Hepatitis-B Surface Antigen (HBsAg) Powder Formulation: Process and Stability Assessment
Yuh-Fun Maa1,2,*, Mahmoud Ameri1,2, Cassandra Shu1,3, Cindy L. Zuleger1, Jenny Che1, Jorge E. Osorio1, Lendon G. Payne1 and Dexiang Chen1,4
1PowderJect Vaccines, Inc., 8551 Research Way Boulevard, Middleton, WI 53562, USA; Current addresses: 2Alza Cor-poration, 1900 Charleston Road, Mountain View, CA 94043, USA; 3Cell Genesys, 342 Lakeside Drive, Foster City, CA 94404; USA; 4Program for Appropriate Technology in Health, 1455 NW Leary Way, Seattle, WA 98107-5136, USA
Abstract: The purpose of this study was to develop a hepatitis-B surface antigen (HBsAg) dry powder vaccine formula-tion suitable for epidermal powder immunization (EPI) via an efficient, scalable powder-formation process. Several HBsAg dry powder formulations were prepared using four different powder-formation methods: freeze-drying/compress/grind/sieve (FD/C/G/S), spray-drying (SD), agarose beads, and spray freeze-drying (SFD). Powder prop-erties and physical stability were determined using particle size analysis, tap density measurement, scanning electron mi-croscopy, optical microscopy, and moisture content analysis. Physical, chemical and biochemical stability of HBsAg was determined by dynamic light scattering, an enzyme immune assay, and immunogenicity in a mouse or hairless guinea pig model. Out of the four powder-formation methods evaluated SFD outperformed other methods in the following considera-tions: good process efficiency, flexible scalability, and desirable particle characteristics for skin penetration. The stress posed by SFD appeared to be mild as HBsAg in the dry form retained its potency and immunogenicity. Notably, the mechanism of fast freezing by SFD actually promoted the preservation of HBsAg nanoparticle size, in good correlation with long-term biochemical stability. Among several formulations screened, the formulation containing 10 µg HBsAg in 1-mg powder with a tertiary mixture of trehalose, mannitol, and dextran, exhibited excellent overall stability performance. In conclusion, HBsAg dry powder formulations suitable for EPI were successfully prepared using SFD. Further, a system-atic formulation development strategy allowed the development and optimization of an HBsAg dry powder formulation, demonstrating excellent long-term physical, biochemical, and immunological stability.
Hepatitis-B virus (HBV) infection in humans may lead to serious health consequences, such as chronic liver disease, cirrhosis, or hepato-cellular carcinoma [1]. Of approximately 300 million carriers of HBV worldwide, up to 30% of these infections might be fatal [2]; thus posing a significant impact on the economy and public health. Vaccination is the pri-mary means to prevent the Hepatitis B disease [3]. The commercially available HBV vaccines are formulated by adsorbing hepatitis-B surface antigen (HBsAg) onto alumi-num hydroxide (commonly known as alum) as a liquid sus-pension and administered by intramuscular (IM) injection using needle and syringe. These vaccine products, although highly immunogenic and protective, are normally adminis-tered following a complex schedule: three-dose regimen over a period of 4-10 months. Aside from the non-compliance issue, this vaccination approach also has some disadvan-tages. Notably many needle injections are considered unsafe due to needle reuse and needle-stick injuries, which lead to the spread of blood-borne diseases (Fact sheet N°231, 2002, World Health Organization).
*Address correspondence to this author at the ALZA Corporation. 1900 Charleston Road, Mountain View, CA 94043, USA; Tel: 650-564-2104; Fax: 650-564-2700; E-mail: [email protected]
Needle-free immunization via the skin appears to be an attractive approach. Skin contains abundant immune cells [4], including Langerhans cells (LCs) in the viable epider-mis, and dendritic cells in the dermis. These antigen present-ing cells (APCs) take up antigens in the skin, migrate into draining lymph nodes, and present processed antigens to the CD8+ and CD4+ T helper cells [5]. Epidermal powder im-munization (EPI) is a needle-free immunization technology designed to deliver antigens to the epidermis, with its feasi-bility being demonstrated in several vaccines in various ani-mal models [6-11] and in humans [12]. EPI uses the motive force of helium gas released from a mechanical device to propel the vaccine powder into the epidermis as previously described [7,8]. Delivery antigens to epidermis was shown to cause no injection site pain because of the sparse sensory nerve endings in the target tissue [12]. Although the immune responses to a variety of vaccines delivered by EPI have been well studied in animal models, formulation process development, a critical component of the drug delivery product development, is not often published [13-15]. This paper attempts to describe the rationale and approaches of developing an HBsAg powder formulation and identifying a scalable manufacturing process for future human clinical investigation. To develop a viable manufacturing process, the associ-ated powder-formation methods should possess the follow-
58 Current Drug Delivery, 2007, Vol. 4, No. 1 Maa et al.
ing characteristics: simplicity, efficiency, scalability, com-patibility with various drug molecules/excipients, etc. Equally important is that the powders derived from the se-lected process should exhibit acceptable particle characteris-tics, stability, and in vivo performance. Established through prior pre-clinical and clinical investigations, the preferred particle characteristics, including particle size of 20-60 µm and tap density of >0.5 mg/mL, will offer balanced particle penetration and skin tolerability [13,14]. The acceptance criteria for stability and in vivo performance are targeted for being equivalent to or better than the marketed product. In this study, three powder-formation methods⎯agarose beads, spray-drying (SD), and spray freeze-drying (SFD)⎯were compared to the method of freeze-dried/compress/ grind/ sieve (FD/C/G/S) for producing HBV dry powder vaccines. Advantages as well as challenges associated with each method were analyzed in this study. After selecting a pre-ferred process, we performed formulation screening to iden-tify formulations with the best overall physical and bio-chemical stability.
