European Journal of Pharmaceutics and Biopharmaceutics xxx (2013) xxx–xxx

Contents lists available at ScienceDirect

European Journal of Pharmaceutics and Biopharmaceutics

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

Research paper

Spray granulation: Importance of process parameters on in vitroand in vivo behavior of dried nanosuspensions

0939-6411/$ - see front matter � 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.ejpb.2013.07.015

Abbreviations: HPC-EXF, hydroxypropylcellulose-EXF grade; SLS, sodium laurylsulfate; DOE, design of experiment; ANOVA, analysis of variance; LOD, loss ondrying; XRD, X-ray diffraction; FaSSGF, fasted state simulated gastric fluid; CMC,critical micelle concentration; FRI, flow rate index; ffc, Jenike flow function; PDI,polydispersity index.⇑ Corresponding author. Pharmaceutical and Analytical Development,

Novartis Pharmaceutical Corporation, 1 Health Plaza, Bldg. 401, Rm. A210A, EastHanover, NJ 07936, USA. Tel.: +1 862 778 5671; fax: +1 973 781 2014.

E-mail addresses: sonali.bose@novartis.com, sonalibose@hotmail.com (S. Bose).

Please cite this article in press as: C.E. Figueroa, S. Bose, Spray granulation: Importance of process parameters on in vitro and in vivo behavior of driedsuspensions, Eur. J. Pharm. Biopharm. (2013), http://dx.doi.org/10.1016/j.ejpb.2013.07.015

Carlos E. Figueroa a, Sonali Bose b,⇑a Department of Chemical & Biological Engineering, Princeton University, Princeton, NJ, USAb Pharmaceutical and Analytical Development, Novartis Pharmaceuticals Corporation, East Hanover, NJ, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 5 December 2012Accepted in revised form 24 July 2013Available online xxxx

Keywords:Wet media millingFluid bed granulationSpray rateAtomizing pressureSpray modeNaproxen

The use of fluid bed granulation for drying of pharmaceutical nanoparticulates on micron-sized granulesubstrates is a relatively new technique, with limited understanding in the current literature of theeffects of process parameters on the physical properties of the dried nanoparticle powders. This workevaluated the effects of spray mode, spray rate and atomizing pressure for spray granulation of drugnanosuspensions through a systematic study. Naproxen and a proprietary Novartis compound were con-verted into nanosuspensions through wet media milling and dried onto a mannitol based substrate usingspray granulation. For naproxen, various physical properties of the granules, as well as the in vitro re-dis-persion and dissolution characteristics of the nano-crystals, were measured. It was found that the spraymode had the most drastic effect, where top spray yielded smaller re-dispersed particle sizes and fasterrelease rates of drug from granules than bottom spray. This was attributed to the co-current spraying inbottom spray resulting in denser, homogenous films on the substrate. Similar in vitro results wereobtained for the proprietary molecule, Compound A. In vivo studies in beagle dogs with Compound Ashowed no significant difference between the liquid and the dried forms of the nanosuspension in termsof overall AUC, differences were observed in the tmax which correlated with the rank ordering observedfrom the in vitro dissolution profiles. These findings make spray granulation amenable to the productionof powders with desired processing and handling properties, without compromising the overall exposureof the compound under investigation.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction reduction based on the Noyes–Whitney equation [4]. This ap-

A significant number of molecules currently under development[1] can be classified as BCS (Biopharmaceutical Classification Sys-tem) Class II compounds [2], with dissolution rate limited solubil-ity resulting in sub-optimal bioavailability. Recently, a revisedclassification system called the Developability Classification Sys-tem (DCS) has been proposed, which further divides BCS Class IIinto two sub classes, IIa for dissolution rate limited and IIb for sol-ubility rate limited compounds [3]. For compounds where dissolu-tion is the rate limiting factor, production of nanoparticles andmicroparticles using wet media milling can improve the dissolu-tion rate by increasing the surface area through particle size

proach to delivering poorly water soluble compounds has beenwell proven, as several products manufactured using this technol-ogy are currently approved and available in the market [5].

Colloidal drug particles produced by wet media milling are typ-ically stabilized against particle agglomeration using either steric(by means of polymers or nonionic surfactants) or electrostatic(by means of ionic surfactants) stabilization mechanisms, or some-times a combination of both mechanisms which is referred to aselectrosteric stabilization [6,7]. Although there is one instance ofa nanosuspension product being marketed in the liquid form (Meg-ace ES), conversion of a nanosuspension into a dried powder formthat can be further filled into capsules or compressed into tablets isoften desirable to ensure the maximum patient compliance. Thedrying step can be further necessitated by the fact that there couldbe possible stability issues, both physical (such as Ostwald ripen-ing and agglomeration) and chemical (such as hydrolysis) associ-ated with nanoparticles in their suspended form [8,9].Additionally, for nanoparticles, there could also be a risk of crystal-lization upon storage if partial surface amorphization occurs dur-ing the wet milling process as has been reported by Sharma et al.[10]. However, it should be pointed out that it is also possible to

nano-

2 C.E. Figueroa, S. Bose / European Journal of Pharmaceutics and Biopharmaceutics xxx (2013) xxx–xxx

generate an amorphous fraction in a dried nanosuspension, if thedrug has some solid solubility in the stabilizer [11].

Freeze-drying and spray drying are the most common processesused to convert a nanosuspension into a solid form [12–16]. Boththese processing methods result in powders that require furtherprocessing to improve the bulk density and flow properties priorto conversion into a tablet or a capsule dosage form. Granulationbased approaches, which typically result in densification with im-proved flow properties of the powder, have also been used but onlyto a limited extent, with most studies focusing only on formulationparameters. For example, Basa et al. reported the layering of aketoconazole nanosuspension formulation onto a lactose monohy-drate substrate, although no evaluation of different formulation orprocess parameters was conducted [17]. Kayaert and co-workersstudied the layering of naproxen and cinnarizine nanosuspensionsonto sugar beads and focused on the effect of different formulationparameters of the nanosuspension, such as stabilizer type, concen-tration and free stabilizer, on coating efficiency and dissolutionprofiles [18]. Recently, we have shown the feasibility of a spraygranulation based process to convert a nanosuspension of a pro-prietary Novartis compound into a solid dosage form, where for-mulation parameters such as the nanosuspension stabilizer, thesubstrate onto which the nanosuspension was coated and the drugloading were studied [19]. However, none of these studies have fo-cused on the processing parameters that would be critical in dryinga nanosuspension using a spray granulation based process. Thislack of information on the critical process parameters of the spraygranulation process for drying of nanosuspensions is a significantgap that needs to be investigated.

Naproxen has been widely used as a model compound to studywet media milling [8,18,20,21] and was selected as the model com-pound in this study to evaluate the processing parameters criticalto drying of nanosuspensions using spray granulation. Naproxen ispractically insoluble in water and has a melting point of 154 �C anda molar mass of 230.26 g mol�1 [22]. The critical parameters iden-tified from the trials with naproxen were then confirmed with aNovartis developmental compound (further referred to as Com-pound A). Compound A is a poorly water soluble free base mole-cule with an equilibrium water solubility of 0.003 mg/mL at25 �C, melting point of 263 �C, molecular weight of 388.4 g mol�1

and mean particle size of 15–20 lm [19].A nanosuspension formulation of naproxen with demonstrated

physical, chemical and dissolution stability for up to 9 months at2–8 �C was used as the starting material for the process optimiza-tion trials. Process parameters such as nozzle atomization pressureand liquid spray rate, typically considered critical in a fluid bedspray granulation process [23], were investigated in this study.Additionally, two different spray modes were also evaluated. Pow-der properties, re-dispersibility, solid-state properties and the rateof dissolution were measured for the naproxen granulations ob-tained from the process optimization trials. The optimized processparameters from these trials were then used to manufacture driednanosuspensions of Compound A. Two dried nanosuspension for-mulations of Compound A, with different in vitro release profiles,were compared in an in vivo study in fasted male beagle dogsagainst the coarse suspension and the parent nanosuspension for-mulations, to assess the implications of differences observed inin vitro dissolution profiles to in vivo exposure.

