenticsample
intoexcipientsisnotnecessary,asthematrixeffectshouldbeinvestigatedin
methodaccuracy.
AcceptanceCriteria.Visualinspectionisthemostsensitivemethodfordetecting
nonlinearity.Therefore,theplothastobelinearbyvisualinspection.Inaddition,
accordingtoICHguidelines,thefollowingresultsshouldbereported:slope,
correlationcoefficient,y-intercept,andresidualsumofsquares.
y-Intercept.Thereareseveralapproachestoevaluatingthesignificanceofthe
y-intercept.
?Intercept/sloperatio.Theintercept/sloperatioisusedtoconvertthey-
interceptfromtheresponseunit(peakarea)totheunitofpercentrelated
substance.Theintercept/sloperatioshouldbecomparedtotheproposed
specificationstodetermineitssignificance.Forexample,iftheshelflife
specificationis2.0%,anintercept/sloperatioof0.2%maybeconsidered
significant,as0.2%represents10%relativetothespecification.
?Statisticalapproach.Thelinearityresultscanbesubjectedtostatistical
analysis(e.g.,useofstatisticalanalysisinanExcelspreadsheet).Thep-
valueofthey-interceptcanbeusedtodetermineiftheinterceptissta-
tisticallysignificant.Ingeneral,whenthep-valueislessthan0.05,the
40METHODVALIDATIONFORHPLCANALYSISOFRELATEDSUBSTANCES
y-interceptisconsideredstatisticallysignificant.Thep-value,whichcom-
paresthey-interceptwiththevariationofresponses,indicatestheprobability
thatthey-intercepttobenotequaltozero.Forexample,whenthep-value
islessthan0.05,thisindicatesthatitis95%confidentthatthey-intercept
isnotequaltozero.Inotherwords,itis95%certainthatthey-interceptis
significant.
Typically,apositivey-interceptindicatestheexistenceofinterferencewith
theresponseorthesaturationofresponsesathighconcentrations.Anegative
y-interceptindicatesthepossibilityofmethodsensitivityproblem(i.e.,alow
responsecannotbedetected)oranalytesgetretainedintheglasswareorHPLC
system(i.e.,acompatibilityissuebetweensamplesolventandmobilephase).
DifferentApproachesforLinearityDetermination.Thefirstapproachisto
weighdifferentamountsofauthenticsampledirectlytopreparelinearitysolu-
tionsofdifferentconcentrations.Sincesolutionsofdifferentconcentrationare
preparedseparatelyfromdifferentweights,iftherelatedsubstancesreachtheir
solubilitylimit,theywillnotbecompletelydissolvedandwillbeshownasa
nonlinearresponseintheplot.However,thisisnotsuitabletopreparesolutions
ofverylowconcentration,astheweighingerrorwillberelativelyhighatsuch
alowconcentration.Ingeneral,thisapproachwillbeaffectedsignificantlyby
weighingerrorinthepreparation.
Anotherapproachistoprepareastocksolutionofhighconcentration,then
performserialdilutionfromthestocksolutiontoobtainsolutionsoflowercon-
centrationsforlinearitydetermination.Thisisamorepopularapproach,asserial
dilutioncanbeusedtopreparesolutionsofverylowconcentrations.Sincethe
lowconcentrationsarepreparedbyserialdilution,thisapproachdoesnotneedto
weighaverysmallquantityofrelatedsubstance.Inaddition,sinceallsolutions
aredilutedfromthesamestocksolution,weighingerrorinpreparingthestock
solutionwillnotaffectthelinearitydetermination.
RelativeResponseFactor.Therelativeresponsefactor(RRF)canbeusedto
correctfordifferencesinrelativeresponsebetweentherelatedsubstancesand
thedrugsubstance.Intheareapercentandhigh–lowmethod,therelatedsub-
stancesarecalculatedagainsttheresponseofthedrugsubstance.Intheexternal
standardcalculation,thestandardcurveofdrugsubstanceisgenerallyusedin
thecalculation.Sincetherelatedsubstancesarecalibratedbytheresponseofthe
drugsubstance,itisnecessarytodeterminetherelativeresponseoftherelated
substancetothatofthedrugsubstance.Afterthelinearityoftherelatedsub-
stancesandthedrugsubstancearedetermined,onecancalculatetherelative
responsefactorbycomparingtheslopeoftherelatedsubstancetothatofthe
drugsubstance.Iftherelativeresponsefactorissignificantlydifferentfromunity,
acorrectionfactormayneedtobeusedinthecalculation.Otherwise,thereported
resultswillbegrosslyover-orunderestimated(Figure3.10).
