Improved storage stability and immunogenicity of hepatitis B vaccine after spray-freeze drying in presence of sugars

European Journal of Pharmaceutical Sciences 55 (2014) 36–45

Contents lists available at ScienceDirect

European Journal of Pharmaceutical Sciences

journal homepage: www.elsevier .com/ locate /e jps

Improved storage stability and immunogenicity of hepatitis B vaccineafter spray-freeze drying in presence of sugars

http://dx.doi.org/10.1016/j.ejps.2014.01.0050928-0987/� 2014 Elsevier B.V. All rights reserved.

⇑ Corresponding author. Tel.: +31 50 3632398; fax: +31 50 3632500.E-mail address: [email protected] (W.L.J. Hinrichs).

W.F. Tonnis a, J.-P. Amorij b, M.A. Vreeman a, H.W. Frijlink a, G.F. Kersten b,c, W.L.J. Hinrichs a,⇑a University of Groningen, Department of Pharmaceutical Technology and Biopharmacy, Antonius Deusinglaan 1, 9713AV Groningen, The Netherlandsb Institute for Translational Vaccinology (Intravacc), P.O. Box 450, 3720 AL Bilthoven, The Netherlandsc Leiden Academic Centre of Drug Research, Drug Delivery Technology, P.O. Box 9502, 2300 RA Leiden, The Netherlands

a r t i c l e i n f o

Article history:Received 18 September 2013Received in revised form 17 December 2013Accepted 15 January 2014Available online 25 January 2014

Keywords:Hepatitis B surface antigenStabilizationPowder formulationInulinDextranTrehalose

a b s t r a c t

The current hepatitis B vaccines need to be stored and transported under refrigerated conditions (2–8 �C).This dependence on a cold-chain is highly challenging in areas where hepatitis B virus infections areendemic. To decrease the cold-chain dependency, powder formulations of the hepatitis B surface antigen(HBsAg) without aluminum were prepared by spray-freeze drying in the presence of either inulin or acombination of dextran and trehalose. The stability of HBsAg in the amorphous powder formulationswas strongly improved during storage both at room temperature and at an elevated temperature(60 �C), compared to a liquid plain and an aluminum hydroxide adjuvanted HBsAg formulation. Immuno-genicity studies in mice showed that reconstituted powder formulations induced higher IgG immuneresponses after intramuscular administration than those induced after administration of unprocessedplain antigen. Although the immune response was not as high as after administration of aluminumadjuvanted HBsAg, the immune response to the reconstituted vaccines shifted towards a more balancedTh1/Th2 response compared to the aluminum containing HBsAg formulation.

� 2014 Elsevier B.V. All rights reserved.

1. Introduction

Each year, approximately 600,000 persons die due to conse-quences of acute or chronic hepatitis B virus (HBV) infection. It isestimated that 350 million people worldwide carry the HBV (Gold-stein et al., 2005), which makes it a serious threat to global health(Wasmuth, 2010). Prevalence of chronic HBV infection variesmarkedly throughout the world. High prevalence rates (10–20%)are observed within Asia and sub-Saharan Africa, where HBV infec-tion is considered to be endemic. In areas like western Europe andNorth America, less than 1% of the population carries HBV (Was-muth, 2010; World Health Organization, 2012).

Vaccination against HBV is known to be a cost effective inter-vention for HBV infection (Beutels, 2001; Maynard et al., 2011).The currently available HBV vaccine formulations are liquid sus-pensions containing virus-like particles (VLPs) of recombinant hep-atitis B surface antigen (HBsAg) adsorbed onto aluminumphosphate or aluminum hydroxide. The exact mechanism of howaluminum acts as an adjuvant has not been fully elucidated, butseems to be related to inflammosome activation and the produc-

tion of uric acid upon injection induced by aluminum (De Gregorioet al., 2008; Kool et al., 2008; Lambrecht et al., 2009).

A disadvantage of these formulations is that the vaccine isunstable when frozen (below �1 �C) and at elevated temperatures.After freezing and thawing, the aluminum salts tend to aggregate(Zapata et al., 1984), which may lead to irritation at the site ofinjection due to the increased size of the aluminum particles. Fur-thermore, in vivo studies in mice showed that freezing reduces theimmunogenicity of aluminum adsorbed HBsAg (Diminsky et al.,1999). According to the manufacturer of Engerix-B, the stabilityof the formulation is guaranteed for 3 years only when it is storedat a temperature between 2 and 8 �C (GlaxoSmithKline, 2013).Although clinical studies showed that the immunogenicity toEngerix-B is not changed after 7 days of storage at 37 �C (Justand Berger, 1988), in vitro studies showed a decrease in antigenic-ity of aluminum adjuvanted HBsAg after 3–4 weeks of storage at37 �C (Braun et al., 2009; Hirschberg et al., 2010) or after one totwo months at 20–25 �C (Galazaka et al., 1998).

Due to their instability, a ‘‘cold chain’’ is required for currentHBsAg vaccines. A ‘‘cold chain’’ means that from manufacturinguntil administration, the formulation has to be kept at refrigeratorconditions (2–8 �C) (Brandau et al., 2003). Transport and storage atrefrigerator conditions is challenging and expensive, especially inhigh prevalence areas like Southeast Asia and sub-Saharan Africa.

W.F. Tonnis et al. / European Journal of Pharmaceutical Sciences 55 (2014) 36–45 37

By drying biopharmaceuticals, like vaccines, their storage stabil-ity can be improved (Saluja et al., 2010). In the dry state the molec-ular mobility of the protein-based antigen is strongly reduced bywhich the vaccine is less prone to degradation. However, duringthe drying process proteins may be subjected to harsh conditionsthat can severely damage them. It is widely known that sugars canbe used as stabilizing excipients to prevent degradation not onlyduring drying but also during storage (Amorij et al., 2008). To actas a stabilizer, it is essential that during drying the sugar becomesglassy and remains in that state during subsequent storage, since su-gar crystallization would damage the antigen. It is generally believedthat three mechanisms are involved in the stabilization of proteinsby sugars, these are water replacement (Chang and Pikal, 2009), par-ticle isolation (Allison et al., 2000b), and vitrification (Chang et al.,2005). The water replacement theory states that during drying thehydrogen bonds between the protein and water molecules are grad-ually replaced by hydrogen bonds between the protein and the hy-droxyl groups of the sugar, by which the structural integrity of theprotein is conserved. According to the particle isolation theory, thesugar molecules form a physical barrier between the individual pro-tein molecules by which the protein molecules cannot interact witheach other. It is proposed that particle isolation prevents degrada-tion pathways like aggregation. Finally, according to the vitrificationtheory, the glassy sugar matrix strongly reduces the molecularmobility of the protein. Since many degradation pathways requiremolecular mobility, the rate at which these degradation pathwaysoccur is strongly reduced. In order to remain in the glassy state dur-ing processing and storage the sugar should preferably have a highglass transition temperature (Tg) and contain low residual moisturelevels, since water acts as a plasticizer decreasing the Tg.

