- Email: [email protected]
Influence of Storage Relative Humidity on the Dispersion of Salmeterol Xinafoate Powders for Inhalation
Influence of Storage Relative Humidity on the Dispersion of Salmeterol Xinafoate Powders for Inhalation SHYAMAL DAS,1 IAN LARSON,1 PAUL YOUNG,2 PETER STEWART1 1
Drug Delivery, Disposition and Dynamics, Monash Institute of Pharmaceutical Sciences, Monash University (Parkville Campus), 381 Royal Parade, Victoria 3052, Australia 2
Advanced Drug Delivery Group, Faculty of Pharmacy, University of Sydney, New South Wales 2006, Australia
Received 17 April 2008; accepted 12 June 2008 Published online 25 July 2008 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21500
ABSTRACT: The in vitro dispersion of salmeterol xinafoate (SX) alone and four SX (2.5%)–coarse lactose (CL) mixtures containing 0%, 5%, 10% and 20% micronised lactose (ML) was monitored during 18-month storage at 33%, 55% and 75% relative humidity (RH) using a twin stage impinger. The surface moisture was monitored over 2 months by thermo gravimetric analysis. The morphology was determined by scanning electron microscopy. An aerosizer was used to compare the agglomerate strengths of formulations before and after storage at 75% RH. Upon storage, no significant difference occurred in fine particle fraction (FPF) of any formulation at 33% and 55% RH. Within 8 weeks, the FPF of mixture containing 20% ML (M20F) significantly decreased from 11.3% to 7.7% at 75% RH ( p ¼ 0.008) and to 4.9% at 95% RH ( p ¼ 0.001). The calculated capillary forces were greater for ML–ML contact than other particle interactions and the propensity of ML–ML contacts was higher in M20F. The agglomerate strength of M20F significantly increased after storage. The study concluded that the critical factors for decreased dispersion of SX formulations were RH of 75% or greater and the presence of high concentrations of ML due to capillary forces and/or solid bridge formation of ML leading to increased agglomerate strength. ß 2008 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 98:1015–1027, 2009
Keywords: dispersion; storage stability; salmeterol xinafoate; relative humidity; dry powder inhaler; agglomerate
INTRODUCTION Drug dispersibility changes during storage present problems for dry powder inhaler (DPI) development.1 Dispersibility is principally governed by the magnitude of interparticulate drug– drug and drug–excipient forces. Interparticulate forces are a combination of various electrical (contact potential and Coulombic) and nonelectrical forces (van der Waals’ forces, capillary
Correspondence to: Peter Stewart (Telephone: 61-3-99039517; Fax: 61-3-9903-9583; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 98, 1015–1027 (2009) ß 2008 Wiley-Liss, Inc. and the American Pharmacists Association
forces, mechanical interlocking and solid bridging).2 These forces can vary in different relative humidities (RHs) depending on material properties and storage duration. Forces like contact potential and Coulombic forces are more dominant at low RH since charge decay is minimised while capillary forces and solid bridging develop at high RH due to the formation of liquid bridges between particles. van der Waals’ forces and mechanical interlocking can contribute at all RHs. Therefore, the strength of interparticulate forces, manifest in DPI formulations by micronised drug– carrier interactions and micronised drug–fine excipient interactions within agglomerates, could be different at different RHs. Furthermore, forces between particles do not develop at the same rate.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
1015
1016
DAS ET AL.
Forces like contact potential, van der Waals’ forces and Coulombic forces develop very quickly, while capillary forces take sometime to develop (30 min)3 and to reach in equilibrium (up to 72 h)4 while solid bridges form after long term storage at high RH and depend on temperature and humidity cycling. Therefore, the hypothesis of this study was that the interparticulate forces may increase during storage, both in the short and long term, with resultant compromised performance of the inhalation powder. There are a number of previous studies1,5–12 which evaluated the influence of storage humidity on dispersion of DPI formulations. A few of the studies were on drug alone formulations (hydrophilic or hydrophobic). For example, Young et al.5 observed that the fine particle fraction (FPF) of disodium cromoglycate and salbutamol sulphate decreased at elevated humidity while that of triamcinolone acetonide increased when stored at 15–75% RH for 12 h. Other studies concentrated on storage humidity effects of DPIs containing carrier particles. For example, Hindle and Makinen6 observed a decrease in mean fine particle dose of micronised cromolyn sodium from its blend with lactose monohydrate from 3.9 0.2 mg at ambient conditions (20 38C and 30 5% RH) to 1.2 0.2, 0.73 0.2 and 0.2 0.04 mg following exposure at 53%, 66% and 76% RH, respectively. Braun et al.7 found a greater decrease in FPF at 55% RH than at 33% RH when mixtures of disodium cromoglycate with two lactose monohydrates and two dextrose monohydrates were stored for 27 days. In Braun et al.’s study, formulations were stored in gelatine capsules which may have a significant effect on dispersibility. In another study, Young and Price8 observed a higher reduction in FPF for the mixture of lactose carrier with micronised salbutamol sulphate than with solution enhanced dispersion by supercritical fluids (SEDS) produced salbutamol sulphate after storage at high RH for 75 h. In vitro inhalation was also studied by Lida et al.9 for the mixture of salbutamol sulphate and lactose carrier covered with varying surface coverage with magnesium stearate and stored at different RH for 7 days. Most recently Zeng et al.10 observed a small decrease in FPF when a mixture of salbutamol and lactose was stored at 75% RH for 6 days. Studies were also carried out to determine the suitable packaging requirements offering better protection from storage humidity.11,12 Only few of these studies5,6,8–10 exposed their formulations in open pans to storage humidity. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
Moreover, these studies concentrated on the effect of RH either on drug alone (D) or on drug–coarse carrier mixtures for short periods of time (maximum 7 days exposed in open pans at storage humidity). The studies did not include drug– coarse carrier mixtures with added fine carriers even though these formulations have greater dispersion capability.13–15 Therefore, the objective of this study was to evaluate the dispersion of formulations of different compositions (drug alone, drug–coarse carrier mixture and drug– coarse carrier mixture containing varying concentrations of additional fines) by exposing the formulations for up to 18 months to a range of RHs existing from hot and dry (35% RH) to hot and humid (75% RH) climatic zones. In addition, the study focussed on understanding the reasons for any changes observed. The study used salmeterol xinafoate (SX) as a model drug, Inhalac 120 as the coarse lactose (CL) carrier and specially micronised lactose monohydrate as the fine added excipient.
