International Journal of Pharmaceutics 287 (2004) 79–87

An investigation into the predictive value of cascade impactorresults for side effects of inhaled salbutamol

M. Wedaa,∗, P. Zanenb, A.H. de Boerc, D.M. Barendsa, H.W. Frijlinkc

a National Institute for Public Health and the Environment, Centre for Quality of Chemical-Pharmaceutical Products,Postbak 40, P.O. Box 1, 3720 BA Bilthoven, The Netherlands

b Heart Lung Center Utrecht, Lung Function Department, Koekoekslaan 1, 3435 CM Nieuwegein, The Netherlandsc Groningen University Institute for Drug Exploration (GUIDE), Department of Pharmaceutical Technology and Biopharmacy, Antonius

Deusinglaan 1, 9713 AV Groningen, The Netherlands

Received 23 March 2004; received in revised form 13 July 2004; accepted 26 August 2004Available online 18 October 2004

Abstract

The aim of this study was to compare the Multistage Liquid Impinger (MSLI) and the Andersen Cascade Impactor (ACI) withrespect to their power to predict differences in side effects of salbutamol delivered by a dry powder inhaler. Three preparations withthe same nominal dose and the same inhaler device but generating aerosols with different aerodynamic particle size distributionswere administered to six healthy volunteers in a randomized, placebo-controlled, four-way crossover study. Cumulative dosesfrom 400 up to 1600�g were given. The serum potassium level (K+-serum) and the heart rate (HR) were measured at baseline and1 butions oft and wass nominald arations.T Cs) wereeb t cascadei sist .©

K

f lop-for

0

5 min after each dose. Both the MSLI and ACI showed large differences between the aerodynamic particle size distrihe three preparations. The decrease in K+-serum revealed significant differences between the three active preparationsignificant for doses of 800�g and higher. The HR results showed differences between the active preparations only at aose of 1600�g and only for the preparation with the highest fine particle dose (FPD) compared to the other two prephe K+-serum appears to be a more sensitive measure for side effects than the HR. In vivo-in vitro correlations (IVIVstablished between the amounts of salbutamol deposited on the various cumulative impactor stages and the K+-serum. Theest IVIVCs were obtained in the FPD range, resulting in correlation coefficients of at least 0.78. It is concluded tha

mpactor results in the FPD range of the MSLI as well as the ACI correlate well with the K+-serum. Cascade impactor analyhus provides a clinically meaningful tool in the development and the quality control of salbutamol inhalation powders

2004 Elsevier B.V. All rights reserved.

eywords:Salbutamol; Inhalation powder; Potassium; Heart rate; Impactor; Correlation

∗ Corresponding author. Tel.: +31 30 274 4214;ax: +31 30 274 4462.

E-mail address:marjolein.weda@rivm.nl (M. Weda).

1. Introduction

Cascade impactors are widely used in the devement and the quality control of medicinal products

378-5173/$ – see front matter © 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.ijpharm.2004.08.017

80 M. Weda et al. / International Journal of Pharmaceutics 287 (2004) 79–87

inhalation and are also proposed for the bridging ofsmall changes in the composition and/or manufacturingprocess (Weda et al., 2000, 2002). Cascade impactorsmeasure the aerodynamic particle size distribution ofthe aerosol cloud generated by the product yielding in-formation about the mass fraction that has the potentialto enter the deeper part of the lung.

Currently two multistage impactors are described inthe European Pharmacopoeia (EP), namely the Mul-tistage Liquid Impinger (MSLI) and the AndersenCascade Impactor (ACI) (European Pharmacopoeia,2003). Despite the widespread use of these apparatus,it is yet unknown whether they are capable of detectingclinically relevant differences between two formula-tions or between batches of the same formulation. Var-ious studies have shown that the aerodynamic particlesize distribution of aerosols generated by inhalationproducts correlates with the amount of drug depositedin the lungs (Laube et al., 1998; Leach, 1998; Seale andHarrison, 1998; Silkstone et al., 2002). Other studiesindicate that there is a significant relationship betweenthe amount of drug deposited in the lungs and the pul-monary effects of preparations for inhalation (Pauwelset al., 1997). So, there is evidence that cascade impactorresults are correlated with efficacy, albeit indirectly.Moreover, most studies refer only to efficacy and notto safety, i.e., side effects.