Various studies have been performed to improve the stability ofHBsAg by drying the antigen in the presence of sugars. In a studyby Chen et al. (2010) and in a study by Hirschberg et al. (2010) tre-halose was used as a stabilizer. The Tg of dry trehalose is around115 �C (Wang, 2000), which seems high enough to remain in theglassy state during storage. However, as described before, wateruptake from the environment or residual water after the dryingprocess can lead to a strongly decreased Tg, which might be evenbelow the storage temperature. In a study by Maa et al. (2007) acombination of trehalose, mannitol and dextran (ratio 3:3:4) wasused to stabilize HBsAg. The addition of dextran might improvethe overall Tg of the combination, but on the other hand, the Tg

of mannitol is low (13 �C), leading to an overall Tg, which is lowerthan that of trehalose alone.

In our study, it was investigated whether sugars with a high Tg

could be used to prepare stable powder formulations of the hepa-titis B vaccine. Inulin and a combination of dextran and trehaloserespectively, were used as stabilizers for HBsAg during dryingand subsequent storage. In various studies, it has been shown thatinulin can act as an excellent stabilizer for a variety of proteins(Amorij et al., 2007; Hinrichs et al., 2001; Saluja et al., 2010). Thestabilizing capacities of inulin can be ascribed to its favorable phys-ico-chemical properties such as high Tg and resistance to crystalli-zation. Polysaccharides like dextran, especially high molecularweight dextrans, exhibit similar favorable physico-chemical prop-erties. However, due to their bulky nature, high molecular weightdextrans cannot accommodate to the irregular surface structureof proteins causing a less tight coating around proteins. In litera-ture, a successful application of a combination of a high molecularweight dextran and trehalose to improve the storage stability of aprotein has been described (Allison et al., 2000a). We hypothesizedthat such a mixture will combine the benefits of the high Tg of dex-tran and the ability of the disaccharide to accommodate to theirregular surface structure of proteins. For this reason, we alsoinvestigated a mixture of a high molecular weight dextran and tre-halose. Spray-freeze drying was chosen as a drying method to pro-

duce small size spherical particles, which would be suitable forpulmonary administration in future research (Amorij et al.,2007). Therefore, HBsAg was spray-freeze dried in the absence ofaluminum, since aluminum salts are not tolerated after pulmonaryadministration (Stacy et al., 1959).

2. Materials and methods

2.1. Materials

Hepatitis B surface antigen (HBsAg) stock solution (1.8 mg/mLin 8 mM phosphate buffer, pH 7.2, purity 98.8%) was purchasedfrom the Serum Institute of India Ltd. (Pune, India). Inulin (4 kDa)was a generous gift from Sensus (Roosendaal, The Netherlands).Dextran (70 kDa), hydrogen peroxide and sodium acetate were ob-tained from Fluka BioChemika (Buchs, Switserland). Trehalose wasobtained from Cargill (Amsterdam, The Netherlands). Alhydrogel2% (wet gel suspension of aluminum hydroxide) was obtained fromBrenntag Biosector (Frederikssund, Denmark). Bovine serum albu-min (BSA) and 3,30, 5,50-tetramethybenzidine (TMB) were obtainedfrom Sigma–Aldrich (Zwijndrecht, The Netherlands). Tween 80 wasobtained from BUFA (IJsselstein, The Netherlands). Sulfuric acidwas obtained from Merck (Darmstadt, Germany). The phosphatebuffered saline (PBS) used in this study consisted of 155 mM so-dium chloride, 1.54 mM potassium dihydrogen phosphate,2.7 mM disodium hydrogen phosphate at a pH of 7.2.

2.2. Formulations

Powder formulations of the different sugars with and withoutHBsAg were obtained by spray-freeze drying. Aqueous solutionsof 2.25 % (w/v) sugar (either inulin or a mixture of dextran and tre-halose at a weight ratio of 1:1) with and without HBsAg in Milli-Qwater at a concentration of 90 lg/mL were sprayed at 5 mL/mininto a vessel of liquid nitrogen using a nozzle of the Büchi 190 MiniSpray Dryer (Büchi, Flawil, Switzerland). The nozzle was placedapproximately 5 cm above the surface of the liquid nitrogen. Thesolution was atomized with nitrogen gas at an airflow of 600 Ln/h. After spraying, the vessel with liquid nitrogen was transferredto a Christ model Epsilon 2–4 freeze dryer (Salm & Kipp, Breukelen,The Netherlands). After evaporation of the liquid nitrogen, theproduct was dried for 24 h at a pressure of 0.220 mBar and a con-denser temperature of �85 �C, while the shelf temperature wasgradually increased from �35 �C to �10 �C. Next, the pressurewas decreased to 0.050 mBar, while the shelf temperature wasgradually increased to 20 �C during 4 h. At these conditions, dryingwas continued for another 20 h. After drying, powders were eitherstored for stability testing (see Section 2.3) or placed in a vacuumdesiccator over dried silica gel at room temperature (20 �C ± 2 �C)until further use.

As a negative control, a HBsAg solution of 90 lg/mL (withoutany excipients) was freeze dried at the same settings as thespray-freeze-dried formulations. HBsAg was freeze dried insteadof spray-freeze dried because spray-freeze drying HBsAg withoutexcipients would imply that the solid content in the feed solutionwould be extremely low (90 lg/mL instead of 22.5 mg/mL for thesugar containing powder formulations), which makes it practicallyimpossible to collect sufficient amounts of the formed powderproperly. Sample collection is not an issue when freeze drying,since the sample remains in the vial during the freeze drying pro-cess. Spray-freeze drying is a more harsh condition than freeze dry-ing as the vaccine is not only subjected to freezing and dehydrationstresses but also to shear stresses. Therefore, compared to freezedrying, spray-freeze drying can be considered as a worst casescenario.

38 W.F. Tonnis et al. / European Journal of Pharmaceutical Sciences 55 (2014) 36–45

Liquid formulations containing aluminum hydroxide werebased on the commercial GlaxoSmithKline product (GlaxoSmithK-line, 2013) and was prepared by adding aluminum hydroxide sus-pension to an HBsAg stock solution at a weight ratio of 25:1(Al3+:HBsAg). The solution was diluted with PBS to a concentrationof 40 lg/mL HBsAg and gently stirred for one hour. The degree ofadsorption of HBsAg onto the aluminum hydroxide particles was98.5% (±0.2%) after one hour, as determined by centrifugation(10 min at 10,000 rpm) of the formulation and analysis of thesupernatant by ELISA. Sedimentation of VLPs that were not ad-sorbed to aluminum during centrifugation did not occur as centri-fugation of a diluted stock solution (40 lg/mL) did not result in adecrease in concentration (data not shown).

A liquid plain formulation, containing only HBsAg and buffer,was prepared by diluting the stock solution of HBsAg (1.8 mg/mL) with PBS to 40 lg/mL.

2.3. Stability testing

The two powder formulations (one with inulin and one with amixture of dextran and trehalose) and both liquid formulations(HBsAg solution with and without aluminum hydroxide) werestored at various conditions to evaluate their stability. First, theformulations were subjected to ten freeze–thaw cycles, which isa worst-case scenario that can be encountered during shipping ofvaccines (Nelson et al., 2007; Techathawat et al., 2007). Onefreeze–thaw cycle consisted of storing 1.5-mL vials, containingthe formulation (powder or liquid), in a freezer at �20 �C for1.5 h and subsequent thawing for 1.5 h at room temperature. Theliquid was frozen within 30 min. This slow freezing rate was cho-sen to mimic the freezing rate that may occur during transport inthe presence of ice packs.