MATERIALS AND METHODS Materials Inhalation grade micronised SX (Glaxo Wellcome, Ware, UK) was used as model drug. a-Lactose monohydrate, CL (Inhalac 1120, Meggle AG, Wasserburg, Germany) was used in ‘‘as supplied’’ form in all the mixtures as the coarse carrier. Inhalac 120 was selected because of its low concentration (1.3% v/v) of inherent fines. Micronised lactose (ML) used in this study was prepared through micronisation of a-lactose monohydrate (Lactose New Zealand, New Zealand) using fluid energy milling.13 Methanol (CH3OH, HPLC grade, Merck KGaA, Darmstadt, Germany), milli-Q grade water (Millipore Corporation, Melsheim, France), ammonium acetate (Analar, BDH Laboratories, Victoria, Australia) were used for HPLC analysis. Magnesium chloride hexahydrate (MgCl2, 6H2O), potassium carbonate (K2CO3), magnesium nitrate (MgNO3), sodium chloride (NaCl) and potassium nitrate (KNO3) (all from BDH Laboratories) were used for generating different RH.5 Methods Storage Conditions Saturated solutions of magnesium chloride, potassium carbonate, magnesium nitrate, sodium DOI 10.1002/jps
INFLUENCE OF RH ON DISPERSION OF SX POWDERS
chloride and potassium nitrate were used to produce 33%, 45%, 55%, 75% and 95% RH, respectively in desiccators.5 The RH was observed to vary by only 2% over the study period. Samples of CL and ML kept in open pans were stored at 45%, 75% and 95% RH for 2 months for the moisture adsorption study. Five formulations: SX alone (D) and four SX–CL mixtures containing 0%, 5%, 10% and 20% ML (M0F, M5F, M10F and M20F, respectively) were stored at 33%, 55% and 75% RH for 18 months for the dispersion study. The mixture containing 20% ML (M20F) was stored at 95% RH for 12 months for a dispersion study. The storage temperature was 25 28C. Determination of Surface Moisture by Thermo Gravimetric Analysis (TGA) The moisture content of the powders was determined by TGA every day for 2 weeks followed by once a week up until 2 months. The sample (approximately 2–5 mg) was placed in a platinum sample pan under a nitrogen purge. The TGA was operated at 58C/min from 40 to 1808C. The moisture content was calculated on dry weight basis and the final weight was taken at 1708C. Surface moisture was calculated by deducting the % moisture for water of crystallisation from the % moisture content. % moisture ¼
ðinitial weight final weightÞ 100 final weight
% surface moisture ¼ % moisture % water of crystallization
Preparation of Powder Formulations Four SX (2.5%)–CL mixtures containing 0%, 5%, 10% and 20% micronised lactose (M0F, M5F, M10F and M20F) were prepared according to a previously validated mixing method.16,17 The micronised powders (SX or SX and ML) were placed between equal amounts of CL in a test tube. Three ceramic beads (approximately 10 mm diameter) were placed inside the test tube which was stoppered and inverted several times to prevent the drug from sticking to the sides of the test tube. The test tube was then shaken vigorously for 5 min by hand. The ceramic beads provided a ball-milling effect for breaking up DOI 10.1002/jps
1017
agglomerates. The batch size for these formulations was 5 g. Homogeneity of Powder Mixtures The average drug content and the variability between samples, expressed by the CV, were used to quantify the homogeneity of each powder mixture. Twenty samples (20 0.5 mg each) were accurately weighed using an analytical balance (Mettler AT 261 Deltarange, Greifensee-Zurich, Switzerland). All samples were dissolved in 40% methanol (HPLC grade) and the amount of SX was determined by a validated UV assay. A mixture with mean drug content within 95–105% of the theoretical value and a CV less than 5% was regarded to have an acceptable degree of homogeneity. These specifications indicated that 95% of samples would fall within 10% of the mean.18 In Vitro Drug Dispersion by Twin Stage Impinger (TSI) In vitro aerosol deposition was determined using a twin stage impinger (TSI, Apparatus, A; British Pharmacopoeia, 2000) (Copley, UK) and a Rotahaler (Glaxo Wellcome, Melbourne, Australia). The airflow was created by a vacuum pump (Model OD 5/2, Dynavac Engineering, Melbourne, Australia) fitted to the mouth piece of TSI and the flow rate was adjusted to 60 L/min before each measurement. The temperature (25 28C) and RH (45 2% RH) of the surrounding environment was measured by a thermohygrometer (Shinyei TRH-CZ, Osaka, Japan). Seven and 30 mL of 40% methanol (HPLC grade) were placed in the stage one and stage two of the TSI, respectively. Size 3 hard gelatine capsules (Capsugel, Sydney, Australia), loaded with the powder formulations (20 0.5 mg) or drug (0.50 0.03 mg), were inserted into a Rotahaler (Glaxo Welcome, UK). The Rotahaler was placed into a moulded mouthpiece attached to the TSI, twisted and the powder was then dispersed by drawing 4 L of air during each measurement (4 s at 60 L/min). Each part (inhaler, stage one and stage two) was then rinsed with 40% methanol (in water) and the rinsed liquid was collected and diluted to an appropriate volume (100 mL). The drug concentration was then determined by a validated high performance liquid chromatographic assay. Five replicates of TSI measurement were performed for each formulation, and the TSI measurements were conducted immediately after removal of the formulations from desiccators. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
1018
DAS ET AL.
The total amount of drug collected from inhaler, stage one (S1) and stage two (S2) was regarded as the recovered dose (RD), and was calculated as a percentage of total dose. The emitted dose (ED) was the amount of drug that was deposited in stage one and stage two and was calculated as the percentage of RD. The FPF was defined as the percentage of RD that was deposited in stage two. ED ¼
ðS1 þ S2 Þ 100 RD
FPF ¼
S2 100 RD
High Performance Liquid Chromatography (HPLC) Assay SX recovered by TSI was analysed by HPLC using a C18 column (5 mm, 4.6 mm 150 mm, Apollo Alltech Associates Inc., Deerfield, IL) and a UV detector (Shimadzu Diode Array Detector, SPD M10A VP, Kyoto, Japan) at a wavelength of 252.2 nm. A freshly prepared mixture of methanol and 0.2% (w/v) ammonium acetate solution in water (55:45, pH 6.9), filtered and degassed using 0.45 mm membrane filter (Millipore, County Cork, Ireland) was used as the mobile phase. A flow rate of 1.0 mL/min at ambient temperature was generated by a HPLC pump (Shimadzu LC-10 AJ, Kyoto, Japan) using an injection volume of 15 mL (Shimadzu Autosampler, SIL-10AD VP, Kyoto, Japan). The retention time of SX was approximately 4 min. The peak area was recorded by integration (Shimadzu Class VP, Kyoto, Japan). A HPLC calibration curve was prepared using five replicates of four concentrations between 0.4 and 10 mg/mL, and the linear regression was performed using Sigmaplot (Jandel Scientific, Chicago, IL). The precision of each assay was determined by analysing five replicates from each of two standard SX solutions (two concentrations: low 1.0 mg/mL and high 4.0 mg/mL). The regression coefficient (R2) was 1.000 and no significant deviation from zero intercept was observed ( p > 0.05). The accuracy ranged from 99.4% to 101.3% and the precision ranged from 0.6% to 1.1%. Scanning Electron Microscopy (SEM) A small amount of powder was sprinkled and glued on a metal sample plates. The samples were gold plated with a sputter coater (BAL-TEC SCD 005, Tokyo, Japan) using an electrical potential JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
of 2.0 kV at 25 mA for 10 min. The surface morphology of the particles was examined at several magnifications under JOEL JSM 6000F scanning electron microscope at 15 kV (JOEL, Tokyo, Japan). Aerodynamic Particle Size Analysis by Aerosizer The aerodynamic particle size of the powders was measured by an Aerosizer (Amherst Process Instruments, Inc., Hadley, MA) using a dry powder dispersion system (aerodisperser). Approximately 5 mg of powder was placed in the sample cup of the aerodisperser. Particle size measurement was carried out at a medium feed rate and a sample run time of 300 s. Different shear pressures (3.4, 10.3, 20.7, 27.6 kPa) were used to understand the effect of increasing shear pressure on the agglomerate strength of the powders. The particle size of the mixture was analysed using API Aerosizer software (LD Version 7.04). The average particle size distribution was determined from five replicates of each sample.