In a previous study, we evaluated the capability ofimpactors to predict differences in the efficacy of in-haled salbutamol (Weda et al., 2002). Three inhalationp e in-h ero-d redt dif-f n thet s inF e of5 cesi is-p ncesi

one sidee theM enC r top ala-t udy

is the first to directly correlate the aerodynamic parti-cle size distribution of preparations for inhalation withsafety data.

2. Materials and methods

2.1. Preparations

All three inhalation powders contained 400�gsalbutamol base per actuation. Two preparations wereadhesive mixtures of salbutamol and lactose. Theywere assigned as coarse (COARSE) and intermedi-ate (INTER), because the aerosol generated by theinhaler with these formulations had large and inter-mediate aerodynamic particle sizes respectively. Onepowder preparation consisted of spherical pellets andwas assigned as fine (FINE), because with this pow-der the inhaler generated an aerosol with the smallestparticles.

The same batch of micronised salbutamol sul-phate (X90 = 3.7�m, X50 = 1.7�m, andX10 = 0.7�m)was used as drug substance for all three preparationsand was a gift of Genfarma B.V., The Netherlands.Sieve fractions from lactose monohydrate 80 Mesh(250–315�m) and 100 Mesh (90–106�m) were ap-plied as carrier excipient for COARSE and INTER re-spectively. Micronised lactose (X50 = 3.4�m) was usedas excipient in the preparation of FINE. All lactosetypes were a gift of DMV International, The Nether-l seds mea-s ara-t atecG

ac-t con-t t9 er-l ing3 wasm pel-l lendi mina l de-fi nesw ero-n h

owders with the same nominal dose and the samaler device but generating aerosols with different aynamic particle size distributions were administe

o asthmatic patients. We showed that despite largeerences in the cascade impactor results betweehree preparations no clinically relevant differenceEV1 were observed in a cumulative dose rang0–400�g. This suggests that rather large differen

n aerodynamic particle size distribution of polyderse aerosols do not necessarily result in differe

n efficacy.In contrast to our previous study which focussed

fficacy, the current investigation is dedicated toffects. The purpose of this study was to compareultistage Liquid Impinger (MSLI) and the Andersascade Impactor (ACI) with respect to their poweredict differences in side effects of salbutamol inh

ion powders. To the best of our knowledge, our st

ands. The particle size distributions of micronialbutamol sulphate and micronised lactose wereured by a Sympatec HELOS laser diffraction appus, using a RODOS dry powder dispenser (SympmbH, Germany).COARSE and INTER were prepared by mixing l

ose and salbutamol sulphate in a stainless steelainer of 160 cm3 during 10 min in a tumbling mixer a0 rpm (Turbula T2C, W.A. Bachofen AG, Switz

and). Batches of 25 g adhesive mixture contain.7% salbutamol sulphate were prepared. FINEanufactured by homogenising, densification and

etising of the salbutamol and micronised lactose bn the same stainless steel mixing container (10t 90 rpm), using stainless steel beads with a welned size distribution as pelletising aid. Next, the fiere removed and the remaining pellets were sphised on a 200�m vibratory sieve for 20 min (Fritsc

M. Weda et al. / International Journal of Pharmaceutics 287 (2004) 79–87 81

Analysette 3, Fritsch GmbH, Germany). Finally, thepellets were classified into different size fractions bymild hand sieving; the fraction 315–630 micron wasused for the experiments. A batch of 20 g pellets con-taining 7.4% salbutamol sulphate was prepared.

All three preparations were filled into a bulk con-tainer of a Novolizer® multi-dose dry powder inhalerhaving an intermediate air flow resistance of 0.028kPa0.5min l−1 (Viatris GmbH, Germany) (Berneret al., 1998; Fenton et al., 2003). A placebo was pre-pared from lactose monohydrate 80 Mesh. Before re-lease by the responsible pharmacist, the quality ofthe preparations was checked by the Centre for Qual-ity of Chemical-Pharmaceutical Products of the Na-tional Institute for Public Health and the Environ-ment, The Netherlands. The dose uniformity of allpreparations was amply within USP and EP standards,with relative standard deviations of 0.4–3.0%. Thepreparations were stored at 18◦ C and 32% relativehumidity.

2.2. Impactors

The aerodynamic particle size distribution of the ac-tive preparations was measured with the MSLI and theACI, both described in the EP. The USP throat wasused for both impactors. For the ACI a preseparatorwas added atop of stage 0.