Furthermore, all four formulations were stored at 20 �C and at60 �C for 1, 2, 4, 6, 9 and 12 weeks. 20 �C was used as the commonroom temperature. Storage at 60 �C was chosen as extreme condi-tion for accelerated stability testing. Furthermore, this temperaturecan occur in tropical countries when exposed to the sun.

2.4. Physico-chemical characterization of the powders

To determine the solid state characteristics, both powders wereanalyzed by X-ray powder diffraction (XRPD) using a D2 PHASERX-ray diffractometer (Bruker, Delft, The Netherlands) with a diver-gence slit of 1 mm and a detector opening of 5�. To prevent directradiation form the X-ray source on the detector, an air-scatterscreen (3 mm) was placed above the sample holder. Samples wereplaced on a silicium zero-background sample holder and rotated at60 rpm. Scans were performed from 5 to 40� 2h with a step size of0.004� 2h and 1 s measurement per step using Cu Ka radiation, at awavelength of 1.5406 Å at 30 kV and 10 mA.

The residual moisture content of the spray-freeze-dried pow-ders was determined by Karl-Fisher coulometric water titrationusing an 831 KF Coulometer (Metrohm Applikon, Schiedam, TheNetherlands). Powders were dissolved in Hydranal

�-Coulomat AG

(Karl Fischer reagent) prior to analysis.Differential Scanning Calorimetry (DSC) was used to determine

the Tg of the powder formulations after spray-freeze drying. Thesamples were analyzed in an open aluminum pan and placed in aQ2000 DSC (TA Instruments, Ghent, Belgium). The samples werepre-heated at 80 �C for 3 min to remove the residual water. Aftercooling to 20 �C, the temperature was raised from 20 to 200 �C at20 �C/min. The inflection point of the step transition in the thermo-graph was taken as the Tg.

Scanning electron microscope (SEM) images were recordedwith a JEOL JSM 6301-F microscope (JEOL, Japan). Powder was ap-plied to a double-sided sticky carbon tape on a metal disk and

coated with 80 nm of gold/palladium in a Balzers 120B sputteringdevice (Balzars Union, Liechtenstein).

2.5. Antigenicity

Antigenicity was measured by ELISA using an antibody directedagainst the ‘a’ determinant epitopes. The ‘a’ determinant is definedas the amino acid residues 120–150 (Karthigesu et al., 1999; Than-avala et al., 1986; Waters et al., 1992). For investigation of the anti-genicity, a Murex HBsAg version 3 ELISA kit (Murex BiotechLimited, Dartford, United Kingdom) was used according to thespecifications of the manufacturer. Samples were, if necessary, dis-persed and diluted in PBS to 1 ng/mL and pre-incubated for 1 h at37 �C in microwells coated with a mixture of mouse monoclonalantibodies specific for the ‘a’ determinant of HBsAg. Next, affinitypurified goat antibodies to HBsAg, conjugated to horseradish per-oxidase, were added to the wells and the wells were incubated at37 �C for 30 min. After washing, a substrate solution containing3,30, 5,50-tetramethylbenzidine (TMB) and hydrogen peroxide wereadded. The conversion of TMB by peroxidase was stopped after30 min with sulfuric acid and measured spectrophotometricallyat 415 nm with a Benchmark Microplate reader (BioRad, Hercules,CA, USA). A calibration curve of a reference HBsAg stock solutionwas used to determine the amount of intact HBsAg in the powderformulations. Antigenicity was expressed as percentage (%) of theoriginal HBsAg amount in the formulation. Aluminum hydroxideinterfered slightly in the ELISA, leading to results up to 110% ofthe theoretical value immediately after preparation of the formula-tion. Therefore, all ELISA results of this formulation during the sta-bility study have been adjusted accordingly.

2.6. Intrinsic tryptophan fluorescence spectroscopy

Intrinsic tryptophan fluorescence spectroscopy was used tostudy conformational changes of HBsAg during storage. Trypto-phan fluorescence spectra were obtained by using a PTI spectroflu-orometer (PTI, Birmingham, AL, USA). Samples were diluted in PBSto 10 lg/mL HBsAg, and 1.5 mL of this solution was placed in a 10-mm quartz cuvette and gently stirred. An excitation wavelength of295 nm with slits of 2.5 nm was used. Emission scans were per-formed from 300 to 400 nm, with slits of 2.5 nm, at a speed of20 nm/s and at 20.0 �C. These emission scans were corrected forbackground caused by PBS and sugar. The emission wavelengthwas plotted against the fluorescence intensity. The maximum fluo-rescence intensity and the wavelength at which the fluorescenceintensity was maximal were used to determine changes in the for-mulations during storage.

2.7. Lowry protein determination

Total protein content determination was performed by a modi-fied Lowry analysis (Lowry and Rosebrough, 1951). In brief, 50 lLof 0.3% (w/v) sodium deoxycholate was added to 500 lL of proteinsample and incubated for 10 min. Next, 50 lL of a 72% (w/v) tri-chloroacetic acid was added and incubated on ice for 30 min toprecipitate the protein. The solution was centrifuged and theremaining pellet was resuspended in 1 mL of Lowry reagent. Final-ly, 100 lL Folin solution was added and incubated for 30 min. Theabsorbance at 750 nm was compared to a calibration curve.

2.8. Dynamic Light Scattering (DLS)

The size of the HBsAg VLPs in the liquid plain formulation dur-ing storage was measured by Dynamic Light Scattering. A 1-cmplastic cuvette containing 1.5 mL of a 10 lg/mL HBsAg solutionwas placed in a Zetasizer Nano Zs (Malvern, Worcestershire, United

W.F. Tonnis et al. / European Journal of Pharmaceutical Sciences 55 (2014) 36–45 39

Kingdom). The scattering of a 633 nm laser beam was detected at a173� angle. Malvern DTS v.5.10 software was used to calculate themean volume diameter of the HBsAg VLPs by an average of 10scans.

2.9. Immunization

Animal experiments were conducted according to the guide-lines provided by the Dutch Animal Protection Act and were ap-proved by the Committee for Animal Experimentation (DEC) ofthe University if Groningen, The Netherlands. For all experiments,6- to 8-weeks-old female BALB/c mice (Harlan, Zeist, The Nether-lands) were used. The mice weighed 18–20 g and were fed ad libi-tum. Mice, seven in each group, were immunized on day 0 and 14.Prior to immunization, all animals were anesthetized with a mix-ture of isofluran/O2. All formulations (see Table 1) were injectedintramuscularly at a volume of 2 � 25 lL, containing 2 lg ofHBsAg, divided over the hind legs. Spray-freeze-dried samples con-taining HBsAg and sugar(s) were reconstituted in PBS, and injectedwithin 30 min after reconstitution. Spray-freeze-dried sugar(s)without HBsAg were reconstituted in a solution of HBsAg in PBSand injected immediately after reconstitution. The non-spray-freeze-dried formulations were prepared by dissolving the sugar(s)in PBS after which HBsAg was added.

2.9.1. Sample collectionOn days 0 and 14, blood samples were taken from the mice by

orbital punction under isoflurane/O2 anesthesia. The animals weresacrificed on day 28. Mice were bled under anesthesia by drainingthe abdominal aorta. Serum samples were obtained by centrifuga-tion at 1200g for 15 min and stored at�20 �C until further analysis.

