2.3.RiskAssessmentofGTICarryover.Ifastartingempiricalapproachestosuchassessmentsbasedonthepointofmaterial,reagent,intermediate,orabyproductisidenti edasaintroductionoftheGTIinquestionandthenumberofstagespossibleGTI,itmayappearthatthebestwaytocontrolitremovedfromtheAPI.Suchanapproach,euphemisticallywouldbetoeliminateitspresencethroughanalternatereferredtoas‘stagecounting’,hasunfortunatelynotbeensynthesisroute.However,inmostcases,thisisanimpracticaluniversallyacceptedbyregulatoryauthorities.Severalregu-approachbecauseanalternatesynthesismayproducelowerlatoryagenciescontinuetoasktheSponsortoprovideyields,havepooreroverallqualitycontrol,andultimatelymayanalyticaldataevenincaseswherecompellingscienti cnotbeeconomicallyfeasible.Moreimportantly,itishighlyargumentswouldsuggestproofofabsence.Suchargumentslikelythatanyalternatesynthesiswillcontainotherimpuritiesshouldbesu cient,particularlywhentheassessmentaddressesthatrequiresimilarcontrolattheTTClevel,andsigni cantthechemistry(reactionsandpuri cations)speci ctoeachstepadditionale ortwillberequiredtodischargetheriskofthosethatisbeing“counted”inaparticularprocess.Nonetheless,itnewimpurities,withoutmateriallyimprovingthesafetyofthewouldbeusefuliftherewereanalternativeapproachthatAPIorreducingtheoverallrisktopatients.ControlofGTIscontainedaquantitative(orsemiquantitative)element,asisratherthaneliminationisthereforestronglyadvocated.Similarpresentedanddescribedinthefollowingsection.considerationsapplytoALARP(aslowasreasonablyItispossibletotakesuchanapproachtoriskassessmentonpracticable)andwereacceptedbyEMAintheQ&Athebasisofthespeci cphysicochemicalpropertiesofaGTI,supplements.8alongwithanunderstandingoftheprocessconditionstowhichTodemonstratecontrol,anassessmentofthepotentialriskitisexposed,toquantitativelyassesstherisk.Theserisk-basedofcarryoverofaGTIatalevelexceedingstagedTTCorTTCassessmentsrelyontheconceptofevaluatingpurgefactorslevelstothedrugsubstanceismade.Thisisalignedwithqualitywhichcanbede nedastheabilityofasyntheticprocesstobydesign(QbD)andtheriskassessmentprinciplesenshrinedpurge(eliminate)aparticularGTI.19,20

