Mathematical approach for understandingdeagglomeration behaviour of drug powder in formulations with coarse carrier

Irene Parisini,James L.Collett,Darragh Murnane*

aDepartment of Pharmacy,University of Hertfordshire,Hat fi eld,AL10 9AB,UK

bPharmaceutical&Preclinical Development Department,Aptuit Srl,Verona,Italy

cScience&Technology Research Institute,University of Hertfordshire,Hat fi eld,AL10 9AB,UK

Original Research Paper

Mathematical approach for understanding
deagglomeration behaviour of drug powder in formulations with coarse carrier

Irene Parisinia,b,James L.Collettc,Darragh Murnanea,*

aDepartment of Pharmacy,University of Hertfordshire,Hat fi eld,AL10 9AB,UK

bPharmaceutical&Preclinical Development Department,Aptuit Srl,Verona,Italy

cScience&Technology Research Institute,University of Hertfordshire,Hat fi eld,AL10 9AB,UK

ARTICLE INFO

Article history:

Received 31 May 2015

Received in revised form 17 August 2015

Accepted 18 August 2015

Available online 24 August 2015

Ternary agents

Deagglomeration of cohesive particles in combination with coarse carrier is a key requirement for inhaled formulations.The aim of the project was to propose a mathematical approach to understand aerosolization behaviour of micronized particles alone and in formulation with carriers.Salbutamol sulphate and salmeterol xinafoate were blended separately with fi ne lactose(ratio 1:4)and fi ne and coarse lactose(1:4:63.5).Laser diffraction was employed to characterize the powder median particle size.The deagglomeration of micronized materials followed an asymptotic monoexponential relationship.When the coarse lactose was added,the relationship fi tted a bi-exponential equation showing an easily and a poorly dispersed fraction.Using model hydrophobic and hydrophilic APIs,this study has demonstrated the utility of an analytical approach that can parameterize deagglomeration behaviour of carrier-free and carrier-based inhalation formulations.The analytical approach provides the ability to systematically study the effect of material,formulation and processing factors on deagglomeration behaviour.

©2015 The Authors.Production and hosting by Elsevier B.V.on behalf of Shenyang Pharmaceutical University.This is an open access article under the CC BY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Micronized particles for inhalation with size between 1 and 5 μm form agglomerates due to the dominance of interparticulate forces.Several dry powder formulations have been marketed without the presence of carrier particles[1,2].The absence of the coarse particle affects the performance of formulations that contain extensive agglomeration[1,2].Budesonide(Bud),for example,is more cohesive than salbutamol sulphate(SS), showing bigger agglomerates under the SEM(scanning electron microscope)[3].It has been suggested[3-5]that FP ( fl uticasone propionate),like SX(salmeterol xinafoate)and Bud, shows greater agglomeration forces than SS.However,the formation of agglomerates of micronizedAPIs leads to an increased entrainment of the drug particles but poor blend fl owability.Toaddress the manufacturing challenge of cohesive agglomerates,drug particles are often blended with coarse lactose carrier to improve fl owability,uniformity of dosing and aerosolization.Many drug particles remain agglomerated in such carrier blends.During aerosolization,particles must be removed from the carrier particles and agglomerates must be disrupted in the air fl ow[6,7].

Micronized material(e.g.FP( fl uticasone propionate))shows highly-cohesive behaviour,forming big agglomerates,especially when carrier is added(median agglomerate size 40.15±6.02 at 90 L min-1)[8].The authors[8]mixed coarse and fi ne lactose with the most cohesive batch of FP and the formulation generated a lower fi ne particle fraction than when the coarse carrier was not used.The authors[8]suggested that it was due to the presence of very large agglomerates of FP in the mixtures.Different types of blending,such as high-or lowshear blending,lead to the re-arrangement of the API on the surface of the carrier.The mechanism of re-distribution of the API on the carrier particles differs from drug to drug(including batch-to-batch variability of the same drug),and depends on how great the cohesive force between agglomerates is[9]. Although all micronized particles are cohesive,when coarse carrier such as lactose is added to the formulation,the micronized material starts to redistribute on the lactose surface exhibiting very different behaviour[8,10,11].When energy is applied to the formulation(such as inhalation strength of the patient),the cohesive forces between agglomerates would be overcome and they would be released together with the fi ne material for lung deposition.

Salbutamol sulphate(SS),which has been widely investigated[10,12,13],presents a rectangular or a plate-like shape when analysed under the SEM[13].Other studies[10,12]suggested that SS exhibits an adhesively balanced behaviour when added to coarse lactose and redistributes evenly on the lactose surface,with lower adhesive energy to the carrier with respect to other particles[10].This is probably due to the relative balance of cohesive and adhesive interparticulate forces in binary(and ternary)formulations with carriers.SS showed a lower cohesive energy compared to Bud,with SX also reported to exhibit a balance in favour of cohesion when added to lactose[10,14,15]. The interaction between lactose and drug particles has to be strong enough to prevent segregation.However,upon aerosolization,the drug must be easily detached to allow lung deposition.