2.9.2. HBsAg antibody ELISASpecific IgG, IgG1 and IgG2a antibodies against HBsAg in sera

were determined by ELISA as described by Hirschberg et al.(2010). In brief, 96-well plates (Greiner Microlon

�600 F-bottom)

were coated with 0.2 lg HBsAg, diluted in PBS, per well. The plateswere incubated overnight at room temperature. The plates werewashed with 0.05% (w/v) Tween 80 in Milli-Q water. Series ofthreefold dilutions of sera were added to the wells. Dilutions wereprepared in PBS containing 0.5% (w/v) BSA and 0.05% (w/v) Tween80. After 2 h of incubation at 37 �C, the plates were washed andgoat-anti-mouse immunoglobulin conjugated to horseradish per-oxidase (IgG/IgG1/IgG2a-HRP, Southern Biotech, Birmingham, AL,USA) diluted 5000 times in PBS containing 0.5% (w/v) BSA and0.05% (w/v) Tween 80 was added. The conjugate was incubatedfor 1 h at 37 �C. The plates were washed and 100 lL of substratesolution containing 0.01% (w/v) TMB, 10% 1.1 M sodium-acetatebuffer (pH 5.5) and 0.02% (v/v) hydrogen peroxide (30%) was addedto each well. The conversion of TMB by peroxidase was stoppedafter 10 min with 2 M sulfuric acid and the absorbance was mea-sured at 415 nm with a Benchmark Microplate reader. Titers are gi-ven as the reciprocal of the calculated sample dilution

Table 1formulations used for immunization. SFD: spray-freeze dried.

Formulation Components

1. plain HBsAg HBsAg in PBS2. HBsAg alum HBsAg and aluminum hydroxide in PBS3. SFD HBsAg-inu SFD HBsAg with inulin and reconstituted4. HBsAg + SFD inu SFD inulin, reconstituted and mixed wit5. HBsAg + inu Inulin dissolved and mixed with HBsAg6. SFD HBsAg-dex/tre SFD HBsAg with dextran/trehalose and r7. HBsAg + SFD dex/tre SFD dextran/trehalose, reconstituted and8. HBsAg + dex/tre Dextran/trehalose dissolved and mixed w

corresponding to an A415 = 0.2 after background correction. Thelowest dilution was 10 times. Therefore, the limit of quantificationis 10Log10 = 1. In this way, a non-responder is defined as a mousehaving a titer below 1 on a 10Log-scale.

2.10. Statistical analysis

Comparisons between formulations and groups were madeusing a two-tailed Student’s t-test. Probability values (P) < 0.05were considered significant.

3. Results

3.1. Physico-chemical characterization of the powders

Fig.1 shows the XRPD patterns of both powder formulations andthe patterns of (partially) crystalline inulin, dextran 70 kDa andtrehalose. Due to periodic arrangement of the molecules, crystal-line materials will give distinct diffraction peaks at certain posi-tions as shown in the diffractograms of the pure sugars. Inamorphous material, the molecules are randomly distributed.Therefore, the X-rays will be scattered in all directions, resultingin the absence of distinct diffraction peaks. The absence of highintensity peaks in the XRPD patterns of the powder formulationsindicates that the sugar(s) in both powder formulations wereamorphous after spray-freeze drying and remained as such during3 months of storage at 60 �C.

Table 2 shows the residual moisture content in and the Tg ofboth powder formulations. The residual moisture content in bothformulations was 2.5–3.0% as measured by Karl-Fisher coulometricwater titration. Immediately after spray-freeze-drying and afterthree months of storage at 60 �C, the Tg of both powder formula-tions after evaporation of the residual water was 165–170 �C.These Tg values are in theory high enough for the sugar(s) to re-main amorphous during storage at 60 �C. Although the residualmoisture acts as a plasticizer, resulting in a strongly decreasedTg, the visual appearance of the powder did not change during stor-age. The powders remained porous and did not collapse upon stor-age as confirmed by scanning electron microscopy (Fig. 2). AlsoXRPD results showed that both powder formulations remainedamorphous during 3 months of storage at 60 �C. In conclusion,even though there was some residual moisture after spray-freeze-drying, the water content was low enough for the Tg to re-main above the storage temperature of 60 �C.

3.2. Process stability

In order to evaluate the stability of HBsAg during spray-freezedrying in the presence of excipients and during freeze drying inthe absence of excipients, ELISA and intrinsic fluorescence wereperformed on the material, both before and immediately after(spray-)freeze drying. ELISA measurements demonstrated thatover 40% of the original antigenicity was lost when HBsAg was

HBsAg Sugar

Not SFD –Not SFD –SFD SFD

h HBsAg Not SFD SFDNot SFD Not SFD

econstituted SFD SFDmixed with HBsAg Not SFD SFDith HBsAg Not SFD Not SFD

Fig. 1. XRPD patterns of (partially) crystalline trehalose, dextran and inulin asreceived and the two amorphous powder formulations immediately after spray-freeze drying (t = 0) and after 3 months of storage at 60 �C (t = 91).

Table 2Tg values of the two spray freeze dried powders measured immediately after spray-freeze drying (SFD) and after 3 months of storage at 60 �C (n = 3; ±SD). RMC: residualmoisture content.

Formulation RMC (%) Tg (�C)

After SFD 3 months storage at60 �C

HBsAg + inu SFD 2.59 (±0.12) 166.70(±2.03)

164.31 (±0.57)

HBsAg + dex/treSFD

3.00(±0.27))

171.36(±0.90)

171.92 (±0.89)

Table 3The effect of spray-freeze-drying on the conformation of the antigen. Post-processvalues are given relative to the value before processing, for ELISA and tryptophanfluorescence measurements. (Mean; n = 3 ± SD).

Analysis HBsAg + inu SFD HBsAg + dex/treSFD

HBsAg FD

ELISA 99.2% (±1.0)(P = 0.39)

98.2% (±0.9)(P = 0.25)

56.1% (±2.1)(P < 0.001)

TryptophanImax

87.5% (±1.7) 84.0% (±0.4) 82.5% (±0.8)

(P = 0.02) (P = 0.0005) (P < 0.0005)

40 W.F. Tonnis et al. / European Journal of Pharmaceutical Sciences 55 (2014) 36–45

freeze dried in the absence of sugars. (see Table 3). In contrast,spray-freeze drying in the presence of sugar(s) did not significantly

Fig. 2. Scanning electron microscopy image of powder formulations before storage. (A) HHBsAg-inulin. (D) HBsAg-dextran/trehalose. Bar in the bottom of each image represents

affect the antigenicity of HBsAg. However, tryptophan fluorescenceintensity of the freeze dried material without excipients and thespray-freeze-dried material with either inulin or the dextran/tre-halose combination showed a small but significant change com-pared to the fluorescence intensity before processing. Thisindicates a slight change in the conformational structure of HBsAgduring both freeze drying andspray-freeze drying. The tryptophanfluorescence was maximal at a wavelength of 326 nm for all for-mulations, before as well as after (spray-)freeze drying.

3.3. Storage stability

To evaluate the storage stability of the two powder formula-tions and the two liquid formulations, they were subjected to tenfreeze–thaw cycles. The samples were also stored at 20 �C and60 �C for three months.