The MSLI consists of four impaction stages and afinal collection filter (Gelmann A/E glass microfibrefi ith2( l-i dropa g ina

lec-t ket with2 ined1 )a n-t ablefl kPaa

di-a(

Table 1Operating conditions and characteristics of the impactors

MSLI ACI

Flow rate (l/min) 80 60Time per actuation (s) 3 4Volume per actuation (l) 4 4Cut-off diameter (�m)Stage 0 5.9Stage 1 11.3 4.1Stage 2 5.9 3.2Stage 3 2.7 2.1Stage 4 1.5 1.4Stage 5 –a 0.62Stage 6 0.35Stage 7 0.15Stage 8 –a

a Stage 5 of the MSLI and stage 8 of the ACI are the final collectionfilters.

2.3. In vitro measurements

For each preparation, two actuations were deliv-ered to the impactors. The amount of salbutamol de-posited in the mouthpiece of the Novolizer® inhaler(M), mouthpiece adaptor (MA), impactor throat (T),and the various impactor stages were determined withHPLC using a 125 mm× 4 mm Hypersil BDS 5�mC18 column with guard column, a detection wave-length of 278 nm and methanol:water 15:85 (v/v) aseluent. This method was validated with respect to ac-curacy, precision, linearity, range, quantification limitand detection limit (0.15 and 0.05% of the nominaldose, respectively), and proved to be suitable accordingto internal GLP standards. The in vitro measurementswere performed in triplicate.

The amounts deposited at the various stages wereexpressed as�g salbutamol base per actuation. Thefine particle dose (FPD) was calculated, defined asthe amount of drug deposited on the stages below thestage with the cut-off diameter with the closest valueto 5.0�m, i.e., stage 3–5 of the MSLI and stage 1–8 ofthe ACI.

For each preparation the mass median aerodynamicdiameter (MMAD) and the geometric standard devia-tion (GSD) were calculated.

2.4. In vivo measurements

Six healthy volunteers (four female, two male) par-t s

lter). The four impaction stages were kept moist w0 ml methanol containing 20�l phosphoric acid 85%m/m) and 10.7�g fenoterol hydrobromide per millitre as internal standard. For the MSLI a pressurecross the inhaler of 4.0 kPa was applied, resultinflow rate of 80 l/min for the MSLI.The ACI consists of eight stages and a final col

ion filter (Gelmann A/E glass microfibre filter). Unlihe MLSI the stages were kept dry, but were coated% viscous oil in hexane. The preseparator conta0 ml methanol with 20�l phosphoric acid 85% (m/mnd 10.7�g fenoterol hydrobromide per millilitre as i

ernal standard. For the ACI the maximum achievow rate of 60 l/min was used, corresponding to 2.5cross the inhaler.

The operating conditions and theoretical cut-offmeters at the flow rates used are shown inTable 1European Pharmacopoeia, 2003; Nichols, 2000).

icipated in the study. The average (± S.D.) age wa

82 M. Weda et al. / International Journal of Pharmaceutics 287 (2004) 79–87

37± 14 years. All subjects were non-smokers anddid not have any history of asthma and/or COPD.The study was approved by the Ethic Committee ofthe St. Antonius Hospital Nieuwegein, The Nether-lands. All subjects gave their written consent toparticipation.

The in vivo study followed a randomised, placebo-controlled, four-way crossover design. On fourseparate sessions with intervals of one week, thethree active preparations and the placebo were ad-ministered with the Novolizer®. At each session,subsequent doses of 400, 400 and 2�g× 400�gwere administered with a dosing interval of 20 min.Before starting each session, the subjects were givendetailed inhalation instructions. The inhalation ma-noeuver consisted of a deep and forceful inhalation,followed by a breath-holding period of 10 s. Theinhalation flow through the Novolizer® was recordedby measuring the pressure drop over the inhaler, fromwhich the flow was calculated. A peak inspiratoryflow of >60 l/min and an inhalation time of >2 sthrough the Novolizer® was necessary in order to starta session.

The heart rate (HR) and the serum potassium level(K+-serum) were measured at baseline and 15 min aftereach dose of salbutamol. The HR was determined bymanual counting over a period of 60 s. The K+-serumwas measured by taking blood samples from a cannulain an antecubital fossa vein, and subsequent analysisby means of ion selective chromatography.

