Td

WAa

b

c

d

e

f

a

ARRAA

KDNILMC

1

ia(sam

ic

0d

Colloids and Surfaces B: Biointerfaces 89 (2012) 29– 39

Contents lists available at SciVerse ScienceDirect

Colloids and Surfaces B: Biointerfaces

jou rn al h om epage: www.elsev ier .com/ locate /co lsur fb

he influence of physical properties and morphology of crystallised lactose onelivery of salbutamol sulphate from dry powder inhalers

aseem Kaialya,b, Gary P. Martinc, Hassan Larhribd, Martyn D. Ticehurste, Ewa Kolosioneka,li Nokhodchia,f,∗

Chemistry and Drug Delivery Group, Medway School of Pharmacy, University of Kent, ME4 4TB Kent, UKPharmaceutics and Pharmaceutical Technology Department, University of Damascus, Damascus 30621, SyriaInstitute of Pharmaceutical Sciences, King’s College London, Franklin-Wilkins Building, 150 Stamford Street, London SE1 9NH, UKSchool of Applied Sciences, Huddersfield University, Queensgate HD1 3DH, UKPfizer Global, R & D, Pharmaceutical Sciences, Pharma Therapeutics, Sandwich, CT13 9NJ Kent, UKDrug Applied Research Center and Faculty of Pharmacy, Tabriz University of Medical Sciences, Iran

r t i c l e i n f o

rticle history:eceived 20 May 2011eceived in revised form 18 August 2011ccepted 24 August 2011vailable online 10 September 2011

eywords:ry powder inhaleron-solvent

nhalation efficiencyactose crystallisationorphology

arrier

a b s t r a c t

The aim of this work was to investigate the mechanistic evaluation of physicochemical properties ofnew engineered lactose on aerosolisation performance of salbutamol sulphate (SS) delivered from drypowder inhaler (DPI). Different crystallised lactose particles were obtained from binary mixtures ofbutanol:acetone. The sieved fractions (63–90 �m) of crystallised lactose were characterised in terms ofsize, shape, flowability, true density and aerosolisation performance (using multiple twin stage impinger(MSLI), Aerolizer® inhaler device, and salbutamol sulphate as a model drug). Compared to commerciallactose, crystallised lactose particles were less elongated, covered with fine lactose particles, and had arougher surface morphology. The crystallised lactose powders had a considerably lower bulk and tap den-sity and poorer flow when compared to commercial lactose. Engineered carrier with better flow showedimproved drug content homogeneity, reduced amounts of drug “deposited” on the inhaler device andthroat, and a smaller drug aerodynamic diameter upon inhalation. Aerodynamic diameter of salbutamolsulphate increased as lactose aerodynamic diameter decreased (linear, R2 = 0.9191) and/or as fine parti-cle lactose content increased (linear, R2 = 0.8653). Improved drug aerosolisation performance in the case

of crystallised lactose particles was attributed to lower drug–carrier adhesion forces due to a roughersurface and higher fine particle content. In conclusion, this work proved that using binary combinationsof solvents in crystallisation medium is vital in modification of the physicochemical and micromeriticproperties of carriers to achieve a desirable aerosolisation performance from DPI formulations. Amongall lactose samples, lactose particles crystallised from pure butanol generated the highest overall DPI

.

formulations desirability

. Introduction

A crucial factor in attaining successful drug delivery to the lungs to have an excellent drug delivery system, i.e. a reliable devicend an efficiently aerosolised formulation. Dry powder inhalersDPIs) have the capacity to fulfil this requirement [1,2]. It has beenhown that DPI performance is dependent on many factors suchs inhaler design [3–5], patient inhalation flow rate and inhalationanoeuvre [6–9], and powder formulation properties [10,11].

However, despite almost over than 40 years of research, achiev-

ng a high degree of drug aerosolisation still provides a majorhallenge with DPIs available in the market producing fine particle

∗ Corresponding author. Tel.: +44 1634 202947; fax: +44 1634 883927.E-mail address: a.nokhodchi@kent.ac.uk (A. Nokhodchi).

927-7765/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.colsurfb.2011.08.019

© 2011 Elsevier B.V. All rights reserved.

fractions (FPFs) ≤ 20% [12–14]. One method that has been employedwith a view to enhancing DPI performance is to incorporateengineered (or modified) drug particles and/or excipient systems[15–18]. Drug–carrier formulations are the most popular means ofpresenting powder for aerosolisation, which comprise interactivemixtures consisting of fine drug particles adhered to coarse car-rier particles. In such formulations, drug particles should have anaerodynamic diameter below 5 �m to promote more peripheraldelivery [19–23], but particles of this size exhibit high cohesiveforces which confer poor flow properties and this can result in highdose variability [24]. Therefore, usually, carrier is blended with drugin order to enhance drug flow properties, metering, and dispersion

by means of reducing drug particle agglomeration and facilitatedrug aerosolisation. In fact, the carrier metering-aid function inDPIs is of extreme importance as the required doses to be dispensedfrom DPIs are usually low.

3 faces B

faoasdamda

i([emtos[tbatscfaeotoobt[wtapaiIilc

2

2

pabcp

2

stt

0 W. Kaialy et al. / Colloids and Sur

Lactose is the most commonly used carrier in DPI formulationsor several reasons which are pharmaceutical, physicochemical,nd historical. Since lactose has been widely used in the designf many oral dosage forms, it is known to be safe and to confercceptable stability. Lactose administered in the mouth also has aweet after-taste which provides an assurance to the patient that aose has been taken [25]. However, the most important factor thatffects the availability of a drug delivered by DPI is the drug detach-ent from the carrier surface during the inhalation process. Poor

etachment of the drug from the carrier surface, can result in only small fraction of the delivered dose reaching the site of action.