3.3.1. Freeze–thaw stabilityFig. 3 shows the antigenicity of HBsAg in the four different for-

mulations after ten freeze–thaw cycles. These results showed thatthe antigenicity of HBsAg was completely lost after ten freeze–thaw cycles when it is formulated with aluminum, as measuredby ELISA. Already after one freeze–thaw cycle a 27.0% (±6.9) lossof the original antigenicity was measured in the aluminum con-

BsAg-inulin. (B) HBsAg-dextran/trehalose and after 3 months of storage at 60 �C. (C)1 lm.

Fig. 3. Antigenicity of HBsAg after ten freeze–thaw cycles compared to theantigenicity before freeze-thawing of the two liquid formulations and two powderformulations. (Mean;n = 3 ± SD) bdl: below detection limit, ***P < 0.001 compared toantigenicity before FT.

W.F. Tonnis et al. / European Journal of Pharmaceutical Sciences 55 (2014) 36–45 41

taining formulation. In the other formulations, the antigenicity wasunaffected after ten freeze-thawing cycles.

The microscopic appearance of the HBsAg aluminum hydroxideformulation before and after freeze-thawing (Fig. 4) showed thatthe size of the aluminum hydroxide particles were increased afterfreeze-thawing from approximately 2–3 lm to 60 lm in diameter.

3.3.2. Storage at 20 and 60 �C3.3.2.1. Antigenicity. During storage at 20 �C for three months, theantigenicity of HBsAg in both powder formulations and in the li-quid plain formulation was unaffected, as shown by ELISA(Fig. 5a). The antigenicity of the liquid HBsAg aluminum hydroxideformulation, however, decreased during the first week after whichit gradually decreased further over time.

After 3 months of storage at 60 �C, spray-freeze-dried HBsAg inboth powder formulations retained approximately 90% of its initialantigenicity (Fig. 5b). In comparison, in freeze dried HBsAg withoutexcipients only 14% of the original antigenicity (before freeze dry-ing) was left. After one week of storage at 60 �C, only 60% antige-nicity was retained in the liquid plain HBsAg formulation andonly 20% in the liquid HBsAg-aluminum hydroxide formulation.We therefore conclude that during storage at 60 �C the antigenicityof HBsAg was substantially better preserved in the powder formu-lations than in both liquid formulations.

The degree of adsorption of HBsAg to aluminum hydroxide wasmonitored during two weeks of storage at 20 �C and 60 �C. This

Fig. 4. Microscopic appearance of HBsAg + alum formulation before (left) and after

was measured by centrifugation of the formulation (10 min at10,000 rpm) and quantification of HBsAg in the supernatant by ELI-SA and Lowry protein determination. To check whether unboundHBsAg would be spun down during centrifugation, a 40 lg/mLsolution of only the antigen (no aluminum) was centrifuged. TheHBsAg concentration before and after centrifugation was the same(data not shown) indicating that free unbound HBsAg would not bespun down by centrifugation. After centrifugation of the aluminumcontaining formulation, no significant amounts of HBsAg were de-tected in the supernatants by Lowry analysis (data not shown),indicating a high degree of adsorption immediately after produc-tion. ELISA, which is a much more sensitive analysis for HBsAg thanLowry, showed that immediately after preparation almost allHBsAg was adsorbed onto the aluminum hydroxide particles (de-gree of adsorption of 98.6% (±0.2), Table 4). The amount of ad-sorbed HBsAg increased significantly to 99.7% (±0.0) and 100.0%(±0.0) after 4 h of storage at 20 �C and 60 �C, respectively,(P < 0.005 at both storage conditions) and did not change upon fur-ther storage.

3.3.2.2. Fluorescence spectroscopy. Due to the opaque appearance ofthe alum-HBsAg suspension, it was not possible to obtain fluores-cence spectra and compare fluorescence intensities of this formu-lation by the two spectroscopic methods used in this study. Bothpowders were reconstituted before analysis and, similar to theplain liquid formulation, diluted to 10 lg/mL. An excitation wave-length of 295 nm was used to selectively excite tryptophan resi-dues in HBsAg. Once excited, electrons of tryptophan can fallback to the ground state and emit light. The intensity of the emit-ted light is strongly dependent on the average local environment ofall the tryptophan residues. A decrease in intensity, a processcalled quenching, can be caused by several factors: e.g. by protontransfer from nearby charged amino acids, by electron acceptors,by electron transfer by disulfides, amides, or peptide bonds inthe protein backbone or by resonance energy transfer among tryp-tophan residues (Lakowicz, 2006). The effect of quenching isdependent on the distance between tryptophan and the quencher.Therefore, the intensity of the emitted light depends on conforma-tion and composition of the protein. When the conformation of theprotein changes, the distance between tryptophan residues anddifferent quenchers changes, causing a change in intensity. Fur-thermore, when the average local environment of the tryptophanresidues changes, energy transfer from excited tryptophan resi-dues to the solvent can occur. This energy transfer results in ared-shift of the maximum fluorescence intensity. Before storagethe wavelength at the maximum fluorescence intensity was 326–327 nm and did not change significantly for any of the formula-tions during storage at both storage conditions (data not shown).

Fig. 6 shows the tryptophan fluorescence maximum intensitiesof both reconstituted powder formulations and the liquid plain for-

(right) ten freeze–thaw cycles. Bar in the top-right corner represents 60 lm.

Fig. 5. HBsAg antigenicity during 3 months storage at (A) 20 �C and (B) 60 �C (% asmeasured by ELISA). d SFD HBsAg-inulin, s SFD HBsAg-dex/tre, . plain HBsAg, 4HBsAg + alum; (n = 3; mean ± SD).

Table 4Degree of adsorption of HBsAg on aluminum at different points in time during thefirst 2 weeks of storage at 20 �C and 60 �C as determined by ELISA. In none of thesupernatants HBsAg was found by Lowry analysis; (n = 3; mean ± SD).

Time point 20 �C 60 �C

Start (no storage) 98.6% (±0.2) 98.6% (±0.2)4 h 99.7% (±0.0) 100.0% (±0.0)1 day 99.5% (±0.2) 99.9% (±0.2)7 days 99.3% (±0.1) 100.0% (±0.0)14 days 99.3% (±0.2) 100.0% (±0.0)

Fig. 6. Maximum intensity of tryptophan emission after 0, 7, 14, 28, 42, 63 and91 days of storage at (A) 20 �C and (B) 60 �C. d SFD HBsAg-inulin, s SFD HBsAg-dex/tre, D liquid plain HBsAg; (n = 3; mean ± SD).

42 W.F. Tonnis et al. / European Journal of Pharmaceutical Sciences 55 (2014) 36–45

mulation, stored at 20 �C and 60 �C. During storage at 20 �C, thefluorescent intensity of tryptophan in the two powder formula-tions hardly changed. The liquid plain formulation, however,showed a substantial decrease in intensity after 28 days of storage.

All formulations stored at 60 �C showed a decrease in maximumfluorescence intensity of the tryptophan residues in HBsAg. The de-crease in intensity in both powder formulations was of the samemagnitude, indicating that in both formulations the rate of confor-mational change was similar. The fluorescent intensity of the liquidplain formulation, however, decreased much faster than that ofboth powder formulations. Furthermore, the intensity of the liquidplain formulation predominantly decreased in the first week ofstorage at 60 �C.