MATERIALS AND METHODS
Materials
The HBsAg/adw2 (genotype A) bulk was obtained from Rhein Biotech (Buenos Aires, Argentina) and was supplied at the concentration of 1.4 mg/mL in saline. The HBsAg was made in yeast Hansenular polymorpha [16, 17] All excipients used in this study were purchased from Sigma (St. Louis, MO) and were of reagent quality or higher grade. These excipients included dextrans of three molecular weights (1, 10, and 37 kDa), lysine, mannitol, poloxamer 188 (Pluronic F68), and trehalose dihydrate. Agarose beads, prepared from 4% agarose gel (i.e., con-taining 96 wt% water when fully hydrated), were obtained from XC Corporation (Tampa, FL).
Methods
HBsAg pre-Formulation
The starting HBsAg (1.4 mg/mL) bulk was dialyzed against 2mM potassium phosphate buffer (pH 6.5) using a 10-kDa molecular weight cut-off, regenerated cellulose membrane (Spectrum/Por 7, Spectrum Labs, Rancho Dom-inguez, CA). After dialysis, the solution was concentrated by ultracentrifugation (Centriprep® with a 10 kDa regenerated cellulose membrane, Millipore, Bedford, MA). The filter was rinsed with 15 ml each of purified water and pH 7.2 sodium phosphate buffer to remove the trace amount of glycerin present in the filter by centrifugation at 3000 rpm at 5-10°C (Allegra 6®, Beckman, Fullerton, CA). The concen-tration step was ended when a pre-determined permeate vol-ume was obtained.
Freeze Drying/Compress/Grind/Sieve (FD/C/G/S)
To prepare high-density powders for EPI, freeze-dried (FD) HBsAg formulations were compressed in a stainless steel dye of 1.3-cm in diameter (Carver Press, Wabash, IN) at a pressure of 5,500–6,800 kilograms for 5-10 minutes. Then the compressed discs were ground manually using a
mortar and pestle and sieved through a stack of 3-in sieves (Fisher Scientific Products, Pittsburgh, PA) to collect parti-cles of 38-53 µm in diameter-the optimal particle size range for skin delivery [13,14].
Spray-Drying
A custom-made spray-dryer was used to produce large particles (>30 µm in median size). This dryer consists of a glass chamber (6 in. in diameter and 5 ft in height) and an ultrasonic atomizer (120 kHz, Sono-Tek Corp., Milton, NY). Directly under the glass chamber was a vacuum paper filter, onto which the dry powder was collected, instead of the con-ventional cyclone mechanism. Spray-drying conditions were as follows: drying air inlet temperature of 130-140oC, liquid feed of 3.5 mL/min, and drying air outlet temperature of 80-85oC. The sprayed formulation droplets were only exposed to this temperature gradient for 1-2 seconds for the water to evaporate. The dry products were never exposed to tempera-tures above 45°C.
Agarose Beads
Agarose beads (4% by weight), supplied as wet beads i.e., fully hydrated, were washed with de-ionized water to remove sodium azide and then freeze dried to produce the dry agarose beads. Similarly, the dry beads were sieved to 38-53 µm to generate the pre-formed agarose beads. To load HBsAg into the beads, the liquid formulation containing HBsAg and the excipients were added dropwise to the dry beads, which in turn hydrated and swelled to facilitate the surface coating and potential inward diffusion of HBsAg and excipients. By mass balance, the liquid (the sum of water, antigen, and excipients) weighed 24-fold the dry bead weight. By a different approach, dry beads could be hydrated using a loading solution in excess followed by filtration to remove the liquid. All loaded beads were then lyophilized.
Spray-Freezing-Drying (SFD)
The SFD method has been previously described [13,14]. Briefly, the HBsAg liquid formulation, prepared by dissolv-ing solid excipients directly in the concentrated HBsAg bulk to a desired solid content, was sprayed using an ultrasonic atomizer (60 kHz, Sono-Tek Corporation, Milton, NY) at a feed rate of 1.5 mL/min into liquid nitrogen contained in a stainless steel pan. The pan was transferred to a lyophilizer (DuraStop, FTS Systems, Stone Ridge, NY) with pre-chilled shelves at –55°C. Primary drying was performed at –10°C for 10 hours and secondary drying at 15°C and then 25°C for 5 hours each. Throughout the cycle, the temperature ramping rate was set at 1oC/minute and the chamber vacuum at 100 mTor.