2

nceo l de-p ll ast us-i icalS onef en asH

ionsa werea vari-a for ad in-d Inc.,C ces

and the absence of a dose by preparation interactionwas chosen as H0-hypothesis.

For each 400�g preparation, cumulative amountsof salbutamol deposited in the impactor stages werecalculated, starting at M up to the final collection fil-ter, subsequently MA up to the final collection filter,then T up to final collection filter, etc. The differencesin K+-serum versus baseline value (delta K+-serum) aswell the HR versus baseline value (delta HR) were cal-culated for the 400, 800, and 1600�g dose. In vivo–invitro correlations (IVIVCs) were developed by plottingthe mean amounts of salbutamol of all three prepara-tions at all three dose levels against the individual deltaK+-serum data. This was done for all cumulative stages,for both impactors. Least squares regression analysiswas used to determine the slope, intercept, and corre-lation coefficient of these relationships, using NCSS.The H0-hypotheses of the correlation being zero andthe intercept being zero were investigated.

In all statistical tests, an� value of 0.05 was con-sidered to be significant.

3. Results

3.1. In vitro results

For the MSLI and ACI, the complete deposition pat-tern of the preparations is presented inFigs. 1 and 2,respectively. For the MSLI, none of the three prepa-r herr them sig-n thei nd 8o if-f ationsw

Iwf heF1 hi hreep

re6C

.5. Statistical analysis

For the three active preparations, the significaf differences between the amounts of salbutamoosited at the various stages of the MSLI as we

he ACI was determined with analysis of variance,ng NCSS for Windows (Number Cruncher Statistystems, Kaysville, Utah, USA). The same was d

or the FPD. The absence of differences was chos0-hypothesis.Dose-effect curves of the three active preparat

nd the placebo were constructed. These curvesnalyzed by repeated measurements analysis ofnce for preparation and dose differences, andose by preparation interaction using SPSS for Wows (Statistical Products and Service Solutionshicago, Illinois, USA). The absence of differen

ations were significantly different from one anotegarding the amounts of salbutamol deposited inouthpiece adaptor and stage 5. For the ACI, noificant difference was seen in the mouthpiece of

nhaler and between INTER and FINE at stage 7 af the ACI. At all other impactor stages, significant d

erences were observed between the three preparith the MSLI as well as with the ACI (P< 0.05).The FPD (mean± S.D.) obtained with the MSL

as 39.8± 1.8�g, 83.5± 10�g, and 204± 7.3�gor COARSE, INTER, and FINE, respectively. TPD measured with the ACI, was 40.7± 0.2�g,44± 1.9�g, and 321± 16�g, respectively. For bot

mpactors, the differences in FPD between the treparations are significant (P< 0.001).

The MMAD (GSD) of the three preparations we.8�m (2.6), 3.4�m (3.1), and 1.7�m (2.2) forOARSE, INTER, and FINE, respectively.

M. Weda et al. / International Journal of Pharmaceutics 287 (2004) 79–87 83

Fig. 1. Deposition pattern of the three preparations, obtained in triplicate with the MSLI, including standard deviations. M, mouthpiece of theinhaler device; MA, mouthpiece adaptor; T, impactor throat.

3.2. In vivo results

All volunteers completed the four sessions. Thedose-effect curves for the delta K+-serum and the deltaHR are shown inFigs. 3 and 4, respectively. The dose bypreparation interaction was significant for both param-eters (p< 0.001), indicating that the dose-effect curvesfor the placebo and the active preparations did not runparallel.

Statistical evaluation of the K+-serum data showedsignificant differences between placebo and the threeactive preparations (p< 0.02). The difference betweenthe three active preparations was also significant(p< 0.03). The dose-effect evaluation revealed that theK+-serum decrease from baseline was significant start-ing from the 800�g dose (p< 0.05).

Statistical evaluation of the HR results showedsignificant differences between the placebo and thethree active preparations (p< 0.001), but not betweenCOARSE and INTER. The dose-effect evaluationshowed that the increase in HR was only significantat the 1600�g dose level (p< 0.01).