The decrease in tryptophan fluorescence of HBsAg in the plainformulation as shown in Fig. 6, could be caused by a decrease in

HBsAg concentration due to adsorption on the wall of the glass vialor by aggregation and subsequent sedimentation of the HBsAg.Therefore, during the first 28 days of the stability study the HBsAgprotein concentration and the size of the HBsAg VLPs in the liquidplain formulation was measured as shown in Table 5. The proteinconcentration was determined by Lowry and the size of the HBsAgVLPs was measured by DLS. Results showed that before storage thesize of the HBsAg VLPs is approximately 20 nm, which is in agree-ment with other studies (Diminsky et al., 1999). After 28 days ofstorage at 20 �C and 60 �C the size of the HBsAg VLPs significantlyincreased to 26 and 27 nm, respectively. At the same time the poly-dispersity index was increased significantly from 0.23 before stor-age to >0.30 for both storage conditions. The protein concentrationafter 28 days was lower although not significant. Since the size andconcentration is not changed after 14 days of storage at 60 �C, butthe tryptophan fluorescence is already decreased by 80% in the first7 days, it is concluded that the drop in tryptophan fluorescenceintensity in the plain formulation as shown in Fig. 6 is not due toa decreased concentration, but is most likely caused by conforma-tional changes of HBsAg.

3.4. Immunogenicity

In order to evaluate the immunogenicity of the prepared formu-lations in vivo, the reconstituted powder formulations, both liquid

Table 5Protein concentration and the mean volume diameters of the HBsAg VLPs in the plain formulation, after storage at 20 �C, and 60 �C. The theoretical protein concentration was40 lg/mL.

Storage time 20 �C 60 �C

Size (nm ± SD) Protein concentration (lg/mL ± SD) Size (nm ± SD) Protein concentration (lg/mL ± SD)

0 days 20.1 (±1.3) 41.2 (±4.4) 20.1 (±1.3) 41.2 (±4.4)14 days 22.6 (±3.0) 41.0 (±4.8) 21.7 (±1.3) 43.8 (±2.4)28 days 25.6 (±0.6)** 39.5 (±3.3) 26.9 (±1.8)* 36.0 (±1.4)

* P < 0.01.** P < 0.005; compared to 0 days.

Fig. 8. Mean serum IgG1 and IgG2a titers at day 28. Titers are shown as the mean of7 animals, unless mentioned otherwise. Error bars represent the standard error ofthe mean.

W.F. Tonnis et al. / European Journal of Pharmaceutical Sciences 55 (2014) 36–45 43

formulations and proper controls were administered to mice. Micewere immunized at day 0 and received a booster dose at day 14.

Fig. 7 shows the serum IgG titers at day 14 and day 28. Admin-istration of the liquid plain formulation resulted in the lowest im-mune response. Serum of one of the seven mice in this group didnot contain detectable IgG antibodies against HBsAg. The highestlevels of antibodies were found after administration of the alumi-num hydroxide containing HBsAg formulation. The average IgG ti-ter after administration of this formulation was almost 100-foldhigher than that after administration of the liquid plain HBsAg for-mulation. The immunogenicity of the formulations where neitherHBsAg nor the sugar was spray-freeze dried (HBsAg + inu andHBsAg + dex/tre) was comparable to the liquid plain formulations:at day 14 also several non-responders were observed and at day 28the IgG titer were not significantly different compared to the IgGtiters after administration of the liquid plain formulation. Remark-ably, mixing unprocessed HBsAg with SFD sugar (either inulin ordextran/trehalose; HBsAg + SFD inu and HBsAg + SFD dex/tre) re-sulted in IgG titers that were significantly higher (p < 0.05) thanthe titers found after administration of the liquid plain formula-tion. Administration of the formulations where both HBsAg and su-gar were spray-freeze dried (SFD HBsAg-inu and SFD HBsAg-dex/tre) resulted in even higher IgG titers, although they were not sig-nificantly higher than the titers after administration of the formu-lations in which only the sugar was spray-freeze dried (eitherinulin or dextran/trehalose; HBsAg + SFD inu and HBsAg + SFDdex/tre). Even though, the titers were higher than those of the plainformulations, they were not as high as after administration of thealuminum hydroxide adsorbed HBsAg formulation.

Fig. 7. Mean serum IgG titers at day 14 and day 28. Titers are shown as the mean of7 animals, unless mentioned otherwise (‘‘x/7’’ indicates the number of responders).Error bars represent the standard error of the mean. The stars represent differencescompared to liquid plain HBsAg administration *P < 0.05, **P < 0.01, ***P < 0.001.

Fig. 8 shows the IgG subtype profile at day 28. In all experimen-tal groups, both IgG1 and IgG2a antibodies were induced. TheIgG2a titers in all groups were of the same magnitude (P > 0.05).The IgG1/IgG2a ratio was the largest in the HBsAg + alum groupand significantly higher than in all the other groups (P = 0.028–0.037). The IgG1/IgG2a ratio of all the other groups were not signif-icantly different from each other, indicating a more balanced Th1/Th2 response in all of these groups compared to administration ofthe HBsAg + alum formulation.

4. Discussion

In this study, we showed that the shelf life of the hepatitis Bsurface antigen can be increased by spray-freeze drying the vac-cine in the presence of either inulin or a mixture of dextran andtrehalose. The incorporation of the HBsAg in a matrix of amor-phous sugar(s) may enable storage and transport without a cold-chain. Secondly, this study indicates that spray-freeze-dried su-gar(s) improves the immune response to HBsAg.

In line with other studies, we found that the current aluminumadjuvanted HBsAg is not stable outside the refrigerator (Hirschberget al., 2010). Ten freeze–thaw cycles that mimicked the slow freez-ing rate in the presence of ice packs during transport, led to a lossof antigenicity of HBsAg when formulated with aluminum hydrox-ide. Optical microscopy of the formulation before and after freeze-thawing revealed that the size of the aluminum hydroxide saltsparticles were increased, which confirms findings by Maa et al.(2003). In that study it was also shown that increased aluminumsalt aggregation correlates with a decrease in immunogenicity.

44 W.F. Tonnis et al. / European Journal of Pharmaceutical Sciences 55 (2014) 36–45

Storage of aluminum adjuvanted HBsAg at both 20 �C and 60 �Cresulted in a decreased antigenicity (50% and 80% loss in the firstweek, respectively). The decrease in antigenicity during storageat 60 �C was larger than the decrease found by Hirschberg et al.at the same storage temperature (60% degraded within one weekat 60 �C) (Hirschberg et al., 2010). The decrease was not due toan increased adsorption onto aluminum, since the degree ofadsorption was already close to 100% immediately after prepara-tion of the formulation. However, the decrease in antigenicitywas steep, compared to the decrease in antigenicity of the plainformulation, which does not contain aluminum. Wittayanukulluket al. showed that aluminum hydroxide can destabilize adsorpedantigens due to an alkaline pH microenvironment at the surfaceof aluminum hydroxide (Wittayanukulluk et al., 2004). Anotherexplanation of the strongly decreased antigenicity might be the in-creased binding strength between HBsAg and aluminum hydroxideduring storage, called ageing or maturation (Matheis et al., 2001).This means that during storage the HBsAg becomes more stronglybound to aluminum hydroxide, which is therefore not totally des-orbed during ELISA analysis.