Statistical Analysis All data were subjected to One Way Analysis of Variance (ANOVA) (SPSS, Chicago, IL, Version 15.0). Probability ( p) values of 0.05 when analysed by post hoc multiple comparisons were considered as statistically significant.
RESULTS AND DISCUSSION Surface Moisture of CL and ML during Storage at Different RH In order to assess the kinetics of surface moisture uptake for both CL and ML, the surface moisture was monitored by a thermo gravimetric method during storage at three different RH (45%, 75% and 95% RH) for 2 months (Fig. 1). The rate of surface moisture uptake and the equilibrium surface moisture content increased with increasing RH for both CL and ML. Moreover, at any specific RH, both the rate of surface moisture uptake and the equilibrium surface moisture content were higher for ML than for CL. For example, the surface moisture of CL increased to the plateau level of 0.25% w/w in 3 weeks at 75% RH and to 0.12% w/w in a month at 45% RH. In DOI 10.1002/jps
INFLUENCE OF RH ON DISPERSION OF SX POWDERS
1019
Powder Formulations
Figure 1. Monitoring of surface moisture (by a thermo gravimetric analysis) of (A) coarse lactose, CL and (B) micronised lactose, ML during storage at three different RHs (45%, 75% and 95%) for 60 days.
contrast for ML, plateau levels of 0.36% w/w and 0.22% w/w were achieved in only 2 weeks at 75% RH and 3 weeks at 45% RH, respectively. The equilibrium moisture contents found for both CL and ML were consistent with previous reports by Callahan et al.19 who reported a lactose monohydrate equilibrium moisture content of 0.2% in the range of 11–93% RH. The time to reach equilibrium for CL was consistent with the report of Berard et al.20 who concluded from adsorption kinetic studies of lactose monohydrate that at least 21 days was required to establish the equilibrium. The extent of water accumulation and condensation depends on surface properties such as the presence of amorphous domains, surface roughness and surface area.3 The higher surface moisture content of ML compared with CL was expected, but the increase was not proportional to the anticipated increase in surface area determined on a particle size basis; this may have been associated with the reduction in expected surface area due to agglomeration of the micronised lactose. DOI 10.1002/jps
Inhalation grade SX particles form cohesive agglomerates because of their small size and predominant cohesion interactions (Fig. 2A). Coarse carriers such as lactose monohydrate are commonly added to DPI formulations to improve flow properties and increase bulk. The SEM of the SX–CL mixture (M0F) revealed that the majority of the SX particles, existing as either individual particles or agglomerates, were adhered to the CL (Fig. 2B). In SX–CL mixtures containing additional ML, agglomerates were observed both on and off the surface of CL (Fig. 2C and D). These agglomerates are likely to consist of interacting particles of SX and ML13 and will vary in the composition of SX and ML because of complex interparticulate interactions occurring among different particles during powder mixing. The homogeneity and consistency in the in vitro TSI deposition of each batch were determined to ensure that the mixing procedure produced powder mixtures approaching target SX content with low variability and reproducible SX dispersion behaviour. The SX content ranged from 99.5% to 100.7% and the coefficient of variation (CV) was below 1.9%. The results indicated that the homogeneity of powder mixtures was excellent, being well within the specifications applied in this study.18 In order to study intra-batch dispersion variability, the ED and the SX content of the stage one and two of the TSI from a single batch of M0F were compared using five replicates. The ED and the SX content of the stage one and two of the TSI were 80.3 0.5%, 74.8 0.5% and 5.3 0.4%, respectively, and showed no significant differences ( p > 0.05). In order to determine inter-batch variability, the in vitro deposition of five batches of M0F were compared. The ED and the SX content of the stage one and two of the TSI were 79.9 0.8%, 74.2 0.7% and 5.4 0.6%, respectively, and showed no significant differences ( p > 0.05). No significant intra- or interbatch variability in ED, and SX content of stage one and two of the TSI was seen for M20F when tested using the same procedure. Homogeneity studies together with intra- and inter-batch dispersion comparisons confirmed that the powder mixing process used in this research produced homogeneous mixtures with reproducible in vitro aerosol deposition. The ED and FPF for all mixtures were consistent with earlier findings from our laboratory.13 Before storage at specific RH, the ED of SX–CL JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
1020
DAS ET AL.
Figure 2. Scanning electron micrographs (SEM) of (A) Salmeterol xinafoate magnification (2306), and SX–CL mixtures containing 0%, 10% and 20% micronised lactose, (B) M0F, magnification (800), (C) M10F, magnification (412) and (D) M20F, magnification (601) at 10 kV, respectively.
mixture, M0F (80.0 3.1%) was higher than the other formulation used in this study. As the concentration of ML in the mixtures increased, the ED decreased, for example, M5F (72.5 3.6%), M10F (68.6 2.9%), M20F (57.6 1.5%) and SX alone formulation (60.4 4%). The minimum ED, shown by the mixture containing 20% micronised lactose, was not significantly different from that of drug alone formulation ( p > 0.05). The addition of the CL to SX decreased dispersion performance with FPF of the SX–CL mixture (5.5 0.7%) decreasing from 10.3 1.01% for the SX alone formulation (Fig. 3B). The addition of ML increased dispersion performance with the FPF significantly increased to 8.1 0.8% for M5F ( p < 0.05) and to 9.1 0.8% for M10F ( p < 0.05). The highest FPF was shown by mixture M20F (11.3 0.6%) though it was not significantly different from the drug only formulation ( p > 0.05). JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
Dispersion Mechanism of Powder Formulations In order to understand the reasons for potential changes observed during storage, the mechanisms of dispersion associated with the mixtures used in this study need to be defined. Micronised powders will usually agglomerate due to the dominance of the interparticulate interaction forces over the gravitational detachment forces.21 When larger carrier particles (such as CL) are mixed with micronised powders, micronised powder particles can exist both as adhered particles to the carrier surface (either individually or as networks of particles like agglomerates), and as agglomerates existing freely in the powder mixture and separated from the carrier surface. For complex mixtures used in this study containing SX, CL and ML, both SX and ML can (a) adhere to the CL surface, (b) cohere to form SX and ML agglomerates and DOI 10.1002/jps
INFLUENCE OF RH ON DISPERSION OF SX POWDERS
1021
process is facilitated by the incorporation of ML particles into the agglomerate since mixed agglomerates of SX and ML have reduced strength, possibly due to the change in the packing fraction of the particles within the agglomerate.13
Dispersion Changes (ED and FPF) of DPI Formulations during Storage The FPF was determined before storage and at 7, 14, 28, 56, 84, 168, 364 and 564 days after storage. For 75% RH, two additional readings were taken at 24 and 48 h. In this article, data are only shown where significant changes in dispersion performance occurred. ED
Figure 3. (A) Emitted dose (ED) and (B) fine particle fraction (FPF) of SX of five formulations: SX alone (D), and its four mixtures M0F, M5F, M10F and M20F containing 0%, 5%, 10% and 20% micronised lactose, respectively during storage at 75% RH for 18 months.