2. Materials and methods

2.1. Materials

Naproxen USP was purchased from Kalchem International(Lindsay, OK) and Compound A was provided by Novartis

Please cite this article in press as: C.E. Figueroa, S. Bose, Spray granulation: Imposuspensions, Eur. J. Pharm. Biopharm. (2013), http://dx.doi.org/10.1016/j.ejpb.2

Pharmaceuticals Corporation. The sources of other materials usedin the study were as follows: hydroxypropylcellulose-EXF grade(HPC-EXF) (Klucel) from Ashland Inc. (Covington, KY, USA), sodiumlauryl sulfate (SLS) from TensaChem S.A. (Ougree, Belgium), man-nitol (DC grade) from Roquette (Lestrem, France) and hard gelatincapsules (size 0) from Capsugel Inc. (Greenwood, SC, USA) respec-tively. Deionized water (18 MX) produced with a Millipore watersystem (Billerica, MA, USA) was used as the dispersion medium.Yttrium stabilized zirconium oxide beads with 0.2 mm diameter(Tosoh Corp., Tokyo, Japan) were used for the milling process. Allother solvents and reagents used were of HPLC or analytical grade.

2.2. Preparation of nanosuspension

Formulations were composed of either 10% w/v of naproxen orCompound A, 2.5% w/v of HPC-EXF and 0.5% w/v of SLS. Suspen-sions were prepared by dissolving the excipients in an aqueoussolution, into which the active was dispersed. The resulting sus-pension was wet milled with 0.2 mm grinding media using a Net-zsch Labstar Mill (Exton, PA, USA) with a Zeta agitator in therecirculation mode. The following process parameters were used:pump speed of 250 rpm, agitator speed of 2000 rpm and millingtime of 4–6 h. Particle size testing of in-process samples was car-ried out at regular time intervals throughout the duration of themilling process. The pH of naproxen and Compound A nanosuspen-sions after wet media milling was measured to be around 5.1 and5.8, respectively.

2.3. Particle size measurement of nanosuspension

The particle size of the nanosuspension formulations was char-acterized using a Delsa Nano™ C particle size analyzer (BeckmanCoulter, Brea, CA, USA). As per manufacturer claims, this instru-ment is capable of measuring particles in the size range of0.6 nm to 7 lm through the use of dynamic light scattering; allsamples measured were in the specified size range. A small amount(about 5 lL) of nanosuspension sample was added to a disposablecuvette and diluted with 5 mL of deionized water. The cuvette wasmanually shaken for about 10 s and placed inside the sampleholder of the particle size analyzer. Once the intensity was withinthe permissible range, analysis was performed to obtain the parti-cle size and the polydispersity index. All particle size measure-ments were performed in triplicate at 25 �C. All reported particlesize data refer to intensity weighted size distributions.

2.4. Fluid bed granulation of nanosuspension

Fluid bed granulation of the nanosuspension was performedusing a Huttlin Mycrolab (Huttlin GmbH, Schopfheim, Germany)fluid bed drier. The equipment can be operated such that the liquidfeed can be sprayed from above into the fluidized bed (top spray)or from below up into the bed (bottom spray). The gas used wasnon-processed air heated to an inlet temperature between 70and 80 �C and fed at a flow rate ranging from 8 to 15 m3/h (ad-justed for each granulation trial to maintain adequate fluidization).The inlet air temperature used in this study is within the range ci-ted in the literature [24]. The mannitol substrate (82.5 g) wasloaded onto the fluid bed prior to spraying and allowed to equili-brate to feed gas temperature. Sufficient fluidization of the sub-strate was maintained to ensure that there was no agglomerationof mannitol particles during the granulation process. The nanosus-pension was sprayed onto the mannitol DC substrate to achieve atheoretical drug loading of 17% in the granulation using the processparameters (spray mode, liquid spray rate, and nozzle atomizationpressure) listed in Table 1. The product temperature was main-tained within the range of 37–42 �C throughout the process. At

rtance of process parameters on in vitro and in vivo behavior of dried nano-013.07.015

Table 1Summary of process results from conducted trials.

Trial Spraymode

Liquid sprayrate (g/min)

Nozzle atomizingpressure (bar)

Processyield (%)

LOD(%)

BET surfacearea (m2/g)

Porosity(%)

Median Porediameter (nm)

Bulk density at0.50 psia (g/mL)

Granule particlediameter (lm)

D10 D50 D90

I Top 1.8 1.0 78.7 0.80 2.55 ± 0.02 54.5 9.1 0.58 30 109 282II Top 1.8 1.5 85.9 0.80 2.27 ± 0.03 53.1 50.3 0.58 32 111 303III Top 2.5 1.0 79.7 0.80 2.38 ± 0.02 60.2 8.9 0.57 46 157 357IV Top 2.5 1.5 83.4 1.00 2.10 ± 0.01 51.8 8.4 0.56 31 104 259V Bottom 1.8 1.0 75.5 1.00 1.95 ± 0.01 48.8 9.1 0.68 27 100 278VI Bottom 1.8 1.5 71.6 0.80 1.90 ± 0.02 47.4 8.2 0.72 25 100 277

C.E. Figueroa, S. Bose / European Journal of Pharmaceutics and Biopharmaceutics xxx (2013) xxx–xxx 3

the end of spraying, the powder was allowed to cool down to roomtemperature, screened through a 18 mesh screen and stored be-tween 2 and 8 �C till further analysis.

2.5. Spray granulation experimental set-up

In order to investigate the effect of process parameters on nano-crystal and granule properties, two two-level factors were evalu-ated using the top spray mode: atomizing air pressure at the spraynozzle (1.0 and 1.5 bars) and spray rate of the liquid nanosuspen-sion (1.8 and 2.5 g/min). Based on these trials, the optimum pro-cess parameters were then selected to compare top spray againstbottom spray. The combination of parameters used is listed in Ta-ble 1. Each trial was characterized for process yield, moisture con-tent and granule properties such as surface area, porosity, porediameter, bulk density and granule size. The powders were alsocharacterized with regard to crystallinity and flow properties.Although characterization of the granular powders included manytypes of measurements, the primary response properties examinedwere particle size after re-dispersion and dissolution profile.

2.6. Powder characterization

The particle size distribution of the dried nanosuspension pow-der was measured using a Malvern Mastersizer� 2000 (MalvernInstruments Ltd., Worcestershire, UK) equipped with a dry powderScirocco disperser. The manufacturer claims the instrument beingable to measure particles in the size range of 20 nm to 2 mmthrough the use of laser diffraction; all samples measured werein the specified size range. An air pressure of 1 bar and a feed rateof 30% were used for the measurements. A screen containing a fewstainless steel balls was attached to the end of the feed tray to facil-itate the dispersion of the more cohesive fines without causingdetectable particle attrition during the measurements.

The moisture content (expressed as loss on drying (LOD)) ofgranulated powder was determined using a HR73 Halogen Mois-ture Analyzer (Mettler-Toledo Inc., Columbus, OH, USA). About1 g of granulated powder sample was spread uniformly in the sam-ple holder and heated to 105 �C, and the loss of water upon dryingmonitored till a stable value was reached.

The flow of the granulated powder was measured using a FT4powder rheometer (Freeman Technology Limited, Gloucestershire,UK) using a previously cited method [25]. The following programsand tests available from the instrument software were used. Thestability and variable flow rate program (REP + VFR) was used tomeasure the flow rate index (FRI) which was calculated as follows:FRI = (energy consumed at 10 mm/s blade tip speed)/(energy con-sumed at 100 mm/s blade tip speed). The shear cell test was usedto determine the Jenike flow function [26]. Finally, the compress-ibility and permeability of the powders were also measured. Alltests are detailed in the reference mentioned above.

Please cite this article in press as: C.E. Figueroa, S. Bose, Spray granulation: Imposuspensions, Eur. J. Pharm. Biopharm. (2013), http://dx.doi.org/10.1016/j.ejpb.2

The porous structures of the dried granulations were character-ized using an AutoPore IV 9500 Series Mercury Porosimeter(Micromeritics Instrument Corporation, Norcross, GA, USA). Thisinstrument uses the mercury porosimetry analysis technique thatis based on the progressive intrusion of mercury into a porousstructure under stringently controlled pressures. Measurementswere performed by placing the sample in the sample cup and thenfilling mercury into the cup and capillary stem at a filling pressureof 0.50 psia and an equilibration time of 10 s. With an increase inthe pressure on the filled penetrometer, mercury intruded intothe pores of the test sample, beginning with those pores of largestdiameter. This movement of mercury from the capillary stem re-sulted in a change in capacitance between the mercury column in-side the stem and the metal cladding on the outer surface of thestem, which was measured by the instrument. The specific surfacearea of the samples was measured by the Micromeritics ASAP�

2420 Accelerated Surface Area and Porosimetry System (Microm-eritics Instrument Corporation, Norcross, GA, USA) using the gassorption technique.