In order to achieve adequate drug–carrier separation duringnhalation (drug re-dispersion), the major forces of interactionincluding Van der Waals forces, electrostatic, and capillary forces)26] must be overcome. These forces are dependent on sev-ral factors including the particle surface properties [27], particleechanical properties [28], particle contact area, and environmen-

al conditions [29]. Several studies have attempted to modify drugr carrier particles in order to achieve lower drug–carrier adhe-ion forces and consequently improved inhalation performance18,30–33]. Jones et al. [30] suggested that lactose could be crys-allised from alcoholic solutions to optimize cohesive adhesivealance (CAB) resulting in a considerably higher FPF of drug. Inddition lactose crystallised from: Carbopol 934 gel [31]; mix-ures of organic solvents [18,32,33], all produced higher FPFs ofalbutamol sulphate (SS) upon aerosolisation. A control of lactoserystallisation to produce tailored lactose for inclusion within DPIormulations appears to provide a sound strategy to promote drugerosolisation. However, relatively few studies have reported theffects of multi-solvent crystallisation of lactose carrier particlesn DPI formulation properties with even less attention been paido binary non-solvent systems. Kaialy et al. showed that the usef lactose particles crystallised from either ethanol:butanol [32]r ethanol:acetone [33] binary mixtures improved the inhalationehaviour of SS from DPIs. It has been shown that higher fine par-icle fraction of SS was obtained when volumes of either butanol32] or acetone [33] were increased from their binary nonsolventsith ethanol. It was clear that nonsolvent polarity in crystallisa-

ion medium is a key factor in determining physical properties anderosolisation performance of crystallised lactose particles. As aart of ongoing investigations, the present study examines differentspects of aerosolisation behaviour of DPI formulations contain-ng lactose crystallised from butanol:acetone cosolvent systems.t was the aim of the present investigation to show how changesn the physicochemical and micromeritic properties of engineeredactose particles can improve the FPF of SS from DPI formulationsompared to formulations containing commercial lactose.

. Materials and methods

.1. Materials

Lactose monohydrate (Pharmatose, 100 M) and salbutamol sul-hate were obtained from DMV International, The Netherlands,nd LB Bohle, Germany, respectively. Analytical reagent grade 1-utanol and acetone were obtained from Fisher Scientific, UK. Allhemicals and materials were used as received without furtherurification.

.2. Crystallisation of lactose particles

A saturated solution of lactose in water was prepared by dis-olving 30 g of lactose in appropriate volume of water at 55 ◦C suchhat final total volume is 100 mL to produce 30% (w/v) lactose solu-ion. Once prepared, the lactose saturated solution was kept under

: Biointerfaces 89 (2012) 29– 39

constant stirring (200 rpm) and temperature (55 ◦C) conditionsuntil the end of experiments. This saturated solution (5 mL) wasadded to different ratios of butanol:acetone non-solvents (1:0, 3:1,1:1, 1:3, and 0:1) at a constant rate (5 mL/min) such that the finaltotal volume was 80 mL.

After adding the whole lactose solution, the resultant suspen-sion was kept under constant stirring (200 rpm) for 10 min at roomtemperature (22 ◦C). Then, the precipitated crystals were collectedby filtration under a vacuum through membrane filter ventilated(<0.45 �m) fitted into a filtration unit. Subsequently, the crystalswere left to dry in a drying oven (50 ◦C) for 24 h. The crystalliseddried lactose samples were then collected and poured into a sealedvial until needed for further investigation. Prior to each study everylactose batch was sieved and only the 63–90 �m size fraction wasemployed (being typical of the size range employed in DPI formu-lations commercially) [34].

2.3. Particle size distribution analysis

Particle size analysis was conducted using a Sympatec(Clausthal-Zellerfeld, Germany) laser diffraction particle size ana-lyzer, as described previously [33].

2.4. Scanning electron microscope (SEM)

Electron micrographs of all lactose samples and lactose–SSblends were obtained using a scanning electron microscope (PhilipsXL 20, Eindhoven, Netherlands) operated at 15 kV. The specimenswere mounted on a metal stub with double-sided adhesive tapeand coated under vacuum with gold, in an argon atmosphere priorto observation.

2.5. Particle image analysis

A small amount of powder (about 20 mg) was homogenouslyscattered onto a microscope slide. The particle size and shape wereassessed using computerized morphometric analyzing system(Leica Q Win Standard Analyzing Software and Leica DMLA Micro-scope; Leica Microsystems Wetzlar GmbH, Wetzlar, Germany). Foreach lactose sample, a minimum 400 particles were detected andmeasured. Several shape descriptors were used including elonga-tion ratio (ER) (Eq. (1)) and shape factor (Fshape) (Eq. (2)) [36] androughness (Eq. (3)) (as the percentage of estimated particle perime-ter (Perimeter) to circumscribed particle perimeter (ConvxPerim)[37])

ER = maximum feret diameterminimum feret diameter

(1)

Fshape = 4 × � × area

perimeter2(2)

Roughness = PerimeterConvxPerim

(3)

2.6. Density and powder flow measurements

The true density (�true) of all lactose samples was measuredusing an ultrapycnometer 1000 (Quantachrom, USA) using heliumgas at an input gas pressure of 19 psi and an equilibrium time of1 min.

Carr’s index (CI) was measured for all lactose powders as anindication of powder flowability. Each powder was filled into a

5 mL measuring cylinder and after recording the volume (bulk vol-ume) the cylinder was tapped 100 times and the new volume wasrecorded (tap volume). A preliminary experiment showed that 100taps was sufficient to attain the maximum reduction in the volume

faces B

oi

C

2

2

so

2

eEftfuAaTo

2

aGNaloSqsec[dAr2

2(

hccmsupawaos(cdtg

W. Kaialy et al. / Colloids and Sur

f powder bed. The bulk density (�b), tap density (�t), and Carr’sndex (CI, Eq. (4)) were measured:

I =(

�t − �b

�b

)× 100 (4)

.7. In vitro formulation assessment

.7.1. Preparing drug–carrier formulation blendsSieved lactose samples (63–90 �m) were blended [32] with

albutamol sulphate (SS) in a ratio of 67.5:1, w/w (being typicalf ratios employed in commercial formulations).

.7.2. Homogeneity of the formulation blendsAfter blending, five randomly selected samples from differ-

nt positions of powder bed were taken for assay of SS content.ach sample, weighing 33 ± 1.5 mg which is the amount of theormulation blend in each capsule, was dissolved in 100 mL dis-illed water. The amount of the active drug (SS) in each powderormulation sample was determined using High Performance Liq-id Chromatography (HPLC) method as previously described [11].verage drug content for each formulation sample was expresseds the percentage of nominal drug dose (% potency, 481 ± 22 �g).he degree of drug content homogeneity was expressed in termsf percentage relative standard deviation (% RSD).