The design of an inhaler and the inhalation strength of the patient affect the geometries and forces available for powder entrainment and drug dispersion.The forces of interaction within the powder formulation additionally play a role in the aerosolization.Studies have demonstrated that the strong adhesion between API and coarse carrier decreased the drug dispersion[11,16].Enhancement of the dispersion process(e.g. aerosol fraction of fi ne particles)and modi fi cation of interparticulate forces within the powder formulation by the presence of ternary agent have been suggested[17,18].Adi et al. [11]showed that adding fi ne lactose improved the dispersion process of SX from the coarse carrier,because the interaction between the fi ne lactose and the SX produced mixed agglomerates.It may be the decreased drug/carrier adhesion energy or the easier break-up of mixed agglomerates with low carrier adhesion that enhances dispersion ef fi ciency.

It has been suggested that the addition of fi ne lactose would increase the respirable fraction[19],whilst the presence of coarse lactose improves the fl owability of the formulation. However,Behara et al.[20]showed that the presence of coarse carrier is not necessary.The aerosolization was improved by doubling the ratio of fi ne lactose to SS(from 1:4 to 1:8).They suggested that increasing the concentration of fi ne lactose reduced the agglomerate strength between SS and fi ne lactose [20].Studying the impact of drug type or the properties of coformulated excipients on deagglomeration with a focus on blend segregation into drug or drug-excipient agglomerates represents a‘blend structure’treatment of deagglomeration.However, focusing blend segregation(which can be assessed through studying the effects of content uniformity on dispersion during aerosolization[21])ignores the fundamental causes of blend structure(or restructuring).These include the effects of the order of material addition in blending[22],of electrostatic charge[23] and tribo-electri fi cation during blending,the blending energy and duration of blending[21,24]that are employed,and of course inhaler device design itself and tribo-electri fi cation[25] behaviour of the aerosol that is being dispersed from the device. However,it remains of merit to consider the resultant interparticulate forces between formulation components,the balance of the agglomerative and deagglomerative forces applied during blending with respect to eventual deagglomeration performance of the blend.The adhesion forces between drug and carrier consist mainly of inter-particulate forces such as van der Waals forces,electrostatic charges,and capillary interactions and in fl uence aerosol dispersion from dry powder inhalers (DPIs)[26].

Inverse gas chromatography,centrifugation and atomic force microscopy have been used to characterize adhesion forces between particles[27,28].However,such techniques do not quantify the cohesion of an actual powder bed as it only measures the interaction between one drug or excipient particle and the probe of the AFM.

Other studies investigated the interaction between individual particles using atomic force microscopy(AFM)[27,28] and suggested CAB(cohesive-adhesive balanced)approach.CAB approach calculated the ratio between API agglomerates and adhesion of API to the carrier.It suggests that CAB ratio＜1 the drug is more adhesive to the carrier than to itself,whilst a CAB＞1 indicates that more drug agglomerates can be found in the formulation,leading to a variable content uniformity[5]. However,the technique does not quantify the adhesion force of the powder bed as it only measures the interaction between one particle or excipient and the probe of the AFM.Other techniques have been proposed to study the interaction between particles.For instance,inverse phase gas chromatography[29] showed potential to understand the interaction between carrier and API,where the formulation is placed inside a packed column and a gas(e.g.helium)is passed through it.The API with higher af fi nity to the carrier would elute later than the one with less af fi nity.This is due to the higher surface energy between API particles and carrier.Studying the retention of a range of probes with different polarity and chemistry,the cohesion strength can be calculated[29].Other techniques to study the different behaviour of blends under applied forces are powder rheometry(where fl ow properties are studied by studying the movement of an impeller through blends under aspeci fi ed force,such as torque)[30]and centrifugation of micronized material[31].

Laser diffraction techniques have been used for dry and liquid dispersion systems to study powder deagglomeration[32]. Air fl ow and pressure drop titration experiments have been performed[4,32]to assess the change in particle size when a fi ne powder bed is subjected to an increasing pressure drop and empirical models have been proposed for micronized particles with or without the common carrier coarse lactose [4,32,33].

When only micronized APIs are present in the formulation alone or in presence of fi ne lactose,studies have suggested that the liberation of fi ne particles＜6 μm shows a sigmoidal trend with increasing air fl ow rate and a three-parameter sigmoidal equation has been used[32].The latter relationship suggested that parameters such as maximum extent of deagglomeration or fl ow rate required to achieve 50%deagglomeration could be used to predict powderdeagglomeration.However,theproblem occurs when the coarse lactose is added to the formulation. Higher fl ow rates are required to deaggregate the mixture fully than if the lactose carrier were not present.Therefore,more energyisrequiredtofullyaerosolizemicronizedAPIsawayfrom the surface of the carrier.An equation similar to the Langmuir adsorption isotherm was proposed with the name of Powder Aerosol Deaggregation Equation[33].However,the authors suggested the use of a cascade impactor together with a dry powder inhaler for the experiments that would affect the deagglomeration of the particles.The use of cascade impaction is not a rapid method to guide formulation choice in early phases of product development.Previous studies have indicatedthebene fi tsofarapidmethodologyusinglaserdiffraction to guide formulation of DPIs.However,the models proposed to explain the deagglomeration process did not show any parameter with a physical meaning.The aim of this study was to develop a data modelling approach to characterize and parameterize powder deagglomeration response to air fl ow for cohesivemicronizedparticlesaloneorinformulationwithcoarse and fi necarrier.Thiswouldbehelpfultopredictdeagglomeration mechanisms of novel formulations or following different pharmaceutical processing.