METHODVALIDATIONEXPERIMENTS41
x
x
x
x
x
x
Drugsubstance
Relatedsubstance(RRF<<1)
%relatedsubstancefound
(underestimatedifnocorrection)
%relatedsubstance
(truevalue)
P
eakarea
Concentration(%relatedsubstance)
Parkarea
found
Figure3.10.Impactofrelativeresponsefactor.
3.3.4Accuracy
Definition:Theaccuracyofananalyticalprocedureexpressestheclosenessof
agreementbetweenthevaluethatisacceptedeitherasaconventionaltruevalue
orasanacceptedreferencevalueandthevaluefound.
GeneralConsiderations
Range.Accuracyfortheareapercentmethodshouldbeestablishedfrom50%
oftheICHreportinglimittothenominalconcentrationofdrugsubstancein
thesamplesolution.Forthehigh–lowandexternalstandardmethods,deter-
mineaccuracyfrom50%oftheICHreportinglevelto150%oftheproposed
shelflifespecificationoftherelatedsubstances.Inaddition,fortheareapercent
andhigh–lowmethods,itisnecessarytodeterminetheaccuracyoftherelated
substancesandthedrugsubstance.Fortheexternalstandardmethod,onlythe
accuracyofrelatedsubstancesisrequired.Sincetheresponseofthedrugsub-
stanceinthesamplesolutionisnotusedintheexternalstandardcalculation,it
isnotnecessarytodetermineaccuracyforthedrugsubstance.
ExperimentalConsiderations.Typically,knownamountsofrelatedsubstances
andthedrugsubstanceinplaceboarespikedtoprepareanaccuracysampleof
knownconcentrationofrelatedsubstance.AccordingtotheICH,accuracyshould
bedeterminedusingaminimumofninedeterminationsoveraminimumofthree
concentrationlevelscoveringtherangespecified.Similartotheexperimentsused
inlinearity(seeSection3.3.3)therelatedsubstancesandthedrugsubstancecan
beweigheddirectlyintotheplaceboorserialdilutionfromastocksolutioncanbe
used.Whennoauthenticsampleisavailableforpreparingthespikedsolutions,
onemaydeterminemethodaccuracybycomparingtheresultsoftheoriginal
methodwiththatoftheorthogonalmethod(e.g.,capillaryelectrophoresis,thin-
layerchromatography,normal-phaseHPLC).
42METHODVALIDATIONFORHPLCANALYSISOFRELATEDSUBSTANCES
IntrinsicAccuracy.Intrinsicaccuracyindicatesthebiascausedbysamplematrix
andsamplepreparation.Inthisapproach,astocksolutionispreparedbyusing
knownquantitiesofrelatedsubstanceanddrugsubstance.Thestocksolution
isfurtherdilutedtoobtainedsolutionsoflowerconcentrations.Thesesolutions
areusedtogeneratelinearityresults.Inaddition,theselinearitysolutionsofdif-
ferentconcentrationsarespikedintoplacebo.Thespikedsolutionsareprepared
accordingtotheprocedureforsampleanalysis.Theresultingsolutions,prepared
fromthespikedsolution,arethenanalyzed.Ifthesamestocksolutionisusedfor
bothlinearityandaccuracyandallofthesesolutionsareanalyzedonthesame
HPLCrun,theresponseoflinearity(withoutspikeintomatrix)andaccuracy
(withspikeintomatrix)canbecompareddirectly.Anydifferencesinresponse
indicatethebiascausedbymatrixinterferenceorsamplepreparation.Todeter-
minetheintrinsicaccuracyateachconcentrationlevel,onecancomparethepeak
areaofaccuracy(withmatrix)withthatoflinearity(withoutmatrix)atthesame
concentration(Figure3.11).Thisisthesimplestapproach,andonewouldexpect
closeto100%accuracyatallconcentrationlevels.
