University of Groningen A Design of Experiment approach to predict product and process

15 Pages·2016·3.97 MB·English

by  Gaurav Kanojia

University of Groningen

A Design of Experiment approach to predict product and process parameters for a spray dried
influenza vaccine
Kanojia, Gaurav; Willems, Geert-jan; Frijlink, Henderik W.; Kesrten, Gideon; Soema, Peter;
Amorij, Jean-Pierre
Published in:
International Journal of Pharmaceutics
DOI:
10.1016/j.ijpharm.2016.08.022
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from
it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2016
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Kanojia, G., Willems, G., Frijlink, H. W., Kesrten, G., Soema, P., & Amorij, J-P. (2016). A Design of
Experiment approach to predict product and process parameters for a spray dried influenza vaccine.
International Journal of Pharmaceutics, 511(2), 1098-1111. https://doi.org/10.1016/j.ijpharm.2016.08.022
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the
author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license.
More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne-
amendment.
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the
number of authors shown on this cover page is limited to 10 maximum.
InternationalJournalofPharmaceutics511(2016)1098–1111

ContentslistsavailableatScienceDirect

International Journal of Pharmaceutics

journal homepage: www.elsevier.com/locate/ijpharm

A Design of Experiment approach to predict product and process

parameters for a spray dried influenza vaccine

Gaurav Kanojiaa,b,1,Geert-Jan Willemsa,Henderik W. Frijlinkb,Gideon F.A. Kerstena,c,
Peter C . Soemaa,*,2, Jean-Pierr e Amorija,2 ,3

aIntravacc(InstituteforTranslationalVaccinology),AntonievanLeeuwenhoeklaan9,3720ALBilthoven,TheNetherlands
bUniversity ofGroni nge n,Departmen tofPharmace uticalTe chno logyandBiopharm ac y,An ton iusDeusin glaan 1,9713AVGroningen,TheNetherlands
cDivisionof D rugDelivery Technology, Le idenAcademicC entreforD rugR esearch,Leide nUnivers ity,Einsteinw eg 55,2 33 3CCLeiden ,The Netherlands

A RTICL E I NFO AB STRACT

Articlehistory:
Receiv ed4May2016 Spraydriedvaccineformulationsmightbeanalternativetotraditionallyophilizedvaccines.Comparedto

Receivedinrevisedform28July2016 lyophilization, spray drying is a fast and cheap process extensively used for drying biologicals. The

Accepted10August2016 current study provides an approach that utilizes Design of Experiments for spray drying process to
Available on line12 August2016 stabilize who le inactiv ate d influenz a vi rus (WI V) vacc ine . The approa ch includ ed sys tematica lly

screeningandoptimizingthespraydryingprocessvariables,determiningthedesiredprocessparameters
Keywords: andpredic ting productqu alit ypara meters .Thepro cesspara metersinleta irte mperat ure,noz zlegasflow
Spraydrying rate andfeed flowrate andth eireffecto nW IVvacci nepowder chara cte risticssucha sparti cle size,
Proces s optimization resid ual moist urec onte nt(R MC) andpo wd eryie ldwere investig ated.Vaccinep owde rs withab road
Process parameters rangeof physicalc haracter istics(R MC 1.2–4.9% ,parti clesiz e2.4–8.5mm andpow deryield4 2–82 % )were
QNuozazliltey gbays dfleoswignrate obtain ed . WIV s howed no sig nifica nt loss i n antige nici ty as r evea led by hem agglu tination test.
Furthermore,descriptivemodelsgeneratedbyDoEsoftwarecouldbeusedtodetermineandselect(set)
spray drying process pa rameter . This was u sed to genera te a drie d W IV powder w ith predefi ned

(predicted)characteristics.Moreover,thespraydriedvaccinepowdersretainedtheirantigenicstability
even after storage for 3 months at 60 (cid:1)C. T he ap proach used her e enable d the generati on of a

thermostable, antigenic WIV vaccine powder with desired physical characteristics that could be

potentiallyusedforpulmonaryadministration.
ã2016Th eAut hor s.Published byElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense

(http://creativecommons.org/licenses/by/4.0/).
1.Introduction Moreover,drypowdervaccineformulationshavethepotentialto

be used for alternative vaccine delivery routes, such as the

Manyexistingvaccinesarecurrentlydistributedandadminis- intranasal, pulmonary or oralroutes (Dicko et al., 2000; Giudice
teredina liquidfo rm.Liqui dva ccinesnee dtobestore dat 2–8(cid:1)Cto andCampb ell,2006;T on nise tal.,20 12;Am or ijet al.,20 10).

remain stable. This dependency on a steady cold chain makes An established method to produce dried biologics is spray

vaccine distribution complex and expensive, especially in devel- drying. Spray drying has the advantage over traditional drying

oping countries (Wang,1999). Dried vaccines can overcome this techniques(suchasfreeze-drying)thatitisrelativelyfastandhas
requir ementfor acoldc hain,a sthey possessa lon gershelfli feat lower oper ating cos ts. Moreover, it re su lts in a disp ersi ble fine

elevatedtemperatures(Geeraedtsetal.,2010;Smithetal.,2015). powder compared to a dry cake as obtained by freeze-drying,

which may enable further powder handling and usage for

alternativedeliveryroutes.