.7.3. Evaluation of drug–carrier adhesion characteristicsAir depression sieving was employed to assess drug–carrier

dhesion forces. A 45 �m sieve (Retsch® GmbhTest Sieve,ermany) was employed and the air jet sieve (Copley Scientific,ottingham, UK) was operated at a gas volume flow that gener-ted a negative pressure of 40 hPa (4 kPa). There was no effectiveoss of particles from each of the 63–90 �m batches after 3 minf air jet sieving, indicating the suitability of 45 �m sieve forS-lactose adhesion assessments. Therefore an accurately weightuantity of 1 g of each formulation blend was placed on top of theieve and four samples (33 ± 1.5 mg) were removed from differ-nt areas of the blend after different functional sieving times. Drugontent in each formulation sample was quantified using HPLC11] and expressed as a percentage of the average drug contentetermined initially in each blend before sieving (Section 2.7.2).dhesion assessments were conducted in an air-conditioned labo-atory where the ambient temperature and relative humidity were2 ◦C and 50%, respectively.

.7.4. In vitro deposition study using multi-stage liquid impingerMSLI)

After blending, each formulation was filled manually into aard gelatine capsule (size 3) with 33 ± 1.5 mg such that eachapsule contained 481 ± 22 �g SS equivalent to the dose in theommercially available Ventolin Rotacaps®. After filling, as mixingay produce electrical charge within the powder, capsules were

tored to allow any charge-relaxation to occur in sealed glass vialsnder laboratory conditions (22 ± 2 ◦C, 50 ± 5% RH) for at least 24 hrior to investigation. Deposition profiles of different blends weressessed in vitro using Aerolizer® inhaler device and MSLI equippedith a USP induction port (Copley Scientific, Nottingham, UK). In

ddition, for comparison purposes, the aerosolisation performancef SS alone (without carrier) formulation was also assessed. Theame amount of SS that was aerosolised from SS-lactose blends10 × 481 �g) was aerosolised as SS alone (2 × 2405 �g). Pharma-

opoeial deposition experiments [38] were performed and severaleposition parameters were employed (recovered dose (RD), emit-ed dose (ED), mass median aerodynamic diameter (MMAD), andeometric standard deviation (GSD)) as described elsewhere [32].

: Biointerfaces 89 (2012) 29– 39 31

The fine particle dose (FPD≤5 �m) was calculated by an interpo-lation of the drug aerodynamic size distribution obtained from theMSLI. FPF≤5 �m was calculated as the ratio of FPD≤5 �m to RD.

In addition the effective inhalation index (EI) was calculated(using Eq. (5)) as an indication of formulation performance [39,40].

EI = (EM × FPF)1/2 (5)

The theoretical aerodynamic diameter (Dae) of SS was calculatedfrom the geometric mean diameter (De) and true density (�ss) usingEq. (6) [41]:

Dae = De ×√

�ss

x(6)

where x is the dynamic shape factor for non-spherical parti-cles which was assumed to be 1 [42,43]. The same equationwas applied to estimate lactose theoretical aerodynamic diame-ter Dae,lactose, based on the lactose particle volume mean diameterand �true. All aerosol deposition experiments were performed inan air-conditioned laboratory where the ambient temperature andrelative humidity were 22 ± 2 ◦C and 50 ± 5%, respectively.

2.8. Statistical analysis

One way analysis of variance (ANOVA) was applied to comparemean results in this study, with P values less than 0.05 being consid-ered as indicative of significant difference. Where ANOVA indicatedsignificant difference, Tukey’s HSD post hoc test was applied.

3. Results and discussion

3.1. Crystallisation procedure

Regardless of the composition of non-solvent mixture usedin the crystallisation procedure, it was found that the crystalsformed immediately after the lactose solution was added to thenon-solvent medium. Anti-solvent crystallisation method is gen-erally regarded as a method where the yield is relatively low [44].However, the present study showed that if binary combinationsof butanol:acetone were used, the percentage yield could be high(up to 94%). This was in good agreement with our previous resultswhere binary non-solvents of ethanol:butanol or ethanol:acetonewere used [32,33].

3.2. Particle size assessments

Characterisation of the particle size distribution (PSD) for alllactose batches was performed by both laser diffraction (volume-weighted size measurement) and microscope image analysis(number-weighted size measurement).

The PSD for all lactose batches showed a symmetric monomodallognormal distribution. The % cumulative undersize distributionsare shown in Fig. 1-I. Despite all lactose powders having been sievedunder the same conditions, it can be seen that crystallised lac-tose powders have a smaller volume mean diameter (VMD) (in therange of 59.7–80.6 �m) than commercial lactose powders (Fig. 1-I) (91.1 �m). Differences in the absolute size of carrier powdersmight have a considerable effect on drug deposition profiles [14].Differences in the size of the sieved lactose powders can be ascribedto possible differences in particle shape. It is known that siev-ing produces a PSD corresponding to particles that pass throughthe mesh according to a settled orientation rather than necessar-ily the smallest cross-sectional dimension. Therefore, elongated

particles tend to produce particles larger than the mesh diameter[45]. This tends to be supported by the commercial lactose havingboth the highest VMD (Fig. 1-I) and elongation ratio (ER) values(P < 0.05) (Fig. 2a). The crystallised lactose particles were found to

32 W. Kaialy et al. / Colloids and Surfaces B: Biointerfaces 89 (2012) 29– 39

Fig. 1. (I) Particle size distribution, (II) volume mean diameter (VMD) (laser diffrac-tion, n ≥ 5) and equivalent diameter (Dequi) (image analysis, n ≥ 400): for salbutamolsua

hlba(arp(

ifS(bTt

halto(siDs(

Fig. 2. (a) (×) Elongation ratio (ER), (�) shape factor (Fshape); (b) (�) span, (�) volume

erogonous distribution in the local concentration of lactose within

ulphate (SS), commercial lactose (a) and crystallised lactose particles obtainedsing different volumes of butanol/acetone: (b) 80/0, (c) 60/20, (d) 40/40, (e) 20/60,nd (f) 0/80 as the precipitating non-solvent (mean ± SD, n ≥ 3).

ave a broader PSD than commercial lactose, as indicated by thearger span values (Fig. 2b). In addition, all the crystallised lactoseatches contained particles with a higher volume specific surfacerea (SSAv) and higher amounts of fine lactose particles <10.5 �mFPL<10.5 �m) compared to commercial lactose product (Fig. 2b). Themount of fines refers to the fine carrier particles that could not beemoved by sieving but rather remained attached to larger carrierarticles, as it is also evident from scanning electron micrographsFig. 3).