OverallAccuracy.Inadditiontothematrixeffectandsamplepreparationerror,
methodaccuracycanalsobeaffectedbycalculationerror:forexample,difference
inrelativeresponse,significanty-intercept,andnonlinearity.Therefore,amore
vigorousapproachistodeterminetheoverallaccuracy,whichincorporatesthe
effectfromallaspectsofthemethod:
?Matrixeffect
?Samplepreparation
?Calculationerror
Inthisapproach,similartothedeterminationofintrinsicaccuracy,theaccuracy
solutionsarepreparedbyspikingaknownquantityofauthenticsamplesanddrug
P
eakarea
Area(withmatrix)Area
(nomatrix)
Accuracysolution
(withmatrix)
Linearitysolution(nomatrix)
x
x
x
x
x
x
Concentration(mg/mL)
Figure3.11.Intrinsicaccuracy.
METHODVALIDATIONEXPERIMENTS43
substanceintoexcipients.TheaccuracysolutionsarethenanalyzedusingHPLC.
Thepercentrelatedsubstanceresultsarecalculatedbytheproposedmethodcalcu-
lationprocedure(e.g.,high–low,areapercent,externalstandard).Anycorrection
factorsareappliedaccordingtotheprocedureproposed.Theoverallaccuracyis
determinedbythefollowingequation(Figure3.12):
overallaccuracy=
%relativesubstance(calculated)
%relativesubstance(theory)
Thisisamorestringentapproach,asthisindicatesthebiascausedbymatrix
interference,samplepreparation,andcalculation.Forexample,relatedsubstance
(found)=1.20%andrelatedsubstance(theory)=1.40%(calculatedfromthe
weightofauthenticsampleusedinthespikedsolution);therefore,
overallaccuracy=
1.20
1.40
×100%=86%
3.3.5Precision
Repeatability.ICHdefinition:Repeatabilityexpressestheprecisionunderthe
sameoperatingconditionsoverashortintervaloftime.Repeatabilityisalso
termedintraassayprecision.
Repeatabilityofamethodcanbedeterminedbymultiplereplicatepreparations
ofthesamesample.Thiscanbedoneeitherbymultiplesamplepreparations
(n=6)inthesameexperimentorbypreparingthreereplicatesatthreedifferent
concentrations.Ingeneral,oneshouldevaluateresultsofindividualrelatedsub-
stances,totalrelatedsubstances,andtheconsistencyofrelatedsubstanceprofiles
inallexperiments.ThepercentRSDandconfidenceleveloftheseresultsare
reportedtoillustratethemethodrepeatability.
Typically,anagedsampleshouldbeusedtoensurethattherearesignificant
levelsofrelatedsubstanceinthesample.Alternatively,ifdifferentsamplesare
availablewithdifferentlevelsofrelatedsubstance(e.g.,freshsampleandsample
x
x
x
Standardcalibration(threepointstandardcurve)
Concentration(%relatedsubstance)
%relatedsubstancefound
P
eakarea
Areafound
Figure3.12.Overallaccuracy(externalstandardmethod).
44METHODVALIDATIONFORHPLCANALYSISOFRELATEDSUBSTANCES
atexpiry),onecandeterminetherepeatabilitybyperformingthreereplicate
preparationsforeachsample.ICHguidelinesrequireaminimumofthreesamples
withthreedifferentlevelsofrelatedsubstance.
Insteadofusingspikesamples(asinaccuracydetermination),drugproduct
lotsthatarerepresentativeofthecommercialproductsshouldbeusedforpreci-
sion(repeatability,intermediateprecision).Thisistoensurethatthecommercial
drugproductisusedinatleastonepartofthemethodvalidationandthatthe
repeatabilityresultsarerepresentativeofthosethatcanbeexpectedinthefuture.