Powders with different physiochemical and morphological
qualAitbybrteavrgiaettiopnrso:d QucbtDp, roqfiulaeli;tWy IbVy, wdehsoilgeni;n aDcotEiv, aDteedsiignnfl uoefn zEaxpveirruims;eXnRtsD; ,QX-TrTaPy, properties c an be o btained b y spray drying. Th e pow der properties
dependontheappliedprocessparametersandcompositionofthe
diffractometry; DSC, differential scanning calorimetry; RMC, residual moisture
content. liquid feed (Crowe et al., 1994; Jain and Roy, 2008). The spray

* Correspondingauthor. drying process consists of nebulization of a liquid product,
E-mailaddress: [email protected](P.C. Soema). genera ting aero sols, int o a heated gase ou s drying medium,

1 Firstauthor.
2 Shar edlastauthors. resulting in a dry powder (Fig. 1). The large surface area of the
3 Current add ress: Virtuvax BV, The Netherlands. aerosols results in a relative rapid drying process. Depending on
http://dx.doi.org/10.1016/j.ijpharm.2016.08.022
0378-5173/ã2016TheAuthors.PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).
G.Kanojiaetal./InternationalJournalofPharmaceutics511(2016)1098–1111 1099

Fig.1. Spraydryingprocess:Aliquidinfluenzavaccine(wholeinactivatedinfluenzavirusvaccine)wasspraydriedwithtrehaloseasanexcipienttoproduceapowdervaccine.
Invest igated proces s parame te rs are indicated in red. Aspirat or capacity was kept fixed at 22 m 3n/h (high est as pirat or setting po ssi ble).
the size of the spray drierand airflowrate, the drying may take requiresalargenumberofexperimentsandinteractionsbetween
betw een 0 .2–3 0s (A non., 1999 ). Duri ng dr ying , the p rote in in paramet er sare frequent ly missed.ADes ign ofExperimen ts(DoE)

evaporating droplets may experience reversible or irreversible approachcanbeusedinsteadinordertosystematicallyscreenand

denaturation. This could be due to the loss or weakening of optimizeprocesses.DoEisastructuredapproachthatcanbeused

hydrogen bonds and simultaneous increase in hydrophobic toidentifycriticalandnon-criticalparameters,andtheirrespective

interactionsduringevaporationofwater.However,theself-cooling interactions,ofaproductionprocess.Moreover,itcanbeusedto

effectofdropletsduetowaterevaporationpreventsthetempera- quantifytheimpactofrawmaterialsandprocessparameterson

ture increase of the droplet surface above the wet bulb theproductcharacteristicsandquality(Cooketal.,2013).Several

temperature (temperature of drying aerosols achieved through studieshaveemployedaDoEapproachtoinvestigateandoptimize

evaporationcooling)(Katja,2011).Thus,spraydryingmaybean thespraydryingprocessofproteins(Prinnetal.,2002;Maltesen

appropriate procedure for drying thermolabile vaccines and has et al., 2008) and liposomal adjuvants (Ingvarsson et al., 2013).

beenusedtoproduceexperimentaldrypowdervaccinesagainst However,thepotentialofutilizingDoEforproducingspray-dried
meas les(L in etal.,201 1;Ohtakeeta l.,2 010),infl uenza(Lo valenti powderv acci neshasn ot beenexp lored so far.

et al., 2016; Saluja et al., 2010; Scherliess et al., 2014; Sou et al., Tomaintainthestructuralintegrityduringandafterthedrying

2015),tuberculosis(Wongetal.,2007)andhepatitisB(Chenetal., process, biological products such as proteins or vaccines often

2010). Moreover, dry powdered measles vaccine has showed requireexcipientsthatactasstabilizersintheirformulation.The

promising results in phase 1 clinical trials (MVDP author group sugar trehalose is an excipient commonly used for stabilizing

etal.,2014).Therefore,spraydryingmightbeasuitablealternative vaccines, due to its good protein-stabilizing characteristics

methodtoobtaindrypowdersofavarietyofvaccines(Amorijetal., (Geeraedtsetal.,2010;Ogainetal.,2011).Includingtrehalosein

2008). the vaccine formulations for spray drying might therefore be

Thespray-dryingprocessusedtoproducethesepowderswith essential to obtain a stable vaccine product after spray drying.
desired product cha racteris tics is u sually o ptimiz ed by a one- Previousl y,s praydr yi ngofin fluenza vaccines hasb eende scribed
factor-a t-a-time (OFAT) approa ch, where the effect of pr ocess using vario us sug ars, M aa et al. (Ma a et al., 200 4) we re first to

parametersontheproductareassessedinalinearfashion,one-at- describetheuseoftrehalose,Souetal.(2015)combinedtrehalose

a-time. This consumes a lot of time and resources. Moreover, it with leucine, whereas Scherliess et al. (2014) showed the
1100 G.Kanojiaetal./InternationalJournalofPharmaceutics511(2016)1098–1111

superiority of trehalose over mannitol. Audouy et al. (2011) and Nitrogen being inert in nature was preferred to avert any

Salujaetal.(2010)describedtheuseofthespraydryingprocess unwantedreactionthatmightoccurinthepresenceofairasdrying
withi nul in asstab ilizer. medium.A two-wa yno zzlew ithor ifi ced iameter of 0.7 m mwas

The goal of this study was to investigate the use of a DoE used in a co-current mode with nitrogen as atomizing gas. The

approachtosystematicallyscreenandoptimizethespraydrying nitrogen pressure was set constant at 5bar. The spray drying
process p ara meters (inlet a ir tem pera ture, nozz le g as flo w rate, paramete rs were varie d in accorda nce w ith the experi mental
andfeed flowrate) andpr edic tingtheproce sssett ings need edto design mat rix (T able 2) . T he feed flo w rat e is displayed in
achi eve the target ed product qu alit y param eters l ike out let percen tage(%) onthee quip men tand feedfl owra te of5,10or1 5%
tempera ture , particle size, pow der yiel d and residu al m oisture correspond ing to exp erimentally de term ined flow r ate s ran ged
content. As a model antig en, whol e-inac tivat ed influe nza virus from1.0,3.4an d4 .5mL/min,respe ctively.Ana tomiz ingai rflowof
(WIV)va cci ne wasus ed,with trehaloseasastabi lizingexci pient. 7.3(cid:3) 17.5 L/m inc orre spondst oasettingfro m3 0to50mm (norm al
Using DoE sof twar e, an expe rimental des ig n was gen erated to lite r[ L ]i sthev olumeat0(cid:1)C a n d1atm ).The as pir ato rra teswere
assess the impact of various proce ss para mete rs (inlet air seta t2n2 m 3 /h inallex pe rime nts; th iscor resp ondedtoi nstru ment
tempe ratur e, nozzl e g as flow rate, an d feed flow rate) and sett in go f10n0% .
trehalose co ncentrat ions on t he fin al p roduc t cha racter istics Afte rs praydryingthespraydriedproductwascollectedand,