The higher SSAv for crystallised lactose particles is attributablen part to the higher fines content (Figs. 2b and 3) [46]. Inact, direct linear relationships were established between lactoseSAv and span (r2 = 0.91) and also between SSAv and FPL<10.5 �mr2 = 0.99), whereas an inverse linear relationship was obtainedetween lactose SSAv and VMD (r2 = 0.88) (figures not shown).hese relationships indicate the strong interdependence betweenhese parameters.

The VMD (as determined by laser diffraction) was markedlyigher than the equivalent diameter (Dequi, determined by imagenalysis) for all the crystallised lactose powders in contrast to theactose particles in the commercial batch (Fig. 1-II). This is likelyo be due in part to the more irregular shape and greater degreef corrugated surface texture of the crystallised lactose particlesFigs. 2b, 3, and 4) which often leads to an overestimation of particleize, as obtained by laser diffraction, due to high-angle light scatter-

ng [47]. Interestingly, there was no difference between VMD andequi for the commercial lactose powder that was the least irregularhaped particles (highest Fshape) and smoothest surface topographysmallest roughness) (Figs. 2a,b, 3, and 4).

specific surface area (SSAv), (©) fine particle lactose FPL<10.5 �m, and (�) roughnessfor commercial lactose and crystallised lactose particles obtained using differentvolumes of butanol/acetone as the precipitating non-solvent (mean ± SE, n ≥ 5).

PSD analysis of the SS sample indicated a monomodal distribu-tion with a VMD of 1.86 ± 0.16 �m, D50% of 1.66 ± 0.06 �m, and 90%of the particles less than 3.14 ± 0.31 �m (Fig. 1-I). As SS particlesexhibited a median diameter below the respirable size of 5 �m, itwas concluded that these particles would be suitable for respiratorydelivery.

3.3. Particle image analysis

Roughness (rugosity) estimations indicated that all crystallisedparticles had rougher surfaces compared to the commercial lac-tose particles (Fig. 2b). This was confirmed by scanning electronmicroscope (SEM) observations (Figs. 3 and 4). Surface roughnessof the different lactose particles followed a similar trend to surfacearea on the basis of volume distribution (SSAv) (Fig. 2) and indeeda clear linear correlation between the two terms was confirmed(r2 = 0.9857) (figure not shown).

A smooth sphere has Fshape and ER values of 1. As seen in Fig. 2a,the crystallised lactose particles had smaller ER and smaller Fshapevalues compared to commercial lactose particles. The smaller ERindicates that crystallised lactose particles are less elongated and/ormore irregular in shape than the commercial grade, which is alsosupported by SEM observations (Figs. 3 and 4). The lower Fshapefor crystallised lactose particles is indicative of the rougher surfaceand/or more irregular shape which correlates well with roughnessestimations (Fig. 2a and b) and the qualitative assessment made onthe basis of SEM data (Figs. 3 and 4). During crystallisation, a het-

crystallisation medium (as a consequence of mechanical stirring)is likely to lead to uneven crystal growth with surface irregularities[35].

W. Kaialy et al. / Colloids and Surfaces B: Biointerfaces 89 (2012) 29– 39 33

F ose sa4

3

ssTwsdmptpalbsi

cstcwlsilt

ig. 3. SEM images for (a) commercial lactose sample and different crystallised lact0/40, (e) 20/60, and (f) 0/80 as the precipitating non-solvent.

.4. Scanning electron microscopy

The use of Fshape and ER as descriptive parameters is notatisfactory to represent completely particle shape, as these mea-urements are function of particle orientation and contact area.herefore, in addition to two dimensional image analyses, SEMas employed to characterise qualitatively the three dimensional

hape and surface morphology of all lactose carriers (Fig. 3) andrug–carrier formulation blends (Fig. 4). The image of the com-ercial lactose confirmed the typical crystalline tomahawk shape

articles of mature crystals reported previously [48]. The crys-allised lactose particles displayed an angular shape with largerotuberances (Figs. 3 and 4) and appeared to comprise densegglomerates of smaller sub-units combined together to formarger clusters (Figs. 3 and 4). Such aggregation is probably inducedy lactose dehydration during the crystallisation process and theubsequent water/solvent evaporation, which occurs as the solid issolated.

The SEM image of micronized SS (Fig. 4a) showed fine parti-les less than 5 �m in size, which is in accordance with laser-lightcattering particle size data (Fig. 1-I). SS particles displayed theypical elongated cuboid shape reported previously. The SS parti-les had a tendency to form aggregates (Fig. 4a) but after blendingith different lactose batches, the drug particles could be morpho-

ogically distinguished in formulation blends adhered to lactose

urface as might be anticipated in the formation of drug–carriernteractive ordered mixtures (Fig. 4b–f). In case of crystallisedactose particles (having higher surface protuberances and asperi-ies (Fig. 3b–f)), it appeared that the drug particles were attached

mples obtained using different volumes of butanol/acetone: (b) 80/0, (c) 60/20, (d)

to uneven regions (microscopic projections, clefts, or pits) on car-rier particle surfaces (Fig. 4b–f). In the case of commercial lactose,drug particles were adhered to flat regions of the carrier particlesurfaces (Fig. 4a). Depending on their particle size, drug particlesadhered to flat regions on carrier surface might exhibit a relativelyhigh degree of adhesion and cohesion due to increased reduceddrug–carrier contact area [49–51]. In contrast, the lower contactarea between drug and carrier in the crystallised lactose–SS blendsmight be expected to allow easier detachment as aerosolisationoccurs.

3.5. Density and flowability assessment

The crystallised lactose samples had significantly (P < 0.05)higher �true but smaller �b and �t compared to commercial lactose(Fig. 5).

Differences in �true between lactose samples are possibly aconsequence of different crystalline nature (�, �-form or pseu-dopolymorph) (since �-lactose has a higher �true than �-lactose)[52] or different particle size (as, for the same material, small par-ticles have different density from larger particles) [53].

This has been attributed to atoms or molecules located near theparticle surface not having true equilibrium positions and with thepercentage of these being greater the smaller the size of the particle

this results in differences in particle �true.