2.7. Powder X-ray diffraction (XRD)

X-ray powder diffraction testing was performed on a Bruker D8Advance (Bruker-AXS, Karlsruhe, Germany) controlled by Diffracplus XRD commander software. Samples were prepared by spread-ing powder samples on PMMA specimen holder rings from Brukerand scanned from 2� to 40� 2h at the rate of 2�/min with 0.02� stepsize and 0.6 s/step at 40 KV and 40 mA. The divergence and anti-scattering slits were set to 1�, and the stage rotated at 30 rpm. Dataanalysis was performed using ‘‘EVA Part 11’’ version 14.0.0.0.

2.8. Scanning electron microscopy (SEM)

The samples were analyzed by using JEOL JSM6301 FXV scan-ning electron microscope (JEOL, Peabody, MA, USA). One drop ofnanosuspension without any further dilution was air dried on thealuminum stubs and then sputter coated with gold/palladiumusing a Denton Bench Desk IV sputter unit, before imaging. Repre-sentative powder samples of the spray granulated powders weremounted onto specimen stubs and then sputter coated withgold/palladium before imaging. Four to five images from differentlocations (bitplanes) were used for particle size distribution. Thenumber of particles in each images varied from 250 to 400. Thesizes were reported in terms of length.

2.9. Determination of drug content in granulated powder

The content of naproxen in the nanosuspension formulationswas determined by using a validated HPLC method (Waters Alli-ance HPLC system, Milford, MA, USA). The granulation was dis-solved in 200 mL of a 90:10 mixture of acetonitrile and water(v/v), shaken on a mechanical shaker for 30 min and sonicated

rtance of process parameters on in vitro and in vivo behavior of dried nano-013.07.015

4 C.E. Figueroa, S. Bose / European Journal of Pharmaceutics and Biopharmaceutics xxx (2013) xxx–xxx

for 5 min. 5 mL of this solution was then pipetted into a 200 mLvolumetric flask, diluted to volume with the mobile phase (aceto-nitrile:water:glacial acetic acid, 50:49:1, v/v/v), and mixed thor-oughly. A portion of this solution was then filtered through a0.45 lm nylon filter and injected into the HPLC. Chromatographicseparation was achieved using a Phenomenex-Inertsil ODS-2,150 mm � 4.6 mm, 5 lm column. An isocratic method with thefollowing parameters was used: Injection volume of 20 lL, flowrate of 1.2 mL/min, ambient column temperature, and a detectionwavelength of 254 nm.

The content of Compound A in the nanosuspension formula-tions was determined by using a validated HPLC method (WatersAlliance HPLC system, Milford, MA, USA). Briefly, an aliquot(5 mL) of the nanosuspension or an equivalent amount of driedpowder was dissolved in 250 mL of methanol and sonicated for30 min. Approximately 13 mL of the sonicated liquid sample wasthen centrifuged at 3000 rpm for 10 min using an accuSpin™3(Fisher Scientific, Pittsburgh, PA, USA). At the end of centrifugation,6 mL of the supernatant was collected, diluted with 200 mL ofmethanol, and then analyzed by HPLC. Chromatographic separa-tion was achieved using a Zorbax SB-C18, 150 mm � 3 mm,3.5 lm column (Agilent Technologies, Foster City, CA, USA). Themobile phase consisted of a mixture of water, acetonitrile and tri-fluoro acetic acid (TFA). An isocratic method with the followingparameters was used: Injection volume of 10 lL, flow rate of0.8 mL/min, column temperature of 30 �C, and detection wave-length of 232 nm.

2.10. Re-dispersion of dried granulation

For the dissolution testing of oral dosage forms, it is common touse a fasted state simulated gastric fluid (FaSSGF), which emulatesthe pH and surface tension of the media found in the stomach.Since it is not fully clear what surfactant(s) are responsible forthe low tension in gastric fluid, sodium lauryl sulfate is generallyused in simulated media to lower the surface tension [27]. A com-monly used medium recipe is that of Dressman et al, which uses8.67 mM SLS in 0.01–0.05 N HCl with 0.2% NaCl [28]. Such a FaSSGFmedium results in a large amount of the SLS being present as mi-celles, due to the concentration of SLS being much higher than itscritical micelle concentration (CMC) as shown by Aburub et al. TheCMC for SLS in 0.01 N HCl/0.2% NaCl is 1.75 mM, which is the con-centration of SLS recommended in the literature to be used in FaS-SGF as this yields a surface tension similar to what is observedin vivo [29].

In order to evaluate the re-dispersed particle size of the naprox-en nano-crystals, 0.01 N HCl with 0.05% SLS (1.75 mM) was used asthe re-dispersion medium in our study. 100 mg of the granulatedpowder was dispersed in 10 mL of the medium and placed on a stirplate for 30 min at 500 rpm. Preliminary experiments were con-ducted at different stirring rates to determine the effect of stirringspeed on the re-dispersed particle size. Stir rates lower than500 rpm were not found to have a significant effect on the re-dis-persed particle size (data not shown). Sonication has been reportedto have the highest impact in breaking the nanoparticle aggre-gates; however, it is not considered to be a very relevant methodin simulating the hydrodynamic conditions in the GI tract [30].Hence, agitation was used to re-disperse the dried nanoparticlesin our study, and the particle size of the re-dispersed samples mea-sured as described previously in the subsection titled ‘‘Particle SizeMeasurement of Nanosuspension.’’

2.11. In vitro dissolution testing

Various analytical techniques can be used to measure the disso-lution rate of nanosized particles [31], out of which the use of

Please cite this article in press as: C.E. Figueroa, S. Bose, Spray granulation: Imposuspensions, Eur. J. Pharm. Biopharm. (2013), http://dx.doi.org/10.1016/j.ejpb.2

syringe filters is one of the simplest and efficient techniques. Thereare references in the literature regarding the use of 0.1 lm or lesserpore size filters to achieve adequate separation of nanosized frommolecularly dissolved drug [32]. In this study, the sole objective ofdissolution testing was to rank order the different batches and todetermine the most optimum processing parameters for dryingof a nanosuspension. The syringe filters used in our study are inagreement with the following literature references [33,34].

Dissolution of the nanosuspensions and fluid bed granulatednanosuspensions (filled into capsules) was carried out in 500 mLof pH 2 (0.01 N HCl) media containing 0.1% sodium lauryl sulfateusing USP Apparatus 1 at 100 rpm and 37 �C (Distek Evolution6100, Distek Inc., North Brunswick, NJ, USA). A 100 mg equivalentdose of the active was used. All dissolution experiments were per-formed in triplicate. The solubility of naproxen and Compound A inthe dissolution medium was determined to be 39.0 lg/mL and95.5 lg/mL, respectively. For dissolution testing of naproxen, SLSin the range of 0.25–2% has been used in the dissolution media[8,35,36]. High amounts of the wetting agent are typically addedto the dissolution medium to create sink conditions for the active,though this leads to high dissolution rates due to solubilization ofthe active by SLS. However, we used a dissolution medium of0.01 N HCl with 0.1% SLS which provided non-sink conditions,intentionally designed for rank ordering of formulations and opti-mum variant selection for further in vivo studies. Samples werewithdrawn at pre-determined time intervals of 10, 20, 30, 45,and 60 min. Between the 45 min and 60 min sample time point, ra-pid stirring at 200 rpm was performed. For naproxen, sampleswere filtered through PVDF filters of 0.2 lm mean pore diameter(Whatman Inc., Clifton, NJ, USA), diluted with a 1:1 mixture of ace-tonitrile:water and then analyzed using UV spectrophotometry at awavelength of 272 nm [37,38] since no interference was detectedfrom the components of the formulation at that particular wave-length. For Compound A, samples were filtered through PTFE filtersof 0.2 lm mean pore diameter (Whatman Inc., Clifton, NJ, USA), di-luted with a 1:1 mixture of methanol:water, and then analyzed byHPLC using the validated method described earlier for drug contentdetermination.