Optical Microscopy
Visual analysis of the particles was performed using an optical microscope (Model DMR, Leica, Germany) with a 10x-eyepiece lens and 10x-objective lens. The system was equipped with a Polaroid camera system for image output.
Scanning Electron Microscopy
The external morphology of coated particles was exam-ined using an Amray 1810T scanning electron microscope
Hepatitis-B Vaccine Powder Formulations Current Drug Delivery, 2007, Vol. 4, No. 1 59
(Amray, Bedford, MA). Powder samples were sputter coated with gold using a Hummer JR Technics unit (Pergamon Corporation, King of Prussia, PA) prior to microscopy.
Particle Size Analysis
The mean geometric/aerodynamic diameter of the parti-cles in the volume distribution was determined using a time-of-flight particle size analyzer (Aerosizer, API, Minneapolis, MN). The mean volumetric size was calculated by the soft-ware using the density of 1.0 g/mL and the particle popula-tion between 10% (D10) and 90% (D90) was reported for par-ticle size distribution. Each analysis required approximately 3-5 mg of powder sample. Triplicate measurement was made for each sample.
Powder Tap Density
Tap density was used to estimate the relative particle density of different formulations. Tap density correlates well with true particle density if particle size/distribution and par-ticle shape/morphology are similar, which were the case with all formulations except for Formulation A reported in this study. Briefly, each powder sample, approximately 100-300 mg, was weighed in a 2-mL glass vial (12 x 35 mm, E & K Scientific Products, Cat. # E025010). The vial was gently tapped 20 times against the lab bench. The tap density of the powder sample could be calculated by comparing the weight of the powder to the weight of water of an equivalent volume (assuming water density = 1 g/mL). Triplicate measurement was made for each sample.
Moisture Content Analysis
Approximately 4-5 mg of the powder sample was weighed into an aluminum-weighing vessel, and the weight was recorded. The sample was then loaded into a Karl Fisher Coulometer (Model 737, Brinkmann) equipped with a drying oven (Model 707). The sample was heated to 150 oC for 150 sec with a gas flow rate of 100 mL/min within the drying oven. The extraction length was set to 120 seconds with a drift level of 10 µg/min. Triplicate measurement was made for each sample.
Light Scattering for HBsAg Nanoparticle Size Analysis
A dynamic light scattering instrument (DynaPro-LSRTM, Protein Solutions Inc., Charlottesville, VA) was used to measure the size of the HBsAg nanoparticles in solution be-cause the preservation of HBsAg’s particle size (22 nm) is important to the immunogenicity of the vaccine [18]. A laser light beam passed through the sample solution, where HBsAg nanoparticles scattered the light. A photomultiplier, positioned at 90° to the laser light direction, collected the scattered light and sent a signal to an autocorrelator, where the scattered-light intensity time correlation functions were established. Analysis of autocorrelation data allowed the hydrodynamic diameter of the light-scattering particles to be calculated. Triplicate measurement was made for each sam-ple.
Protein Assay
A micro-BCA (Bicinchoninic acid) protein assay kit (Catalog #23233) and a Bovine Serum Albumin (BSA at 2
mg/mL) protein reference standard (Catalog #23210) were obtained from Pierce (Rockford, Illinois). One mL of the BCA reagent was mixed with 1-mL of a BSA standard (5-20 µg/mL) or an unknown sample in a glass tube. The mixture was vortexed and incubated at 60°C in a water bath for 30 minutes. After incubation, the sample solution (300 µL) was pipetted into a 96-well plate and scanned at 562 nm using a Plate Reader (SpectraMax 340PC, Molecular Dynamics, Sunnyvale, CA) to measure the color intensity of each sam-ple. Thus, the BCA standard curve was established, against which the concentration of the unknown samples could be calculated. This assay determined the total protein content but was not necessarily related to the potency of HBsAg. Triplicate measurement was made for each sample. This assay is known to have 30% sample-to-sample variations.
In Vitro Antigenicity of HBsAg
The potency of HBsAg formulations was determined by a quantitative enzyme immune assay using an enzyme im-munoassay (EIA) kit (AUSZYME® Monoclonal diagnostic kit, Abbott Laboratories, Abbott Part, IL). Each sample was examined at three different dilutions against a linear fitting to the responses of control standard dilutions. The potency of the formulations was expressed as the quantity of HBsAg (µg).
Immunogenicity Test
Six to eight week-old female Balb/c mice (n=4, Charles River, Wilmington, MA) were used to assess the immuno-genicity of the HBsAg formulations. Formulations were re-constituted in saline and injected into the deltoid muscle us-ing a 26-gauge needle on day 0 and 28. Each animal received 0.05 ml of solution containing 0.1 µg HBsAg. Blood was collected via retro-orbital bleeding prior to each immuniza-tion and 4 weeks after the final immunization and the anti-body responses were analyzed using ELISA. For EPI, the previously described methods were used [7-12]. Briefly, approximately 0.5 mg of the dry powder con-taining 5 µg HBsAg (the amount of powder was adjusted to ensure the HBsAg is the same for all formulations) was dis-pensed into a trilaminate cassette and the cassette inserted into a PowderJect ND device. The device was placed against the shaved abdominal skin of hairless guinea pigs (n=5, Charles River, Wilmington, MA) and actuated by releasing the compressed helium at 40-bar pressure from the gas cyl-inder. Control animals were immunized with 0.05 ml (5µg HBsAg alone or adsorbed to 100µg of aluminum hydroxide) of liquid vaccine in saline by IM injection using a 30-gauge needle. Immunizations were performed on days 0 and 28. Blood was collected via the carotid vein under anesthesia on days 0, 28 and 42.