The smaller �b and �t for crystallised lactose powders indicatea lower number of contact points between particles within thesepowders compared to those present in commercial grade lactose.

34 W. Kaialy et al. / Colloids and Surfaces B: Biointerfaces 89 (2012) 29– 39

Fig. 4. SEM images for formulation blends containing salbutamol sulphate (SS) mixed with (a) commercial lactose, crystallised lactose obtained using different volumes ofb tating

c(lpip

tcoappblp

pttbatf00

utanol/acetone: (b) 80/0, (c) 60/20, (d) 40/40, (e) 20/60, and (f) 0/80 as the precipi

Except for lactose powder crystallised from pure butanol,rystallised lactose powders showed higher compressibility23.9 ± 0.6–30.0 ± 1.3%, n ≥ 4, poor flowability) than commercialactose sample (18.8 ± 1.9%, fair flowability) (Fig. 5). A higher com-ressibility index (CI) for crystallised lactose powders confirms the

ncreased cohesiveness (due to higher interparticulate forces) com-ared to commercial lactose.

The irregular shape and higher surface roughness of the crys-allised lactose samples compared to the commercial lactose willontribute to the poorer flow properties, higher �b, and higher �t

f the former due to expected increased interlocking propensitynd frictional forces between the particles. In addition, the smallerarticles present in the crystallised batches (Fig. 1), especially theroportion <10 �m (Fig. 2b) will also lead to poor powder flowa-ility. Carr’s index of lactose powder could be directly correlated to

actose particle VMD (r2 = 0.90) and inversely correlated to lactosearticle surface roughness (r2 = 0.80) (figure not shown).

The results showed that �b and �t for crystallised lactoseowders were dependent on the ratio of non-solvents used duringhe crystallisation of lactose. These results are in agreement withhose reported previously when lactose was crystallised usinginary mixtures of ethanol:butanol [32] or ethanol:acetone [33]s non-solvents. The �b and �t of the lactose crystals produced in

he presence of the various non-solvent combinations were in theollowing rank order: ethanol:butanol (�b = 0.21 ± 0.04 g cm−3, �t =.27 ± 0.05 g cm−1 n = 15) < ethanol:acetone (�b = 0.29 ±.04 g cm−3, �t = 0.39 ± 0.06 g cm−3, n = 15) < butanol:acetone

non-solvent. FPA: fine particle aggregate; FPL: fine particle lactose.

(�b = 0.32 ± 0.11 g cm−3, �t = 0.42 ± 0.13 g cm−3, n = 26). Thechanges in densities of resultant lactose with different combi-nations of non-solvents could be attributed to the fact that thecomposition of non-solvent influences the amounts of intrinsicfine particle lactose (FPL) content (Fig. 2b) and this in turn influ-ences average contact points between particles within crystallisedlactose powders. Indeed, the rank order of �b and �t for crys-tallised lactose particles matched the rank order of FPL<10 �m

indicating that �b and �t for crystallised lactose powders increaseas FPL<10 �m increases.

3.6. Deposition studies

3.6.1. Homogeneity testRegardless of the volume fraction of non-solvent used during

the crystallisation process, when compared to commercial lactoseblend, the blends containing the crystallised lactose produced sig-nificantly lower (P < 0.05) potency (81.8–95.2% for all crystallisedsamples versus 98.7% for commercial lactose) but with improveddrug content homogeneity (RSD values of 3.4–8.9% versus 10.9%)(Fig. 6a). A RSD less than 6% can be taken as indication of adequatehomogeneity for DPI systems, but achieving acceptable variabilityof aerosol delivered dose is always challenging [54]. It is known

that mixing time has a significant effect on drug–carrier interac-tions (interparticulate forces) within the DPI formulation [36] andconsequently fine particle fraction (FPF) of drug during inhalation[55] but for comparison purposes, all drug–carrier formulations

W. Kaialy et al. / Colloids and Surfaces B: Biointerfaces 89 (2012) 29– 39 35

Fig. 5. (a) True density (�true), bulk density (� ), tap density (�t); (b) Carr’s index (CI) for commercial lactose and crystallised lactose samples obtained using different volumeso datafl

wo

dfabiw

ctptlcct

3M

w

b

f butanol/acetone as the precipitating non-solvent (mean ± SE, n ≥ 4) (true densityow properties, respectively).

ere subjected to same blending conditions in this study withoutptimization for each powder blend.

In drug–carrier interactive mixtures, drug uniformity is depen-ent on both cohesive forces between fine particles and adhesiveorces between fine and coarse particles. The rougher surfacend higher content of fines in the lactose crystal containingatches are likely to reduce drug–carrier contact area resulting

n lower drug–carrier interparticulate forces and consequentlyeaker drug–carrier adhesion.

It is known that the reduced drug content homogeneity of blendsan be affected by higher PSD polydispersity which is likely result inhe formation of drug rich areas within the blend due to enhancedercolation segregation (segregation by size). Good linear rela-ionships were established between the RSD of drug content withactose powder CI (r2 = 0.9648) and particle VMD (r2 = 0.8332) indi-ating an improved drug content homogeneity when incorporatingarrier powders with better flowability and larger particle size intohe blend (Fig. 6b).

.6.2. In vitro assessment of salbutamol sulphate deposition usingSLI

On the basis of the results from the deposition studies (Fig. 7), itould appear that the blends containing crystallised lactose would

, �true, was plotted on the basis of right y-axis; F and P correspond to fair and poor

produce higher amounts of drug depositing on the deep lung air-ways (i.e. <2 �m), peripheral alveolar airways (2–4 �m), and centralbronchial airways (4–6 �m) but lower amounts on the upper air-way regions (>6 �m) than the formulations containing commerciallactose. Powdered salbutamol sulphate per se formulation (carrier-free) produced the highest sub-2 �m size fraction (Fig. 7).

The higher amounts of drug deposited in MSLI stages 3, 4, andfilter generated from formulations containing crystallised lactoseindicated their better aerosolisation performance in comparisonto the blend containing commercial lactose. However, when thedrug was aerosolised without carrier then the deposition occurredevenly across the mouthpiece adaptor and MSLI stages 1, 3 and 4,with again the least amount reaching the MSLI-filter (Fig. 8). Also,the highest variation in deposition at all stages was apparent whenthe carrier free powder was aerosolised.

