Merck(Online-Blending)默克(在线混料)

On-lineblenduniformityofpharmaceuticalformulationsbyNear-IRNathanPixleyContributingauthors:JeffGivand,RobertL.Green,JohnP.Higgins,ErnestinaLuna,ArthurMateos,MarkSullivan,RobertA.Reed,GertThurauMerckResearchLaboratories,MerckandCo.,Inc.,P.O.Box4,WP78-110,WestPoint,PA19486利用近红外测定药剂的在线混料均匀度OutlineBackground–blendingunitoperation&characterizationofblendsNIRimplementation,acquisitionofspectratriggeredvs.continuoussamplingconsiderationsDataanalysis–movingblock,wavelengthcorrelationReproducibility,endpointdeterminationApplytoprocessingdevelopment&scale-upFillvolume,scale-up,activevs.blankformulationOutline背景–混料仪器操作和混料特征NIR应用，光谱获取触发vs.连续采样注意事项数据分析–移动块,波长相关性重现性,终端判定应用于处理开发和混料机尺寸增大情况填料体积，混料机尺寸增大,活性vs.无效配方BlendingofpharmaceuticalpowdersProcessoverview:SequentialadditionofmaterialsSpinblenderEquipment:Bohletoteblenders(binblenders)Scale:Pilot(300L)&Full-scale(1800L)Drypowderblending(API&excipients)&lubrication(ofpowderblends,orgranules)Processconditions–fillvolume,RPM’s,&formulationattributes药品粉末混合过程概述:连续添加原料旋转混料机设备：Bohletote混料机(桶式混料机)尺寸:试验尺寸(300L)&全尺寸(1800L)干粉末混合(API&辅料)&润滑(粉末混合，或颗粒)过程条件–填料体积、转数和药剂属性CharacterizationofblendingprocessesSamplethieving:spatialsamplingatdiscretetimepoints->analyticalassayProvidesvariabilityofmaterialvs.timeCanidentifylocation-specificblendperformanceDrawbacks:TimeresolutionRepresentativesamplingResourcesrequiredAnopportunityforPAT…Near-IRSpectraateachrevolutionSampleprovidedbyprocessReal-timeinformation;automation混料过程特征取样:在离散时间点内进行空间取样->分析含量提供有差异的原料vs时间可以区分特定地点的混料性能缺陷:时间分辨率代表性取样需要资源是PAT的机会…近红外每次旋转的光谱过程中进行采样实时信息，自动化操作ImplementationapproachBrimroseLuminarNear-IRspectrometerBattery-powered,wirelesscommunicationInterfacedtolid–measurementthroughsapphirewindowAcquisition:1100-2300nm,6nmincrementPilotscale Full-scale实施方法BrimroseLuminar近红外光谱仪电池供电，无线通讯直接检测–通过蓝宝石视窗检测获取:1100-2300nm,6nm增量试验尺寸

全尺寸AcquisitionofspectraContinuousacquisition1spectrum~20%Rev.SeparationofspectraSinglewavelengthabsorbancevalueMinimalimpact…Thresholdvalue#spectra/revolutionincludedDirectindicationofwindowfouling(lackof)ApplytowardoptimizingtriggeredacquisitionTriggeredacquisitionS/NiftriggeroptimizedMoreamenabletoautomation,routineuseToteuprightandstationarySpectraofformulationSpectraofairChangesinblenderdirectionBlendingstopped.ToteuprightandstationarySpectraofair+formulation光谱获取连续获取1光谱~20%转.光谱分离

单波长吸光度值最小影响…阈值#包含光谱/转数窗口直接显示(无)应用于最佳触发获取触发获取触发优化后，S/N更适合自动化，常规应用ToteuprightandstationarySpectraofformulationSpectraofairChangesinblenderdirectionBlendingstopped.ToteuprightandstationarySpectraofair+formulationNear-IRsamplingofbatch~10mgformulationinterrogatedformulation&instrumentdependentStationaryvs.dynamicmeasurementCer-revolutionsComparisonwithsamplethief(example)NIR:10mgx100revolutions=1gAseriesof“micro-assessments”withhightimeresolutionThief:100mgsamplex10locations=1gMacro-assessmentatonerevolutionStopblenderbasedonasinglemeasurement(revolution)?近红外批次取样~10mg需测定的配方药取决于配方和仪器静态vs.动态检测考虑外部-vs.内部-转数与取样对比(实例)NIR:10mgx100转=1g一系列“微观评估”，高时间分辨率