2.12. Pharmacokinetic study with Compound A

The study protocol was reviewed and approved by the IACUC(Institutional Animal Care and Use Committee). Formulations ofCompound A were orally administered to fasted male beagle dogs(n = 5, body weight between 10 and 11 kg) in a crossover study at adose of 20 mg/kg. No pentagastrin was administered to the dogsprior to dosing. A washout period of 1 week was kept between dos-ing of each formulation. Approximately 2 mL of blood sample wascollected from each animal at 0 (pre-dose), 0.25, 0.5, 1, 1.5, 2, 3, 4,6, 8, 12, 24, and 48 h post-dose. Plasma was obtained by centrifu-gation of the blood samples and stored at �70 �C prior to analysis.The parent compound in the plasma samples was analyzed using aliquid–liquid extraction method followed by evaporation of thesupernatant to dryness and analysis of the reconstituted sampleby liquid chromatography–tandem mass spectrometry (LC–MS/MS) in selected reaction monitoring mode using electrospray neg-ative ionization (ESI) as an interface. Chromatographic separationwas achieved using an ACE C8, 50 mm � 2.1 mm, 3 lm column(Advanced Chromatography Technologies). The mobile phase con-sisted of a mixture of methanol and water with 0.1% formic acid(65:35, v/v). An isocratic method with the following parameterswas used: Injection volume of 10 or 20 lL, flow rate of 0.2 mL/min, column temperature of 45 �C, and detection in the MS/MSESI negative ion mode. The lower limit of quantification of the ana-lytical method was 10.0 ng/mL. Pharmacokinetic analysis was per-formed using WinNonlin� Professional Version 5.2 (Pharsight,

rtance of process parameters on in vitro and in vivo behavior of dried nano-013.07.015

C.E. Figueroa, S. Bose / European Journal of Pharmaceutics and Biopharmaceutics xxx (2013) xxx–xxx 5

Cary, NC, USA), and statistical analysis was performed using Stat-graphics Plus 5.1 (Statpoint Technologies Inc., Warrenton, VA,USA).

3. Results and discussion

3.1. Naproxen nanosuspension characterization and stability

The milling process yielded naproxen nano-crystals stabilizedby HPC-EXF and SLS at a naproxen concentration of 10% wt/vol.The nanosuspension initially exhibited a mean particle size of211 nm with a polydispersity index of 0.178. The particle size ofthe nanoparticles measured using the Delsa Nano™ C particle sizeanalyzer was in agreement with the particle size obtained fromSEM images (data not shown). Assay testing confirmed the theoret-ical naproxen content in the nanosuspension to be around 98%.After 9 months of storage at 2–8 �C, the nanosuspension demon-strated very little change in particle size. The mean particle sizewas 256 nm with a polydispersity index of 0.183. Furthermore,analytical assay testing yielded 97% of the theoretical naproxencontent after 9 months. The excellent physical and chemical stabil-ity allowed for the use of this nanosuspension formulation over thecourse of 3 months for the spray granulation trials and character-ization studies described in this manuscript.

3.2. Top spray granulation

Top spray granulation has been one of the most prevalent tech-niques of fluidized bed granulation since the 1960s [39]. Tangentialand bottom (or Wurster) spray modes have been evaluated only toa limited extent for granulation [40,41]. Thus, for our study, wechose to first evaluate the effects of both the liquid spray rateand the nozzle atomizing pressure using the top spray mode.

3.2.1. Effect of liquid spray rateIt is known that increasing the spray rate of a liquid binder in a

fluidized bed will generally result in the faster growth of particlesand faster agglomeration kinetics [42–44]. The spray rate will alsoaffect the moisture content on the surface, which can impact theuniformity of the layer that forms, and thus, the dissolution ratesonce rehydrated [45]. Tzika et al. observed a decrease in the disso-lution rates of polymer latex-coated fertilizer granules, as the latexspray rate was increased from 0.7 to 2.6 mL/min, due to the forma-tion of denser and more homogeneous granules of fertilizer parti-cles which were confirmed by SEM. In the same study, when thespray rate was increased from 2.6 mL/min to 7 mL/min, the disso-lution rate increased due to the increased amount of solvent vol-ume present on the particle surfaces creating porous,heterogeneous coatings [46]. Therefore, depending on the rangeof spray rates that are used in the process, very different behaviorscan be observed.

Our work (presented in Fig. 1a) showed that for both atomizingpressures evaluated (1.0 and 1.5 bars), the mean re-dispersed par-ticle size increased as the spray rate was increased from 1.8 g/minto 2.5 g/min, with the differences being statistically significant(p < 0.05). This increase in the re-dispersed particle size with an in-crease in the spray rate might be explained by the higher surfacemoisture content on the mannitol particles as the spray rate wasincreased. While in all cases, the same volume of nanosuspensionwas sprayed, in the transient state, a higher spray rate is equivalentto a larger number of droplets (or equivalently larger mass of nano-crystals) being deposited on the fluid bed per unit time. It is rea-sonable that the drying behavior of the nanoparticles will be differ-ent when the kinetics of droplet deposition is different, since

Please cite this article in press as: C.E. Figueroa, S. Bose, Spray granulation: Imposuspensions, Eur. J. Pharm. Biopharm. (2013), http://dx.doi.org/10.1016/j.ejpb.2

different moisture levels will affect the uniformity and structureof the film.

However, in all four trials, the increase in mean particle sizecompared to the initial size was only 10–30% larger, indicatinggood re-dispersibility. Interestingly, the PDI did not change whenconsidering trials run at the same atomizing pressure, as the gran-ulations at 1.0 bar both resulted in PDIs of �0.30 and at 1.5 bars thePDIs were �0.22, corresponding to roughly 60% and 20% increase,respectively, from the initial nanosuspension. In these experi-ments, using a slower spray rate of 1.8 g/min resulted in overallbetter re-dispersibility.

3.2.2. Effect of nozzle atomizing pressureThe nozzle atomizing pressure is a measure of the amount of air

used in the atomization nozzle head when the feed liquid and airconverge to form droplets that will be sprayed; the more air usedin comparison with liquid, the finer the droplets. Gao and co-work-ers [47] have shown that for the fluid bed granulation of micron-ized drug substance with a cellulose binder, an increase inatomizing pressure led to an increase in granule fines. This wasattributed to the finer droplets forming weaker liquid bridges withgranules that allowed for attrition to be more dominant thanagglomeration. In the aforementioned study, the spray rate andthe atomizing pressure were shown to be the process parametersto have the greatest effect on the granule properties of the param-eters evaluated. In another study, the degree of atomization wasreported to affect the structure and mechanical strength of theformed granules [48]. Based on these references, the atomizationpressure was selected as one of the process parameters for evalu-ation in our trials.

The re-dispersion data (Fig. 1b) showed a significant decrease inre-dispersed mean particle size and PDI as the atomizing pressurewas increased at a constant spray rate (p < 0.05). At the spray rateof 1.8 g/min, there was a 7% decrease in mean particle size, whilefor a spray rate of 2.5 g/min, the decrease was 13%; for both sprayrates, the PDI dropped by about 25%. Overall, a higher atomizingpressure of 1.5 bars resulted in better re-dispersibility of the driednanosuspension. Comparison of dissolution profiles of trials runusing top spray at a spray rate of 1.8 g/min showed that at anatomizing pressure of 1.5 bars, the granulation yielded a slowerbut more complete dissolution compared to an atomization pres-sure of 1.0 bar (Fig. 2b).

The atomizing air pressure will affect the droplet size which isan important parameter to consider in how the droplet will wetand spread on the granulation substrate, which could in turn affectfilm porosity. Since a higher atomizing pressure will result in smal-ler droplets, this will result in more droplet surface area that willbe exposed to the mannitol, allowing for a more uniform spreadingof the liquid on the particle surface. The more homogeneous filmwill provide more surface area over which the nano-crystals candistribute, thus reducing the chances of particle aggregation. Whenthe droplets are larger, the total droplet surface area that can wetthe mannitol is less, leading to spotty films with less area for thenano-crystals to disperse through. This could lead to nanoparticleaggregation, since there is less distribution of the particles. Fur-thermore, the non-uniform film on the granule surface will likelyhave exposed mannitol surfaces that can allow for faster disinte-gration of the granule and more exposed nano-crystal surfaceswhich could possibly explain the faster dissolution rates obtainedat 1.0 bar. Hence, the differences observed in the re-dispersionand dissolution profiles comparing the different atomization pres-sures using the top spray mode could be related to the surfacestructure of the film which is consistent with reports of drug re-lease from material coated using the top spray method [39]. Whilethe porosity and pore diameter values among the batches pro-cessed at different spray rates and atomizing air pressures were

rtance of process parameters on in vitro and in vivo behavior of dried nano-013.07.015

(b)(a)

Fig. 1. Re-dispersability of spray granulated naproxen nanosuspension produced using top spray at spray rates of 1.8 and 2.5 g/min and atomizing pressures of 1.0 and 1.5bars. Particle size data shown are in terms of mean intensity diameter (bars) and polydispersity index (PDI) (points). Asterisks indicate statistical significance with 95%confidence between the two data points over which they are situated (p < 0.05) for: (a) mean diameters, (b) mean diameters and PDIs.