2. Materials and methods

2.1. Materials

Micronized salbutamol sulphate(SS,batch number B027798) was obtained from GlaxoSmithKline Research and Development(Ware,UK).Salmeterol xinafoate(SX)was purchased from Vamsi Labs Ltd(Solapur,India).Lactose monohydrate,Span 80 and cyclohexane were purchased from FisherScienti fi c(Loughborough,UK).Fine lactose(LH300)was donated from Friesland Foods(The Netherlands).Adhesive carbon tabs and aluminium pin stubs were purchased fromAgar Scienti fi c Ltd(UK).

2.2. Blend preparation

Lactose monohydrate(Fisher Scienti fi c,UK)was sieved to separate the 63-90 μm particle fractions using a mechanical sieve shaker(model-AS 200 digit,Retsch,Germany)with the 90 μm sieve and the 63 μm sieve.It was operated for 5 min at 100 amplitudes.The fraction of lactose between the sieves was collected to prepare the blends.A total of 3 g of blends were prepared. The drugs were blended separately with either fi ne lactose(FL) in ratio 1:4,coarse lactose(CL)in ratio 1:67.5 or combination of both in ratio 1:4:63.5.The drug was sandwiched with coarse lactose(CL)in equal volumes.When the FL was used,it was fi rst blended with the drug and the pre-mixture sandwiched with coarse lactose.Glass vials were used for the blending and vortexed(Vortex Genie 2,model G-560E,Scienti fi c Industries LTD,New York)for 2 min.The blending was carried out in a Turbula mixer(Turbula 2583,type+2C,Glen Mills,Clifton) for 40 min.

2.3. Particle size analysis

Salbutamol sulphate(SS), fi ne lactose(FL),salmeterol xinafoate (SX)and coarse lactose(CL)were tested for the homogeneity of the particle size using the laser diffraction technique.The Sympatec Rodos module(Inhaler Helos/KF,Sympatec Limited, Bury,UK)with Aspiros feeder was used.The speed of the feeder was set at 25 mm/s.The pressure drop was set between 0.1 and 5.0 Bar in order to titrate the dispersion pressure.The PSD(particle size distribution)was collected every 0.1 Bar when tested from 0.1 to 1.0 Bar and every 1.0 Bar when tested from 1.0 to 5.0 Bar.The duration of the measurements was 5 s.The optimum concentration of the powder to be detected was in the range of 1.0 and 1.1%.The technique was used to detect the particle size of the blend as a dry dispersion.

A liquid laser diffraction technique using a Malvern Mastersizer X(Malvern Instruments Ltd,UK)was also used to determine the‘true’fully dispersed particle size.The blends were suspended in 0.5%(w/v)Span 80 in cyclohexane(for SX blends,SX,FL,CL)and 1%(w/v)Span 80(for SS blends and SS). Malvern Mastersizer X(Malvern Instruments Ltd,UK)was used with 100 mm focal length lenses(0.5-180 μm)and MS7 magnetically stirred cell.Prior to measurement,the solvents were saturated with the appropriate particles and sonicated for 30 min followed by overnight stirring.For the measurements, approximately 1 mg of powder was added to 2 mL fi ltered dispersant(0.2 μm cellulose acetate syringe fi lter,Sartorius Stedim Biotech.,UK)and sonicated(Sonicleaner,DAWE,Ultrasonics Ltd, USA)for 1 min for SS,FL and CL and 5 min for SX.The stirring was set at 3 min for all the powders with sweeps of 2500 for SX and 3500 for the other powders.A background reading was taken and the suspension was added to the sample cell until the obscuration was 10-30%.Following equilibration(60 s for SX and 30 s for the others),ten individual measurements were taken for n=3 samples.From the instruments,Dv10,Dv50, and Dv90(corresponding to the cumulative percentage particle undersize values for 10%,50%and 90%of the particles by volume),and the%volume＜5 μm were used for analysis.

2.4. Particle morphology assessment

Powder samples were transferred to adhesive carbon tabs, mounted onto aluminium pin stubs(Agar Scienti fi c Ltd, England).To assess blend interactions,scanning electron microscopy(SEM,Jeol Carry Scope JCM 5700,Welwyn Garden City,UK)was performed.Samples were sputter coated with gold under argon for 4 min to achieve a thickness of approx.30 mm using an Emitec SC 7620 coater(QuorumTechnologies Limited, West Sussex,England).To view the particle morphology,the SEM was operated at 15 kV in low vacuum mode and a working distance of 13 mm.

2.5. Equation of deagglomeration and mathematical model

The%relative deagglomeration for micronized particles was calculated using the model proposed by Behara et al.[32]in Equation(1).

where%RD is the relative deagglomeration.