ELISA
The antibody responses to HBsAg were determined using a modified ELISA method [11]. A 96-well plate (Costar, Fisher Scientific Products, Pittsburgh, PA) was coated with 0.1 µg of HBsAg antigen in 30 mM phosphate buffered sa-line (PBS), pH 7.4, per well overnight at 4°C. Plates were washed 3 times with tris-buffered saline (TBS), pH 7.4, con-taining 0.1% Brij-35, and incubated with test sera diluted in
60 Current Drug Delivery, 2007, Vol. 4, No. 1 Maa et al.
PBS containing 5% dry milk for 1.5 hr. The plates were then washed and incubated with biotin-labeled goat anti-mouse antibodies (1:8,000 in PBS, Southern Biotechnology Associ-ate, Birmingham, AL) for 1 hr at room temperature. Finally, plates were washed and developed with TMB substrate (Bio-Rad Laboratories, Melville, NY). The endpoint titers of the sera were determined by 4-parameter analysis using the Softmax Pro 4.1 program (Molecular Devices, Sunnyvale, CA) and defined as the reciprocal of the highest serum dilu-tion with an OD reading above the background by 0.1. A standard serum, containing a known level of antibodies to HBsAg, was added to each plate and used to standardize the titer in the final data analysis and to adjust assay-to-assay and plate-to-plate variation. The titer was expressed as ELISA unit.
RESULTS AND DISCUSSION Preformulation of HBsAg
HBsAg was in the form of 22 nm lipoprotein particles suspended in saline at 1.4 mg/mL with purity >95%. Alto-gether there were ten representative formulations prepared in this study (Table 1) with two different objectives: the as-sessment of the four powder-formation processes (Formula-tions A-D) and screening formulations prepared by the most promising spray-freeze-drying process (Formulations E-J). In the first study, the HBsAg bulk was dialyzed with 2 mM sodium phosphate buffer (pH 6.4). For the second study, the HBsAg bulk was concentrated into 20 mg/mL, approxi-mately 15-fold increase, using an ultracentrifugation device (Centriprep® with a 10 kDa membrane) after dialysis. Macroscopically no significant difference was noted for the bulk vaccine before and after dialysis although the con-centrated bulk was slightly opalescent due to light scattering of the HBsAg nanoparticles. By protein analysis, approxi-mately 90% of the protein was recovered after concentration.
The AUSZYME® analysis indicated that HBsAg potency was preserved (within ±30% sample-to sample variation). The DLS analysis showed that the HBsAg particle size (22 nm in theory) increased from 25 ± 5 nm to 33 ± 8 nm after dialysis and concentration (Table 3). It was undetermined whether the particle size increase was caused by aggregation or due to the extremely high protein concentration. Overall, HBsAg’s physical and biochemical properties were main-tained during preformulation.
Formulation of liquid HBsAg
Different excipients and excipient compositions were employed to prepare the ten powder formulations (Table 1). These formulated HBsAg solutions remained clear due to the absence of insoluble particulates (other than the original HBsAg nanoparticles) as detected by light scattering. Anti-gen stability (potency) at 2-8oC based on the AUSZYME® assay was maintained for up to 10 weeks (data not shown), considered a sufficiently long interval between two process steps in the manufacturing setting. The excipients tested in this study were selected because they have either been used in parenterally injected products or are suitable for such use, and these excipients are previ-ously known to provide physical and biochemical stability as well as appropriate mechanical properties to the powder formulation [13-15,19]. In detail, all formulations included an amorphous disaccharide sugar, trehalose, which is com-monly used as a protein stabilizer. Other common sugars, such as sucrose, lactose, and maltose, have been evaluated. However, sucrose is highly hygroscopic, making particles very sticky, and in turn became less stable physically. Lac-tose and maltose are also more hygroscopic than trehalose. Further, both sugars are reducing sugars and can potentially react with proteins by glycosylation.
Table 1. Summary of All Formulations Used in the Study
Formulation Powder formation methods
HBsAg conc. (mg/mL)
Solid con-tent (%)
Excipient ratio HBsAg dose (µg) per powder weight (mg)
Purpose
A FD/C /G/S 1.4 10 T :M :D(10)=3 :3 :4 2/0.5
B 4%Agarose beads 1.4 10 T :D(10)=7 :3 2/0.5
C Spray drying 14 10 T :M :D(10)=3 :3 :4 2/0.5
D SFD 1.4 15 T :M=7 :3 2/0.5
Powder feasibil-ity and process
evaluation
E SFD 20 35 T :M :D(37)=3 :3 :4 10/1
F SFD 20 35 T :M :D(10)=3 :3 :4 10/1
G SFD 20 35 T :M :D(1)=3 :3 :4 10/1
H SFD 20 35 T :M=7 :3 (1% HBsAg; 7% lysine) 10/1
I SFD 20 35 T:M=7 :3 (1% HBsAg; 5% Pluronic
F68P) 10/1
J Spray drying 20 30 T :M :D(10)=3 :3 :4 20/1
Formulation screening
a T: trehalose; M: mannitol; D (37): dextran (37 kDa); D (10): dextran (10 kDa); D (1): dextran (1 kDa).