(a) (b)

Fig. 2. Reconstitution metrics of spray granulated naproxen nanosuspension, run at a spray rate of 1.8 g/min, comparing the spray mode. (a) Re-dispersability is showncomparing top spray and bottom spray. Particle size data shown are in terms of mean intensity diameter (bars) and polydispersity index (PDI) (points). Asterisks indicatestatistical significance with 95% confidence between the two top spray trials (p < 0.05). (b) Dissolution profiles for top spray (j) and bottom spray (d), at both 1.0 bar (solidline) and 1.5 bars (dotted line) atomizing pressures.

6 C.E. Figueroa, S. Bose / European Journal of Pharmaceutics and Biopharmaceutics xxx (2013) xxx–xxx

not drastically different (Table 1), these measurements are nottruly representative of the surface coating. These values indicatethe porosity of the whole granule and thus are primarily reflectingon the internal structure of the mannitol, which should not be dif-ferent among all trials since the quality of mannitol was main-tained constant. Further characterization of the films on thesubstrate would be necessary to elucidate whether the structureis responsible for the observed differences in re-dispersibility anddissolution.

3.3. Comparison of spray mode

Various spray modes have been reviewed in the literature forspray coating processes, particularly for the coating of substrates,such as pellets, crystals, or granules, with a polymeric solution orsuspension [24,49]. As previously mentioned, top spray has histor-ically been the most commonly used spray mode for pharmaceuti-cal applications. On the other hand, the Wurster-type bottomspray, which most closely resembles the bottom spray mode oper-able on the Huttlin Mycrolab, has been cited as being more effi-cient than top spray at producing thin uniform polymer coatingson particles. This has been observed for coating potassium chloridecrystals with ethylcellulose [50] as well as unspecified pellets witha cellulosic binder [51]. Since spray granulation of nanosuspen-sions is still a relatively new field of research with few publicationson this subject, there have been no reports in the literature of theeffect of spray mode on re-dispersion and dissolution of the dry

Please cite this article in press as: C.E. Figueroa, S. Bose, Spray granulation: Imposuspensions, Eur. J. Pharm. Biopharm. (2013), http://dx.doi.org/10.1016/j.ejpb.2

granules coated with nano-crystals. Hence, a comparison of thetwo spray modes was selected as one of the parameters of thisstudy. The effects of different atomizing pressures at a single sprayrate (1.8 g/min) were compared for top and bottom spray granula-tions, since previous screening had shown better re-dispersibilityat the lower spray rate.

The results from the spray mode comparison are shown inFig. 2. The results showed a statistically significant drop in meanparticle size and PDI as the atomizing pressure was increased forthe top spray batches, and no significant difference was observedfor the bottom spray trials (Fig. 2a). Faster dissolution rates wereobserved for granules produced using top spray, at least duringthe first 30 min of the experiments (Fig. 2b). Although the dissolu-tion profiles of the granulations produced using the top spraymode at different atomizing pressure were not statistically differ-ent for most time points, the trial produced at 1.5 bars resultedin a statistically higher amount of drug dissolved (p < 0.05). Onthe other hand, for bottom spray, while the profiles for both atom-izing pressure were statistically different for most time points,both achieved complete dissolution.

Table 1 shows that powders produced using top spray had BETsurface area values that were, on an average, 20% larger than thoseproduced using bottom spray. This might explain the faster disso-lution rates obtained for granules obtained from the top spraymode, since the mass of film exposed per unit water during disso-lution was greater for these granules. Furthermore, the bulk den-sity measurements showed that granules produced using top

rtance of process parameters on in vitro and in vivo behavior of dried nano-013.07.015

Fig. 3. Scanning electron micrographs of naproxen and Compound A nanosuspension granulation powders produced at a nanosuspension spray rate of 1.8 g/min and at anozzle atomizing pressure of 1.5 bars at (a) naproxen, top spray; (b) naproxen, bottom spray; (c) Compound A, top spray; (d) Compound A, bottom spray. The granules appearto be more porous in (a and c) as opposed to (b and d), which correlates with the measured BET surface areas (larger surface area for (a and c)).

C.E. Figueroa, S. Bose / European Journal of Pharmaceutics and Biopharmaceutics xxx (2013) xxx–xxx 7

spray had bulk densities that were roughly 23% lower than thoseproduced through bottom spray. These findings are in agreementwith the literature reports where granules produced from topspray granulation have been characterized by low bulk densityand porous surfaces that facilitate improved wicking of liquid intothe granules leading to improved dispersibility and disintegration[39]. While the dissolution behavior might be explained by the dif-ferences in porosity of the nanosuspension film on the substratestructure, it is not clear whether this is relevant to the re-dispersi-bility of nanoparticles. Clearly, no drastic difference was observedin the re-dispersed particle size data among the top spray and bot-tom spray trials (Fig. 2a). However, as Bhakay and co-workers havepointed out, it is difficult to correlate re-dispersibility of driednanosuspension to dissolution when different media are used inthe two experiments [30].

The observed differences between the two spray modes, wheretop spray was found to be more sensitive to atomizing pressure ascompared to bottom spray, are most likely influenced by the pro-cess operation. In the top spray mode, the liquid nanosuspensionis being sprayed counter-current to the fluid bed flow, while forbottom spray, the flow of the liquid and the fluid bed are co-cur-rent. In the counter-current operation, since the nozzle is not im-mersed in the bed, the distance travelled by the sprayed dropletsprior to impact with the fluidized particles is random and impossi-ble to control [52]. During this time, enhanced liquid evaporationdue to longer mass transfer is possible, leading to a decrease indroplet size, an increase in droplet viscosity and possibly spraydrying of the droplet [24,49]. Once the viscous droplet hits a man-nitol particle, uniform spreading of the liquid as a film on the par-ticle surface is difficult [51], possibly leading to increasedpatchiness or porosity of the film. On the other hand, co-currentflow allows for the liquid to contact the particles immediatelyupon spraying. This allows for improved film formation on themannitol surfaces, since there is relatively little water evaporationfrom nozzle exit to particle impact. This is in agreement with pre-vious findings regarding more uniform films being formed throughbottom spray, as measured by film porosity [50]. SEM imaging ofthe granules (Fig. 3) showed that the granules produced using

Please cite this article in press as: C.E. Figueroa, S. Bose, Spray granulation: Imposuspensions, Eur. J. Pharm. Biopharm. (2013), http://dx.doi.org/10.1016/j.ejpb.2

top spray had more porous surfaces compared to granules pro-duced using bottom spray. A more uniform film with a less poroussurface will slow the wicking of water when hydrated, thus slow-ing down primary particle breakup and dissolution. Since top spraywill generally form more porous coatings, it is possible that varia-tions in the starting droplet size arising from differences in theatomization pressure, as well as the number of droplets sprayedcontrolled by the spray rate, might lead to varying degrees ofimperfections in the surface nano-crystal film, allowing for differ-ent rates of water penetration and thereby differences in the rate ofdissolution.

3.4. Characterization of granulation powder

Each trial was characterized for process yield (or mass recovery)and moisture content in terms of the loss on drying. Overall, theaverage yield was approximately 80%, though the bottom spray tri-als resulted in slightly lower yields due to some powder agglomer-ation from poorer fluidization. The LOD was similar amongst alltrials (0.80–1.00%).

With regard to the powder crystallinity, it was important todetermine the physical state of the pharmaceutical active, naprox-en. A crystalline drug substance was desired because amorphousmaterial will compromise stability and can possibly lead to particlesize growth and altered dissolution and absorption profiles. X-raydiffraction was conducted on neat mannitol DC and naproxenmaterial, as well as several different trial powders produced atthe different process parameter levels. The diffraction patternsare displayed in Fig. 4. It is obvious that amorphous material can-not be detected from the measured patterns, indicating a large de-gree of crystallinity, if not purely crystalline material. The spraygranulated powders all show an overlap of both naproxen andmannitol patterns, confirming the recovery of both the drug sub-stance and the granulation substrate. While the characteristicpeaks of naproxen are not as easily distinguishable as those ofthe mannitol in the spray granulated material, this is due to thedilution of the drug substance in the sugar carrier (1:4.7naproxen:mannitol by mass).

rtance of process parameters on in vitro and in vivo behavior of dried nano-013.07.015

10 15 20 25 30 35 40

Fig. 4. X-ray diffraction patterns of neat mannitol DC, naproxen drug substance andfour selected trial batches from the spray granulation trials.