2.6. Statistical analysis

Statistical analysis was performed in Minitab(version 15)using one-way ANOVA and post-hocTukey’s test(multiple comparisons)or Student’s two-tailed t test for pair-wise comparisons, both at 95%con fi dence intervals.Non-linear regression analysis of the powder dispersion data was performed using OriginPro(ver.8)and MatLab software(ver.R2013a).

3. Results and discussion

The aim of the current study was to propose a model in order to understand the deagglomeration mechanisms of micronized material for inhalation in combination with fi ne and/or coarse carrier.The design of an inhaler and the inhalation strength of the patients allow the dispersion of inhaled formulations and lung deposition.However,upon aerosolization,the API(active pharmaceutical ingredient)might deaggregate in a different manner depending on the presence or absence of a coarse carrier or also on its distribution on the lactose surface prior to release from the inhaler[4].Coarse lactose is well known to be used to improve aerosolization and fl owability of micronized drug[12,34].However,a strong adhesion of the API to the carrier would decrease the drug dispersion[11,16].Therefore, ternary agents(or fi ne lactose)are usually added to the formulation to compete with the drug on the active sites of the carrier. Saturating high adhesive-energy surface sites,the drug can be detached easily from the carrier during aerosolization[34,35]. Behara et al.[20]showed that the presence of coarse carrier was not necessary to improve aerosolization.The addition of lactose fi nes in increasing concentration successfully reduced the cohesive strength of drug agglomerates,leading to better aerosolization.Moreover,Islam et al.[36]indicated that SX dispersion increased as the amount of FL increased.Fine lactose and cohesive micronized particles created agglomerates[20].

Using laser diffraction techniques,the effect of the addition of either fi ne lactose or coarse lactose to theAPI was shown [36].On the other hand,adding coarse carrier to the mixture shifted the distribution to a larger size range[36].Studying the deagglomeration process of the dispersed powder at different fl ow rates has featured widely in the literature.A few studies have investigated the deagglomeration trend of either the powder bed[4]or the aerosol emitted from a Rotahaler[32]. These studies proposed a quick methodology to assess deagglomeration of the formulation upon increasing fl ow rate or pressure drop.Laser diffraction analyses such as Sympatec and Malvern instrument[4,32]were used.However,the studies seem to focus on the fi ne particles,rather than on the formulation with coarse carrier.

3.1. Laser diffraction analysisIn order to understand the deagglomeration mechanism,blends with either SX or SS were manufactured in-house with different grades of lactose and their aerosolization was studied using laser diffraction techniques.All formulations were regarded uniformity(%CV＜10%);however,content uniformity was poorer for blends with fi ne lactose(Table 1).For the purpose of the study and to understand the different deagglomeration mechanism of SX and SS,the former was considered cohesivelybalanced when added to the lactose,due to its tendency to create agglomerates on the carrier surface[15].On the other hand,SS was considered due to its adhesively balanced behaviour when added to lactose as previously described[12]. The concentration of the SS and SX was 1.48%w/w and they were blended with either FL(1:4),coarse lactose(1:67.5)or both (1:4:63.5).

Table 1-Particle size distribution for pure material and blends of salbutamol sulphate(SS)and salmeterol xinafoate(SX) with lactose using liquid dispersion laser diffraction(mean±SD,n=3)and content uniformity for the inhalation blends (coef fi cient of variance(%CV)of n=6 determinations).FL= fi ne lactose,CL=coarse lactose.

When the APIs were dispersed using the Sympatec/RODOS for pressure drop titration,both SX and FL showed a bi-modalPSD at 0.5 Bar(Fig.1A and 1B).The same trend was seen also for SS that exhibited a similar behaviour at 1 Bar(Fig.1C),suggesting a greater energy is required for the deagglomeration mechanism to occur for SS compared to SX and FL.However SS,SX and FL showed a bell-shaped distribution at higher pressure drops as con fi rmed in previous studies[22,37].When FL was added to the APIs,a more consistent distribution was seen for both drugs when the pressure was increased.However,SSFL deagglomeration at low pressure(0.5 Bar)did not occur completely(shoulder in the large particle size range in Fig.2A). The same trend was seen by Jaffari et al.[4]with SX,and by Behara et al.for SS:FL[32].However,the bi-modal distribution was not present for SS:FL at 1 Bar unlike for SS alone, probably due to the addition of the FL leading to disruption of SS agglomerates and enhanced deagglomeration[20].

When the formulations containing either CL or FL:CL were analysed,the lactose carrier showed a distribution in the 60-90 μm range(Fig.3A).However,in blends also containing FL,a shoulder was seen in the micronized region due to a higher amount of fi ne particles present in the mixture(Fig.3B).The formulations showed a bi-modal distribution due to the presence of easily-dispersible agglomerates(small shoulder in the respirable range)and a poorly-dispersible fraction of drug still attached to the carrier(large shoulder)[4,35].

Liquid dispersion laser diffraction was used to determine the fully dispersed particle size of the API and blends.The results con fi rmed the dry dispersion trend of the blends as the Dv50 of the blends with the carrier were shifted to higher values than when the drugs were blended with fi ne lactose only (Table 1).On the other hand,the micronized API possessed a PSD within the micron size range(Table 1).