3.3. In vivo–in vitro correlations

The IVIVCs between the cascade impactor, re-sults and the changes in K+-serum are given inTables 2 and 3, respectively, including the correlationcoefficients and the parameters describing the regres-sion lines. The intercept did not differ significantlyfrom zero for the K+-serum regression lines, with theexception of the IVIVC obtained for stage 5 of the

F in tripl ce of thei

ig. 2. Deposition pattern of the three preparations, obtainednhaler device; MA, mouthpiece adaptor; T, impactor throat.

icate with the ACI, including standard deviations. M, mouthpie

84 M. Weda et al. / International Journal of Pharmaceutics 287 (2004) 79–87

Fig. 3. K+-serum 15 min after administration of nominal doses of 400, 800, and 1600�g placebo and three salbutamol inhalation powders withdifferent aerodynamic particle size distribution, including standard errors of the mean. The K+-serum is expressed as difference vs. baselinevalues.

Fig. 4. HR 15 min after administration of nominal doses of 400, 800, and 1600�g placebo and three salbutamol inhalation powders with differentaerodynamic particle size distribution, including standard errors of the mean. The HR is expressed as difference vs. baseline values.

Table 2Slope, intercept, and correlation coefficients (r) of the IVIVCs be-tween the amount of salbutamol deposited on the various impactorstages of the MSLI and delta K+-serum

Stages Intercept Slope (× 10−4) r

M up to 5 0.119 −4.38 −0.660MA up to 5 0.104 −4.56 −0.638T up to 5 0.099 −5.10 −0.6341 up to 5 0.051 −4.96 −0.5532 up to 5 0.012 −9.82 −0.8033 up to 5 0.003 −11.7 −0.7984 and 5 0.032 −21.9 −0.8085 0.126 −67.2 −0.791

MSLI (p< 0.03). For all IVIVCs, significant correla-tions were seen (p< 0.0001).

4. Discussion

For the in vitro experiments, the two multistage im-pactors described in the EP were used, i.e., the MSLIand the ACI. The MSLI was fully operated accordingto the instructions for dry powder inhalers included inthe EP, applying a pressure drop of 4.0 kPa across theinhaler. Due to the relatively high resistance of the ACI,it was not possible to obtain a pressure drop of 4.0 kPa.

M. Weda et al. / International Journal of Pharmaceutics 287 (2004) 79–87 85

Table 3Slope, intercept, and correlation coefficients (r) of the IVIVCs be-tween the amount ofsalbutamol deposited on the various impactorstages of the ACI and delta K+-serum

Stages Intercept Slope (× 10−4) r

M up to 8 0.107 −3.83 −0.633MA up to 8 0.105 −4.06 −0.632T up to 8 0.131 −5.07 −0.6840 up to 8 0.111 −5.16 −0.6421 up to 8 −0.016 −7.17 −0.7952 up to 8 −0.012 −7.45 −0.7963 up to 8 −0.001 −8.28 −0.8004 up to 8 0.015 −10.1 −0.8045 up to 8 0.033 −19.0 −0.8056 up to 8 0.035 −55.7 −0.8117 and 8 0.107 −242 −0.7878 0.096 −382 −0.775

The maximum achievable flow rate of 60 l/min wastherefore used. This difference in flow rate between theMSLI and ACI is not considered relevant, because thepurpose of the in vitro study was not to compare theaerodynamic particle size distributions obtained withthe MSLI and ACI, but to compare the three prepara-tions within one impactor and to evaluate the capabilityof both impactors to predict clinically relevant differ-ences between these preparations.

For the in vivo study, a starting dose of 400�gwas used, since it is expected that below this dose theK+-serum and HR will not noticeably be influenced(Kleerup et al., 1996; Seppala et al., 2001). The maxi-mum dose was set at 1600�g, because this is the maxi-mum daily dose registered in the Netherlands. The K+-serum level and the HR were selected as response pa-rameters for safety, reflecting the systemic�2-responseof salbutamol. The study has not been carried out withasthmatic patients, but with healthy volunteers, becausethere are indications regarding the development of tol-erance with respect to�2-responses in asthmatic pa-tients when using�2-agonists during longer periods(Hancox et al., 2002; Woude et al., 2001). However,the development of tolerance is not applicable to asth-matic patients using a�2-agonist for the first time. Iftolerance is an issue, any differences in systemic�2-response detected with healthy volunteers will there-fore also be relevant for asthmatic patients in general.The discriminatory power of the study was thus maxi-mized by selecting healthy volunteers.