(particle size, powder yield and residual moisture content). aliquoted(100mg)invials(3mLvial,NuovaOmpi)inaglovebox
Regressio n mo dels wer e fitte d on the meas ured outp ut param- underare lativ ehu mi dityo f< 3% (Ter raUniv ersalIn c, S eries1 00)
eters, and the pred iction powe r o f th e model w as asse ssed by andse al ed.They ieldwasd efi ned asthe ratiobetw een theam ount

selecting three untested combinations of process parameters ofpowderobtainedandtheamountofsubstanceintroducedinthe

within the investigated design space. Finally, the antigenic liquidfeed(Eq.(1))(Maltesenetal.,2008).Theweightofbuffer

recoveryandthermostabilityoftheobtainedspray-driedvaccines saltswasnotincludedinyieldcalculationastheywereconstantfor

wasassessed. allformulations.

Collectedpowderweight½g(cid:4)
PowderYield½%(cid:4)¼ (cid:5)100 ð1Þ
2.Materialsandmethods Trehaloseinthefeed½g(cid:4)

2.1.Influenzavaccine
2.3.Targetproductprofile

Influenza A/PR8/34 WIV was obtained by inactivating egg-
propagated influenza virus with b-propio lact one as desc ribed The quality target product profile (QTTP) was defined, and
processparametersthatmayhaveanimpactonresidualmoisture
previously (Hendriks et al., 2011). The bulk vaccine was
concentrat ed with Cen trip rep centrifu gal fil ters (M illipore) with content (RMC), particle size, powder recovery and outlet temper-
ature were determined from previous studies on spray drying
amolecularweightcut-off(MWCO)of10kDa,formulatedinHBS
(2 0 mM HEP ES, 125 mM Na Cml, 9 mM Ca Cl2, 5 mM MgCl2). Th e final (QPTrPinPnd eets acrl.i,b 2e0s0t2h;e Inchgvaararscsteorni settic asl.o, f20th13e;fi Ongaalinp reotd aulc.,t 2(0T1a1b)l.e T1h)e.
WIV stock contained 800 g/mL HA that was determined by
Theparametersinvestigatedwereinlettemperature,atomization
surface plasmon resonance as described previously (Hendriks airfl ow rate (no zzle pressure ), fee d flow rate and fe ed excipient
etal.,2011).Thevaccinesolutiontobespraydriedwaspreparedby
concentration. Additional factors such as raw materials and
mixing the WIV stock with D-(+)-trehalose dihydrate (Sigma-
operatorwerekeptconstant.Theinvestigatedspacewouldresult
Aldrich) solution in PBS (resulting in weight ratio hemagglutinin inadesig nspa cewh ereoneor mo reofthetarge tedpr oductp rofile
protein(HA)/trehalose:1/400).Theratioofproteintosugarwas
couldbeachieved.
chosenbasedonpreviousresearchexperiencewithspray-freeze
drying ofinflu enz avaccine s(Geerae dtsetal.,20 10).T hetrehalose
solutio nw asfiltere dusinga 0.45mmM il lex- HVfilt er(M illipore) 2.4. Design of Experiments (DoE)

priortomixingwithWIV.
TheDoEmodelwaspreparedandevaluatedusingMODDE10.0
2.2. Spray drying of influenza vaccine formulation ((UMmLRe)tr aicnsd A Bad).j uMsote ddelsb ywererem fiotvtien dg wn iothn -msiuglntiifiplc ea nlitnem aro dreeglrtees rsmiosn.

Priortoourstudy,screeningexperimentswereperformedusinga
WIVpowderswereproducedusingaBüchiminispray-drierB-
fullfactorialdesigntodeterminethemostrelevantinputprocess
290inconjunctionwithahighperformancecycloneandaB-296
dehu m idifier (both from B üchi Labortechnik AG). Al l the e xperi- parameters that affected the output process and product
ments were perfo rmed in a closed loop confi gur ation using wpaarsamdeecteidrse.d Sitnocefi tnhde twheeigahptp rraotpiori aotfe HpAr/otrceehsasloinsep uwtaps a1r/a4m00et, eirt
nitrogenasdryingmedium.
rangeswithoutusingtheantigen.Thereby,assumingthatsuchlow

Table1
Qualit ytargetedproductprofile.DescribesthedesiredqualitytargetprofileforadrypowderWholeinactivatedinfluenzavirusvaccine.

Desired Target Rationale

response
Particlesize 1–5mm Particlesizethatissuitableforinhalationalandparenteraladministrationafterreconstitution.

Residual 3%orless Withintheprescribedlimitsfordriedbiologicals(Mayetal.,1992;TheFoodandDrugAdministration,

moisture 1990)
content
PowderYield 70%ormore Powderyieldshouldbesufficienttomaketheprocesseconomicallyfeasible.