inICHQ8andQ9.11,12Aseeminglysimplebutpotentially2.4.PurgeFactors.Theprinciplesupportingthemyopicwaytoaddressthisistodevelopasensitiveanalyticalcalculationofapurgefactorisasimpleone.ItistoidentifymethodandtestforthepossibleGTIinquestionatthepointthekeyphysicochemicalparametersthatin uencetheremovalafterintroduction,atthe naldrugsubstance,oratsomeofaspeci cGTIinaprocess(inQbDterms:criticalqualityintermediatepoint,i.e.QualitybyTesting(QbT).Thisattributes(CQAs))andtousethese,incombinationwiththeapproach,however,canbeatechnicallychallengingandspeci cprocessconditions,todetermineapurgefactor,i.e.howresource-intensiveactivity,especiallywhenappliedtoallthemuchoftheGTIislikelytoberemoved.GTIsthatareassociatedwiththesyntheticprocessandrunsThekeyparametersidenti edarereactivity,solubility,contrarytothetenantsofQbD.Moreoverthisapproachfailstovolatility,ionisability,plusanyadditionalphysicalprocessrecognizethatreactiveGTIswilloftenbedestroyedordesignedtoeliminateimpurities,e.g.chromatography.Theeliminated nalinthesubsequentprocessstepsleadinglatterapproachisoftenemployedinmedicinalchemistrysyntheticroutesemployedforpreparingtestmaterialsfor rstThedrugtotheusesubstance.ofscienti cunderstandinginrelationtoevaluationhumanclinicaltrials.InordertoensurethataconsistentofGTI-relatedriskisgrowing.Severalarticles13,14haveoutlinedapproachistaken,ascoringsystemhasbeendevisedwhichistheuseofQbDprinciplesintheassessmentofGTI-relatedrisk,outlinedinTable2.thesebeingcenteredontheuseofspiking/purgefatestudiesto2.4.1.Reactivity.Thepurposeofthisspeci cparameteristoprovetheabilityofthesyntheticprocessinquestiontoremoveassessthereactivityoftheGTIinquestioninrelationtospeci cGTIs.Suchapproachesareextremelyusefulwhenprocessingconditionstowhichitissubsequentlyexposed.Onappliedinthecontextofa xedprocess.thebasisofreactivitytowardtypicalreagentsWorkconductedbytheProductQualityResearchInstituteencounteredduringworkuporstorage(e.g.,otherchemicals(PQRI)relatingtosulfonateesters15,16(Teasdaleetal.)usedwithintheprocess,water,acid,base,alcohols,etc.)itiscompellinglydemonstratedhowagoodscienti cunder-proposedthatGTIscouldbegenerallyplacedintooneofthreestandingofthepropertiesofacompound,combinedwithacategories(seeTable3).riskassessmentoftheimpactofprocessconditions,canbeusedItiscritical,though,thatanysuchclassi cationmusttakee ectivelytocontrol,oreveneliminatethepotentialGTIriskposed.Suchapproachesasthosedescribedabove,althoughoneforcomprehensive,areimpracticaltoapplytoallpossibleGTIsahighlyreactiveGTI,e.g.acylhalide,andthesecondforalesswithineachandeverysyntheticprocess.Whatissoughtinsteadreactivearylnitrocompound.isarisk-basedmechanismwherebythepotentialriskofThionylchlorideistypicallyusedtoconvertnonreactivecarryovercanbequicklyandsimplyassessed.Formanycarboxylicacidsintoreactiveacylhalides,whichcanthenbeimpurities,chemistry-basedargumentsalonemaybesu cientfurtherprocessed(e.g.,formationofesters,amides,etc.).21toconcludethatanimpuritywouldnotresideinthe nalAPIHowever,duetotheirhighreactivitytheyareverye cientlywithouttheneedforlaboratoryexperimentation.Thisapproachpurgedfromthesubsequentdownstreamchemistryreac-isoftenusedwhenprocessingconditionsforthedownstreamtion(s);thisalsoappliestothionylchlorideitselfwhichreactspurgeordestructionoftheimpurityareobvious.Aconsistentviolentlywithwater.22

frameworkforthistypeofrisk-basedassessmentwherebytheIncontrast,arelativelystablenitrocompoundwouldshowpotentialforcarryovercanbeassessedthroughconsiderationofhighreactivity,undermostreductionconditions,toformanthephysicochemicalpropertiesoftheGTIandtherelevantaromaticamine,butitmaywellbeunreactiveduringprocessingconditionsisrecommended,thepotentialbene tssubsequentstagesoftheprocess.toregulatoryreviewersbeingobvious.SuchanapproachisnotExperimentaldatarelatingtotheprocess(e.g.inprocessanewconcept.BothDobo17andPierson18havedescribedreactionmonitoring)canbeusedtore nethisparameter.For

Table2.Purgefactorsisolationoftheproductasasolid,thegenotoxicreagent/

physicochemicalintermediateshouldremaininthereactionmotherliquorsandparameterspurgefactoraberemovedwhentheliquorsare lteredo .Assignmentofareactivityhighlyreactive=100purgefactorforthisparameteristhusbasedonthesolubilityof

moderatelyreactive=10theGTIinquestionwithintheprocesssolvent.Table4

lowreactivity/unreactive=1outlinesthede nitionsusedwithintheUSPrelatingto

solubilitybfreelysoluble=10solubility.Forthepurposesofthepurgefactorcalculations

moderatelysoluble=3thesehavebeengroupedtomatchthethreeclassesde nedin

sparinglysoluble=1Table1.