(c) adhere to each other to form mixed agglomerates with different ratios of SX and ML. In the study of the formulations used in this research, SX dispersion will occur due to (a) particle detachment from the CL surface and (b) deagglomeration. SX alone mixtures will disperse through deagglomeration of the SX agglomerates. The major mechanisms for dispersion for mixtures containing SX and CL only (where the SX particles are essentially located at the CL surface as individual particles or agglomerates—see Fig. 2B) are SX detachment from the CL surface and SX deagglomeration from the agglomerates on the surface. These mixtures tend to perform poorly due to the relatively strong SX attachment to the CL carrier particles.13 However, when ML is incorporated into the SX– CL mixtures, previous studies13 have demonstrated that deagglomeration of the SX–ML agglomerates is the dominant mechanism of SX dispersion in the powder mixtures. The deagglomeration DOI 10.1002/jps
During storage at 75% RH (Fig. 3A), the ED of mixture M20F sharply increased from 57.6 3.1% to 77.3 4% within a week and did not significantly change over the remaining period. The ED for the mixture M10F gradually increased from 69.1 2.8% before storage to 76.3 3.7% at the end of the study. These two points were statistically different ( p < 0.05); however, no significant change was observed in the ED of all the other formulations (D, M0F and M5F) for the whole study period. At 33% and 55% RH, there was no significant change in ED for any formulation during the study period except for the mixture M20F at 55% RH (data are not shown in figures). The ED of M20F significantly increased from 57.6 2.9% to 65.5 3.2% within 3 days ( p < 0.05), and did not change significantly over the remaining period ( p > 0.05). FPF During storage at 75% RH (Fig. 3B), the FPF of mixture M20F significantly decreased from 11.3 0.7% to 7.7 0.6% within 8 weeks ( p ¼ 0.008). The FPF of mixtures M5F and M10F did not change over the first 6 months ( p > 0.05); however, the FPF had significantly decreased at 18 months ( p < 0.05). Other formulations (D and M0F) at 75% RH and all formulations at 33% and 55% RH showed no significant change in FPF ( p > 0.05) for the whole study period. When M20F was stored at 95% RH, the ED increased from 57.6 2.2% to 77.9 3.2% just in 3 days (Fig. 4A). Then, it gradually rose to 87.0 3.4% in 12 months. Accordingly, the FPF JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
1022
DAS ET AL.
hydrophobic powder having a contact angle of 67.78 with water.25 Previous studies have demonstrated that capillary forces do not develop in SX even at RH of 75%.26 Lactose monohydrate is hydrophilic.25 During storage, CL and ML were found to gain only a small amount of water, around 0.1% w/w and 0.2% w/w, respectively, at a RH of 45% (Fig. 1). Capillary forces in lactose monohydrate have not been detected at less than 55% RH.26 Therefore, the contribution of capillary forces due to lactose at 33% RH was unlikely to be significant in the mixtures studied (M0F, M5F, M10F and M20F). Electrical forces such as contact potential and Coulombic forces might be generated during powder handling and contribute to interparticulate interactions of both SX and lactose at this low RH. Charge decay would be expected to occur, but any change in the extent of electrostatic particle interactions did not manifest itself in any observable changes in ED and FPF for the formulations during storage at this RH. Figure 4. (A) Emitted dose (ED) and (B) fine particle fraction (FPF) of SX of mixture containing 20% micronised lactose (M20F), during storage at 95% RH for 12 months.
significantly declined ( p ¼ 0.001) from 11.3 0.7% to 4.9 0.5% within the first 8 weeks of storage (Fig. 4B). The FPF gradually decreased to 3.4 0.3% over the remaining storage period.
Possible Particle Interactions at Low to Medium Environmental RH Conditions That Lead to No Significant Change in Dispersion RH 33% During storage at 33% RH, there was no significant change in dispersion of SX for any formulation. This means that during storage there has been little change in the magnitude of interaction between particles in these mixtures. The obvious force components likely to change during storage are associated with capillary and electrostatic interactions. Though not commonplace, there are a few examples of formation of capillary forces in some systems at low RH.22–24 The emergence and extent of capillary forces mainly depend on material properties. For example, capillary forces are almost negligible for hydrophobic materials, while they can be considerable for hydrophilic substances. SX is a JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
RH 55% At 55% RH, there was no significant change in ED for any formulation during the study period, except for an increase for the mixture M20F. The FPF of all formulations did not significantly change for the whole study period. At 55% RH, a balance between diminishing electrostatic forces and emerging capillary forces was expected. The outcome of storage of the SX formulations was consistent with an earlier report26 which observed only a minimal increase in auto adhesion force in lactose monohydrate up to a RH of 55%. A small increase in ED of M20F within the first 3 days could be due to slightly higher amount of moisture associated with higher percentage of ML facilitating increased agglomeration. However, this did not have any effect on FPF of M20F.
Possible Particle Interactions at High Environmental RH Conditions That Lead to Change in Dispersion RH 75% There was no significant change ( p > 0.05) in both ED and FPF during storage of the SX alone formulation at 75% RH (Fig. 3). This could be attributed to the very low likelihood of capillary interaction at this RH by hydrophobic SX.27 This outcome was supported by the earlier findings that auto adhesion forces of SX were less sensitive to storage humidity and the absence of capillary DOI 10.1002/jps
INFLUENCE OF RH ON DISPERSION OF SX POWDERS
forces in SX when stored at 75% RH or greater for 72 h.26 Jashnani et al.28 also concluded that hydrophobic powders were less sensitive to high storage humidity than hydrophilic powders during a study of the emptying and aerosolisation properties of DPIs. The FPF and ED of SX–CL mixture with no additional micronised lactose (M0F) did not significantly change at 75% RH for 18 months (Fig. 3). In general, capillary forces are not important at a RH less than 50%; however, they start to dominate interparticulate interactions at a RH greater than 60%.3 Capillary forces were observed by other scientists to contribute to adhesion forces between lactose monohydrate and SX when stored at a RH of 75% or 95%, and an equilibrium value of the capillary force was reached within 72 h.29 Thus, capillary interactions between the SX and CL were likely to increase the total interparticulate interaction during storage at 75% RH. However, any increase in interaction that might have occurred did not lead to ED and FPF change during storage. Since the dominant mechanism of dispersion for mixtures of SX and CL at the SX concentration used in this study has been shown to be deagglomeration of the surface agglomerates of SX14 and since capillary interactions within SX particles were less likely to be affected by storage at 75% RH, changes in adhesion that might occur due to any capillary interaction between SX and CL may be less important in the overall dispersion process. At 75% RH, no significant change was observed in the ED of mixture containing 5% micronised lactose (M5F) for the whole study period (Fig. 3). The ED for mixture containing 10% micronised lactose (M10F) gradually increased throughout the study period. On the other hand, the ED of M20F sharply increased from 57.6 3.1% to 77.3 4% within a week (Fig. 3A) though it did not significantly change ( p > 0.05) over the remaining period. The ED increase of M20F and M10F during storage at 75% RH could be due to increased agglomeration caused by increased capillary interactions between the micronised powders leading to better flow properties and entrainment of the mixtures. Significant difference ( p < 0.05) in dispersion behaviour demonstrated by the changes in FPF was observed during storage when micronised lactose (ML) was added to the formulation. The onset of the decrease in FPF during storage depended on the concentration of additional ML. For example, a significant decrease ( p > 0.05) in the FPF for DOI 10.1002/jps
1023
mixtures containing 5% and 10% additional micronised lactose (M5F and M10F) was first observed at 18 months storage while that for M20F was observed at 4 weeks (Fig. 3B). The FPF of SX for M20F significantly decreased ( p < 0.05) from 11.3 0.7% to 8.2 0.6% within the first 4 weeks and continued to decrease to 7.7 0.6% until week 8 of storage, and then it did not significantly change ( p > 0.05) for the remaining storage period (Fig. 3B). It is interesting to note that the FPF of M20F after 8 weeks storage at 75% RH was not significantly different ( p > 0.05) from the FPF of mixture M5F before storage. RH 95% When M20F was stored at 95% RH, the ED significantly increased in just 3 days (Fig. 4A) while the FPF significantly declined ( p ¼ 0.001) from 11.3 0.7% to 4.9 0.5% within the first 8 weeks of storage (Fig. 4B), which is almost a 57% reduction from the initial value. The FPF gradually decreased to 3.4 0.3% over the remaining storage period. At 95% RH, surface moisture of ML and CL increased to 0.45% (w/w) and 0.3% (w/w) in 1 and 2 weeks, respectively. A more pronounced decreasing effect was observed more quickly at 95% RH than it was at 75% RH because of higher capillary forces followed by solid bridging.