Stability Lessthan10%lossinHAtitersduring Powdervaccineshouldbeatleastasstableasthedescribedstabilityforwholeinactivatedliquid
stora gefo r12 mon th sat 2–8(cid:1)C influenz avaccin es(Farn sw or thet al., 2011; Ku mru etal.,201 4).Theh um andos e15mgofH Acanbe

calculatedtheoreticallybasedonpowderyieldofthevaccine.
G.Kanojiaetal./InternationalJournalofPharmaceutics511(2016)1098–1111 1101

Table2
Exper imental design matrix. The design matrix showsthe input parameters setfor spray drying (inlet temperature, nozzle gasflowrate, feed flowrate, trehalose

concentration) and the output parameters that were experimentally determined [Particle size(X50median particle size) and span, outlet temperature, RMC and powder
yield].Aspiratorwassetatmaximuminstrumentsetting.

Exp Inlettemp Nozzlegasflowrate Feedflowrate Trehaloseconc Particlesize Size(span) Outlet Moisture PowderYield
No. ((cid:1)C) (L/min) (mL/m in) (mg/mL) [X50] (m m) (X90 (cid:3) X10)/X50 temp ((cid:1)C) content (%) (%)
1 110 7.3 1.0 100 6.1 1.7 67 4.2 63

2 160 7.3 1.0 100 6.9 2.0 91 1.4 72

3 110 17.5 1.0 100 2.4 2.1 58 2.4 82

4 160 7.3 4.5 100 7.3 2.0 78 3.8 53

5 110 17.5 4.5 100 5.7 2.0 48 7.7 64

6 160 17.5 4.5 100 3.2 1.5 63 3.1 72

7 110 7.3 1.0 150 6.8 1.3 63 4.1 61

8 110 17.5 1.0 150 2.9 1.6 61 2.6 75

9 160 17.5 1.0 150 2.9 1.2 83 1.9 69

10 110 7.3 4.5 150 8.5 1.8 53 2.9 42

11 160 7.3 4.5 150 7.9 1.8 76 3.6 52

12 160 17.5 4.5 150 3.1 1.5 73 2.3 69

13 110 12.4 3.4 125 4.5 1.8 55 3.7 69

14 160 12.4 3.4 125 4.3 2.0 78 2.4 70

15 135 7.3 3.4 125 7.5 1.9 68 4.9 57

16 135 17.5 3.4 125 3.3 2.1 65 3.1 76

17 135 12.4 1.0 125 3.7 1.8 74 2.4 77

18 135 12.4 4.5 125 4.2 1.6 61 3.3 69

19 135 12.4 3.4 100 3.8 2.1 67 2.9 78

20 135 12.4 3.4 150 4.2 2.1 62 2.8 62

21 135 12.4 3.4 125 4.1 1.6 63 3.1 67

22 135 12.4 3.4 125 4.5 2.0 66 3.2 72

23 135 12.4 3.4 125 4.4 2.1 66 3.5 73

antigenweightratiowouldnotsignificantlyinfluencetheoutput measurements were done in triplicate with 0.2mL of the

parameters. The model was further optimized using a Central reconstitutedWIVsamples.Sampleswerepreparedwithvaccine
Composite Fami ly (CCF ) des ign, inc luding WI V as t he model powderrecon stitu tedbyge ntleshak ingin purifiedw ater (MiliQ)
antigen. A reduce d CCF design was use d for o ptim ization, (HA0.1m g/mL)atano per atingte mperatu re of25(cid:1)C. Homo geneity
consistin g of in tota l 23 experim enta l runs . Th e choice of a of th e s ize dist rib uti onwas re flectedin the p olydis persity index

reduced CCF design was made due to fewer experimental runs (PdI).

giving the same amount of information as the CCF. To reduce

systematic errors,all the experiments were completelyrandom- 2.6.Residualmoisturecontent(RMC)

ized.
The RMC of spray dried influenza vaccine samples was

2.5.Antigencharacterization determined using a C30 Compact Karl Fischer Coulometer

(Mettler-Toledo). Samples of approximately 100mg of dried

2.5.1.WIVhemagglutinationtiter powder vaccine in vials were reconstituted in 1mL HYDRANAL

To determine the hemagglutination titer of the WIV in dried Coulomat A (Sigma-Aldrich) and subsequently injected into the

vaccine powder, a hemagglutination assay was performed as titration vessel. Each sample was measured in triplicate. The

describedpreviouslybySoemaetal.(2014).Spray-driedvaccines relativeandabsolutemoisturecontentwerecalculatedfromthe
were reco nstituted a cco rdingly in p urified water (MilliQ ) corre- standar dplo t,basedo nthewei ghtofth edried productin thev ial,
m m
sponding to a HA concentration of 0.1 g/ L. Finally, PBS was volumeofthereconstitutedreagent,volumeofextractedsample

added to the reconstituted vaccine to obtain a 1:10 dilution. injectedintothetitrationvesselandtheblanktitration.

Dilutedvaccinesolutionweretransferredtoa96-wellsV-bottom

plate(Greiner)andseriallydilutedtwo-foldwithPBS.Allthewells 2.7.Physicalcharacterizationofthevaccinepowders
m
consisted of 50 L after dilution. Next, an equal volume of a 1%

suspensionofturkeyerythrocytes(Harlanlaboratories)wasadded 2.7.1.Geometricparticlesize
to the wel ls and th e plates we re incu bated for 1h at room Th egeometr icpartic lesize(X definedasthemedianparticle
50
temperature. The titer was determined by visual observation of size) of spray dried powder product was analyzed by laser

agglutinationofRBCinthewells andsubsequentlyexpressedas diffraction with a Helos system (Sympatec GmbH). The powder

the reciprocal of the highest dilution that yielded complete was dispersedintotheHelossystemusinganaspirosdispersing

hemagglutination. The titers were determined on reconstituted system operated at a dispersing pressure of 1.0bar. The vaccine

powdersjustafterspraydrying(t=0)andwererepeatedonstored powderwasmeasuredwithlenshavingameasuringrangeof0.1/
vaccine p owd er o ver a period o f 3 mon ths a t interva ls of one 0.18–35 mm. Furtherm ore, t o che ck for an y aggregat es the d ried