Freeze-drying of HBsAg showed that the antigenicity was de-creased when dried in the absence of any excipients. Furthermore,the remaining antigenicity decreased even further upon subse-quent storage at 60 �C. Both the process stability and the storagestability of HBsAg was improved by incorporation of the antigenin an amorphous matrix of either inulin or a combination of dex-tran and trehalose. Optimal stability will be obtained when the su-gar matrix is and remains in its glassy state during storage. Thisrequires formulations with a high Tg like the Tg (165–170 �C) ofthe powder formulations prepared in this study. Residual waterin the powder may act as a plasticizer, reducing the Tg. However,as shown by XRPD, the sugars remained fully amorphous in bothformulations during storage at 60 �C indicating that the Tg was wellabove the storage temperature. Powder formulations subjected tostrong temperature changes (freeze–thaw cycles) did not resultin a decrease in antigenicity. During 3 months of storage at 20 �C,neither the antigenicity nor the conformation of HBsAg changed.At 60 �C, both powder formulations showed an improved stabilitycompared to the liquid formulations. Some conformationalchanges occurred based on the changes in fluorescence spectros-copy signal, but the antigenicity hardly changed at this high tem-perature. We hypothesize that during storage at 60 �C,conformational changes occurred that had an effect on the hydro-phobic parts of HBsAg (tryptophan environment), but did not affectthe hydrophilic parts of the protein where the epitopes are locatedand, therefore, the antigenicity did not change. By these conforma-tional changes it is clear that HBsAg is not fully stable under theseharsh conditions.

Even though the antigenicity of both the liquid formulations de-creased rapidly upon storage, this does not necessarily mean theimmunogenicity is decreased (Chen et al., 2010; Hirschberg et al.,2010). However, this study shows that the powder formulationswere more stable than the liquid formulations in terms of remain-ing their antigenicity.

Spray-freeze drying HBsAg in the presence of sugars not onlyincreased the storage stability of the antigen, but surprisingly alsoseems to have a potentiating effect on the immune response toHBsAg. The IgG titers in sera of mice that received reconstitutedspray-freeze-dried HBsAg formulations were higher than afteradministration of the liquid plain formulation and also the numberof responders was higher. The results suggest that the improvedimmune response compared to the liquid plain formulation mightbe related to the processing of the antigen but it is more likely to becaused by spray-freeze drying of the sugar(s). This is confirmed bythe lower immune response after administration of the formula-tions where neither the sugar nor the antigen was spray-freeze

dried compared to the immune response after administration ofthe formulations where only the sugar was spray-freeze dried. Ina study by Amorij et al. it was found that the immune responseafter pulmonary administration of spray-freeze-dried influenzavaccine in the presence of inulin was better than after pulmonaryadministration of a liquid plain formulation (antigen in buffer). Theadjuvating effect could be due to the formation of insoluble sugarparticles during the spray-freeze drying process. From the litera-ture it is known that some crystalline forms of inulin can act asan adjuvant (Silva et al., 2004). Although XRPD measurementsshowed that inulin was fully amorphous after spray-freeze drying,a minor amount (below detection limits) of crystalline inulin maystill have been present. The same principle might be true for theformulation containing dextran and trehalose. Stenekens et al.showed that insoluble particles of dextran can be formed in aque-ous supersaturated dextran solutions (Stenekes et al., 2001). At thismoment the exact mechanism behind the apparent adjuvating ef-fect of spray-freeze-dried sugars is unclear. Further research isneeded to unravel this phenomenon.

Although the immune response was improved when the sugarwas spray-freeze dried, the immune response was not as high asafter administration of the aluminum adjuvanted HBsAg formula-tion. However, for all formulations where the sugar was spray-freeze dried and for the aluminum containing formulation, all miceshowed antibodies against HBsAg at day 14. The potency assay ofthe hepatitis B vaccine described in the European Pharmacopoeiastates is solely based on the number of responders after one doseand not on the extent of the immune response. In this light bothpowder formulations and the aluminum containing HBsAg formu-lation are equally potent and they are all more potent than the li-quid plain formulation.

A final advantage of the reconstituted powder formulations isthat they elicit a more balanced Th1/Th2 immune response thanthe liquid HBsAg formulation containing aluminum hydroxide.This could lead to a better protection against HBV, but should ide-ally be supported by clinical studies focusing on protection.

5. Conclusion

Our study shows that incorporation of HBsAg in an amorphousmatrix of inulin or a combination of dextran and trehalose byspray-freeze drying, results in a formulation that is more stablethan the aluminum hydroxide containing formulation, based onthe current GlaxoSmithKline product. Both powder formulationscan be stored for at least 3 months at room temperature, withoutany change in antigenicity or conformation of the HBsAg protein.Next to this improved stability, administration of the reconstitutedpowder formulations appears to elicit an improved immune re-sponse compared to administration of the unprocessed antigenwithout processed sugars.

References

Allison, S.D., Manning, M.C., Randolph, T.W., Middleton, K., Davis, a., Carpenter, J.F.,2000a. Optimization of storage stability of lyophilized actin using combinationsof disaccharides and dextran. J. Pharm. Sci. 89, 199–214.

Allison, S.D., Molina, M.C., Anchordoquy, T.J., 2000b. Stabilization of lipid/DNAcomplexes during the freezing step of the lyophilization process: the particleisolation hypothesis. Biochim. Biophys. Acta 1468, 127–138.

Amorij, J.-P., Saluja, V., Petersen, a H., Hinrichs, W.L.J., Huckriede, a., Frijlink, H.W.,2007. Pulmonary delivery of an inulin-stabilized influenza subunit vaccineprepared by spray-freeze drying induces systemic, mucosal humoral as well ascell-mediated immune responses in BALB/c mice. Vaccine 25, 8707–8717.

Amorij, J.-P., Huckriede, a., Wilschut, J., Frijlink, H.W., Hinrichs, W.L.J., 2008.Development of stable influenza vaccine powder formulations: challenges andpossibilities. Pharm. Res. 25, 1256–1273.

Beutels, P., 2001. Economic evaluations of hepatitis B immunization: a globalreview of recent studies (1994–2000). Health Econ. 774, 751–774.

W.F. Tonnis et al. / European Journal of Pharmaceutical Sciences 55 (2014) 36–45 45

Brandau, D.T., Jones, L.S., Wiethoff, C.M., Rexroad, J., Middaugh, C.R., 2003. Thermalstability of vaccines. J. Pharm. Sci. 92, 218–231.

Braun, L.J., Jezek, J., Peterson, S., Tyagi, A., Perkins, S., Sylvester, D., Guy, M., Lal, M.,Priddy, S., Plzak, H., Kristensen, D., Chen, D., 2009. Characterization of athermostable hepatitis B vaccine formulation. Vaccine 27, 4609–4614.

Chang, L., Pikal, M., 2009. Mechanisms of protein stabilization in the solid state. J.Pharm. Sci. 98, 2886–2908.

Chang, L.L., Shepherd, D., Sun, J., Ouellette, D., Grant, K.L., Tang, X.C., Pikal, M.J., 2005.Mechanism of protein stabilization by sugars during freeze-drying and storage:native structure preservation, specific interaction, and/or immobilization in aglassy matrix? J. Pharm. Sci. 94, 1427–1444.