Proposed Mechanisms for the Change in Dispersion of the SX Mixtures during Storage at High RH The critical parameters in the decrease in dispersion of the SX mixtures during storage were (a) the presence of ML in sufficient concentrations (clearly evident at 20%) and (b) the presence of sufficient atmospheric moisture during storage (75% RH and greater). The dispersion of SX from mixtures containing micronised lactose has been shown to be mainly controlled by the deagglomeration of mixed agglomerates of SX and ML. The presence of ML was well known to increase dispersion performance due to its influence on the agglomerate structure,13 and its contribution to decreased dispersion during storage at RH of 75% and greater was also likely to be associated with its influence on agglomerate structure. Micronised lactose was seen to absorb moisture during storage at high RH (Fig. 1B); therefore, capillary interactions will contribute particularly to greater cohesion between ML particles and also to greater adhesion between JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
1024
DAS ET AL.
ML and SX. In order to understand the extent of capillary interaction influence within the agglomerate, capillary forces ( Fc) were theoretically calculated using the equation mentioned by Zimon.3 Fc ¼ 2prg cos u where, g is surface tension, r is the radius of particle in contact, and u is contact angle. For the contact between two particles, capillary forces ( Fc) can be calculated extending the above equation to the following: Fc ¼ 2pr1 g cos u1 þ 2pgr2 cos u2 where, g is surface tension, r1, r2 are the radii of particles in contact, and u1, u2 are contact angles of the contacting particles. The surface tension of water is 0.0729 N/m3, while the radii for SX and micronised lactose particles are 1.10 and 1.55 mm, respectively. The contact angle of water for SX is 67.78. Assuming the contact angle for lactose is zero, the calculated capillary forces for SX–SX, SX–ML and ML–ML contacts were 0.382, 0.836 and 1.42 mN, respectively. Capillary forces have fivefold higher effect on the adhesion forces for ML–ML contacts compared with that of the SX–SX contact. In order to understand the relative abundance of different types of contacts, a very simple randomised distribution matrix of SX and ML within the mixed agglomerate was designed. In mixtures containing 5%, 10% and 20% ML, the ratio between SX and ML are 1:2, 1:4 and 1:8, respectively. A hypothetical randomised distribution of SX and ML in the agglomerate (Fig. 5) shows that the abundance of ML–ML contact increased while SX–SX contact or SX–ML contact decreased with increasing concentration of micronised lactose in mixtures. For example, the hypothetical number of ML–ML contacts in the randomly distributed agglomerate of the M5F was 88, whereas 186 contacts occurred in the agglomerates of M20F. Moreover, there was no SX–SX contact in M20F, although six SX–SX contacts were present in M5F. Therefore, during storage at high humidity conditions, the extent of ML–ML contact and the magnitude of capillary forces increased with the increased concentration of ML in mixtures. These additional capillary forces would increase the extent of particle interactions within the agglomerates, increasing the agglomerate strength and decreasing the FPF during storage. Although the randomisation process can be JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
Figure 5. Random distribution of micronised lactose (ML) and salmeterol xinafoate (SX) in mixtures of 2.5% SX with (A) 5%, (B) 10% and (C) 20% micronised lactose.
criticised as it did not consider the polydispersity of SX and ML particles and variation in particle shape, this simplistic approach was still useful to view the propensity of different contact. Moisture also has some plasticising effect and will affect the interaction in other ways.25 Surfaces of ML are softened after absorbing moisture at high RH and contact area is increased through particle deformation. Therefore, in addition to the contribution of the increased capillary interactions, higher intermolecular interactions between the micronised particles of the agglomerate could occur and lead to greater interaction and stronger agglomerates with a resultant decreased dispersion performance. In addition to increased capillary and intermolecular interactions during storage, slight dissolution at the surface of micronised particles was likely to occur in presence of absorbed moisture. The temperature of storage fluctuated between 23 and 278C during study period and was conducive to the formation of solid bridges formed through the crystallisation of lactose during temperature variations. Solid bridges formation is not uncommon in pharmaceutical systems. Lactose monohydrate was found to dissolve under the influence of high RH.29 Liquid bridges are often followed by solid bridges due to crystallisation to adjust evaporation of moisture from liquid bridges.30 Weak solid bridges of lactose after recrystallisation were also claimed earlier by other scientists.21 Solid bridges have been also observed in model sulphathiazole–Emdex DOI 10.1002/jps
INFLUENCE OF RH ON DISPERSION OF SX POWDERS
and succinyl sulphathiazole–Emdex1 mixtures during storage of dry powders over 7 days.31 Thus, during storage at high humidity, interparticulate interactions within the agglomerate increased through possible increased capillary and intermolecular interaction forces and solid bridging. These magnitudes of the interactions were greater for ML–ML contacts particularly as the concentration of ML increased within the mixtures and the SX particles within the agglomerates were encaged within the agglomerates by the micronised lactose matrix. Thus dispersion performance decreased during storage. Comparison of Agglomerate Strengths of Mixtures before and after Storage In order to support the above predicted mechanism, the aerodynamic particle sizes of mixtures containing 0%, 5%, 10% and 20% micronised lactose before and after storage at 75% RH for 3 months were determined using an Aerosizer. Each mixture was dispersed at four different shear pressures (3.4, 10.3, 20.7 and 27.6 kPa) to develop particle size-shear pressure profiles to provide an understanding of the extent and propensity of the mixtures to deagglomerate.13 No significant difference in the extent or rate of deagglomeration was observed for the SX–CL mixture (M0F) before and after storage (Fig. 6A).