month.HAtiters(activity)ofthesamplesateachconditionwere vaccine powder was measured again with a lens having a
calculat ed relativ e to the st artin g liquid m ixtur e of WIV with measur ingrange of4.5 –875mm.M oreov er,inc rea sing thedispe r-
excipienta t4(cid:1)C.Th eL og2 HAtiters weree xpressed in perce ntage sion pressu re to 5b ar did n ot re sult in cha nge of the me asured

forcomparisonbetweendifferentsamples. particlesizedistribution,whichindicatesthatthesizedistribution

oftheprimaryparticleswasobtainedat1bar.Resultsarethemean

2.5.2.Dynamiclightscattering(DLS) of three measurements. The span of the produced powders was
Th e size of WIV after rec onstitution was measured using a ca lculate dusingtheequa tion :span = (X (cid:3)X )/X .Thespan isa
90 10 50
Zetasizer Nano-ZS system (Malvern Instruments). DLS
1102 G.Kanojiaetal./InternationalJournalofPharmaceutics511(2016)1098–1111

measure for the homogeneity/polydispersity of the particle size with a LynxEye Si strip one-dimensional detector (both from
a
distribution. BrukerAXS). Thesampleswere exposed toCu K (X-rays)atan
angular rangi ngw asfrom 5(cid:1)to6 0(cid:1)2&z.T he ta; with asteps ize of
2.7.2.Differentialscanningcalorimetry 0.01(cid:1)an dadwel ltim eof0 .5s.T hecrystallinest atuso f thep owd er

Modulated differential scanning calorimetry (mDSC) was was assessed qualitatively by examination of the resulting

conducted using a TA DSC Q100 (TA instruments). Samples diffraction patterns. The upper limit of detection of crystallinity
weighing b etwee n 10.0 –15.0 mg w ere c rimped in her metically is2%ofth etotalsam ple volum e.

sealed pans for measurement. The glass transition temperature

(Tg)wasdeterminedbymodulatedDSC(MDSC);thesampleswere
cool edto 10(cid:1)Candth en heatedto1 80(cid:1)C atarate of2 .0(cid:1)C/min .The 3. Results
modula ti on am pli tude was s et at (cid:6) 0.3 1 8(cid:1)C ev ery 60s. The
midpoint in deflection in the r ever se h eat flow w as take n as th e Tg. effeAc tsDooEf faepedprfloaocwh rwataes, iunsleedt tteom spyesrtaetmuraet,icnaollzyz lienvgeasstiflgoawte rtahtee

(nozzle pressure) and trehalose concentration (which were
2.7.3.Scanningelectronmicroscopy(SEM)
establishedtobetheimportantinputprocessparametersinprior
The morphology of the particles was visualized using a JSM
screeningstudies)ontheoutputparameters(outlettemperature)
6301Fscanningelectronmicroscope(JEOL).Samples(Experiment
and vaccine powder characteristics. To gain more insight in the
1,7,8,11and12)werepreparedbyplacingthepowdersondouble-
relation between input parameters on the responses, a reduced
sided sticky carbon tape on a metal disk. Subsequently, the
CCF design was adopted for further optimization of process
particles were coated with a gold layer of approximately 10nm
parameters (Table 2). After spray drying and analyzing the
uusseindg fao Braalnzaelryss 1is20w Ba ssp1u0tkteVriwngit dheaviscpe o(Btaslizzeero UfN7I–O8N.)I.m Thagee vsowltaegree formulation s of the CCF desig n, MLR models were fitted for each
taken at amagni ficati on of 1000(cid:7) and5 000 (cid:7). output parameter. Valid models were obtained for outlet
temperature,particlesize,residualmoisturecontentandpowder
yield, descri bed in mod el fit ( R2), pred iction p owe r (Q2),
2.7.4.PowderX-raydiffractometry(XRD)
reproducibilityandavalidmodel.
Vaccinepowdersamples(Experiment1,7,8,11and12)were

analyzedbyanD2PhaserdesktopX-raydiffractometerequipped

3Fi.1g .m 2m. )P, oBwodtteorm m: oRrupnh o8l o(gXy5 0o:f 2s.p9r maym d)r]i. eMd aWgnIVifi pcoawtiodne rism. Sacgaens nlienftg (e1l0ec0t0r(cid:7)on) manidcr roiggrhatp (h5s0 o0f0 r(cid:7)e)p.resentative spray dried trehalose-stabilized WIV vaccines [Top: Run 12 (X50:
G.Kanojiaetal./InternationalJournalofPharmaceutics511(2016)1098–1111 1103

3.1.Physicalcharacteristicsofobtainedvaccinepowders resultedinbiggerparticles,whereasanincreaseinnozzlepressure

resultedinadecreasedinparticlesize(summarizedinTable6).

3.1.1.Particleshapeandsize

Alltheexperimentsinthecurrentdesignyieldedwhitetooff- 3.1.2.Powderyield
white pow dersaftersp ray dr ying.No visible aggrega tesco uld be Th einflue nceofspraydryingprocessparametersontheyieldof

observed in the powders by eye. SEM was performed on the thevaccinepowderwasassessed.Thepowderyieldsrangedfrom

powders to provide particle form and surface morphology. The 42%to82%,andapproximatelyhalfoftherunsinthedesignhada

powders showed a relatively homogenous surface morphology. yieldof>70%,whichwasthetargetedyieldintheQTPP(Table2).
Typically , they we re spherical with a smoo th surfa ce (Fig. 2). In Apply in g a M LR mo del r esu lted in a good m ode l fit, p redict ion