Hepatitis-B Vaccine Powder Formulations Current Drug Delivery, 2007, Vol. 4, No. 1 61
In addition, one or two additional bulking agents were added to fortify mechanical strength of the particle and/or alleviate the hygroscopicity of the powder. Mannitol, a crys-tallizing sugar, may offer both functions to the particle. A polymer, such as dextrans of different molecular weights, may impart elasticity to the formulation to reduce particle brittleness. Dextran may be capable of enhancing the glass transition temperature of the dry protein formulation, thereby improving the protein’s biochemical stability [20]. All for-mulations except B and D contained a mixture of three ex-cipients. Instead of dextran, the third compound in Formula-tions H and I included lysine and poloxamer 188, respec-tively. Lysine, a highly water soluble, crystalline amino acid, may serve two purposes: enhancing the solid content of the
solution and providing additional stabilizing effects. Pluronic F68 is a hydrophilic detergent (HLB approx. 29) with a wa-ter solubility of greater than 10 %. As a solid surfactant at room temperature, we anticipate it may form a less hygro-scopic layer on the surface of the SFD particle, and may im-prove particle flowability and physical stability.
Process Evaluation and Powder Characterization
Freeze-Dried/Compressed/Ground/Sieved (FD/C/G/S)
This method produced Formulation A with a low yield and low potency recovery, <30% (Tables 2 and 3). Indeed, as a random pulverizing process that lacks effective particle size control, grinding tended to produce powders of a broad
Table 3. Effect of Formulation Processes on HBsAg Particle Sizes Determined by Dynamic Light Scattering and Potency Meas-ured by AUZYME Assay
Formulation or process HBsAg particle hydrodynamic diameter (nm) Potency recovery
Theoretical value 22 -
Starting HBsAg bulk 25 ± 5 -
Dialyzed and concentrated to 20 mg/mL 33 ± 8 90%
Formulation A 140 ± 20 20-30%
Formulation B Not determined 60%
Formulation C 150 ± 50 50%
Formulation D 36 ± 6 90%
Formulation E 31 ± 7 90%
Formulation F 39 ± 9 90%
Formulation G 36 ± 6 90%
Formulation J 170 ± 40 50%
Table 2. Summary of the Physical Properties of the Powder Formulations in Table 1
62 Current Drug Delivery, 2007, Vol. 4, No. 1 Maa et al.
particle size distribution. Sieving such a powder into a nar-row particle size range of 38-53 µm would compromise the yield and tended to create irregularly shaped particles (Fig. 1A). Despite shape irregularity, the measured particle size was consistent with the theoretical (sieved) particle size. The powder’s moisture content was measured to be 3.3%, dic-tated probably by the collective effect of lyophilization and subsequent powder processing. The tap density of this pow-der was 0.49 g/mL, but this measurement may underestimate the tap particle density measured by this method because the irregularly shaped particles will impede more compact pack-ing compared to spherical particles. From the process viewpoint, freeze-drying normally es-tablishes a porous dry cake in a vial. Compression is needed to densify the cake, which needs to be pulverized by a sec-ondary dispersive force such as mechanical milling or grind-ing to create particles. Furthermore, each step of this process may render a specific stress event to HBsAg. Although in vitro potency analysis (AUSZYME®) did not clearly show potency change probably due to the wide sample-to-sample variation (30%), HBsAg nanoparticle formed 140 ± 20 nm aggregate as indicated by DLS (Table 3), which might have a long-term implication of stability and/or immunogenicity loss. Therefore, despite the benefit as a quick research tool, the potential of this multi-step, laborious process to be a vi-able manufacturing method is not promising.
Agarose Beads
As pre-formed particles, agarose beads featured a powder with pre-determined properties suitable for skin penetration. After being sieved to the size range of 38-53 µm, the dry agarose beads were loaded with an HBsAg solution (Formu-lation B). When fully hydrated with a vaccine formulation, the 4% agarose gel could take up water 24 times the body weight of the dry beads. After lyophilization, the composi-tion of Formulation B was estimated to be 30% agarose, 1% HBsAg, 42% trehalose, and 27% dextran. Based on the po-tency assay, only 60% of HBsAg was found in the loaded agarose beads (Table 3). It is likely that the HBsAg particles were coated mostly on the bead surface along with excipi-ents when lyophilized. Passive diffusion of the HBsAg parti-cles into the pores may be limited by steric hindrance be-cause the pore size of the agarose beads might not be sub-stantially larger than the HBsAg particles. As to particle characteristics, the measured particle size (Table 2) matched with the sieved particle size. The loaded agarose beads were spherical in shape but displayed some levels of particle fusion arising possibly from the loading solution after dehydration (Fig. 1B). Such fusion, if strong, may shift the particle size distribution. One of the major ad-vantages of the agarose beads was the high process yield (~90%), due to the fact that the powder was pre-sieved to the same narrow size range. The moisture content of the agarose beads was relatively high, 5.5%, compared to other lyophi-lized powders probably because of the agarose’s hydrogel nature. For large-scale manufacturing, effective loading is the key to ensure uniform and consistent loading. There are two issues with the agarose formulation; aga-rose is not biodegradable which may present a safety concern for parental applications, and a significant HBsAg potency
loss, ~40%, was detected by AUSZYME® analysis. Al-though the reason for this loss remains to be clarified, it might be due to ineffective loading, inhibited/delayed release of antigen from the hydrated beads, or both. Therefore, the agarose beads method is considered viable only after these two hurdles are overcome.