The in vivo data show that the three preparationsare significantly different with respect to the decreasein K+-serum. On the other hand, COARSE and INTERdo not produce differences in HR. The difference in HRwith FINE is significant, but only a dose of 1600�g re-sults in a considerable HR increase. At 800�g, the dif-ference in HR versus baseline is below 10 bpm for allthree preparations. Apparently, the HR is a less sen-sitive parameter for safety evaluation of salbutamolinhalation powders. This is confirmed by other stud-ies (Janssen et al., 2002; Fowler and Lipworth, 2001).Summarizing the in vivo data, we conclude that dif-ferences in the aerodynamic particle size distributionof the aerosols generated from different formulationsresult in significant differences in side effects of salbu-tamol. The clinical relevance of impaction data for thedevelopment and the quality control of salbutamol in-halation powders with respect to side effects is thusconfirmed. Furthermore, the decrease in K+-serum ap-pears to be a more sensitive parameter to detect differ-ences in safety. For determination of the IVIVCs onlythis parameter is therefore taken into account.

The in vitro results show large differences betweenthe three active preparations with both the MSLI andACI. The results of the adhesive mixtures are in linewith other studies investigating the effect of carrier sizeon the dispersion of salbutamol sulphate (Zeng et al.,2000; Louey et al., 2003). Smaller lactose particles re-sult in higher fine particle doses of salbutamol sulphate.

The relationship between the impactor measure-m mi-n old ainstt cal-c as-s l ab-s fora reti-c wnt bedf tot in-h -t achs tioni d allt plot.T g

ents and the K+-serum was investigated by deteration of IVIVCs. Cumulative amounts of salbutameposited on the impactor stages were plotted ag

he K+-serum and the correlation coefficients wereulated. For the construction of these IVIVCs, theumption was made that the amount of salbutamoorbed via the gastrointestinal tract was negligiblell three preparations, since this fraction could theoally contribute to the systemic side effects. It is knohat after inhalation of salbutamol the fraction absorrom the gastrointestinal tract contributes only 0.3%he overall bioavailability over the first 30 min postalation (Chrystyn et al., 1996). Since all three salbu

amol doses were administered within 40 min and eession was completed within 60 min, our assumps reasonable. The data of all three preparations anhree dose levels were therefore combined in onehe results given inTables 2 and 3show a breakin

86 M. Weda et al. / International Journal of Pharmaceutics 287 (2004) 79–87

point for the IVIVCs at stage 2 of the MSLI (cut-offdiameter 11�m) and at stage 1 of the ACI (cut-offdiameter 5.9�m), with correlation coefficients higherthan 0.78.

It is generally accepted that particles with an aerody-namic diameter below 5–7�m have the highest proba-bility of reaching the deep or peripheral lungs (Heyderet al., 1986). In this respect, it is remarkable that a goodcorrelation is obtained with the cumulative amount ofdrug on the stages 2–5 of the MSLI, since particlesbetween 7–11�m are included in this range. It is how-ever conceivable that for the preparations used in ourstudy the amount of salbutamol deposited at stage 2mostly consists of particles with an aerodynamic di-ameter that approaches the cut-off diameter of 5.9�mof this stage. This explanation is supported by the factthat the micronised salbutamol sulphate batch used forthe production of the preparations contains more than98% particles smaller than 6.0�m. If a larger fractionof particles with a size close to 11�m would be present,the correlation coefficient might be less favourable. Inview of the broad range of particles that are likely todeposit at stage 2 and taking into account the fact thatonly particles below 5–7�m will reach the lungs, it issensible not to use stage 2–5 for comparison of dif-ferent formulations of a product or for quality controlof product batches. In addition, the advantage of in-cluding the mass on stage 2 is negligible in terms of ahigher correlation coefficient compared with stage 3–5.So, for salbutamol the FPD range is the most relevanti cts.

co-e agesw nsd R >F t thea epo-s ea ar-r thea ys-t test mics dis-p icless nie rsef

holding chamber showed that the latter preparationcaused more systemic side effects, suggesting that thevery small particles (0.5–2�m) of the polydispersepMDI could be responsible for this difference (Zanenand Lammers, 1999). Our data demonstrate that alower MMAD results in more systemic side effects ofsalbutamol. The fact that within the entire FPD rangethe same correlation coefficient is found, is probablydue to the polydisperse nature of the preparations.Taking into consideration the GSD there will be a con-siderable overlap in the particle size distributions. It ispossible that a comparable study with monodisperseaerosols would reveal a better correlation for the lowerimpactor stages compared to the upper stages. Our cor-relation data are therefore only valid for polydisperseaerosols. It is remarked that all salbutamol productsthat are currently on the market are polydisperse.