3.2. Empirical model

The DV50was plotted for all micronized material and formulations(Fig.4A)and the descending part of the graph prior the plateau shows different deagglomeration behaviour between particles and formulations(e.g.SS vs.SX and FL in Fig.4A).The difference,especially at low pressure drops,suggests variability in agglomerate size and strength as suggested from the density distributions of size in Fig.1.The%relative deagglomeration(equation 1)was also plotted(Fig.4B)and SX reached the maximum deagglomeration earlier than SS,which on the other hand,did not reach a constant particle size at the highest pressure(Fig.4B).When FL was added to both APIs,an increase in the degree of deagglomeration was seen at 2 Bar, which then decreased at 3 Bar(Fig.4B).This was probably due to the instability of the formulation at high pressure drops.

The%relative deagglomeration for micronized particles and micronized formulations showed a mono-exponential trend when a pressure drop was applied to the agglomerates.The trend could be explained by Equation(2)which is a modi fi cation of the Johnson-Mehl-Avrami-Kolmogorov(JMAK)equation [38]:

Due to the high variability in the replicate measurements, an arti fi cial error was added when normalizing the data.

The deagglomeration mechanism depends on the distribution of the APIs either alone or in combination with FL in formulation.The slope of the curve(Table 2)is the maximum gradient and represents how readily the deagglomeration process occurs before the drug reaches the plateau size (deagglomeration sensitivity).The plateau size represents the maximum deagglomeration(“m”parameter in Table 2 de fi nes the positive asymptote of the curve)which varies with the type of particle(Table 2).This was in accordance with Jaffari et al. [4],where formulations with fi ne particles reached a monoexponential plateau when the pressure drop was increased.SS showed the lowest maximum gradient(indicating a deagglomeration process that is resistant to increased air fl ow) when pressure is applied compared to the other micronized materials(Table 2,P＜0.05).The difference,especially at low pressure drops,suggests variability in agglomerate size(i.e.x0 in Equation 2)and strength between agglomerates[5,14].In order to compare the drugs and blends,the critical pressure point(CPP),the pressure point corresponding to reaching 95% of the maximum deagglomeration,was calculated.SS started to reach the plateau at 1.25±0.19 Bar,at a much greater pressure than SX or FL(Table 2,P＜0.05).However,when FL was added to SS,the CCP95was reached at lower pressure than when SS was analysed alone(0.84±0.08 Bar,P＜0.05)and the greater maximum gradient was seen for FL than SS(Table 2,P＜0.05). The maximum gradient is a vector of the derivative of the function.This indicates the rate of ascent of the curve, therefore,the rate of the degree of deagglomeration.If a small tangent is drawn in a selected point(in this case x0;y0), then the maximum gradient refers to the slope of the tangent.

The enhanced degree of deagglomeration seen for SS due to addition of FL could be explained by creation of mixed SSFL agglomerates of lower cohesive strength than SS-agglomerates[20,32].Therefore,the cohesion forces between API(or API-FL)agglomerates are reduced.SX:FL had a poor deagglomeration compared to FL or SX,suggesting that the addition of FL to very cohesive agglomerates does not improve the deagglomeration mechanism[39].In accordance with this theory,Adi et al.[15]showed that increasing the shear pressure to formulations of SX containing FL with different size range(3.0 and 7.9 μm)was required especially for formulation containing FL 3.0 μm.This might be a result of segregation in the formulation.The parameters‘x0’and‘n’are interdependent.The parameter‘x0’has a similar meaning to the parameter proposed by Behara et al.[32].The authors used a Rotahaler with Spraytec to study the deagglomeration behaviour of either SS or FL.They suggested that the parameter was the dispersion energy required to overcome internal interaction of the agglomerates.In the current work we suggested that‘x0’could be seen as an in fl ection point and therefore the pressure at which the deagglomeration attempts to reach the maximum point.In accordance with Behara et al.[32],the blends with fi ne particles and FL alone showed similar‘x0’values(Table 2,P＞0.05)with the exception of SS.The parameter‘n’is the trend of the fi rst part curve between 0.1 and x0 Bar.SS:FL exhibited the highest deagglomeration exponent whilst SX:FL showed the smallest.This indicates a different deagglomeration mechanism when the fi ne lactose overcomes SS-SS and SX-SX interactive forces.Moreover,the blendstructure plays a role.Adding the fi ne lactose would modify the adhesive strength between particles and enhance the detachment of the API(Table 2,maximum gradient was greater for SS:FL than SX:FL).

When CL was added to the formulation,the plateau was not reached for the blends containing CL.Instead,the formulations showed a bi-exponential trend(Fig.5)highlighting an easily dispersed fraction and a second poorly dispersible fraction of respirable particles.This response is described by Equation(3).

‘A’represents the amplitude of the easily dispersed fraction(i.e.,extent of deagglomeration),whereas 100-‘A’represents the deagglomeration extent of the poorly dispersed fraction;‘xw’is the pressure when the poorly dispersible fraction starts to deagglomerate;‘H’is the Heaviside-function(1 if x＞xw,0 if x＜xw);x1and x2are pressure points when the deagglomeration response is at maximum for the easily and poorly dispersed fractions,respectively.