addition, the particle size of the acquired powder vaccines was power,reproducibilityandvalidity(Fig. 4A).Threeinputparam-
measured by laserdiff racti on .The powders hadam edianvol ume eters(n ozzlepressure, feed flowrat eand treh alosec oncen tration)
m m
diameter (X ) between 2.4 m and 8.5 m, with narrow size were found to affect the yield (Fig. 4B). An increase in nozzle
50
distributions(Table2).Toinvestigatetherelationshipbetweenthe pressureincreasedthepowderyield,whereasincreasingthefeed
inputparame tersan dt he particlesiz e,a MLRmodelw asfitted on flowrate andtreha lose concent ration decrease dthepowd ery ield.
theda ta(Fig.3A) .The mo delfitan dpr ed ictio npowe r(R 2=0.9 65, Thee ffec toft womosti nfluentialfact orsnozzle pres sureand feed
Q2= 0.91 7)we rego od, andmo de lrep roducibility andva lidity were flow rate on pre dicted powder yield is shown in cont our plot,
suffi cient. There gress ionc oeffici entplot(Fig.3B )sh owstha tthe Fig.4 C.In ge neral,high nozzlep ressu re andlow er feedflow rate
nozzlegas flow rate(noz zlepressur e), in leta irte mpera ture and with ot her parame tersfi xedfa vorshigh erpo wder yield .
feedflo wr atew erer elevant toparticle size. The effectofthe two
most influ enti alfact ors,feed flo wratea ndno zzle press ure ,on the 3.1.3.Residualmoisturecontent
partic le size is shown i n Fig . 3C . An incr ease in feed flo w r ate Th e moistu reconte nt of the produced powders was between

1.2%and4.9%(Table2).Oneoutlierwasdetected(experimentNo

Fig.3. Regressionmodelforparticlesize.A:Summaryoffitplotforparticlesize.R2(Goodnessoffit,1=perfectmodel)(R2=0.964).Q2(Goodnessofprediction,valuesgreater
than 0 .5isagood fit)(Q 2= 0.917).M ode lv alidity(va lue g reate rt han0.25 ind icatesgoodfi t). Re pr o ducibili ty(grea tert han0.5 indicatesasm a llexperime ntale rror)B:
Regre ssio n co effici ents forp article size.C .Respon secont ourplot sfor parti clesize.T heeff ecto fthetwomost influenti alfac tors (asobser ve dfrom FigB)wasta kenin to
accountand otherfacto rsw erekept cons tan t.Thecolo rregion srepr esen tthepr edicte dre spons e(p arti cles ize)fo rdefinedp aramete rse ttings.Asp irato rwa ss etat maxim um

instrumentsetting.
1104 G.Kanojiaetal./InternationalJournalofPharmaceutics511(2016)1098–1111

Table3
Therm ostabilityofspraydriedformulations.Spraydriedformulationswerestoredupto3monthsat60(cid:1)C,andtheirantigenicitywassubsequentlydeterminedby
hemagglutinatio n( HA).L iquidW IVwasalsos tored at60(cid:1) Casacontrol .The HAtite rsw er em easured in triplic ate; since nodifferenc eint hetriplicateo fthetitersw as
observednostand ardd eviatio nsar esho wn. Numbe r1 –23r ep re sentthe diffe ren tform ulatio nssprayd rie daccordin gto the designdes cri bed inTable2 .

Type Day0HAtiters(%) Day30HAtiters(%) Day60HAtiters(%) Day90HAtiters(%)

LiquidWIVstoredat60(cid:1)C(negativecontrol) D0 D1 D2 D3 D4 D5

100 72 56 45 34 0

Experiment1 100 100 100 96

Experiment2 100 104 106 96

Experiment3 100 104 100 96

Experiment4 94 104 104 96

Experiment5 90 101 100 92

Experiment6 86 105 104 102

Experiment7 87 98 98 96

Experiment8 100 108 105 96

Spraydried9 93 105 112 102

Experiment10 96 105 100 98

Experiment11 93 108 115 105

Experiment12 100 101 100 92

Experiment13 100 105 105 102

Experiment14 96 105 105 98

Experiment15 96 107 100 98

Experiment16 100 104 98 96

Experiment17 100 104 104 96

Experiment18 100 112 105 102

Experiment19 100 108 98 96

Experiment20 94 104 98 96

Experiment21 87 95 98 96

Experiment22 96 94 100 98

Experiment23 100 102 105 102

Table4

PredictionpowerofDoEmodel.Threeexperimentalinputparameterswerechosenfromthepredicteddesignspace(Fig.9).PointAwasselectedforproof-ofprinciple
purposesf orafor mu latio nthatw ould resultintarge tedpr oductspecifi cation s(with inop timu mareaof predict eddes igns pac e).Poi nt Ban dCdon otm eetallc riteriafor
targetedp rod u ct(outsideo ptim umare aofde sig nspace). Therati onaleforchoo singBan dCwasto con firm thepred ictiona bilities ofthe m ode lw ith int hede sign space but

outsidetheoptimumarea.Aspiratorwassetatmaximuminstrumentsettingforthethreeexperimentalruns.

ExpNo. Inlettemp((cid:1)C) Nozzlegasflowrate(L/min) Feedflowrate(mL/min) Trehaloseconc(mg/mL)

Withinoptimumarea(A) 120 12.4 1.0 100

Outsideoptimumarea(B) 135 10.5 1.0 125

Outsideoptimumareaspot(C) 132 9.0 1.0 125

5),whichwassubsequentlyexcludedfromtheregressionmodel. processparametersinfluencetheRMCofthespray-driedvaccine
Fitt ingaM LRr egressionmod elonthe RMC data resultedin agood powder s(Fig.5B).T heeffecto ftw omo sti nflu entialfactor snozzle
model fi t, pr ediction po wer, m od el r epro ducib ility and v a lidity pressure and inle t tem pera tur e on pred icted RMC is sho wn in
(Fig.5A ).T heresulting regress ioncoe fficientplotshow sth atmany contour plot, Fig. 5 C. An increas e in inlet tem peratu re or noz zle

Table5
Assess ment/qualificationofthepredictionmodels.TheacquiredmodelfromexperimentaldesigndescribedinTable2predictedtheparticlesize,residualmoisturecontent,
outlettemperatureandpo w der yield.Pred ictionsar eex presseda smean (cid:6)95 %confidencei nterval s.Theuppe ra ndlo w errangesw ere takenf romt hemode l.Datafor particle
sizean dRMCrepre sen tmean(cid:6) SDo fonesampl em easuredin tr iplica te N= 3(observed ).