1025
The aerodynamic particle size was around 100 mm at 3.4 kPa. This size was dominated by the particle size of CL as the mixture contained 97.5% CL. The aerodynamic particle sizes slightly decreased with increasing shear pressures both before and after storage, possibly because of dispersion of drug particles/agglomerates on the surface of CL. It should be noted that when the CL alone was stored under these conditions, the particle sizeshear pressure profile was not significantly different before and after storage ( p > 0.05) and that there was no significant change in particle size between the shear pressures ( p > 0.05). This indicated that no comminution of the carrier occurred. The slightly higher aerodynamic particle size of the mixture containing 5% micronised lactose (M5F) was an indication of increased agglomerate formation in M5F because of the presence of micronised lactose. However, no significant difference in the extent and propensity of deagglomeration was observed before and after storage. This finding supports the proposed mechanisms discussed above that the particle interactions in M5F were not sufficient to bring about any significant changes in interparticulate interactions after storage (Fig. 6B). The aerodynamic particle size of mixture containing 10% micronised lactose (M10F) was higher after storage than before at the lowest shear pressure (3.4 kPa) indicating that agglomerates were not
Figure 6. (A–D) Comparison between before and after storage mixtures containing 0%, 5%, 10% and 20% micronised lactose, respectively. DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
1026
DAS ET AL.
as dispersible at this low shear pressure after storage (Fig. 6C). However, there was no significant difference in deagglomeration at higher shear pressures (10.3, 20.7 and 27.6 kPa) between before and after storage ( p > 0.05). In contrast, the particle size-shear pressure profile of the mixture containing 20% micronised lactose (M20F) was higher after storage than before at all shear pressures (Fig. 6D). There was no significant difference between particle sizes at 20.7 kPa and at 27.6 kPa after storage ( p > 0.05). Thus, the extent of deagglomeration was less after storage, although the propensity to deagglomerate over the shear pressure range was similar. This is a clear indication of the formation of stronger agglomerate in M20F after storage at 75% RH and the outcomes of this Aerosizer study provided strong support to the mechanistic proposals derived from hypothetical distribution matrix and capillary force calculation.
CONCLUSIONS The dispersion performance of SX mixtures for inhalation, shown by changes to the FPF, was influenced predominately by the presence of micronised lactose in the formulation and the RH of the storage condition. A significant decrease in FPF on storage over 18 months was observed in formulations containing 10% and 20% ML when the RH was 75% or greater. All other formulations showed no change in FPF during storage at 75% and all formulations showed no change in FPF at 33% and 55% RH. For mixtures containing high concentrations of micronised powders, such as 2.5% SX–CL mixtures containing 10% and greater ML, previous studies have demonstrated that the dominant dispersion mechanism was deagglomeration of SX–ML agglomerates formed during mixing.13 At 75% RH, capillary forces between particles in the agglomerate contributed to increased particle interactions within the mixture giving rise to stronger agglomerates. In particular, a calculation of the capillary interactions demonstrated that the capillary interaction between ML particles was significantly greater than SX–ML or SX– SX capillary interactions. Thus, the increase in agglomerate strength occurred predominately in mixtures containing high concentrations of ML. The use of a simple hypothetical two-dimensional model of a random distribution of SX and ML particles in an agglomerate demonstrated that JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
the increased likelihood of ML–ML interactions within the agglomerate as the concentration of ML increased. Therefore, while the presence of moisture in the storage environment increased all capillary interactions within the agglomerate giving rise to stronger agglomerates, the propensity for increased extent and magnitude of ML– ML interactions at high concentrations of ML was likely to result in the development of a ‘‘cage-like’’ structure with the SX being trapped within the micronised lactose matrix. The increased capillary interactions may also have been supplemented with increased intermolecular interactions due to particle deformation at high moisture conditions and the development of solid bridges due to temperature variations. The use of a novel particle size-shear pressure profile generated by the Aerosizer provided confirmation of increased agglomerate strength for mixtures containing high concentrations of ML after storage at high RH. The study was conducted on a laboratory scale using small batches (5 g). The mixing conditions, both in relation to batch size and type of mixer, will influence the magnitude of shear generated during mixing procedures and thus will influence the extent and structure of agglomerates. This research provides an explanation of the changes observed during storage of SX mixtures for inhalation based on the agglomerate characteristics developed during laboratory mixing. These general principles will apply to other mixing conditions where agglomerates are produced, but these laboratory scale findings should be carefully interpreted for industry scale operation.
ACKNOWLEDGMENTS Shyamal Das would like to acknowledge the scholarship support through the Faculty of Pharmacy, Monash University and thank to Dr Handoko Adi and Dr Herbert Chiou for their support during using SEM and Aerosizer. He also acknowledges Dr Richard J. Prankerd and Dr Joseph A. Nicolazzo for their help during TGA use and statistical analysis, respectively.
REFERENCES 1. Borgstrom L, Asking L, Lipniunas P. 2005. An in vivo and in vitro comparison of two powder inhalers following storage at hot/humid conditions. J Aerosol Med 18:304–310. DOI 10.1002/jps
INFLUENCE OF RH ON DISPERSION OF SX POWDERS
2. Stewart PJ. 1986. Particle interaction in pharmaceutical systems. Pharm Int 7:146–149. 3. Zimon AD, editor. 1982. Adhesion in a gas medium. Sources of adhesion. In: Adhesion of dust and powder. 2nd edition. New York: Consultants Bureau. pp 93–144. 4. Podczeck F, Newton JM, James MB. 1996. The influence of constant and changing relative humidity of the air on the autoadhesion force between pharmaceutical powder particles. Int J Pharm 145: 221–229. 5. Young PM, Price R, Tobyn MJ, Buttrum M, Dey F. 2003. Effect of humidity on aerosolization of micronized drugs. Drug Dev Ind Pharm 29:959–966. 6. Hindle M, Makinen GM. 1996. Effects of humidity on the in vitro aerosol performance and aerodynamic size distribution of cromolyn sodium for inhalation. Eur J Pharm Sci 4:S142. 7. Braun MA, Oschmann R, Schmidt PT. 1996. Influence of excipients and storage humidity on the deposition of disodium cromoglycate (DSCG) in the twin impinger. Int J Pharm 135:53–62. 8. Young PM, Price R. 2004. The influence of humidity on the aerosolisation of micronised and SEDS produced salbutamol sulphate. Eur J Pharm Sci 22: 235–240. 9. Lida K, Hayakawa Y, Okamoto H, Danjo K, Luenberger H. 2004. Influence of storage humidity on the in vitro inhalation properties of salbutamol sulphate dry powder with surface covered lactose carrier. Chem Pharm Bull 52:444–446. 10. Zeng XM, MacRitchie HB, Marriott C, Martin GP. 2007. Humidity-induced changes of the aerodynamic properties of dry powder aerosol formulations containing different carriers. Int J Pharm 333:45–55. 11. Harjunen P, Lankinen T, Salonen H, Lehto VP, Jarvinen K. 2003. Effects of carriers and storage of formulation on the lung deposition of a hydrophobic and hydrophilic drug from a DPI. Int J Pharm 263:151–163. 12. Maggi L, Bruni R, Conte U. 1999. Influence of moisture on the performance of a new dry powder inhaler. Int J Pharm 177:83–91. 13. Adi H, Larson I, Chiou H, Young P, Traini D, Stewart P. 2006. Agglomerate strength and dispersion of salmeterol xinafoate from powder mixtures for inhalation. Pharm Res 23:2556–2565. 14. Islam N, Stewart P, Larson I, Hartley P. 2004. Effect of carrier size on the disperson of salmeterol xinafoate from interactive mixtures. J Pharm Sci 93:1030–1038. 15. Louey MD, Stewart PJ. 2002. Particle interactions involved in aerosol dispersion of ternary interactive mixtures. Pharm Res 19:1524–1531. 16. Always B, Sangchantra R, Stewart PJ. 1996. Modelling the dissolution of diazepam in lactose interactive mixtures. Int J Pharm 130:213–224.