The bimodal distribution of certain blends was attributed to the presence of coarse carrier in the formulation that,when pressure was applied led to a release of easily dispersible agglomerates( fi ne particles)and a poorly dispersible fraction of large cohesive agglomerates.The trend of the latter seemed to be the same for all the formulations.However,differences were observed in the initial exponential region of the deagglomeration curves of the blends.SS:CL seemed to have the steepest curve compared to the other API:CL blends.As expected,API:FL:CL blends showed a steeper slope in the fi rst fraction than API:CL blends(Fig.5).They showed also a greater %of deagglomeration expressed by the parameter“A”in equation(3).SX:FL:CL for example showed 170.97±12.74 bar-1and SX:CL presented 71.64±7.42 bar-1.At high pressures there was an increase in the release of fi ne particles.The presence of aternary agent would help the detachment of the micronized API from the surface of the carrier,resulting in greater respirable fraction for the patient than when the formulation was composed of CL only[39](higher‘A’parameter,Table 3).The greatest‘A’-value was observed for SS:FL:CL(Table 3).Jones et al. [39]also con fi rmed improved detachment when alternative ternary agents were added to binary formulations by showing an increased mode in the respirable fraction for Bud ternary erythriol-lactose formulations.

Equation(3)shows the Heaviside function(H)which is usually used in engineering to represent a parameter that changes abruptly at speci fi ed values of time t.In the current model the step function(or Heaviside function)was applied to the bi-exponential trend.A small plateau can be observed in Fig.5,between 1.0 and 2.0 Bar,between the end of the deagglomeration of the easily dispersed fraction and the beginning of the deagglomeration of the poorly dispersed fraction [40].‘H’was applied to the deagglomeration mechanism of the blends with coarse carrier,where the“H”parameter of the function was 1 if x＞xw,with xwbeing the pressure when the poorly dispersed fraction starts to deagglomerate(the delayed unit), or 0 if x＜xw.All the formulations with the exception of SX:CL showed the plateau between the easily and the poorly dispersed fraction,with SS:CL showing the highest pressure value. This suggested that this particular formulation required the highest pressure to deagglomerate.The addition of FL led to a greater deagglomeration as the pressure(xw)decreased by 1 unit.Furthermore,the corresponding%deagglomeration to xw was the highest between all the formulations(Table 3).

3.3. Particle morphology

Microscopy was employed to examine blend structure.Unlike what some studies have suggested[4,10],SS did not exhibit a plate-like shape(Fig.6A),whereas both APIs created elongated agglomerates without clear differentiation in their shape(Fig.6).Moreover,when added to coarse lactose,no difference in the API distribution was seen on the surface of the carrier(Fig.7).This particle behaviour is different to what was proposed previously in the literature.Begat et al.suggested that SS distributes evenly on the surface of the carrier[10].SX created agglomerates due to the fact that SX-lactose blends are dominated by SX cohesive interactions[11].However,this was not shown in the current work.Nevertheless,different redistribution behaviour of the two differentAPIs was clearly seen when FL was added to the API-CL mixtures(Fig.8).The addition of FL to SX:CL(Fig.8B)led to more agglomerates of FLSX being heterogeneously distributed on the lactose surface, presenting larger agglomerates on the extremities of the carrier compared to when FL was absent in the formulation.This was probably due to the weak adhesion of FL to the carrier surface, and therefore,its tendency to agglomerate with SX[35].This suggests that SX-FL have stronger adhesion than FL:FL.Adi et al. [35]showed that when the concentration of fi nes(i.e.SX and FL)was increased,a better detachment of agglomerates from the lactose was observable because of saturation of the activesites on the carrier.This led to a better interaction between SX and FL causing agglomeration and surface detachment.The authors proposed that this was due to the fact that adhesion force between SX and CL was lower than SX-SX and SX-FL[35]. This potentially shows the strong cohesive behaviour of SX particles that is likely to create agglomerates.

Table 2-Mono-exponential deagglomeration parameters for salbutamol sulphate(SS),salmeterol xinafoate(SX), fi ne lactose(FL)and SS-FL or SX-FL blends.

The same tendency to create agglomerates was seen for FLSS,and“free”agglomerates of FL:SS were visible in the space between carrier particles(Fig.8B).When energy is applied,these agglomerates would be dispersed more easily compared to those adhered to the surface of the carrier(e.g.in SX:FL:CL),leading to the higher%deagglomeration for the SS:FL:CL compared to SX:FL:CL(“A”parameter,Table 3).In fact,the maximum gradient value for SS:FL(Table 2)was greater than SX:FL,suggesting that the aerosolization of SS was enhanced in presence of the FL.However,the adhesive strength predicted(Table 3)of the CL-SS interactions in the agglomerates would require high pressure to achieve full deagglomeration(xw=3.0±0.3 for SS:CL in Table 3).