IntermediatePrecision.ICHdefinition:Intermediateprecisionexpresses,within
laboratoriesvariations,differentdays,differentanalysts,differentequipment,and
soon.
Intermediateprecisionistodeterminemethodprecisionindifferentexperiments
usingdifferentanalystsand/orinstrumentsetup.Similartothatofrepeatability,
oneshouldevaluatetheresultsofindividualrelatedsubstances,totalrelated
substances,andtheconsistencyofrelatedsubstanceprofilesinallexperiments.
ThepercentRSDandconfidenceleveloftheseresultsarereportedtoillustrate
theintermediateprecision.
Reproducibility.ICHdefinition:Reproducibilityexpressestheprecisionbetween
laboratories(collaborativestudiesaregenerallyused,forstandardizationof
methodology).
Thisisanoptionalvalidationparameterthatrequiresdemonstrationoflaboratory-
to-laboratoryvariationonlyifmultiplelaboratoriesusethesameprocedure.The
reproducibilitydatacanbeobtainedduringmethodtransferbetweenlaboratories.
3.3.6Range
ICHdefinition:Therangeofananalyticalprocedureistheintervalbetween
theupperandlowerconcentrations(amounts)ofanalytesinthesample(includ-
ingtheseconcentrations)forwhichithasbeendemonstratedthattheanalytical
procedurehasasuitablelevelofprecision,accuracy,andlinearity(Figure3.13).
Typically,linearityandaccuracydeterminationcoversawideconcentrationrange
(e.g.,50%oftheICHreportinglimitto150%ofspecification).However,the
concentrationrangeforprecisionwillbelimitedbytheavailabilityofsample
ofdifferentrelatedsubstancelevels.Therefore,toensureanappropriatemethod
validationrangewithrespecttoprecision,itiscriticaltousesamplesoflow
andhighlevelsofrelatedsubstanceinprecisionexperiments(e.g.,freshand
stressedsamples).
3.3.7Robustness
ICHdefinition:Therobustnessofananalyticalprocedureisameasureofits
capacitytoremainunaffectedbysmallbutdeliberatevariationsinmethodparam-
etersandprovidesanindicationofitsreliabilityduringnormaluse.
METHODVALIDATIONEXPERIMENTS45
Concentration(%relatedsubstance)
Linearityandaccuracy
Precision
Range
P
eakarea
Figure3.13.Range.
GeneralConsiderations.Thisistoverifythatthemethodperformanceisnot
affectedbytypicalchangesinnormalexperiments.Therefore,thevariationin
methodconditionsforrobustnessshouldbesmallandreflecttypicalday-to-
dayvariation.Experimentaldesign(e.g.,Plackett–Burmanscreening,factorial
design)isveryusefultoinvestigatemultipleparameterssimultaneously.Critical
parametersareidentifiedduringthemethoddevelopmentprocess.Onlythese
criticalmethodparametersshouldbeinvestigatedforrobustness.Commoncritical
methodparameterscanbedividedintotwocategories:
1.HPLCconditions
a.HPLCcolumn(lot,age,brand)
b.Mobile-phasecomposition(pH±0.05unit,percentorganic±2%)
c.HPLCinstrument(dwellvolume,detectionwavelength±2nm,column
temperature±5
?
C,flowrate)
2.Samplepreparation
a.Samplesolvent(pH±0.05unit,percentorganic±2%)
b.Samplepreparationprocedure(shakingtime,differentmembranefilters)
c.HPLCsolutionstability
OtherConsiderations.Typically,thevariationsinrobustnessresultsarecom-
paredtotheintermediateprecisionresultstodemonstratethatrobustnessis
notaffectedsignificantlywithinnormalday-to-dayvariation.Whentherelated
substanceresultsareaffectedbysomecriticalexperimentalparameters,apre-
cautionarystatementneedstobeincludedintheproceduretoensurethatthis
parameteristightlycontrolledbetweenexperiments.Forexample,ifpercent
organicofmobilephaseaffectstheresultssignificantly,theprocedureshould
indicatetheacceptablerangeforpercentorganic(e.g.,50%organic±2%)
46METHODVALIDATIONFORHPLCANALYSISOFRELATEDSUBSTANCES
Built-inRobustnessinMethodProcedure.Thefollowingaresomesuggestions
toimprovemethodrobustness:
?Weighingerror.Weighingerrorisusuallythemainsourceoferror.Analyt-
icalprocedureshouldaskforaweighingsampleorstandardofmorethan
10mgtominimizeweighingerror.Inaddition,usetwotothreeindepen-
dentweighingsinthestandardcurveandverifythenominalresponsesof
thesestandardpreparationstoensurethatthereisnosignificantweighing
error.Alternatively,useareapercentcalculationtoeliminatetheneedfor
weighingasmallquantityofreferencestandard.