The results of this study also allows us to considerthe definition of the FPD. The EP requires the FPDto be calculated by interpolation as the mass of activeingredient less than 5.0�m from the cumulativemass versus stage cut-off diameter plot. Our resultsshow no clinical evidence that the exact definitionadopted in the EP is necessary for development andquality control, since the correlation coefficients ofour IVIVCs do not change much around 5.0�m. We,therefore conclude that our definition of the FPD asbeing the amount of drug deposited on the stagesbelow the stage with the cut-off diameter with theclosest value to 5.0�m is adequate.

isc SLIa ectt TheF toi

5

tingc ctsb Csw ndt Ina to beo usel dis-

mpactor part for comparison of systemic side effeUnexpectedly, no difference in the correlation

fficients is seen between the various impactor stithin the FPD range. The MMAD of the preparatioecreases in the ranking order of COARSE > INTEINE. When particles get smaller the deposition alveoli increases at the expense of the bronchial dition (Heyder et al., 1986). In view of the large surfacrea of the alveolar epithelium, the thin air-blood bier, and the high blood flow, it is conceivable thatlveolar fraction will be rapidly absorbed into the s

emic circulation. This fraction no longer contribuo the desired clinical effect, but may cause systeide effects. Indeed, efficacy studies with monoerse aerosols indicate that the size of the parthould not be too small (Zanen et al., 1994; Usmat al., 2003). A safety study comparing monodispe

enoterol particles of 2.8�m and a pMDI with valved

Summarizing the in vivo and in vitro data, itoncluded that measuments obtained with the Ms well as the ACI are clinically relevant with resp

o systemic side effects of inhaled salbutamol.PD range is the most suitable impactor fraction

nvestigate.

. Conclusion

We showed that impactors are capable of prediclinically relevant differences in systemic side effeetween salbutamol inhalation powders. Good IVIVere found in the FPD range for both the MSLI a

he ACI. The MSLI and ACI are equally predictive.previous study the same impactors were shownverdiscriminating with respect to efficacy, beca

arge differences in the aerodynamic particle size

M. Weda et al. / International Journal of Pharmaceutics 287 (2004) 79–87 87

tribution did not lead to clinically relevant differencesin efficacy. The value of these apparatuses for thedevelopment and the quality control as well as thebridging of post-approval changes in the formulationand/or production of salbutamol inhalation powders isthus confirmed.

Acknowledgments

The authors wish to thank Paul Hagedoorn, AhmedAjaoud, Johan S. Toren, Rob Janssen, and Wim Ger-ritsen for their technical assistance.

References

Berner, B., Fyrnys, B., de Boer, A.H., 1998. ASTA Medica multidose dry powder inhaler. In: Dalby, R.N., Byron, P.R., Farr, S.J.(Eds.), Respiratory Drug Delivery VI. Interpharm Press, BuffaloGrove, pp. 475–477.

Chrystyn, H., Corlett, S.A., Silkstone, V., 1996. Lung bioavailabilityof generic and innovator salbutamol MDIs. Thorax 51, 658.

European Pharmacopoeia, 2003. Preparations for inhalation: aero-dynamic assessment of fine particles. In: European Pharma-copoeia Convention, Council of Europe, Strasbourg, France, pp.209–219.

Fenton, C., Keating, G.M., Plosker, G.L., 2003. Novolizer® A mul-tidose dry powder inhaler. Drugs 63, 2437–2445.

Fowler, S.J., Lipworth, B.J., 2001. Pharmacokinetics and systemic�2-adrenoreceptor-mediated responses to inhaled salbutamol. Br.J. Clin. Pharmacol. 51, 359–362.

H reave,cise-165,

H , W.,ct in

J ., Za-vivo

K Cu-134athma.

L vivo. J.

Leach, C.L., 1998. Improved delivery of inhaled steroids to the largeand small airways. Respir. Med. 92, 3–8.