volatilitycboilingThescaleusedforthesolubilityfactorhasbeensetovera

processpointsolvent>20°=Cthatofthereaction/range1 10.Experiencehasshownthatthesolubilityfactor

boilingreaction/processpointwithinsolvent.±10°C=of3thatofthecouldmuchhigher,perhapsjustifyingarangeof1 100aswas

boilingusedforreactivity.Howeverwebelievethatthatthemoreprocesspointsolvent>20°=C1thatofthereaction/conservativescaleof1 10shouldberetainedtocompensate

ionisabilityionisationforvarianceinprocessconditionssuchasuncontrolled

fromthatpotentialofthedesiredofGTIproductsigni cantlyddi erentcrystallization,poorwashingand/orine cientdeliquoringof

physicalchromatographyprocesses:chromatography:separation10 100basedonextentoftheisolatedproduct.

physical2.4.3.Volatility.Manylow-molecularweight,potentiallyotherscavengerprocesses:resinse.g.evaluatedonanindividualbasis.genotoxicimpurities(PGIs)suchasaldehydesandalkylhalidesaPurgefactor=concentrationbeforepurging/concentrationafterarevolatile;e.g.methylchlorideformedthroughthereactionofpurging.bmethanolandHCl(duringHClsaltformation)hasaboilingrecrystallisationThisprocessrelateswherebytosolubilitytheimpuritywithininthequestion,contextifhighlyofapointof 23°C.Manysyntheticprocessesemploysolventsoluble,distillationorsolventexchange.Asaresultofsuchaprocess,fromanyvolatileGTIpresentmayalsoberemoved,dependingonsolventthewill

throughdesiredremain

solventproduct.withinc

distillationThisthemotherrefersorsolventtoliquorsthedeliberateandhencebepurged

exchange.removaldThisrelatesofathevolatilityoftheGTIrelativetoeithertheboilingpointof

tothesolventorthetemperatureofthereactionprocess.aqueousadeliberate

ofpHandtochangeorganicattemptlayer,topartitionthedesiredproduct/GIbetweenan

theionised/un-ionisedtypicallyachievedthroughstatetheofmanipulationoneofthe2.4.4.Ionisability.WheretheionisabilityofaGTIandthat

components.ofthematrixinwhichitispresentdi er,forexampleanionisableGTIwithinanonionisableintermediateoraTable3.Genotoxiccompoundsclassi edonthebasisofnonionisableGTIwithinanionisabledrugsubstance,thereactivitypotentialexiststoreducetheleveloftheGTIthroughliquid/liquidextraction,byemployingatwo-phasesystemwithreactivityclassgenotoxicgroupsappropriatemanipulationofthepHoftheaqueousphase.highlyreactiveepoxides/aldehydes/sulfonateTheremovalofanalkylhalideGTIfromanionisabledrug

aziridines/hydrazinesesters/acylhalides/substanceprovidesoneexampleofsuchanapproach.pH

moderatereactivityNprimaryorSmustards/Michaelhalidesreactiveacceptors/halo-alkenes,adjustmentoftheaqueousphasecouldallowthedruglowreactivityaminosubstancetobeextractedintotheaqueousphase;thecarbamatesaryls,nitrocompounds,purinesorpyrimidines,nonionisablealkylhalideGTIremainsintheorganicphasethatcanthensimplybediscarded.Thereafter,thepHoftheaqueousphasecanbeadjustedandthedrugsubstanceback-examplewhereanin-processcontrolshowscompletionofextractedintoanappropriatefreshorganicsolvent.Inadditionreaction(<1%GTIremaining)thiscanbeusedtoclassifythetoclassicalliquid/liquidextractions,solid-phaseextractionGTIconcernedashighlyreactiveforthatstageoftheprocess.(SPE)canalsobeemployed.