Chen, D., Kapre, S., Goel, a., Suresh, K., Beri, S., Hickling, J., Jensen, J., Lal, M., Preaud,J.M., Laforce, M., Kristensen, D., 2010. Thermostable formulations of a hepatitisB vaccine and a meningitis A polysaccharide conjugate vaccine produced by aspray drying method. Vaccine 28, 5093–5099.

De Gregorio, E., Tritto, E., Rappuoli, R., 2008. Alum adjuvanticity: unraveling acentury old mystery. Eur. J. Immunol. 38, 2068–2071.

Diminsky, D., Moav, N., Gorecki, M., Barenholz, Y., 1999. Physical, chemical andimmunological stability of CHO-derived hepatitis B surface antigen (HBsAg)particles. Vaccine 18, 3–17.

Galazaka, A., Milstien, J., Zaffran, M., 1998. Thermostability of Vaccines, Report byWolrd Heatlh Organization, URL: <http://whqlibdoc.who.int/hq/1998/WHO_GPV_98.07.pdf> (accessed 11.09.13).

GlaxoSmithKline, 2013 Full Prescribing Information of Engerix-B URL: <http://us.gsk.com/products/assets/us_engerixb.pdf> (accessed 17.12.13).

Goldstein, S.T., Zhou, F., Hadler, S.C., Bell, B.P., Mast, E.E., Margolis, H.S., 2005. Amathematical model to estimate global hepatitis B disease burden andvaccination impact. Int. J. Epidemiol. 34, 1329–1339.

Hinrichs, W.L., Prinsen, M.G., Frijlink, H.W., 2001. Inulin glasses for the stabilizationof therapeutic proteins. Int. J. Pharm. 215, 163–174.

Hirschberg, H.J.H.B., van de Wijdeven, G.G.P., Kraan, H., Amorij, J.-P., Kersten, G.F.a.,2010. Bioneedles as alternative delivery system for hepatitis B vaccine. J.Control. Release 147, 211–217.

Just, M., Berger, R., 1988. Immunogenicity of a heat-treated recombinant DNAhepatitis B vaccine. Vaccine 6, 399–400.

Karthigesu, V.D., Allison, L.M., Ferguson, M., Howard, C.R., 1999. A hepatitis B virusvariant found in the sera of immunised children induces a conformationalchange in the HBsAg ‘‘a’’ determinant. J. Med. Virol. 58, 346–352.

Kool, M., Soullié, T., van Nimwegen, M., Willart, M.A.M., Muskens, F., Jung, S.,Hoogsteden, H.C., Hammad, H., Lambrecht, B.N., 2008. Alum adjuvant boostsadaptive immunity by inducing uric acid and activating inflammatory dendriticcells. J. Exp. Med. 205, 869–882.

Lakowicz, J.R., 2006. Protein fluorescence. In: Lakowicz, J.R. (Ed.), Principles ofFluorescence Spectroscopy, third ed. Springer, New York, pp. 529–575.

Lambrecht, B.N., Kool, M., Willart, M.A.M., Hammad, H., 2009. Mechanism of actionof clinically approved adjuvants. Curr. Opin. Immunol. 21, 23–29.

Lowry, O., Rosebrough, N., 1951. Protein measurement with Folin phenol reagent. J.Biol. Chem. 193, 265–275.

Maa, Y.-F., Zhao, L., Payne, L.G., Chen, D., 2003. Stabilization of alum-adjuvantedvaccine dry powder formulations: mechanism and application. J. Pharm. Sci. 92,319–332.

Maa, Y.-F., Ameri, M., Shu, C., Zuleger, C.L., Che, J., Osorio, J.E., Payne, L.G., Chen, D.,2007. Hepatitis-B surface antigen (HBsAg) powder formulation: process andstability assessment. Curr. Drug Deliv. 4, 57–67.

Matheis, W., Zott, a., Schwanig, M., 2001. The role of the adsorption process forproduction and control combined adsorbed vaccines. Vaccine 20, 67–73.

Maynard, J.E., Kane, M.A., Hadler, S.C., 2011. Global control of hepatitis B throughvaccination: role of hepatitis B vaccine in the Expanded Programme onImmunization. Rev. Infect. Dis. 11(Suppl. 3), S574-8.

Nelson, C., Froes, P., Van Dyck, A.M., Chavarría, J., Boda, E., Coca, A., Crespo, G., Lima,H., 2007. Monitoring temperatures in the vaccine cold chain in Bolivia. Vaccine25, 433–437.

Saluja, V., Amorij, J.-P., Kapteyn, J.C., de Boer, a.H., Frijlink, H.W., Hinrichs, W.L.J.,2010. A comparison between spray drying and spray freeze drying to producean influenza subunit vaccine powder for inhalation. J. Control. Release 144,127–133.

Silva, D.G., Cooper, P.D., Petrovsky, N., 2004. Inulin-derived adjuvants efficientlypromote both Th1 and Th2 immune responses. Immunol. Cell Biol. 82, 611–616.

Stacy, B.D., King, E.J., Harrison, C.V., Nagelschmidt, G., Nelson, S., 1959. Tissuechanges in rats’ lungs caused by hydroxides, oxides and phosphates ofaluminium and iron. J. Pathol. Bacteriol. 77, 417–426.

Stenekes, R.J., Talsma, H., Hennink, W.E., 2001. Formation of dextran hydrogels bycrystallization. Biomaterials 22, 1891–1898.

Techathawat, S., Varinsathien, P., Rasdjarmrearnsook, A., Tharmaphornpilas, P.,2007. Exposure to heat and freezing in the vaccine cold chain in Thailand.Vaccine 25, 1328–1333.

Thanavala, B.Y.M., Brown, S.E., Howard, C.R., Roitt, I.M., Steward, M.W., 1986. Asurrogate hepatitis b virus antigenic peptide represented internal by a syntheticand an image antiidiotype antibody. J. Exp. Med. 164, 227–236.

Wang, W., 2000. Lyophilization and development of solid protein pharmaceuticals.Int. J. Pharm. 203, 1–60.

Wasmuth, J.-C., 2010. Hepatology, a clinical textbook. In: Maus, S., Berg, T.,Rockstrof, J., Sarrazin, C., Wedemeyer, H. (Eds.), Hepatology, A Clinical Textbook.Flying Publisher.

Waters, J.a., Brown, S.E., Steward, M.W., Howard, C.R., Thomas, H.C., 1992. Analysisof the antigenic epitopes of hepatitis B surface antigen involved in the inductionof a protective antibody response. Virus Res. 22, 1–12.

Wittayanukulluk, A., Jiang, D., Regnier, F.E., Hem, S.L., 2004. Effect ofmicroenvironment pH of aluminum hydroxide adjuvant on the chemicalstability of adsorbed antigen. Vaccine 22, 1172–1176.

World Health Organization, 2012. Hepatitis B fact sheet URL: <http://www.who.int/mediacentre/factsheets/fs204/en/> (accessed 11.09.13).

Zapata, M., Peck, G., Hem, S., 1984. Mechanism of freeze–thaw instability ofaluminum hydroxycarbonate and magnesium hydroxide gels. J. Pharm. Res. 73,3–8.