DOI 10.1002/jps
1027
17. Liu J, Stewart PJ. 1998. Deaggregation during the dissolution of Benzodiazepines in Interactive mixtures. J Pharm Sci 87:1632–1638. 18. Crooks MJ, Ho R. 1976. Ordered mixing in direct compression of tablets. Powder Technol 14:161– 167. 19. Callahan JC, Cleary GW, Elefant M, Kaplan G, Kensler T, Nash RA. 1982. Equilibrium moisture content of pharmaceutical excipients. Drug Dev Ind Pharm 8:355–369. 20. Berard V, Lesniewska E, Andres C, Pertuy D, Laroche C, Pourcelot Y. 2002. Affinity scale between a carrier and a drug in DPI studied by atomic force microscopy. Int J Pharm 247:127–137. 21. Visser J. 1995. Particle adhesion and removal: A review. Particulate Sci Technol 12:169–196. 22. Eastman T, Zhu DM. 1996. Adhesion forces between surface-modified AFM tips and a mica surface. Langmuir 12:2589–2862. 23. Price R, Young PM, Edge S, Staniforth JN. 2002. The influence of relative humidity on particulate interactions in carrier-based dry powder inhaler formulations. Int J Pharm 246:47–59. 24. Xu L, Lio A, Ogletree DF, Salmeron M. 1998. Wetting and capillary phenomena of water on mica. J Phys Chem B 102:540–548. 25. Podczeck F, Newton JM, James MB. 1997. Influence of relative humidity of storage air on the adhesion and autoadhesion of micronized particles to particulate and compacted powder surfaces. J Colloid Interface Sci 187:484–491. 26. Podczeck F, Newton JM, James MB. 1996. The influence of constant and changing relative humidity of the air on the autoadhesion force between pharmaceutical powder particles. Int J Pharm 145: 221–229. 27. Podczeck F, Newton JM, James MB. 1995. Assessment of adhesion and autoadhesion forces between particles and surfaces. Part II. The investigation of adhesion phenomenoa of salmeterol xinafoate and lactose monohydrate particles in particle-onparticle and particle-on surface contact. J Adhes Sci Technol 9:475–486. 28. Jashnani RN, Byron PR, Dalby RN. 1995. Testing of dry powder aerosol formulations in different environmental conditions. Int J Pharm 113:123–130. 29. Podczeck F, Newton JM, James MB. 1997. Variations in the adhesion force between a drug and a carrier particles as a result of changes in the relative humidity of the air. Int J Pharm 149:151– 160. 30. Padmadisastra Y, Kennedy RA, Stewart PJ. 1994. Influence of carrier moisture adsorption capacity on the degree of adhesion of interactive mixtures. Int J Pharm 104:R1–R4. 31. Padmadisastra Y, Kennedy RA, Stewart PJ. 1994. Solid bridge formation in sulphonamide-Emdex interactive systems. Int J Pharm 112:55–63.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
Drug Delivery, Disposition and Dynamics, Monash Institute of Pharmaceutical Sciences, Monash University (Parkville Campus), 381 Royal Parade, Victoria 3052, Australia 2
Advanced Drug Delivery Group, Faculty of Pharmacy, University of Sydney, New South Wales 2006, Australia
Received 17 April 2008; accepted 12 June 2008 Published online 25 July 2008 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21500
ABSTRACT: The in vitro dispersion of salmeterol xinafoate (SX) alone and four SX (2.5%)–coarse lactose (CL) mixtures containing 0%, 5%, 10% and 20% micronised lactose (ML) was monitored during 18-month storage at 33%, 55% and 75% relative humidity (RH) using a twin stage impinger. The surface moisture was monitored over 2 months by thermo gravimetric analysis. The morphology was determined by scanning electron microscopy. An aerosizer was used to compare the agglomerate strengths of formulations before and after storage at 75% RH. Upon storage, no significant difference occurred in fine particle fraction (FPF) of any formulation at 33% and 55% RH. Within 8 weeks, the FPF of mixture containing 20% ML (M20F) significantly decreased from 11.3% to 7.7% at 75% RH ( p ¼ 0.008) and to 4.9% at 95% RH ( p ¼ 0.001). The calculated capillary forces were greater for ML–ML contact than other particle interactions and the propensity of ML–ML contacts was higher in M20F. The agglomerate strength of M20F significantly increased after storage. The study concluded that the critical factors for decreased dispersion of SX formulations were RH of 75% or greater and the presence of high concentrations of ML due to capillary forces and/or solid bridge formation of ML leading to increased agglomerate strength. ß 2008 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 98:1015–1027, 2009
Keywords: dispersion; storage stability; salmeterol xinafoate; relative humidity; dry powder inhaler; agglomerate
INTRODUCTION Drug dispersibility changes during storage present problems for dry powder inhaler (DPI) development.1 Dispersibility is principally governed by the magnitude of interparticulate drug– drug and drug–excipient forces. Interparticulate forces are a combination of various electrical (contact potential and Coulombic) and nonelectrical forces (van der Waals’ forces, capillary
Correspondence to: Peter Stewart (Telephone: 61-3-99039517; Fax: 61-3-9903-9583; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 98, 1015–1027 (2009) ß 2008 Wiley-Liss, Inc. and the American Pharmacists Association
forces, mechanical interlocking and solid bridging).2 These forces can vary in different relative humidities (RHs) depending on material properties and storage duration. Forces like contact potential and Coulombic forces are more dominant at low RH since charge decay is minimised while capillary forces and solid bridging develop at high RH due to the formation of liquid bridges between particles. van der Waals’ forces and mechanical interlocking can contribute at all RHs. Therefore, the strength of interparticulate forces, manifest in DPI formulations by micronised drug– carrier interactions and micronised drug–fine excipient interactions within agglomerates, could be different at different RHs. Furthermore, forces between particles do not develop at the same rate.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
1015
1016
DAS ET AL.