2.4.2.Solubility.Whereagenotoxicreagent/intermediateisNote:Aspeci cpurgefactorisonlyassignedinrelationtointroducedintothesyntheticprocesstheprocessisgenerallythisparameterwheresuchanapproachisspeci callyapplied.optimizedtomaximizebothyieldandproductquality.One2.4.5.Chromatography.Althoughpotentiallycostlyincriticalfactorinanyprocessistherequirementthatthecomparisontothesimpleprocessingtechniquesalreadyintendedreactionactuallyoccurs.Thisgenerallymeansthatthedescribed,chromatographyneverthelessdoesprovideaverygenotoxicreagent/intermediateinquestionislikelytobehighlypowerfuland exibletoolforremovalofaGTIwhererequired,solubleinthesolventselectedfortheprocessinquestion.Thisparticularlyforearly-phase,nonoptimized,medicinalchemistry,solubilitywillmeanthatwheretheprocessconcernedinvolvessyntheticroutes.

Table4.Solubilityde nitions23

descriptivetermsolubilityclasspartsofsolventrequiredfor1partofsolutesolubilityrange(mg/mL)solubilityassigned(mg/mL)verysolublefreely<1>10001000freelysoluble1 10100 1000100solublemoderately10 3033 10033sparinglysolublesparingly30 10010 3310slightlysoluble100 10001 101veryslightlysoluble1000 100000.1 10.1practicallyinsolubleorinsoluble>10000<0.10.01

PreparativeHPLC,normallyinanormalphasemode,isanneeded.Inthiscontexttherequiredpurgefactorwouldbeestablishedtechniqueappliedtothereductionorremovalof5000,asbasedonthemaximumlevelofmethylchlorideimpurities,atamultikilogramscale,usuallyatthedrugpotentiallypresentis5%=50,000ppm.Thus,inthisinstanceasubstancestage.TheremovalofaGTIcouldsimplybepurgefactorof>500,000associatedwithmethylchlorideconsideredasubsetofthestandardchromatographicchallengeshouldprecludetheneedforanyfurtherinvestigation.ofimpurityremoval.Thisisexploredinfurtherdetailthroughaseriesofexamples.Analternativetotheuseofchromatographyistheuseofresins.Leeetal.(2010)recentlydemonstratedtheabilityof3.RISKASSESSMENTCASESTUDIES

certainnucleophiliccation-exchangeresinstoe ectivelypurgesulphonateestersfromthe nalstageofanAPI.24Thefollowingsectionoutlinesanumberofexamplesthatserve

Againaspeci cpurgefactorisonlyassignedinrelationtotoillustrateboththeutilityofthepurgetoolanditsrobustness.thisparameterwherechromatographyisspeci callyapplied.2.4.6.CalculationofPurgeFactors.Asoutlined,scoresforScheme1.SynthesisofBAY