Louey, M.D., Razia, S., Stewart, P.J., 2003. Influence of physico-chemical carrier properties on the in vitro aerosol deposition frominteractive mixtures. Int. J. Pharm. 252, 87–98.

Nichols, S.C., 2000. Andersen Cascade Impactor; calibration andmensuration issues for the standard and modified impactor.Pharmeuropa 12, 598–602.

Pauwels, R., Newman, S.P., Borgstrom, L., 1997. Airway depositionand airway effects of antiasthma drugs delivered from metered-dose inhalers. Eur. Respir. J. 10, 2127–2138.

Seale, J.P., Harrison, L.I., 1998. Effect of changing the fine particlemass of inhaled beclomethasone dipropionate on intrapulmonarydeposition and pharmacokinetics. Respir. Med. 92, 9–15.

Seppala, O.R., Aalto, E., Annila, T., Hakonen, T., Lukkari-Lax, E.,Jouhikainen, T., Nieminen, M.M., Liipp, K., 2001. The effi-cacy of a new salbutamol metered-dose powder inhaler in com-parison with two other inhaler devices. Respir. Med. 95, 949–953.

Silkstone, V.L., Dennis, J.H., Pieron, C.A., Chrystyn, H., 2002. Aninvestigation of in vitro/in vivo correlations for salbutamol neb-ulized by eight systems. J. Aerosol. Med. 15, 251–259.

Usmani, O.S., Biddiscombe, M.F., Nightingale, J.A., Underwood,S.R., Barnes, P.J., 2003. The effect of brochodilator particle sizein asthmatics using monodisperse aerosols. J. Appl. Physiol. 95,2106–2112.

van der Woude, H.J., Winter, T.H., Aalbers, R., 2001. Decreasedbronchodilating effect of salbutamol in relieving methacholineinduced moderate to severe bronchoconstriction during highdose treatment with long acting beta2 agonists. Thorax 56, 529–535.

Weda, M., Geuns, E.R.M., Vermeer, R.C.R., Buiten, N.R.A.,Hendriks-de Jong, K., Bult, A., Zanen, P., Barends, D.M., 2000.Equivalence testing and equivalence limits of metered-dose in-halers and dry powder inhalers measured by in vitro impaction.

W A.,of

rsus

Z rticlearm.

Z of in-tion.

Z ectsamolarm.

ancox, R.J., Subbarao, P., Kamada, D., Watson, R.M., HargF.E., Inman, M.D., 2002. Beta2-agonist tolerance and exerinduced bronchospasm. Am. J. Respir. Crit. Care Med.1068–1070.

eyder, J., Gebhart, J., Rudolf, G., Schiller, C.F., Stahlhofen1986. Deposition of particles in the human respiratory trathe size range 0.005–15�m. J. Aerosol. Sci. 17, 811–825.

anssen, R., Weda, M., Ekkelenkamp, M.B., Lammers, J.-W.Jnen, P., 2002. Metal versus plastic spacers: an in vitro and incomparison. Int. J. Pharm. 245, 93–98.

leerup, E.C., Tashkin, D.P., Cline, A.C., Ekholm, B.P., 1996.mulative dose-response study of non-CFC propellant HFAsalbutamol sulfate metered-dose inhaler in patients with asChest 109, 702–707.

aube, B.L., Jashnani, R., Dalby, R.N., 1998. Predicting inlung deposition of cromolyn sodium from in vitro estimatesAerosol. Med. 11, S35–S42.

Eur. J. Pharm. Biopharm. 49, 295–302.eda, M., Zanen, P., de Boer, A.H., Gjaltema, D., Ajaoud,

Barends, D.M., Frijlink, H.W., 2002. Equivalence testingsalbutamol dry powder inhalers: in vitro impaction results vein vivo efficacy. Int. J. Pharm. 249, 247–255.

anen, P., Go, T.L., Lammers, J.-W.J., 1994. The optimal pasize for�-adrenergic aerosols in mild asthmatics. Int. J. Ph107, 211–217.

anen, P., Lammers, J.-W.J., 1999. Reducing adverse effectshaled fenoterol through optimization of the aerosol formulaJ. Aerosol. Med. 12, 241–247.

eng, X.M., Martin, G.P., Marriott, C., Pritchard, J., 2000. The effof carrier size and morphology on the dispersion of salbutsulphate after aerosolization at different flow rates. J. PhPharmacol. 52, 1211–1221.