4. Conclusion

The study revealed a rapid technique to characterize the deagglomeration behaviour of micronized particles during formulation development.The micronized particles showed different deagglomeration processes to each other suggesting variability in agglomerate size and strength between agglomerates.The empirical parameterization was descriptive of agglomerate and powder structure and deagglomeration mechanisms.The micronized particles(including one model hydrophobic API and one model hydrophilic API)showed different deagglomeration processes to each other suggestingvariability in agglomerate size and strength between agglomerates.Co-formulating with a coarse carrier and fi ne lactose did modify the deagglomeration process suggesting that two different types of fractions of dispersed particles can be identi fi ed when increasing dispersing air fl ow.Moreover,SEM studies suggested a greater number of free agglomerates for SS:FL:CL than SX:FL:CL,leading to higher dispersion of the micronized material.Inhaler devices play a major role in the aerosolization of the formulation.Previous powder deagglomeration models have been proposed with speci fi c devices(such as the Rotahaler)and additional work is required to expand our understanding of the implications of adding lactose(or other excipients)for the aerosolization of blends when used in other device platforms.However,we believe that the outcomes of the current study(albeit using a limited range of API and excipient materials and low shear blending)provide for good starting point for future rapid screening experiments into the effects of multiple material,formulation,blend and processing properties and provide functionally-relevant parameters of deagglomeration performance for DPI blends.

Table 3-Parameters from a bi-exponential equation for salbutamol sulphate(SS)blends(FL= fi ne lactose,CL=coarse lactose)and salmeterol xinafoate(SX)blends.

Acknowledgements

The authors would like to acknowledge the fi nancial support of the University of Hertfordshire and the Society of Electron Microscopy Technology.

REFERENCES

[1]Wetterlin K.Turbuhaler:a new powder inhaler for administration of drugs to the airways.Pharm Res 1988;5:506-508.

[2]Martin GP,MacRitchie HB,Marriott C,et al.Characterisation of a carrier-free dry powder aerosol formulation using inertial impaction and laser diffraction.Pharm Res 2006;23:2210-2219.

[3]Begat P,Morton DA,Shur J,et al.The role of force control agents in high-dose dry powder inhaler formulations.J Pharm Sci 2009;98:2770-2783.

[4]Jaffari S,Forbes B,Collins E,et al.Rapid characterisation of the inherent dispersibility of respirable powders using dry dispersion laser diffraction.Int J Pharm 2013;447:124-131.

[5]Jones MD,Harris H,Hooton JC,et al.An investigation into the relationship between carrier-based dry powder inhalation performance and formulation cohesive-adhesive force balances.Eur J Pharm Biopharm 2008;69:496-507.

[6]Zeng XM,Martin GP,Marriott C,et al.Lactose as a carrier in dry powder formulations:the in fl uence of surface characteristics on drug delivery.J Pharm Sci 2001;90:1424-1434.

[7]Hooton JC,Jones MD,Price R.Predicting the behavior of novel sugar carriers for dry powder inhaler formulations via the use of a cohesive-adhesive force balance approach.J Pharm Sci 2006;95:1288-1297.

[8]Le VN,Robins E,Flament MP.Agglomerate behaviour of lf uticasone propionate within dry powder inhaler formulations.Eur J Pharm Biopharm 2012;80:596-603.

[9]Tsukada M,Irie R,Yonemochi Y,et al.Adhesion force measurement of a DPI size pharmaceutical particle by colloid probe atomic force microscopy.Powder Technol 2004;141:262-269.

[10]Begat P,Morton DAV,Staniforth JN,et al.The cohesiveadhesive balances in dry powder inhaler formulations II: in fl uence on fi ne particle delivery characteristics.Pharm Res 2004;21:1826-1833.

[11]Adi H,Larson I,Stewart PJ.Adhesion and redistribution of salmeterol xinafoate particles in sugar-based mixtures for inhalation.Int J Pharm 2007;337:229-238.

[12]Kaialy W,Ticehurst M,Nokhodchi A.Dry powder inhalers: mechanistic evaluation of lactose formulations containing salbutamol sulphate.Int J Pharm 2012;423:184-194.

[13]Kaialy W,Martin GP,Larhrib H,et al.The in fl uence of physical properties and morphology of crystallised lactose on delivery of salbutamol sulphate from dry powder inhalers.Colloids Surf B Biointerfaces 2012;89:29-39.

[14]Begat P,Morton DAV,Staniforth JN,et al.The cohesiveadhesive balances in dry powder inhaler formulations I: direct quanti fi cation by atomic force microscopy.Pharm Res 2004;21:1591-1597.

[15]Adi H,Larson I,Chiou H,et al.Agglomerate strength and dispersion of salmeterol xinafoate from powder mixtures for inhalation.Pharm Res 2006;23:2556-2565.

[16]Kawashima Y,Serigano T,Hino T,et al.A new powder design method to improve inhalation ef fi ciency of pranlukast hydrate dry powder aerosols by surfacemodi fi cation with hydroxypropylmethylcellulose phthalate nanospheres.Pharm Res 1998;15:1748-1752.

[17]Louey MD,Razia S,Stewart PJ.In fl uence of physico-chemical carrier properties on the in vitro aerosol deposition from interactive mixtures.Int J Pharm 2003;252:87-98.

[18]Young PM,Price R,Tobyn MJ,et al.The in fl uence of relative humidity on the cohesion properties of micronized drugs used in inhalation therapy.J Pharm Sci 2004;93:753-761.