43-9006

eachparameterareassignedonthebasisofthephysicochemicalpropertiesoftheGTIrelativetotheprocessconditions.ThescaleisbasedonthepremisethatahighpurgefactorequatestohighGTIclearance.Thus,ahighpurgefactorvalueindicatesalowprobabilitythataGTIwillbeobservedonthebasisofknowledgeofproduct’sphysicochemicalpropertiesandunderstandingofthesyntheticprocess.Foreachstagetheseindividualpurgefactorsarethensimplymultipliedtogethertodeterminea‘purgefactor’forthatstage.TheoverallpurgefactoristhensimplyamultipleofthefactorscalculatedpurgefactorwouldindicatethelevelofaGTItobe>100timesbelowtheappropriateTTClimit,thennofurtheractionshouldtypicallyberequired.Wherethevalueofthecalculatedpurgefactorwouldindicatetheleveltobebetween10and100timesbelowtheappropriatelimit,thenitshouldpotentiallybesupportedbyconductofappropriateanalyticaltesting/furtherprocessinvestigation.Thismaytaketheformof,forexample,periodictestingoruseofpurgeandspikeexperimentsascon rmationforremovinglong-termtesting.WherethecalculatedpurgefactorindicatesthatthelevelcouldexceedtheappropriateTTC,thenanalyticaltestingasaItshouldbenotedthattheexamplesincludebothknownminimumwouldberequired.Itmayalsoprovideanindicationmutagensandpotentialmutagens.Thereisnointenttoinferoftheneedtomodifytheprocessingconditionstofurtheranythingaboutthespeci cstatusoftheimpuritiesinquestion,reducetheriskofcarryover.merelytousethemasexamplesforthespeci cpurposeofHowthismightbeappliedisaddressedinthefollowingdemonstratingtheutilityofthepurgetool.

hypotheticalexamples.Theinitialcasestudiesdescribeindetailtheprocessby

2.4.6.1.Example1.Foraproductdosedat150mgadaywhichpurgevaluesareselectedservingtoillustratetherigorthatisintendedforchronicuse,theTTClimitof1.5ug/dayinvolvedinthisprocess.Casestudies3 5focusonthewould,inconcentrationterms,representalimitof10ppmforaccuracyofthepurgefactorcalculation,ineachcase,throughanyGTIpresent.Shouldtherebeagenotoxicintermediate,comparisonwithactualdata.

thenthecalculationofriskofcarryoverwouldstartatapointof3.1.CaseStudy1:PurgingofHighlyReactiveThionyl100%,i.e.,1,000,000ppmoftheintermediateinthestepinChloridefromSynthesisofBAY-43-9006.Thesyntheticwhichitisproduced.Thus,inthisinstanceapurgefactorofschemefortheproductionofBAY-43-9006isshowninScheme100,000wouldindicatethattheprocessislikelytoreducethe1.Thionylchloride(SOCl2)isintroducedintostage1(ofa4-leveloftheGTIinquestiontoalevelequivalenttotheTTC.stageprocess)toconvertnonreactivepicolinicacid(4)intothe(i.e.,[concentrationbeforepurging,1,000,000ppm]/[purgecorrespondingreactiveacylhalideintermediate(4-chloropyr-factor,100,000]=concentrationafterpurging,(10ppm).idine-2-carbonylchloride)(5).Although,thionylchlorideisThus,inthisinstancecalculatedpurgefactorsof>10,000,000reportedtoAmespositive,thereisincreasingevidencethatthisshouldprecludetheneedforanyfurtherinvestigation.isanartifactofthetestsystem.AmestestsperformedinDMSOCalculatedpurgefactorsbetween1,000,000and10,000,000givepositiveAmes ndings,whereasthesametestperformedinsignalthatfurtherinvestigationmayberequired,and nallyacetonitrileisAmesnegative.25Itiswell-knownthatsul nylchloridescalculatedpurgefactorsof<100,000wouldrequireasaminimumanalyticaltestingoftheGTIinquestion. de(CDMS),canreactaknownwithmutagen.DMSOto26producechlorodimethylsul-TheCDMSisprobablytheagentresponsiblefortheAmespositiveresultofthetestarticle2.4.6.2.Example2.Forthesameproductdescribedabove,inDMSO.reactionmonitoringhasshownthatduringasaltformationTheacylhalide(5)isreactedwithmethylaminetoproducestep,alevelofapproximately5%methylchlorideisformed.thecorrespondingamide(7)in88%yield.TheamideisthenThismeasurementcanbeusedtocalculatethepurgefactorcoupledwith4-aminophenol(8)inthepresenceofbasetorequiredandallowforadjustingthepurginge ciency,ifproducethepenultimateintermediate([4 4-aminophenoxy)-