Retention of particles by the inhaler and mouthpiece is affectedby the flow properties of the final formulation, which in turn arecontrolled by the addition of carrier particles. Interestingly, plottingamounts of SS deposited on I&M + IP against CI for different lac-

tose samples resulted in positive linear relationship (R2 = 0.8593)(Fig. 9) indicating that the poorer the carrier flowability the higherthe amounts of drug deposited on inhaler device plus throat.In addition, higher �true and lower ER for crystallised lactose

36 W. Kaialy et al. / Colloids and Surfaces B: Biointerfaces 89 (2012) 29– 39

Fig. 6. (a) Percentage potency and relative standard deviation (RSD) of salbutamol sulphate content; (b) relationship between RSD with lactose particle volume mean diameter( and crt compa

pd

Mcfftmt

Di

i[pt

b

VMD) and powder Carr’s index (CI) for DPI blends containing commercial lactose

he precipitating non-solvent (mean ± SE, n ≥ 4) ( indicates significant difference

articles (Fig. 5) might contribute to higher drug deposition on IPue to higher inertial forces generated during aerosolisation.

The crystallised lactose-containing blends produced higherMADs but with smaller GSDs for SS than the commercial lactose-

ontaining blend (Fig. 10a). However, no significant difference wasound between commercial lactose formulation and without carrierormulation in terms of drug MMAD and GSD (P > 0.05). Never-heless, for an inhaled aerosol powder, neither the total deposited

ass nor the aerosolisation efficiency can be predicted accordingo MMAD and GSD alone.

Carriers with poorer flowability, higher FPL<10.5 �m, and smallerae,lactose produced a higher MMAD of SS (Fig. 10b), which may be

ndicative of a higher drug aggregation after aerosolisation [56].The higher experimental MMAD of SS compared to the theoret-

cal Dae (Fig. 10a), found in this study concurs with earlier studies57] and may be attributed to the high cohesive forces between SS

articles which resulted in the SS depositing as aggregates ratherhan single particles.

Altered aerosolisation performance of lactose–salbutamollends particles has been ascribed in part to the different

ystallised lactose powders obtained using different volumes of butanol/acetone asred to commercial lactose).

morphologies of the carrier employed, with carriers having a higherER producing a higher FPF of the drug upon aerosolisation [58].However, this was not the case in this study where the crystallisedlactose carriers having a lower ER (Fig. 2a) than the commerciallactose produced higher FPFs.

Upon air-jet sieving of DPI formulation blends, drug particlesare expected to separate from carrier surfaces. After only 5 s ofsieving, both the commercial and crystallised lactose-containingblends showed that drug separated from the powder mass and theamounts of drug remaining in the blends decreased as the siev-ing time increased (Fig. 11). The drug was detached more readilyfrom the crystallised lactose particles than from the commerciallactose. This suggests that in the case of crystallised lactose thatthe drug–carrier physical interactions were weaker than the adhe-sion forces between the carrier and the drug in the commercialsample. Such stronger adhesion is likely to account in part for the

lower drug FPF from the latter formulation.

Another key factor which affects DPI performance is the densityof the drug particles. However, there is a little information in the lit-erature showing the effect of carrier density on drug aerosolisation

W. Kaialy et al. / Colloids and Surfaces B: Biointerfaces 89 (2012) 29– 39 37

Fig. 7. Mass distribution of salbutamol sulphate when aerosolised from DPI blends containing commercial lactose (a) and crystallised lactose powders obtained using differentvolumes of butanol/acetone: (b) 80/0, (c) 60/20, (d) 40/40, (e) 20/60, and (f) 0/80 as the precipitating non-solvent assessed by MSLI and aerosolised using an Aerolizer® atflow rate of 92 L/min. Regional lung deposition was expected according to Stahlhofen et al. [66] and Mitchell and Nagel [67].

Fig. 8. The amount of salbutamol sulphate (% RD) deposited on the inhaler andmouthpiece adaptor (I&M), induction port (IP) and different MSLI stages after theaerosolisation of DPI blends containing either commercial or crystallised lactosepowder, or salbutamol sulphate alone (designated—without) using an Aerolizer® atflow rate of 92 L/min (S indicates no significant difference).

Fig. 9. Cumulative amounts of salbutamol sulphate deposited on the Aerolizer®

inhaler with its fitted mouthpiece adaptor (I&M) plus induction port (IP) plotted asa function of the Carr’s index of the carrier lactose (mean ± SE, n ≥ 3).

Fig. 10. (a) Mass median aerodynamic diameter (MMAD), theoretical aerodynamicdiameter (Dae), geometric diameter (De), and geometric standard deviation (GSD);(b) relationship between MMAD and lactose powder Carr’s index (CI), fine particlelactose FPL<10.5 �m, and theoretical aerodynamic diameter (Dae,lactose) of salbutamolsulphate aerosolised from DPI blends containing commercial lactose and crystallisedlactose powders generated using different volumes of butanol/acetone as the pre-cipitating non-solvent (mean ± SE, n ≥ 3).

38 W. Kaialy et al. / Colloids and Surfaces B: Biointerfaces 89 (2012) 29– 39

Ffc

pc

tacfc

oc[ecd

d

wYi

Fp

Fig. 12. Relationship between FPF of salbutamol sulphate, lactose bulk density (�b),

ig. 11. The amount of salbutamol sulphate (%) remaining on a 45 �m sieve as aunction of the time of sieving DPI blends containing commercial lactose or lactoserystallised using acetone as a non-solvent.

erformance. The drug FPF obtained in this study correlated witharrier �b and �t supporting higher drug FPF (Fig. 12).

Other factors that impact upon and can enhance the aerosolisa-ion performance of SS from crystallised lactose formulation blendsre the smaller carrier size [59] (Fig. 1), higher amounts of fines onarrier surface [60] (Figs. 2b, 3, and 4), higher carrier specific sur-ace area [61] (Fig. 2b), higher carrier �true [47] (Fig. 5a) and higherarrier porosity [62] (Fig. 5a).