取样:100mg样品x10个位置=1g一转处的微观评估

基于单次检测（转数），停止混料机?DataanalysisapproachesMeanspectralstandarddeviationNocalibrationSpectralvariability->formulationvariabilityPre-processing:SNVWavelengthcorrelationCorrelationvalueofindividualspectratolibrary(dotproduct)Librarybasedonend-pointspectrafrompreviousbatchesPre-processing:2ndderivative123456RevolutionNumberSpectra数据分析方法平均光谱标准偏差无校准光谱变异性->配方变异性预处理:SNV波长相关系数单光谱与光谱库（点积）的相关值光谱库基于先前批处理的终点光谱预处理:二次微分123456转数光谱Blendprofiles:movingblockspectralstandarddeviationLargeblocksizeSuitableendpointbasedony-axisvalue(1point)orprofileslopeSmallblocksizeSuitableendpointbasedonconsecutivey-axisvalues(multiplepoints)byminimalprofilevariability混料图:移动块光谱标准偏差大移动块适当终点

基于Y轴值（1点）或图示斜率小移动块适当终点

基于连续的Y轴值（多点）byminimalprofilevariabilityProfilereproducibility:Movingblock9batchesSameprocessingconditionsAPIblendingBlocksize=3Blocksize=20Profile重现性:移动块9批相同处理条件API混合移动块尺寸=3移动块尺寸=20Profilereproducibility:wavelengthcorrelation9批相同处理条件API混合EachprofilevaluerepresentsasinglerevolutionEndpoint–multipleprofilepointsProfile重现性:波长相关性9批相同处理条件API混合EachprofilevaluerepresentsasinglerevolutionEndpoint–multipleprofilepointsMovingblockvs.wavelengthcorrelationWavelengthcorrelationThresholdbasedonendpointvariabilityacross9batches17-26Revrepresents10consecutive“passing”valuesCorrespondencetomovingblockprofilevalue移动块vs.波长相关性波长相关性9批内阈值，基于终点可变性17-26转代表10连续“传递”值与移动块对应CorrespondencetothiefdataSamplethiefandNIRappliedtosamebatchEarlytimepointsamplethievingThiefdatabasedon10locations/timepoint对应取样数据在相同批次中使用取样和NIR方法早期时点取样10个位置/时点的取样数据Applicationtoprocessdevelopmentandscale-upFillvolumeAPIvs.blankformulationAPI/excipientblendingLubricationPilotvs.full-scaleAPI/excipientblendingLubricationBlenddynamics(largefillvolume)应用于处理开发和混料机体积增加填料体积

APIvs.无效配方API/辅药混合润滑试验尺寸vs.全尺寸API/辅药混合润滑混合动力(大填料体积)FillvolumeeffectAPIblending,pilotscale5%differenceinfillvolumeMovingblockWavelengthcorrelation填料体积效果API混料,试验尺寸填料体积，5%差异移动块

波长相关性Activevs.blankformulationActivevs.blankformulationPilotscale活性vs.无效配方活性vs.无效配方试验尺寸Activevs.blankformulation:lubricationLubricationofactivevs.blankformulationPilotscale活性vs.无效配方:润滑活性vs.无效配方的润滑试验尺寸Processscale-upPilotvs.full-scaleSameformulation过程混料机尺寸增大试验vs.全尺寸相同配方Processscale-up-lubricationPilotvs.fullscaleLubricationofblankformulation过程混料机尺寸增大-润滑试验vs.全尺寸无效配方润滑Blenddynamics–largefillvolumeLubricationofgranuleformulation~95%fillvolumeLubricantaddedthroughbottomofblender混料动力–大的填料体积颗粒配方的润滑~95%填料体积通过混料机底端添加润滑剂Summary&ConclusionsArapid,non-invasive,&information-richapproachtoblenduniformityhasbeenappliedtowardprocessdevelopment,scale-up,&monitoringTwodataanalysisapproachesconsidered:MovingblockCalibration-freeEachpointcorrespondstomultiplerevolutionsWavelengthcorrelationRequirespre-definedlibraryEachpointcorrespondstoonerevolutionEndpointcriteria--basedonsamplingconsiderationsanddataanalysisapproachNear-IRandtraditionalcharacterization(samplethief)arecomplementarytoolsDifferentinformationValueduringdevelopment总结与结论在处理开发，体积增加和监控过程中应用了一个快速无损和信息丰富的方法测定混料均匀性使用了两种数据分析方法:移动块

免校准每个点对应多转数