Outputparameters Predicted Predictedlowerrange Predictedupperrange Observed

Particlesize(mm)
A 3.38 3.00 3.77 3.08(cid:6)0.1
B 4.23 3.88 4.57 4.07 (cid:6) 0.1
C 5.88 5.55 6.21 5.78 (cid:6) 0.1

Residualmoisturecontent(%)
A 2.1 1.8 2.5 2.4(cid:6)0.03
B 2.7 2.4 3.0 2.7 (cid:6) 0.02
C 3.6 3.3 4.0 2.8 (cid:6) 0.02

Outlettemperature((cid:1)C)

A 67 65 69 67

B 74 73 76 76

C 75 73 77 75

Powderyield(%)

A 80 76 84 77

B 73 70 77 73

C 66 63 69 64
G.Kanojiaetal./InternationalJournalofPharmaceutics511(2016)1098–1111 1105

Table6
Summ aryofmainfindings.Descriptionoftherelationshipsbetweenprocessinputparametersandproductoutputparametersinthisstudyforthespraydryingprocess
(Büchi, B- 290 and 0.7 mm o rifice nozzle ) o f a w hole inactiva ted influe nza vacc ine. A spirator capa city was kep t fixed a t 22 m3n/h [ hig hes t aspir ator set ting po ssible (of used
equipment)].
Parameter/ Nozzlepressure" Inletairtemperature" Feedflowrate" Trehaloseconcentration"

Dependence
Particle ###Greaterenergyfordispersion, #Moreevaporationaftershell "Morefluidneedstobedispersed Noeffectintheinvestigated

Size smallerparticlesafter formation range

atomization
Powder "Smallerparticles,improved ""Eventuallydryerproductbut #Higherwatercontenttoevaporate, #Biggerdropletsfordrying,

Yield drying,lesserdepositionindrying chancesofstickingathigh morestickingbutdependentonother moredepositionindrying

chamber temperatures variables chamber
Residual #Smallerdropletsthusbetter ##Lowerrelativehumidityinair ""Morewaterforevaporationleadsto #Lesswaterevaporated,lower

moisture drying higherpartialpressure partialpressure

content
Outlet #Morenitrogeninflowinsystem """Directlyproportional ##Morewaterneedstobeevaporated "Lesswaterneedstobe

Temperature forheating. asavailableenergyfordryingincreases evaporatedasmoreexcipientin

proportionallywithriseininlet thedroplet

temperature
Antigenicity Noeffectintheinvestigatedrange Noeffectintheinvestigatedrange Noeffectintheinvestigatedrange Noeffectintheinvestigated

range
pressureresultedinadecreaseofRMC,whileanincreaseinfeed indexes below 0.08 (Supplementary Table 1), indicating that the
flowrate increase d th eRMCof th epow ders. WIVpar ticlesw erei ntactandnotagg regate dd uringoraf tersp ray

dryingorreconstitution.

3.1.4.Amorphousnatureofdriedpowdervaccine To assess the antigenicity of the WIV antigens after spray

WehadobservedsomesampleswithhighRMCandsincewater drying, a hemagglutination assay was performed on the recon-
isapot ent plasticize rfors ugarglas ses.W ew anted to confir mthe stituted v accinepowdersdir ectlya fters praydrying. Ov erall ,lossof

amorphousnessofthesamples.Theamorphousnatureofthespray antigenicity was minimal after spray drying, with antigenic
dried vaccine p ow der s was d eterm ined by a ssessin g the glass recoveries r angin g from 86 % to 1 00% (a fter 2log transf ormation)

transition temperatures (Tg) with DSC and crystallinity by X-ray (Table 3).Due totheseminimal changes in antigenicity, no MLR
diffraction (XRD).Tgana lysis ofth evac cine powdersga vev alues model cou ldb efi ttedo ntheant igenicre cov erydata.
between 4 5(cid:1)C a nd 83(cid:1)C f or the highes t and lo west RMC- In a dditio n t o the rec over y of the antigen im mediately after

containing powders, respectively (Supplementary Table 2). The spraydrying,anacceleratedstabilitystudywasperformedonthe
X-ray diffr action (X RD) profiles of the excipie nt (tr eha lose vaccin e powd er s. Liquid W IV vaccin e stor ed u nder refrig era ted
dihyd rate) and s pray dr ied influ enz a va ccines are shown in conditio ns was ta ken as a co ntrol gr oup. Th e pow der vaccine
Fig. 6. The pro files in dicate that solid trehalose dih ydrate w as samples fr om the ex per im ental d esign w ere sealed under a
crys tal line priortos praydry ing,a spea kswereob served.On the nitrogen enviro nme nt(100mgpe rvial)a ndsto redat6 0(cid:1)Cfor 3

other hand, no peaks were detected in the spray dried vaccine months. The change in hemagglutinin titers upon storage was

powders,indicatingthatthepowderswereamorphous. assessed by hemagglutination assay after reconstitution of the

powdervaccines.Minimalornolossinantigenicitywasobserved
3.2.Outlettemperature after sto rage for 3 month s a t 6 0(cid:1)C fo r all spray d ried samples