Overall desirability function parameter (DF) is helpful in termsf evaluating several parameters simultaneously and thereforeould be applied for experimental design-based optimization63,64]. In order to perform quantitative comparison of differ-nt DPI formulation blends containing commercial lactose, lactoserystallised from butanol:acetone; a desirability function (di),efined in Eq. (7), was used [65]:

imax = Yi − Ymin

Ymax − Ymin(7)

here Yi is the actual observed value for the response variable,max is the maximum acceptable value of the variable, and Ymins the minimum acceptable value of the variable. It is well known

ig. 13. Overall desirability function (DF) for DPI blends containing commercial lactose, arecipitating non-solvent.

lactose tap density (�t) obtained from DPI blends containing crystallised lactosepowders obtained using different volumes of butanol/acetone as the precipitatingnon-solvent (mean ±SE, n = 3).

that a higher ED, FPD, and EI are important requirements for effec-tive inhalation performance through DPI. In addition a lower RSDand CI for DPI formulation are desirable in terms of producing areduced dose variability and easier powder handling, respectively.Therefore, ED, FPD, EI, RSD, and CI parameters were selected askey parameters in evaluating different DPI formulation blends bymeans of DF. The maximum (Ymax) response variables were definedas follows: ED = 481 �g, FPD = 481 �g, EI = 100%, CI = 40%, RSD = 20%.The minimum (Ymin) response for the measured variables were setas follows: ED = 10 �g, FPD = 10 �g, EI = 10%, CI = 5%, and RSD = 1%.Overall desirability function (DF) was calculated as the geometricmean of different desirability parameters by Eq. (8)

DF =(

dED + dFPD + dEI + 1dCI

+ 1dRSD

)1/5(8)

Using this equation, the lowest DF was obtained for the formulation

nd crystallised lactose obtained using different volumes of butanol/acetone as the

containing no carrier, whilst the highest DF was obtained for formu-lation blend containing lactose crystallised using butanol:acetone(80:0) as the precipitating non-solvent (Fig. 13). These observa-tions suggest that this batch of lactose produced the best overall

faces B

pta

4

baorotgabitf

A

IIS

R

[

[

[[

[[[[

[

[[[[

[

[[[

[[[

[

[[

[

[

[[

[

[

[

[

[

[[[[

[[

[

[

[

[[

[

[[

[[

[

[[[[

W. Kaialy et al. / Colloids and Sur

erformance in terms of highest amounts of drug that is poten-ially deliverable to the lower airways with satisfactory flowabilitynd acceptable dose variability.

. Conclusion

The performance of carrier–drug based DPI formulations coulde enhanced when employing engineered lactose carrier presenteds secondary particles composed of aggregates. The crystallisationf lactose particles using different ratios of 1-butanol and acetoneesulted in crystals with smaller particle size with higher amountsf fines, higher �true, higher surface rugosity, lower flow proper-ies, and a better DPI aerosolisation performance than commercialrade lactose. A linear relationship was established between FPFnd carrier �b and �t. In conclusion, this work proved that usinginary combinations of solvents in crystallisation medium is vital

n modification of the physicochemical and micromeritic proper-ies of carriers to achieve a desirable aerosolisation performancerom DPI formulations.

cknowledgments

The authors gratefully appreciate the assistance of Mr. Stevengram (King’s College London) in obtaining image analysis data, Mran Slipper (School of Science, University of Greenwich) in obtainingEM images.

eferences

[1] E. Bronsky, G.A. Bucholtz, J. Allergy Clin. Immunol. 79 (1987) 741.[2] M.P. Timsina, G.P. Martin, C. Marriott, D. Ganderton, M. Yainneskis, Int. J. Pharm.

101 (1994) 1.[3] R.N. Dalby, S.L. Tiano, A.J. Hickey, Lung Biology Health Disease, vol. 94, 1996, p.

441.[4] I.J. Smith, M. Parry-Billings, Pulm. Pharmacol. Ther. 16 (2003) 79.[5] S.P. Newman, W.W. Busse, Respir. Med. 96 (2002) 293.[6] H.W. Frijlink, A.H. De Boer, Dry powder inhalers for pulmonary drug delivery,

Expert Opin. Drug Deliv. 1 (2004) 67–86.[7] I. Ashurst, A. Malton, D. Prime, B. Sumby, Pharm. Sci. Technol. Today 3 (2000)

246.[8] J.S. Patton, J. Bukar, S. Nagarajan, Adv. Drug Deliv. Rev. 35 (1999) 235.[9] H.K. Chan, J. Aerosol Med. 19 (2006) 21.10] P.M. Young, D. Traini, M. Coates, H.K. Chan, Phys. B: Condens. Matter 394 (2007)

315.11] W. Kaialy, G.P. Martin, M.D. Ticehurst, M.N. Momin, A. Nokhodchi, Int. J. Pharm.

392 (2010) 178.12] D. Ganderton, J. Biopharm. Sci. 3 (1992) 101.13] E. Guenette, A. Barrett, D. Kraus, R. Brody, L. Harding, G. Magee, Int. J. Pharm.

380 (2009) 80–88.14] H. Steckel, B.W. Müller, Int. J. Pharm. 154 (1997) 31.15] Y.F. Maa, S.J. Prestrelski, Curr. Pharm. Biotechnol. 1 (2000) 283.16] H.K. Chan, Expert Opin. Ther. Patents 13 (2003) 1333.17] H. Schiavone, S. Palakodaty, A. Clark, P. York, S.T. Tzannis, Int. J. Pharm. 281

(2004) 55.18] H. Larhrib, G.P. Martin, D. Prime, C. Marriott, Eur. J. Pharm. Biopharm. 19 (2003)

211.

19] P. Zanen, L.T. Go, J.W.J. Lammers, Br. Med. J. 51 (1996) 977.20] P. Zanen, L.T. Go, J.W.J. Lammers, Int. J. Pharm. 114 (1995) 111.21] P. Zanen, L.T. Go, J.W.J. Lammers, Int. J. Pharm. 107 (1994) 211.22] O.S. Usmani, M.F. Biddiscombe, J.A. Nightingale, S.R. Underwood, P.J. Barnes, J.

Appl. Physiol. 95 (2003) 2106.

[[[[[

: Biointerfaces 89 (2012) 29– 39 39

23] J. Elversson, A. Millqvist-Fureby, G. Alderborn, U. Elofsson, J. Pharm. Sci. 92(2003) 900.