Table 3Summary of pharmacokinetic parameters of in vivo study in male beagle dogs (n = 5).Reported values are the mean followed by the standard deviation.

Formulation Cmax (ng/mL)

tmax (h) AUC0–48

(h ng/mL)AUC0-inf

(h ng/mL)t1/2 (h)

Coursesuspension

68.7 ± 19.3 0.9 ± 0.3 262 ± 132 353 ± 160 3.1 ± 1.8

Nanosuspension 318.3 ± 88.5 0.8 ± 0.3 1204 ± 370 1340 ± 413 4.1 ± 1.8Top spray 237.5 ± 65.4 1.5 ± 0.4 1038 ± 218 1393 ± 361 6.5 ± 1.8Bottom spray 226.3 ± 58.0 3.0 ± 2.2 1327 ± 437 1410 ± 420 5.3 ± 0.8

Table 2Flow characterization of selected spray granulated trial batches. Experimental FRIvalues were obtained through stability tests through variable flow rates (REP + VFR)and ffc values were obtained through a shear cell test. Please refer to the section fordetails on the tests.

Trial Spraymode

Liquid sprayrate (g/min)

Nozzleatomizingpressure (bar)

Flow rateindex(FRI)

Jenike flowfunction(ffc)

I Top 1.8 1.0 1.44 40.6II Top 1.8 1.5 1.28 89.5III Top 2.5 1.0 1.46 14.8VI Bottom 1.8 1.5 1.33 25.2

8 C.E. Figueroa, S. Bose / European Journal of Pharmaceutics and Biopharmaceutics xxx (2013) xxx–xxx

The flow properties of the spray granulated powders were alsoevaluated. It is important that the granules not be cohesive afterspray granulation, as this can be problematic for further processingsteps such as encapsulation or tabletting. When loaded on hoppers,powders are desired to have good flow properties such that pro-duction is not held up because of consolidation due to cohesion.Therefore, stability tests at variable flow rates (REP + VFR) andshear cell tests, as well as compressibility and permeability tests,were conducted on the various trial batches. The REP + VFR testshowed that all batches tested exhibited flow rate indices (FRI),which is a measure of powder cohesion, below 3 (Table 3). AnFRI value below 3 indicates that the sensitivity to flow rate changeis average or characteristic of powders that are not cohesive. Fur-thermore, the Jenike flow function (ffc), which is used to gaugethe flowability, was also measured. Typically, values below 4 indi-cate cohesive powders, values between 4 and 10 indicate an easyflowing material and values greater than 10 indicate free flowingmaterial [26]. As seen in Table 2, all batches tested were describedas free flowing (ffc > 10). The compressibility for all the batches

Please cite this article in press as: C.E. Figueroa, S. Bose, Spray granulation: Imposuspensions, Eur. J. Pharm. Biopharm. (2013), http://dx.doi.org/10.1016/j.ejpb.2

tested was 1.03 ± 0.01%, while the pressure drop across the powderbed was 0.71 ± 0.01, 0.73 ± 0.01, 0.38 ± 0.00, and 2.47 ± 0.06 for tri-als I, II, III, and VI, respectively. The compressibility and permeabil-ity tests both confirm the observations that the spray granulatedpowders are non-cohesive, as there is insignificant change in vol-ume and pressure drop as the normal stress is increased from1 kPa to 15 kPa.

Among the batches tested, all exhibited the same flow behavior,with some variation in the actual numerical values. The biggest dif-ference amongst batches was in the permeability for bottom spraybatch VI as opposed to the three top spray batches, where VI pro-duced at least triple the pressure drop of the top spray batches.This could also be due to the aforementioned characteristics of bot-tom spray, where the nano-crystal layer will be more uniform onthe substrate surface, leading to less specific surface area and in-creased bulk density. A denser material will experience a largerpressure drop when exposed to the same air flow rate as a lessdense material. It should also be noted that trial batch III displayeda slightly larger particle size as compared to the other batches (Ta-ble 1); however, this is consistent with the lower pressure drop inthe permeability test.

3.5. Application to developmental drug

Finally, it is important to determine whether and how the ef-fects of the process parameters will affect the spray granulatednanosuspension formulations in vivo. However, naproxen, whileBCS class II, has been shown to be absorbed readily as indicatedby its high bioavailability [53,54]. Thus, naproxen was not consid-ered as an ideal compound for evaluation in in vivo animal studies.Hence, the findings attributed to the effect of processing parame-ters from the naproxen nanosuspension were tested using a Novar-tis developmental drug (Compound A) which is also BCS class II,but has poor permeability. Of particular interest was the effect ofthe spray mode. Thus, nanosuspensions of Compound A were pro-duced using a similar formulation to that of naproxen in terms ofcomposition, production method, particle size and stability. Thesenanosuspensions were then dried using the same spray granula-tion procedure used for naproxen, using mannitol as the carriersubstrate.

In vitro dissolution testing on liquid and dried nanosuspensionsamples of Compound A was carried out in non-sink conditionsto observe whether the spray mode resulted in drastic differencesbetween dissolution profiles. As shown in Fig. 5a, the same behav-ior as observed with the dried naproxen nanosuspensions (Fig. 2b)was exhibited for the dried Compound A nanosuspensions, wheredried powders produced through top spray yielded faster dissolu-tion profiles than those produced through bottom spray. Based onthis result, it appears that the spray mode affected the structure ofthe nano-crystal film on the substrate to a greater extent than theactive when comparing similar formulations. This can be sup-ported by the SEM images of naproxen and Compound A granulesobtained from trials using the top and bottom spray mode (Fig. 3).For both naproxen and Compound A, top spray produced more por-ous granules compared to bottom spray where a more uniformcoating of the nanosuspension on the substrate was observed.Additionally, the BET surface area for the top spray granulationsfor both naproxen and Compound A was higher as opposed tothe bottom spray. For naproxen, top spray yielded a surface areaof 2.55 ± 0.02 m2/g versus 2.27 ± 0.03 m2/g (Table 1), while forCompound A, the comparison was 2.49 ± 0.03 m2/g to2.11 ± 0.02 m2/g.

Furthermore, an in vivo study was conducted to determinewhether the effect of spray mode is relevant in the GI tract. Usingfasted male beagle dogs, Compound A was administered as acoarse suspension, nanosuspension, and dried nanosuspension

rtance of process parameters on in vitro and in vivo behavior of dried nano-013.07.015

(a) (b)

Fig. 5. Release profiles of Compound A. (a) Results for in vitro dissolution showing the initial nanosuspension (j) as well as spray granulated powders made through top spray(d) and bottom spray (N). (b) Results for in vivo in male beagle dogs (n = 5) showing a course suspension (j), the liquid nanosuspension (d), and spray granulated powdersmade through top spray (N) and bottom spray (.). Although the in vitro test would seem to indicate that there would be drastic differences in the absorption of the drugin vivo, in reality, there is little difference between the liquid and dry forms of the nanosuspension. This suggests that the increase in surface area over the course suspension iswhat helps to increase the absorption in vivo.