Spray-Drying
Spray-drying, the most common powder-generating methodology, allowed Formulations C & J to be rapidly transformed into a dry particulate form by atomizing into a hot drying medium. Its utility mainly results from the advan-tage that the final dry powder product can be collected in a matter of a few seconds and because the commercially avail-able bench-top spray dryers have the ability to process small batches of high-valued biopharmaceuticals. Nevertheless, this advantage also limits the particle size to <10 µm [21-24], which is not ideal for EPI. Certainly, larger dryers offer longer drying times for larger droplets to be dehydrated. But this comes at the expense of the small-batch-size capability. To overcome this, a custom-made spray dryer was designed with a long, narrow drying chamber and a straight-down filter collection system. The custom-made spray dryer produced larger particles, 50-µm in medium size (Formulations C & J in Table 2). The size distribution was broader, 28-80 µm, compared to the sieved Formulations A & B. Other powder characteristics included: a free flowing powder due perhaps to the less con-tent of hygroscopic trehalose; and a moisture content of 3.5%, comparable to other freeze-dried powders due to the higher air outlet temperature of 130-140oC. The particles had a tap density of 0.48 mg/mL, suggesting that the particle might be slightly porous although such porosity was not de-tected as indicated by the perfectly smooth, spherical particle morphology (Fig. 1C) a unique property primarily deter-mined by the protein [21]. Aided by the filter unit instead of the conventional cyclone system, powder collection effi-ciency improved to 60-70%. Finally, as illustrated by the size of HBsAg nanoparticles, spray drying appeared to affect HBsAg potency due possibly to the multiple stress events⎯large air-water interfacial area, shear stress upon atomization, and thermal stress. The HBsAg nanoparticles formed aggregates of 150 ± 50 and 170 ± 40 nm for Formulation C and J, respectively (Table 3)-a possible instability sign that warrants attention (to be dis-cussed later). Monitoring HBsAg potency showed that the spray-dried formulation lost approximately 50% (Table 3) of its original potency. Overall, spray-drying appeared to be a viable method but its use for producing HBsAg powder for-mulations needs to be optimized for long-term stability and immunogenicity preservation.
Spray Freeze Drying (SFD)
Of the six SFD formulations prepared by SFD (Formula-tion D-I), no differences in either particle size or moisture content were noted despite different formulation composi-tions (Table 2). Particle size appeared to be dictated by the frequency of the ultrasonic atomizing nozzle: 60-kHz in this case to generate particles mainly in the range of 30-60 µm.
Hepatitis-B Vaccine Powder Formulations Current Drug Delivery, 2007, Vol. 4, No. 1 63
Fig. (1). Scanning electron micrographs of the seven powder formulations (Formulation A-G) listed in Table 1 (The magnification bar for A, D, E-F represent 100 µm while the bar for C is 1 µm. The length of the magnification bar for B is like A but it represents 2 mm).
After investigating a series of lyophilization experiments involving different drying temperatures and durations, we selected an aggressive drying cycle, 20 hours of total drying time (10 hours each for primary and secondary drying), to
lyophilize all powder formulations. At the end of each ex-periment, we determine how the cycle parameters affected particle properties, such as morphology and tap density, and certainly the final moisture content. In the end several cycles