(Table3).Incontrast,liquidWIVvaccinelostallitshemagglutinin
The outlet air temperature was determined during the spray bindin gca pa cityafter 5days ofst orageat 60(cid:1) C.T hi sindicatesthat

drying process of each experiment in the design. During spray spraydriedWIVpowdervaccinesaremuchmoreresistanttoheat
drying ofthein flu enza vaccineform ula tion stheou tlettem pera- stress than conv entional liquidva ccin es.
ture w as foun d to be b etween 48(cid:1)C and 91 (cid:1)C ( Table 2 ). A MLR
regre ssio nmode lw as fittedtot heout lett emper atured ata ,w hich 3.4.PredictionpoweroftheDoEmodel
resulted in a goo d m odel fi t (Fig . 7A). The obtained regr ession
coefficie nt pl ot sho wed th at the outle t tem perature was most AnadvantageofthegeneratedMLRmodelsisthattheycanbe

affected by the inlet temperature (Fig. 7B). In addition, the usedtopredictprocessorproductparameters(inthiscaseoutlet
relations hip betw een i nlet air temp eratur e and fe ed flow ra te is temp era ture, p owder p art icle size , powder y ield , and RM C) for

depictedinresponsecontourplotforoutlettemperature(Fig.7C). combinationsofthedifferentinputprocessparametersthathave
It was o bs erved th at on i ncrea sin g the feed flow rate and notbeen teste d yet. As previo usly describe d, the study obta ined

decreasinginletairtemperature(keepingotherfactorsconstant) valid models for outlet temperature, particle size, powder yield

resultedindecreaseoftheoutlettemperature. and moisture content. The structured approach helped in

identifying a design space where the established QTPP can be

3.3.Antigenrecoveryandstabilityafterspraydrying achieved(Fig.9).Thepredictionabilityofthemodelwastestedby
selecting inpu tpa ram etersthatw ouldr es ulti nprede fine dprodu ct
Next to the physical powder characteristics, the antigen specificat ionsf orthreeunt este dform ulation s. Thecalcula tionfor

characteristics were determined after spray drying. First, the prediction are based on statistical values from the model

particle size of WIV was determined for each spray dried calculated by MODDE software(Lundstedta et al.,1998). Experi-

formulation. Dynamic light scattering was performed on recon- mentAwasselectedinthesweetspotregion,whichwouldresult
stitutedvacc inepowd ers.A llsamples disso lvedreadily an dgave inap ro duct accordin g tot hedefi ned QTPP.E xperim entsB andC

clear opalescent liquid, which did not contain any visible wereselectedoutsidethesweetspot,whichwouldyieldpowder

aggregates. The particle size of WIV remained unchangedfor all vaccinesthatwouldnotmeetalltheQTTPcriteria(Table4).Using

formulations, with an average size of 175nm and polydispersity the selected targetproductcharacteristicsasinput, the acquired
1106 G.Kanojiaetal./InternationalJournalofPharmaceutics511(2016)1098–1111

Fig.4. Regressionmodelforyield.A:Summaryoffitplotforpowderyield.R2(Goodnessoffit=0.825).Q2(Goodnessofprediction=0.688).B:Regressioncoefficientsof
pow de ryieldafter model refi neme nt. Aspiratorw a sse tat max imumc apacit y.C.Response co nto u rplots forpowderyie ld. Theeffect of thetw om ostinfluen tialfactors( as
observe dfrom B)w astak enintoacco untando ther fac tor swerekep tconstan t. Thecolor regions repres ent thepre dicted resp onse (po wd eryi eld)f ordefined parame ter

settings.Aspiratorwassetatmaximuminstrumentsetting.

DoE model gave spray-drying parameters that would yield the stable spray dried influenza vaccines. The main findings are

desired product (Table 4). For experiment A, the experimentally summarizedinTable6.

acquireddataindeedcorrelatedinrangewiththepredictedvalues
for parti cle s ize (pr edicted 3.3 8 mm, o bser ved 3.08mm) , RMC 4.1.Effectofnozzlepressure(atomizationgasflowrate)

(predicted 2.1%, observed 2.4%), powder yield (predicted 80.2%,
observed7 7%)an doutlett empera ture(pre dicte d67.1(cid:1)C,ob served In general, the nozzle pressure (7.3–17.5L/min) was the most
67(cid:1)C)(Ta ble5 ).Th epred ictedparame tersthusy ielded apowder influe ntialpro cess parame tersforth estudied charac teris tics ofthe

vaccinethatmetallthecriteriaoftheQTTP.Theoutputparameters obtainedpowdervaccine.Anincreaseingas(nitrogen)pressureon

(particle size, powder yield and outlet temperature) of experi- thenozzleresultedindecreasedparticlesize,whichcanberelated

mentsBandCalsoconcurredwiththeresponsespredictedbythe totheincreasedenergyavailableforbreakinguptheliquidjetby

model,withtheexceptionoftheRMCforexperimentC,whichwas thenozzle,thusformingsmallerparticlesduringatomization.This

lower (2.8%) than the predicted (3.6%) RMC. These data indicate wasinlinewithpreviousstudiesfromMaltesenetal.(Maltesen

thatthepredictionsmadebythemodelwereaccurate. etal.,2008),whoreportedadecreaseinparticlesizeofspraydried
m m
product from 26 m to 9.8 m when increasing the nozzle

4.Discussion pressure in the range from 7.3 to 17.5L/min. In addition, the

interaction factor (Noz*Noz) (Fig. 3B) showed an opposite effect

The present study demonstrates the usefulness of a DoE comparedtothemainfactornozzlepressure,indicatingthatthe

approach to understand the effects of spray drying process relationship between nozzle pressure and particle size could be

parameters and excipient concentration on the production of non-linearoutsidethestudiedrange.