24] N. Rasenack, Pharm. Dev. Technol. 9 (2004) 1.25] D. Prime, P.J. Atkins, A. Slater, B. Sumby, Adv. Drug Deliv. Rev. 26 (1997) 51.26] W.C. Hinds, Aerosol Technology: Properties, Behavior, and Measurement of

Airborne Particles, John Wiley & Sons, Inc., New York, 1999, p. 3.27] N.M. Ahfat, G. Buckton, R. Burrows, M.D. Ticehurst, Int. J. Pharm. 156 (1997) 89.28] G. Buckton, Adv. Drug Deliv. Rev. 26 (1997) 17.29] V. Berard, E. Lesniewska, C. Andres, D. Pertuy, C. Laroche, Y. Pourcelot, Int. J.

Pharm. 232 (2002) 213.30] M.D. Jones, H. Harris, J.C. Hooton, J. Shur, G.S. King, C.A. Mathoulin, K. Nichol,

T. Smith, M.L. Dawson, A.R. Ferrie, R. Price, Eur. J. Pharm. Biopharm. 69 (2008)496.

31] H. Larhrib, G.P. Martin, C. Marriott, D. Prime, Int. J. Pharm. 257 (2003) 283.32] W. Kaialy, G.P. Martin, M.D. Ticehurst, P. Royall, M.A. Mohammad, J. Murphy,

A. Nokhodchi, AAPS J. 13 (2011) 30.33] W. Kaialy, M.D. Ticehurst, J. Murphy, A. Nokhodchi, J. Pharm. Sci. 100 (2011)

2665.34] H. Steckel, M. Borowski, F. Eskandar, P. Villax, Pharm. Technol. Eur. 16 (2004)

2.35] R.S. Dhumal, S.V. Biradar, A.R. Paradkar, P. York, Pharm. Res. 25 (2008) 2835.36] B.H.J. Dickhoff, A.H. de Boer, D. Lambregts, H.W. Frijlink, Int. J. Pharm. 327 (2006)

17.37] V. Janoo, Special Report 98-1, US Army Corps of Engineers, Cold Regions

Research & Engineering Laboratory, 1998.38] USP Pharmacopeia, The official compendia of standards, USP 26/NF 21; United

States Pharmacopeia Convention, Inc., Rockville, MD, Aerosols, 601, 2003, p.2114.

39] T. Nakate, H. Yoshida, A. Ohike, Y. Tokunaga, R. Ibuki, Y. Kawashima, J. Control.Release 97 (2004) 19.

40] T. Hino, T. Serigano, H. Yamamoto, H. Takeuchi, T. Niwa, Y. Kawashima, Int. J.Pharm. 168 (1998) 59.

41] J. Fiegel, L. Garcia-Contreras, M. Thomas, J. VerBerkmoes, K. Elbert, A. Hickey,D. Edwards, Pharm. Res. 25 (2008) 805.

42] M.S. Hassan, R.W.M. Lau, AAPS PharmSciTech 10 (2009) 1252.43] S.P. Velaga, S. Bergh, J. Carlfors, Eur. J. Pharm. Sci. 21 (2004) 501.44] R.S. Dhumal, S.V. Biradar, A.R. Paradkar, P. York, Int. J. Pharm. 368 (2009) 129.45] P.M. Young, S. Edge, D. Traini, M.D. Jones, R. Price, D. El-Sabawi, C. Urry, C. Smith,

Int. J. Pharm. 296 (2005) 26.46] V.V. Sethuraman, A.J. Hickey, AAPS Pharm. Sci. Tech. 3 (2002) 7.47] A.P. Tinke, A. Carnicer, R. Govoreanu, G. Scheltjens, L. Lauwerysen, N. Mertens,

K. Vanhoutte, M.E. Brewster, Powder Technol. 186 (2008) 154.48] S.L. Raghavan, R.I. Ristic, D.B. Sheen, J.N. Sherwood, L. Trowbridge, P. York, J.

Phys. Chem. B: Condens. Phase 104 (2000) 12256.49] K. Iida, Y. Hayakawa, H. Okamoto, K. Danjo, H. Leuenberger, Chem. Pharm. Bull.

49 (2001) 1326.50] M. Tsukada, R. Irie, Y. Yonemochi, R. Noda, H. Kamiya, W. Watanabe, E.I. Kaup-

pinen, Powder Technol. 141 (2004) 262.51] F. Podczeck, Aerosol Sci. Technol. 31 (1999) 301.52] E. Berlin, P.G. Kliman, B.A. Anderson, M.J. Pallansch, Thermochim. Acta 2 (1971)

143.53] S. Lowell, Characterization of Porous Solids and Powders: Surface Area, Pore

size, and Density, Springer, 2004, p. 326.54] A.L. Grzesiak, M. Lang, K. Kim, A.J. Matzger, J. Pharm. Sci. 92 (2003) 2260.55] K. Iida, H. Leuenberger, L.M. Fueg, R. Müller-Walz, H. Okamoto, K. Danjo, Yaku-

gaku zasshi: J. Pharm. Soc. Jpn. 120 (2000) 113.56] M.D. Louey, M. Van Oort, A.J. Hickey, Pharm. Res. 21 (2004) 1207.57] L. Van Campen, G. Venthoye, Inhalation, dry powder, in: Encyclopedia of Phar-

maceutical Technology, 2nd ed., Marcel Dekker Inc., New York, 2002, p. 1529.58] W. Kaialy, M.N. Momin, M.D. Ticehurst, J. Murphy, A. Nokhodchi, Colloids Surf.

B: Biointerfaces 79 (2010) 345.59] M.A. Braun, R. Oschmann, P. Schmidt, Int. J. Pharm. 135 (1996) 53.60] M.D. Jones, R. Price, Pharm. Res. 23 (2006) 1665.61] D. Cline, R. Dalby, Pharm. Res. 19 (2002) 1274.62] C. Bosquillon, C. Lombry, V. Preat, R. Vanbever, J. Pharm. Sci. 90 (2001) 2032.

63] P.K. Gupta, C.T. Hung, F.C. Lam, J. Microencapsul. 6 (1989) 147.64] Y.M. Wang, H. Sato, I. Adachi, I. Horikoshi, J. Pharm. Sci. 85 (1996) 1204.65] E.E. Hassan, R.C. Parish, J.M. Gallo, Pharm. Res. 9 (1992) 390.66] W. Stahlhofen, J. Gebhart, J. Heyder, Am. Ind. Hyg. Assoc. J. 41 (1980) 385.67] J.P. Mitchell, M.W. Nagel, J. Aerosol Med. 16 (2003) 341.