C.E. Figueroa, S. Bose / European Journal of Pharmaceutics and Biopharmaceutics xxx (2013) xxx–xxx 9

formulations manufactured using the top and bottom spray modes.The results are displayed in Fig. 5b and in Table 3. The in vivo dataclearly demonstrate the improvement in drug absorption when theparticle size of the formulation is reduced from the micron to thenanometer scale, which is consistent with the existing theory(Noyes–Whitney equation). Table 3 shows the Cmax and AUC valuesfor the nanosuspensions to be significantly higher compared tothose of the course suspension (p < 0.05). Comparing the variousnano-formulations, a slightly higher Cmax value was observed fromthe parent nanosuspension compared to both the dried formula-tions. The overall AUC (AUC0-inf) can be considered to be compara-ble among all the three formulations, the parent nanosuspensionand the two dried formulations. This is probably due to the factthat when administered in vivo, the formulations will encountera medium with bile salts and other unidentified surfactants whichmight allow for significant wetting and solubilization and thus bet-ter disintegration and re-dispersion of the primary particles overtime, which is sufficient to nullify any potential effect on the ob-served overall AUC. The PK parameter which clearly differentiatesthe three nano-formulations in vivo is the tmax where the samerank ordering as seen for the in vitro dissolution profiles is ob-served, with tmax values increasing in the following order, parentnanosuspension < top spray dried nanosuspension < bottom spraydried nanosuspension. The faster tmax observed from the top spraydried formulation could be related to the higher porosity of thesurface film for these granules, which leads to faster disintegrationof the granules and subsequent absorption. For the bottom spraydried formulation, a more uniform film with a less porous surfacewill slow the wicking of GI fluid when hydrated, thus slowingdown primary particle breakup and absorption, translating to alonger tmax. This difference in tmax was not found to be critical forCompound A due to the relatively long absorption window of themolecule, which made it possible for even the bottom spray formu-lation to be well absorbed. However, this could be a critical factorfor compounds which have a very narrow absorption window,where the absorption window could be potentially missed if a pro-longation of tmax is observed in vivo. For such compounds, top spraygranulation might be more amenable to obtaining a reliable in vivoexposure profile. From an overall perspective, additional com-pounds would need to be evaluated in vivo to recommend onespray mode over the other for a broader range of compounds.

4. Conclusions

This work evaluated the effect of critical process parameterssuch as spray mode, spray rate and nozzle atomization pressure

Please cite this article in press as: C.E. Figueroa, S. Bose, Spray granulation: Imposuspensions, Eur. J. Pharm. Biopharm. (2013), http://dx.doi.org/10.1016/j.ejpb.2

for spray granulation of drug nanosuspensions through a system-atic study using naproxen as a model compound. Various physicalproperties of the granules, as well as the in vitro re-dispersion anddissolution characteristics of the nano-crystals, were measured.Spray mode was found to have the biggest impact, where top sprayyielded smaller re-dispersed particle sizes and faster release ratesof drug from granules compared to bottom spray. This was attrib-uted to the co-current spraying in bottom spray that resulted indenser, homogenous films on the substrate. Furthermore, a higherliquid spray rate was shown to result in more particle aggregation,while increasing the atomization pressure decreased the re-dis-persed particle diameters. The optimum process parameters iden-tified from the trials with naproxen were then used to manufacturedried nanosuspensions of Novartis Compound A, which were fur-ther evaluated in an in vivo study in fasted male beagle dogs. Invivo studies with Compound A showed no significant difference be-tween the parent nanosuspension and dried forms manufacturedusing the top and bottom spray modes for the overall AUC; how-ever, a distinct prolongation of tmax was observed for the bottomspray formulation compared to the parent nanosuspension andthe top spray formulation. Based on these findings, spray granula-tion could be an attractive option for the production of powderswith desired processing and handling properties without compro-mising the enhanced pharmacokinetic profiles of the active com-pound in nanoparticle form. Further studies with additionalmodel compounds (including compounds that are in the free acidform) will be evaluated in future to expand the applicability of thisformulation platform for a broader range of compounds within BCSClass IIa.

Acknowledgements

This work was carried out within the framework of a summerinternship project at Novartis Pharmaceuticals Corporation. Theauthors would like to acknowledge the following colleagues: AlHollywood for providing training on manufacturing equipment,Mark Mecadon for providing input to the spray granulation exper-imental set-up, Paul Karpinski and Ester Maulit for analytical sup-port, Marilyn Alvine and Radha Vippagunta for XRDmeasurements, Greg Argentieri for the SEM analysis, Shau Yenfor providing training on the FT4 powder rheometer, Tycho Heim-bach and Handen Hee for providing input to the in vivo animalstudy design and Colleen Ruegger and Michael Motto from Novar-tis management for their support to this summer project.

rtance of process parameters on in vitro and in vivo behavior of dried nano-013.07.015

10 C.E. Figueroa, S. Bose / European Journal of Pharmaceutics and Biopharmaceutics xxx (2013) xxx–xxx

References

[1] C. Lipinski, Poor aqueous solubility – an industry wide problem in drugdiscovery, Am. Pharm. Rev. 5 (2002) 82–85.

[2] G.L. Amidon, H. Lennernas, V.P. Shah, J.R. Crison, A theoretical basis for abiopharmaceutic drug classification: the correlation of in vitro drug productdissolution and in vivo bioavailability, Pharm. Res. 12 (1995) 413–420.

[3] J.M. Butler, J.B. Dressman, The developability classification system: applicationof biopharmaceutics concepts to formulation development, J. Pharm. Sci. 99(2010) 4940–4954.

[4] A.A. Noyes, W.R. Whitney, The rate of solution of solid substances in their ownsolutions, J. Am. Chem. Soc. 19 (1897) 930–934.

[5] R.H. Muller, S. Gohla, C.M. Keck, State of the art of nanocrystals – specialfeatures, production, nanotoxicology aspects and intracellular delivery, Eur. J.Pharm. Biopharm. 78 (2011) 1–9.

[6] L. Peltonen, J. Hirvonen, Pharmaceutical nanocrystals by nanomilling: criticalprocess parameters, particle fracturing and stabilization methods, J. Pharm.Pharmacol. 62 (2010) 1569–1579.

[7] B. Van Eerdenbrugh, J. Vermant, J.A. Martens, L. Froyen, H.J. Van, P. Augustijns,G. Van den Mooter, A screening study of surface stabilization during theproduction of drug nanocrystals, J. Pharm. Sci. 98 (2009) 2091–2103.

[8] B. Van Eerdenbrugh, L. Froyen, H.J. Van, J.A. Martens, P. Augustijns, G. Van denMooter, Drying of crystalline drug nanosuspensions – the importance ofsurface hydrophobicity on dissolution behavior upon redispersion, Eur. J.Pharm. Sci. 35 (2008) 127–135.

[9] B. Van Eerdenbrugh, G. Van den Mooter, P. Augustijns, Top-down production ofdrug nanocrystals: nanosuspension stabilization, miniaturization andtransformation into solid products, Int. J. Pharm. 364 (2008) 64–75.

[10] P. Sharma, W.A. Denny, S. Garg, Effect of wet milling process on the solid stateof indomethacin and simvastatin, Int. J. Pharm. 380 (2009) 40–48.

[11] P. Kayaert, G. Van den Mooter, Is the amorphous fraction of a driednanosuspension caused by milling or by drying? A case study withNaproxen and Cinnarizine, Eur. J. of Pharm. Biopharm. 81 (2012) 650–656.

[12] J. Beirowski, S. Inghelbrecht, A. Arien, H. Gieseler, Freeze-drying ofnanosuspensions, 1: freezing rate versus formulation design as criticalfactors to preserve the original particle size distribution, J. Pharm. Sci. 100(2011) 1958–1968.

[13] B. Van Eerdenbrugh, L. Froyen, J.A. Martens, N. Blaton, P. Augustijns, M.Brewster, G. Van den Mooter, Characterization of physico-chemical propertiesand pharmaceutical performance of sucrose co-freeze-dried solidnanoparticulate powders of the anti-HIV agent loviride prepared by mediamilling, Int. J. Pharm. 338 (2007) 198–206.

[14] M.V. Chaubal, C. Popescu, Conversion of nanosuspensions into dry powders byspray drying: a case study, Pharm. Res. 25 (2008) 2302–2308.

[15] D. Mou, H. Chen, J. Wan, H. Xu, X. Yang, Potent dried drug nanosuspensions fororal bioavailability enhancement of poorly soluble drugs with pH-dependentsolubility, Int. J. Pharm. 413 (2011) 237–244.

[16] V. Nekkanti, R. Pillai, V. Venkateshwarlu, T. Harisudhan, Development andcharacterization of solid oral dosage form incorporating candesartannanoparticles, Pharm. Dev. Technol. 14 (2009) 290–298.

[17] S. Basa, T. Muniyappan, P. Karatgi, R. Prabhu, R. Pillai, Production and in vitrocharacterization of solid dosage form incorporating drug nanoparticles, DrugDev. Ind. Pharm. 34 (2008) 1209–1218.

[18] P. Kayaert, M. Anne, G. Van den Mooter, Bead layering as a process to stabilizenanosuspensions: influence of drug hydrophobicity on nanocrystalreagglomeration following in-vitro release from sugar beads, J. Pharm.Pharmacol. 63 (2011) 1446–1453.