?Dilutionerror.Pipetteavolumeofmorethan5mL,andavoidusingvolu-
metricflasksoflessthan25mL.
?Sonication.Theefficiencyofsonicationishighlyvariableanddepends
onvariousfactors(e.g.,conditionofthesonicationbath,levelofwater,
andpositionofflaskinthesonicationbath).Mechanicalshakingisrecom-
mended,instead,andismuchmorereproducible.
?Mobilephaseassamplesolvent.Ifpossible,alwaysusemobilephaseasthe
samplesolvent.Thisensuresthecomposition(e.g.,percentorganic,pH)of
samplesolutionmatchesthatofmobilephaseandreducesthechanceof
anyproblemduetoincompatibilityofsamplesolventandmobilephase.
Alternatively,alwaysusesamplesolventweakerthanthatofthemobile
phasetoensurethatthechromatographyisnotdeteriorated.Forexample,
inreversed-phaseHPLC,uselessorganicsolventinthesamplesolventthan
inthemobilephase.
?Buffer.Ensurethatthebuffer(pK
a
)isappropriateforthepHofthesolution.
Ingeneral,toprovideappropriatebufferingcapacity,thepHofthesolution
shouldbewithin±1pHunitofthepK
a
valueofthebuffer.
?Isocraticmethod.Wheneverpossible,useisocraticHPLCcondition,asthis
isaffectedlessbythevariationinflowrate,temperature,anddwellvol-
ume.IfgradientHPLCconditionshavetobeused,asimplelineargradient
ispreferredovermultistepgradients.Complicatedgradientconditionsare
moresusceptibletodifferencesbetweenHPLCinstruments(e.g.,flowrate,
dwellvolume).
3.4COMMONPROBLEMSANDSOLUTIONS
1.Presentationofmethodvalidationdata.Table3.2providesaquickover-
viewofthevalidationdata.
2.Systemsuitability.Duringtherobustnesstestingofmethodvalidation,criti-
calmethodparameterssuchasmobilephasecompositionandcolumntemperature
arevariedtomimictheday-to-dayvariability.Therefore,thesystemsuitability
resultsfromtheserobustnessexperimentsshouldreflecttheexpectedrangeforthe
systemsuitabilityresults.Asaresult,systemsuitabilityresultsinthesemethod
validationexperimentsareveryusefulindeterminingthesystemsuitability
COMMONPROBLEMSANDSOLUTIONS47
Table3.2.SummaryofValidationResults
ICHValidation
CharacteristicAnalysisValidationResults
SpecificityRepresentativechromatograms
todemonstratespecificity.
Alldrugsubstancesandmajor
relatedsubstances(AandB)are
resolvedfromeachother.There
isnosignificantinterferencefrom
excipients.
LinearityDatafromregressionanalysis
(correlationcoefficient,
y-intercept,slope,residual
sumofsquares)andplot.
LinearityisevaluatedforA
from0.05to1.0%andforB
from0.05to2.0%ofthe
nominalsample
concentration.
Datafromregressionanalysis:
CompoundAB
Correlation
coefficient
1.0001.000
y-intercept(%
rel.sub.)
0.022?0.013
Slope(area
units/%rel.
sub.)
87.2275.3
Residualsum
ofsquares
177.81290.7
RangeTheprocedureprovidesan
acceptabledegreeof
linearity,accuracy,and
precisionwhenappliedto
samplescontaininganalytes
withinorattheextremesof
thespecifiedrangeof
procedure.