[19]Jones MD,Price R.The in fl uence of fi ne excipient particles on the performance of carrier-based dry powder inhalation formulations.Pharm Res 2006;23:1665-1674.

[20]Behara SR,Kippax P,McIntosh MP,et al.Structural in fl uence of cohesive mixtures of salbutamol sulphate and lactose on aerosolisation and de-agglomeration behaviour under dynamic conditions.Eur J Pharm Sci 2011;42:210-219.

[21]Le VN,Hoang Thi TH,Robins E,et al.Dry powder inhalers: study of the parameters in fl uencing adhesion and dispersion of fl uticasone propionate.AAPS PharmSciTech 2012;13:477-484.

[22]Zeng XM,Martin GP,Tee SK,et al.Effects of particle size and adding sequence of fi ne lactose on the deposition of salbutamol sulphate from a dry powder formulation.Int J Pharm 1999;182:133-144.

[23]Staniforth JN.Performance-modifying in fl uences in dry powder inhalation systems.Aerosol Sci Technol 1995;22:346-353.

[24]Grasmeijer F,Hagedoorn P,Frijlink HW,et al.Mixing time effects on the dispersion performance of adhesive mixtures for inhalation.PLoS ONE 2013;8:1-18.

[25]Telko MJ,Hickey AJ.Aerodynamic and electrostatic properties of model dry powder aerosols:a comprehensive study of formulation factors.AAPS PharmSciTech 2014;15:1378-1397.

[26]Zeng XP,Martin GP,Marriott C.Particulate interaction in dry powder formulation for inhalation.Taylor&Francis;2001.p. 1-272.

[27]Louey MD,Stewart PJ.Particle interactions involved in aerosol dispersion of ternary interactive mixtures.Pharm Res 2002;19:1524-1531.

[28]Young PM,Tobyn MJ,Price R,et al.The use of colloid probe microscopy to predict aerosolization performance in dry powder inhalers:AFM and in vitro correlation.J Pharm Sci 2006;95:1800-1809.

[29]Behara SR,Larson I,Kippax P,et al.An approach to characterising the cohesive behaviour of powders using a fl ow titration aerosolisation based methodology.Chem Eng Sci 2011;66:1640-1648.

[30]Xu Z,Mansour HM,Mulder T,et al.Dry powder aerosols generated by standardized entrainment tubes from drug blends with lactose monohydrate:1.Albuterol sulfate and disodium cromoglycate.J Pharm Sci 2010;99:3398-3414.

[31]Buckton G.Characterisation of small changes in the physical properties of powders of signi fi cance for dry powder inhaler formulations.Adv Drug Deliv Rev 1997;26:17-27.

[32]Lucas P,Anderson K,Potter UJ,et al.Enhancement of small particle size dry powder aerosol formulations using an ultra low density additive.Pharm Res 1999;16:1643-1647.

[33]Nguyen TT,Rambanapasi C,de Boer AH,et al.A centrifuge method to measure particle cohesion forces to substrate surfaces:the use of a force distribution concept for data interpretation.Int J Pharm 2010;393:88-95.

[34]Iida K,Hayakawa Y,Okamoto H,et al.Evaluation of fl ow properties of dry powder inhalation of salbutamol sulfate with lactose carrier.Chem Pharm Bull 2001;49:1326-1330.

[35]Adi H,Larson I,Chiou H,et al.Role of agglomeration in the dispersion of salmeterol xinafoate from mixtures for inhalation with differing drug to fi ne lactose ratios.J Pharm Sci 2008;97:3140-3152.

[36]Islam N,Stewart P,Larson I,et al.Lactose surface modi fi cation by decantation:are drug- fi ne lactose ratios the key to better dispersion of salmeterol xinafoate from lactose-interactive mixtures?Pharm Res 2004;21:492-499.

[37]Young PM,Edge S,Traini D,et al.The in fl uence of dose on the performance of dry powder inhalation systems.Int J Pharm 2005;296:26-33.

[38]Zhang H,Ban fi eld JF.Kinetics of crystallization and crystal growth of nanocrystalline anatase in nanometer-sized amorphous titania.Chem Mater 2002;14:4145-4154.

[39]Jones MD,Hooton JC,Dawson ML,et al.An investigation into the dispersion mechanisms of ternary dry powder inhaler formulations by the quanti fi cation of interparticulate forces. Pharm Res 2008;25:337-348.

[40]Wright DL,Yu S,Kasibhatla PS,et al.Retrieval of aerosol properties from moments of the particle size distribution for kernels involving the step function:cloud droplet activation. J Aerosol Sci 2002;33:319-337.

*Corresponding author.Department of Pharmacy,University of Hertfordshire,College Lane,Hat fi eld,AL10 9AB,UK.Tel.:+44(0)1707 285904; fax:+44(0)1707 284506.

E-mail address:d.murnane@herts.ac.uk(D.Murnane).

Peer review under responsibility of Shenyang Pharmaceutical University.

http://dx.doi.org/10.1016/j.ajps.2015.08.007

1818-0876/©2015 The Authors.Production and hosting by Elsevier B.V.on behalf of Shenyang Pharmaceutical University.This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Deagglomeration

Cohesive powders

Non-linear regression modelling