[19] S. Bose, D. Schenck, I. Ghosh, A. Hollywood, E. Maulit, C. Ruegger, Applicationof spray granulation for conversion of a nanosuspension into a dry powderform, Eur. J. Pharm. Sci. 47 (2012) 35–43.

[20] A. Monteiro, A. Afolabi, E. Bilgili, Continuous production of drug nanoparticlesuspensions via wet stirred media milling: a fresh look at the Rehbinder effect,Drug Dev. Ind. Pharm. (2012) 1–18.

[21] L. Sievens-Figueroa, A. Bhakay, J.I. Jerez-Rozo, N. Pandya, R.J. Romanach, B.Michniak-Kohn, Z. Iqbal, E. Bilgili, R.N. Dave, Preparation and characterizationof hydroxypropyl methyl cellulose films containing stable BCS Class II drugnanoparticles for pharmaceutical applications, Int. J. Pharm. 423 (2012) 496–508.

[22] C.P. Mora, F. Martinez, Thermodynamic quantities relative to solutionprocesses of Naproxen in aqueous media at pH 1.2 and 7.4, Phys. Chem. Liq.44 (2006) 585–596.

[23] S.M. Iveson, P.A.L. Wauters, S. Forrest, J.D. Litster, G.M.H. Meesters, B. Scarlett,Growth regime map for liquid-bound granules: further development andexperimental validation, Powder Technol. 117 (2001) 83–97.

[24] F.N. Christensen, P. Bertelsen, Qualitative description of the Wurster-basedfluid-bed coating process, Drug Dev. Ind. Pharm. 23 (1997) 451–463.

[25] R. Freeman, Measuring the flow properties of consolidated, conditioned andaerated powders – a comparative study using a powder rheometer and arotational shear cell, Powder Technol. 174 (2007) 25–33.

Please cite this article in press as: C.E. Figueroa, S. Bose, Spray granulation: Imposuspensions, Eur. J. Pharm. Biopharm. (2013), http://dx.doi.org/10.1016/j.ejpb.2

[26] J. Tomas, Fundamentals of cohesive powder consolidation and flow, GranularMatter 6 (2004) 75–86.

[27] M. Efentakis, J.B. Dressman, Gastric juice as a dissolution medium: surfacetension and pH, Eur. J. Drug Metab. Pharmacokinet. 23 (1998) 97–102.

[28] J.B. Dressman, G.L. Amidon, C. Reppas, V.P. Shah, Dissolution testing as aprognostic tool for oral drug absorption: immediate release dosage forms,Pharm. Res. 15 (1998) 11–22.

[29] A. Aburub, D.S. Risley, D. Mishra, A critical evaluation of fasted state simulatinggastric fluid (FaSSGF) that contains sodium lauryl sulfate and proposal of amodified recipe, Int. J. Pharm. 347 (2008) 16–22.

[30] A. Bhakay, R. Dave, E. Bilgili, Recovery of BCS Class II drugs during aqueousredispersion of core–shell type nanocomposite particles produced via fluidizedbed coating, Powder Technol. (2012).

[31] D. Junemann, J. Dressman, Analytical methods for dissolution testing ofnanosized drugs, J. Pharm. Pharmacol. 64 (2012) 931–943.

[32] D. Juenemann, E. Jantratid, C. Wagner, C. Reppas, M. Vertzoni, J.B. Dressman,Biorelevant in vitro dissolution testing of products containing micronized ornanosized fenofibrate with a view to predicting plasma profiles, Eur. J. Pharm.Biopharm. 77 (2011) 257–264.

[33] J.J. Guo, P.F. Yue, J.L. Lv, J. Han, S.S. Fu, S.X. Jin, S.Y. Jin, H.L. Yuan, Developmentand in vivo/in vitro evaluation of novel herpetrione nanosuspension, Int. J.Pharm. 441 (2013) 227–233.

[34] Q. Ma, H. Sun, E. Che, X. Zheng, T. Jiang, C. Sun, S. Wang, Uniform nano-sizedvalsartan for dissolution and bioavailability enhancement: influence ofparticle size and crystalline state, Int. J. Pharm. 441 (2013) 75–81.

[35] M.K. Gupta, Y.C. Tseng, D. Goldman, R.H. Bogner, Hydrogen bonding withadsorbent during storage governs drug dissolution from solid-dispersiongranules, Pharm. Res. 19 (2002) 1663–1672.

[36] D.B. Shenoy, G.B. Sukhorukov, Engineered microcrystals for direct surfacemodification with layer-by-layer technique for optimized dissolution, Eur. J.Pharm. Biopharm. 58 (2004) 521–527.

[37] C. Kindermann, K. Matthee, J. Strohmeyer, F. Sievert, J. Breitkreutz, Tailor-maderelease triggering from hot-melt extruded complexes of basic polyelectrolyteand poorly water-soluble drugs, Eur. J. Pharm. Biopharm. 79 (2011) 372–381.

[38] F.J. Otero-Espinar, S. Anguiano-Igea, N. Garcia-Gonzalez, J.L. Vila-Jato, J.Blanco-Mendez, Interaction of naproxen with b-cyclodextrin in solution andin the solid state, Int. J. Pharm. 79 (1992) 149–157.

[39] D.M. Jones, Factors to consider in fluid bed processing, Pharm. Technol. 9(1985) 50–62.

[40] S. Flogel, H. Egermann, Fluid bed granulation of lactose using bottom spraymethod, Eur. J. Pharm. Sci. 4 (Supplement 1) (1996) S185.

[41] Z.H. Loh, D.Z. Er, L.W. Chan, C.V. Liew, P.W. Heng, Spray granulation for drugformulation, Expert Opin. Drug Deliv. 8 (2011) 1645–1661.

[42] K. Dewettinck, L. Deroo, W. Messens, A. Huyghebaert, Agglomeration tendencyduring top-spray fluidized bed coating with gums, LWT – Food Sci. Technol. 31(1998) 576–584.

[43] M. Hemati, R. Cherif, K. Saleh, V. Pont, Fluidized bed coating and granulation:influence of process-related variables and physicochemical properties on thegrowth kinetics, Powder Technol. 130 (2003) 18–34.

[44] X. Hu, J.C. Cunningham, D. Winstead, Study growth kinetics in fluidized bedgranulation with at-line FBRM, Int. J. Pharm. 347 (2008) 54–61.

[45] K. Dewettinck, W. Messens, L. Deroo, A. Huyghebaert, Agglomeration tendencyduring top-spray fluidized bed coating with gelatin and starch hydrolysate,LWT – Food Sci. Technol. 32 (1999) 102–106.

[46] M. Tzika, S. Alexandridou, C. Kiparissides, Evaluation of the morphological andrelease characteristics of coated fertilizer granules produced in a Wursterfluidized bed, Powder Technol. 132 (2003) 16–24.

[47] J.Z. Gao, A. Jain, R. Motheram, D.B. Gray, M.A. Hussain, Fluid bed granulation ofa poorly water soluble, low density, micronized drug: comparison with highshear granulation, Int. J. Pharm. 237 (2002) 1–14.

[48] X. Wang, F. Cui, Y. Yonezawa, H. Sunada, Preparation and evaluation of highdrug content particles, Drug Dev. Ind. Pharm. 29 (2003) 1109–1118.

[49] E. Teunou, D. Poncelet, Batch and continuous fluid bed coating – review andstate of the art, J. Food Eng. 53 (2002) 325–340.

[50] P. Bertelsen, F.N. Christensen, P. Holm, K. Jorgensen, Comparison of organicsolvent-based ethylcellulose coatings of KCl crystals applied by top andbottom spraying in fluidized-bed equipment, Int. J. Pharm. 111 (1994) 117–125.

[51] A.M. Mehta, D.M. Jones, Coated pellets under the microscope, Pharm. Technol.9 (1985) 52–60.

[52] K. Dewettinck, A. Huyghebaert, Fluidized bed coating in food technology,Trends Food Sci. Technol. 10 (1999) 163–168.

[53] H.H. Frey, B. Rieh, Pharmacokinetics of naproxen in the dog, Am. J. Vet. Res. 42(1981) 1615–1617.

[54] E. Samara, D. Avnir, D. Ladkani, M. Bialer, Pharmacokinetic analysis ofdiethylcarbonate prodrugs of ibuprofen and naproxen, Biopharm. DrugDispos. 16 (1995) 201–210.

rtance of process parameters on in vitro and in vivo behavior of dried nano-013.07.015