64 Current Drug Delivery, 2007, Vol. 4, No. 1 Maa et al.
could offer acceptable final powder properties but this ag-gressive cycle was unique compared to the common lyophi-lization cycles for freeze drying in the vial. The powder’s final moisture content, prescribed by the temperature and duration of primary and secondary drying in the lyophiliza-tion cycle, hovered at 3-4% based on the short drying time (10 hours). Thus, prolonging the secondary drying time could possibly trim down the moisture content to less than 2%. SFD dictates particle density in a unique way. The result-ing particles (Formulation D) showed abundant internal pores of diverse sizes (Fig. 1D), a manifestation of a typical freeze-dried cake of a high void volume [14-16, 25-29]. In-deed, the tap density was low at 0.3 g/mL. Since high-density particles (tap density ≥0.5 g/mL) are a prerequisite for effective skin penetration, we first chose to increase the solid content of the spraying solution because a greater mass per unit volume would theoretically impart higher density to particles. Indeed, the tap density increased to ~0.6 mg/mL, as reflected by the “dry raisin” or “shrunken” morphology (Figs. 1E-1G), when the solid content of the spraying solu-tion was escalated to 35% (Formulations E-I in Table 2). It remains unclear which factor affects particle shrinkage more⎯the solid content of the spraying solution, drying conditions, or formulation composition. Since increasing the solid content is limited by each excipient’s solubility maxi-mum and by the solution viscosity maximum where atomiza-tion ceases, we attempted to increase particle density by in-ducing particle shrinkage via formulation composition. As previously reported [14, 29], formulations containing treha-lose alone or the binary mixture of trehalose/mannitol still produced lower-density particles even at high levels of solid content (>25%) whereas the mixture of three excipients, tre-halose/mannitol/dextran, resulted in significant particle shrinkage. The working hypothesis was that either the freeze-concentrate structure collapses during ice sublimation, or possibly during secondary drying, when the freeze-concentrate mixture softens and can flow to fill the void left behind by ice crystals. Certainly, it must be related to the maximally frozen freeze-concentrate’s glass transition tem-perature (Tg’), which drops below the drying temperature. Yet the collapse occurs only at the intra-particle level, be-cause the powder would lose particle characteristics if the collapse took place at the inter-particle level. More specifi-cally, the composition of trehalose:mannitol:dextran at the 3:3:4 weight ratio (e.g. Formulations E-G with different mo-lecular weights of dextran) was particularly effective in plas-ticizing the mixture and promoting particle shrinkage, thereby resulting in a higher tap density at 0.63-0.68 g/mL. Nevertheless, the addition of lysine or Poloxamer 188 to this formulation composition (Formulations H & I) did not cause particle shrinkage as much as the dextran-containing formu-lations. From the process viewpoint, SFD afforded outstanding attributes as compared to other particle formulation tech-nologies [7-13]. The process yield was high, 82-85% at small batch sizes. SFD allowed fine control on particle char-acteristics (described above) and the process was highly compatible with a wide range of excipients and biopharma-ceuticals. The process offered flexibility in scalability, espe-
cially in small-scale capabilities for initial feasibility re-search and formulation screening and manufacturability in the pharmaceutical setting with a proven utility in manufac-turing commercial products (e.g. Nutropin Depot™). More importantly, the SFD process exposed HBsAg only to mild stress conditions since HBsAg potency was un-changed. Approximately 90% potency was recovered in the Auzyme assay (Table 3). Likewise, the size of HBsAg nanoparticles was unchanged after SFD, remaining in the range of 30-40 nm for Formulations D-G, which was similar to that of the dialyzed and concentrated HBsAg bulk (Table 3). The significance of HBsAg nanoparticle size preservation lies in a possible correlation with antigen potency and im-munogenicity. Mechanistically, a gain in HBsAg nanoparti-cle size during dehydration could stem from the nanoparti-cles being brought to close proximity, overcoming repulsive forces, and resulting in nanoparticle coagulation. Once co-agulated, the original particle size might not be reproduced upon rehydration. The SFD process may avoid this coagula-tion. As previously shown [13,14], the freezing rate played an essential role in minimizing the large-scale coagulation of nanoparticles during freeze-drying. Faster freezing caused a greater rate of nucleation, a lower rate of ice crystal growth, and the formation of smaller ice crystals. The tiny ice crys-tals divided the HBsAg nanoparticle-containing freeze con-centrate into much smaller compartments as compared to the compartments formed by large ice crystals generated by slower freezing rates. Based on this concept, since SFD could achieve the greatest freeze rate possible by atomizing liquid formulation into liquid nitrogen (-196 °C), it was most effective in preventing large-scale HBsAg nanoparticle co-agulation. Overall, SFD offered both a viable powder manufacturing process and presentation of HBsAg stability. Therefore, a series of SFD powder formulations were prepared to select a preferred formulation for further development.
Formulations Screening
Long-Term Stability Evaluation
This study screened five SFD powders (Formulations E-I) and one spray-dried powder (J) by measuring their long-term stability of the physical properties, the potency, and the immunogenicity. Good powder flowability and low hygro-scopicity were critical to the long-term physical stability of the powder and were important for efficient delivery by EPI as well. Powder samples of these formulations were weighed and sealed in glass vials. The vials were stored at 40 oC, an accelerated condition commonly used for formulation screening. HBsAg potency was monitored by AUSZYME® over a period of 16 weeks (Fig. 2). All SFD formulations appeared to maintain good biochemical stability while the spray-dried formulation suffered a steady potency decrease, almost 90% loss at Week 16. This is much greater than the 30% batch-to-batch variation that is typical for this bioassay. The physical stability of all powders was assessed by optical microscopy, and is exemplified with three formula-
Hepatitis-B Vaccine Powder Formulations Current Drug Delivery, 2007, Vol. 4, No. 1 65
tions (Formulations F, H, and J) in Fig. (3). After a 16-week storage at 40oC, particles in SFD Formulations F and H re-mained discrete and free flowing whereas particles in For-mulation J (spray-dried)) showed a slight level of agglomera-tion, as confirmed by particle size analysis (data not shown). This experiment suggested that all SFD formulations are acceptable and showed no appreciable differences in stability among them. When both stability and particle properties are considered, powders containing dextran (Formulation E-G) seemed to be superior. Dextran containing powders were, therefore, further assessed for their ability to elicit antibody responses in vivo.