ThevalidatedrangeforAis0.1to
0.4%,andBis0.05to1.0%.
AccuracyAssessedusingnine
determinationsoverthree
concentrationlevelscovering
therangefrom0.1to0.4%
forAand0.05to1.0%
forB.
Compound%Accuracy
A90–100%
B92–105%
PrecisionTheaverageandstandarddeviationfortheindividualandtotal
relatedsubstances(TRSs)foreachdrugsubstancearereportedfor
eachtypeofprecisioninvestigated.Theoverallmethodprecision
wasevaluatedusingacombinedvariancecomponentanalysis.
Repeatability
AverageStd.Dev.
(n=16)(%)(%)
A0.110.001
B0.080.002
TRSs0.180.013
(continuedoverleaf)
48METHODVALIDATIONFORHPLCANALYSISOFRELATEDSUBSTANCES
Table3.2.(continued)
ICHValidation
CharacteristicAnalysisValidationResults
Intermediate
precision
AverageStd.Dev.
(n=16)(%)(%)
A0.110.005
B0.080.020
TRS0.150.030
Detection
limit(DL)
and
quantitation
limit(QL)
Basedonthepeakareafrom
thedilutedsolutionof
relatedsubstancesinthe
samplematrix,thedetection
andquantitationlimitare
calculatedfromthe
followingequations:
DL(as%ofnominalsample
conc.)=
(3×std.dev.)
slope
QL(as%ofnominalsample
conc.)=
(10×std.dev.)
slope
DLandQLweredeterminedtobe
0.004and0.02%,respectively.
TheQL(0.02%)islowerthan
thecorrespondingICHreporting
threshold(0.1%).
RobustnessTypicalvariationsinmobile
phase:pH,organic
composition,SDS
a
and
EDTA
a
concentrations.
Typicalvariationsinsample
preparation:pH,organic
composition,SDS
concentration,samplesize,
sampletreatment,andEDTA
concentration.
Resultsfromrobustness
experimentswereanalyzedby
statisticalanalyses.Variationsof
allexperimentalparametershave
nosignificanteffectonthe
procedurebasedonanalysisof
themaineffectsofthefactors
evaluatedformobilephaseand
samplepreparation.
Solutionstabilityofstandards
andsampleswereassessed
at5
?
Candatroom
temperature.
Thesamplesolutionsarestablefor
4.5hoursatroomtemperatureand
50hours(approximately2days)
at5
?
C.Thestandardsolutionsare
stablefor24hoursatroom
temperatureand7daysat5
?
C.
a
SDS,sodiumdodecylsulfate;EDTA,ethylenediaminetetraceticacid.
acceptancecriteria.Thisisaveryeffectiveapproachsincetherequiredsystem
suitabilityresultscanbegeneratedduringmethodvalidationandnootherspecial
studyisrequired.However,theseresultsreflecttheexpectedperformanceofthe
system,butnotnecessarilytheminimumperformancestandardforacceptable
results.Forexample,theminimumresolutionofthecriticalpairfrommethod
REFERENCES49
validationmaybe3.5;however,aresolutionof2.0maystillbeacceptableas
longastheyarebaselineresolvedandallotherchromatographicparameters
remainacceptable.
REFERENCES
1.FDAGuidanceforIndustry;AnalyticalProceduresandMethodsValidation(draft),
2000.
2.ICHHarmonizedTripartiteGuideline,ICHQ2A,TextonValidationofAnalyticalPro-
cedures,Mar.1995.
3.ICHHarmonizedTripartiteGuidelineICHQ2B,ValidationofAnalyticalProcedures:
Methodology,May1997.
4.ICHHarmonizedTripartiteGuidelineICHQ3B(R),ImpuritiesinNewDrugProducts,
Oct.1999.
5.UnitedStatesPharmacopoeia,USP25,Chapter<1225>,ValidationofCompen-
dialMethods.
6.Snyder,J.Kirkland,andJ.Glajch,PracticalHPLCMethodDevelopment,2nded.,
Wiley,NewYork1997.
|
|