Immunogenicity Testing
The immunogenicity of dextran formulations was as-sessed in mice (four per group) before (T=0) and after (T=16 weeks) storage by intramuscular injection of reconstituted powder formulations. Animals on weeks 0 and 16 received the same amount of powders, which contained 0.1 µg of HBsAg at week 0. The total IgG titers elicited by Formula-tions E-G four weeks post boost vaccination (i.e. 8 weeks post prime immunization) are summarized in Fig. (4). In terms of the mean titers, all un-stored (T=0) powder formula-tions showed equivalent immunogenicity compared to the untreated HBsAg bulk (p>0.05, t test). More importantly, the same powders after 16-week storage at 40°C also elicited equivalent levels of immune responses (p>0.05). In this study we wanted to determine if the SFD process generated antigen particles that would penetrate the kerati-nized superficial barrier of the epidermis and enter the im-mune competent, deeper, viable epidermis. In this experi-ment, hairless guinea pigs were employed because they have a thicker skin than mice and therefore posed a more stringent test of the immunogenicity of SFD antigen particles. EPI is achieved by using a PowderJect ND device to accelerate
particles into the epidermis of target skin as described previ-ously (6-12). EPI of hairless guinea pigs with 5 µg of HBsAg prepared by the SFD process (Formulation E) was compared to intramuscular needle injection of the same dose of HBsAg in liquid form (Fig. 5). At four weeks post-immunization, the antibody titers to HBsAg in animals immunized by EPI were statistically higher than animals immunized by IM needle injection of HBsAg (p<0.05). The differences between the two routes disappeared after a booster immunization (P>0.05). EPI with HBsAg without an adjuvant was also compared to IM injection of commercial vaccine (HBsAg adsorbed to aluminum hydroxide). EPI with HBsAg alone elicited similar antibody titers to IM injection with HBsAg adsorbed to aluminum hydroxide on both weeks four and six (p>0.05), demonstrating both the utility of SFD antigen par-ticles in EPI and the superiority of the epidermal route over the standard IM route of immunization.
CONCLUSIONS
We have successfully investigated the utility of four powder-formation processes to produce HBsAg powder for-mulations for use in EPI. These processes were assessed in multiple ways with a focus on particle maneuverability, process efficiency, scalability, and HBsAg stability. Not only did SFD demonstrated good process performance but also it resulted in powder formulations meeting the criteria for EPI, including particle characteristics for penetration, short-term and long-term potency and immunogenicity preservation of the powder formulation. Among several SFD formulation parameters we investigated, the most critical factors for pro-ducing powders of the desired particle properties, particu-larly high-density, were total solid content in the spraying solution and formulation composition. High-temperature formulation screening determined that the treha-lose:mannitol:dextran (3:3:4) formulations, regardless of
Fig. (2). Formulation screening for biochemical stability of six powder formulations (Formulations E-J in Table 1). Powder samples were weighed into and sealed in glass vials. Vials were incubated at 40oC up to 16 weeks. HBsAg potency at each time-point was analyzed by AUSZYME® assay.
66 Current Drug Delivery, 2007, Vol. 4, No. 1 Maa et al.
Fig. (3). Formulation screening for physical stability of three powder formulations (Formulations F, H, and J in Table 1). Powder samples were weighed into and sealed in glass vials. Vials were incubated at 40 oC up to 16 weeks. Particle integrity was observed under optical mi-croscopy. Pictures of the powder samples at T=0 and T=16 weeks are shown on the left and right, respectively. Formulation designation is labeled directly on the top left corner of the T=0 pictures.
Fig. (4). The immunogenicity of three powder formulations (E-G in Table 1) before (T=0) and after 16-week storage (T=16) at 40°C (Fig. 4) and a HBsAg bulk. Powder samples were reconstituted in water and injected into mice. Two vaccinations were given with the boost immuni-zation at 4 weeks after the prime vaccination. Total IgG titers (4 weeks after the boost shot) were determined by ELISA and presented in the titers in log-scale.
Hepatitis-B Vaccine Powder Formulations Current Drug Delivery, 2007, Vol. 4, No. 1 67
Fig. (5). Immune responses to HBsAg in guinea pigs. Animals (five per group) were immunized with 5 µg of HBsAg by epidermal powder immunization. The dried material was prepared as an SFD T:M:D 37K formulation (Table 1; Formulation E). The animals were immunized on days 0 and 28 and serum samples were collected on days 28 and 42. The antibody response was compared to the response induced by IM delivery (same dose and schedule) of HBsAg alone or HBsAg adsorbed to aluminum hydroxide (commercial vaccine). The “*” indicate a statistical significant difference when compared to EPI group (p<0.05, t test).
dextran’s molecular weight, exhibited excellent physical and biochemical stability. More importantly, the dextran formu-lations maintained the immunogenicity, as measured by IM injection of the reconstituted powder solution and EPI of the powder formulation that could penetrate the epidermis and elicit an immune response.
ACKNOWLEDGEMENTS
The authors thank Drs. Steve Prestrelski and Hansi Dean for project support.