Forces like contact potential, van der Waals’ forces and Coulombic forces develop very quickly, while capillary forces take sometime to develop (30 min)3 and to reach in equilibrium (up to 72 h)4 while solid bridges form after long term storage at high RH and depend on temperature and humidity cycling. Therefore, the hypothesis of this study was that the interparticulate forces may increase during storage, both in the short and long term, with resultant compromised performance of the inhalation powder. There are a number of previous studies1,5–12 which evaluated the influence of storage humidity on dispersion of DPI formulations. A few of the studies were on drug alone formulations (hydrophilic or hydrophobic). For example, Young et al.5 observed that the fine particle fraction (FPF) of disodium cromoglycate and salbutamol sulphate decreased at elevated humidity while that of triamcinolone acetonide increased when stored at 15–75% RH for 12 h. Other studies concentrated on storage humidity effects of DPIs containing carrier particles. For example, Hindle and Makinen6 observed a decrease in mean fine particle dose of micronised cromolyn sodium from its blend with lactose monohydrate from 3.9 0.2 mg at ambient conditions (20 38C and 30 5% RH) to 1.2 0.2, 0.73 0.2 and 0.2 0.04 mg following exposure at 53%, 66% and 76% RH, respectively. Braun et al.7 found a greater decrease in FPF at 55% RH than at 33% RH when mixtures of disodium cromoglycate with two lactose monohydrates and two dextrose monohydrates were stored for 27 days. In Braun et al.’s study, formulations were stored in gelatine capsules which may have a significant effect on dispersibility. In another study, Young and Price8 observed a higher reduction in FPF for the mixture of lactose carrier with micronised salbutamol sulphate than with solution enhanced dispersion by supercritical fluids (SEDS) produced salbutamol sulphate after storage at high RH for 75 h. In vitro inhalation was also studied by Lida et al.9 for the mixture of salbutamol sulphate and lactose carrier covered with varying surface coverage with magnesium stearate and stored at different RH for 7 days. Most recently Zeng et al.10 observed a small decrease in FPF when a mixture of salbutamol and lactose was stored at 75% RH for 6 days. Studies were also carried out to determine the suitable packaging requirements offering better protection from storage humidity.11,12 Only few of these studies5,6,8–10 exposed their formulations in open pans to storage humidity. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
Moreover, these studies concentrated on the effect of RH either on drug alone (D) or on drug–coarse carrier mixtures for short periods of time (maximum 7 days exposed in open pans at storage humidity). The studies did not include drug– coarse carrier mixtures with added fine carriers even though these formulations have greater dispersion capability.13–15 Therefore, the objective of this study was to evaluate the dispersion of formulations of different compositions (drug alone, drug–coarse carrier mixture and drug– coarse carrier mixture containing varying concentrations of additional fines) by exposing the formulations for up to 18 months to a range of RHs existing from hot and dry (35% RH) to hot and humid (75% RH) climatic zones. In addition, the study focussed on understanding the reasons for any changes observed. The study used salmeterol xinafoate (SX) as a model drug, Inhalac 120 as the coarse lactose (CL) carrier and specially micronised lactose monohydrate as the fine added excipient.
MATERIALS AND METHODS Materials Inhalation grade micronised SX (Glaxo Wellcome, Ware, UK) was used as model drug. a-Lactose monohydrate, CL (Inhalac 1120, Meggle AG, Wasserburg, Germany) was used in ‘‘as supplied’’ form in all the mixtures as the coarse carrier. Inhalac 120 was selected because of its low concentration (1.3% v/v) of inherent fines. Micronised lactose (ML) used in this study was prepared through micronisation of a-lactose monohydrate (Lactose New Zealand, New Zealand) using fluid energy milling.13 Methanol (CH3OH, HPLC grade, Merck KGaA, Darmstadt, Germany), milli-Q grade water (Millipore Corporation, Melsheim, France), ammonium acetate (Analar, BDH Laboratories, Victoria, Australia) were used for HPLC analysis. Magnesium chloride hexahydrate (MgCl2, 6H2O), potassium carbonate (K2CO3), magnesium nitrate (MgNO3), sodium chloride (NaCl) and potassium nitrate (KNO3) (all from BDH Laboratories) were used for generating different RH.5 Methods Storage Conditions Saturated solutions of magnesium chloride, potassium carbonate, magnesium nitrate, sodium DOI 10.1002/jps
INFLUENCE OF RH ON DISPERSION OF SX POWDERS
chloride and potassium nitrate were used to produce 33%, 45%, 55%, 75% and 95% RH, respectively in desiccators.5 The RH was observed to vary by only 2% over the study period. Samples of CL and ML kept in open pans were stored at 45%, 75% and 95% RH for 2 months for the moisture adsorption study. Five formulations: SX alone (D) and four SX–CL mixtures containing 0%, 5%, 10% and 20% ML (M0F, M5F, M10F and M20F, respectively) were stored at 33%, 55% and 75% RH for 18 months for the dispersion study. The mixture containing 20% ML (M20F) was stored at 95% RH for 12 months for a dispersion study. The storage temperature was 25 28C. Determination of Surface Moisture by Thermo Gravimetric Analysis (TGA) The moisture content of the powders was determined by TGA every day for 2 weeks followed by once a week up until 2 months. The sample (approximately 2–5 mg) was placed in a platinum sample pan under a nitrogen purge. The TGA was operated at 58C/min from 40 to 1808C. The moisture content was calculated on dry weight basis and the final weight was taken at 1708C. Surface moisture was calculated by deducting the % moisture for water of crystallisation from the % moisture content. % moisture ¼
ðinitial weight final weightÞ 100 final weight
% surface moisture ¼ % moisture % water of crystallization
Preparation of Powder Formulations Four SX (2.5%)–CL mixtures containing 0%, 5%, 10% and 20% micronised lactose (M0F, M5F, M10F and M20F) were prepared according to a previously validated mixing method.16,17 The micronised powders (SX or SX and ML) were placed between equal amounts of CL in a test tube. Three ceramic beads (approximately 10 mm diameter) were placed inside the test tube which was stoppered and inverted several times to prevent the drug from sticking to the sides of the test tube. The test tube was then shaken vigorously for 5 min by hand. The ceramic beads provided a ball-milling effect for breaking up DOI 10.1002/jps
1017
agglomerates. The batch size for these formulations was 5 g. Homogeneity of Powder Mixtures The average drug content and the variability between samples, expressed by the CV, were used to quantify the homogeneity of each powder mixture. Twenty samples (20 0.5 mg each) were accurately weighed using an analytical balance (Mettler AT 261 Deltarange, Greifensee-Zurich, Switzerland). All samples were dissolved in 40% methanol (HPLC grade) and the amount of SX was determined by a validated UV assay. A mixture with mean drug content within 95–105% of the theoretical value and a CV less than 5% was regarded to have an acceptable degree of homogeneity. These specifications indicated that 95% of samples would fall within 10% of the mean.18 In Vitro Drug Dispersion by Twin Stage Impinger (TSI) In vitro aerosol deposition was determined using a twin stage impinger (TSI, Apparatus, A; British Pharmacopoeia, 2000) (Copley, UK) and a Rotahaler (Glaxo Wellcome, Melbourne, Australia). The airflow was created by a vacuum pump (Model OD 5/2, Dynavac Engineering, Melbourne, Australia) fitted to the mouth piece of TSI and the flow rate was adjusted to 60 L/min before each measurement. The temperature (25 28C) and RH (45 2% RH) of the surrounding environment was measured by a thermohygrometer (Shinyei TRH-CZ, Osaka, Japan). Seven and 30 mL of 40% methanol (HPLC grade) were placed in the stage one and stage two of the TSI, respectively. Size 3 hard gelatine capsules (Capsugel, Sydney, Australia), loaded with the powder formulations (20 0.5 mg) or drug (0.50 0.03 mg), were inserted into a Rotahaler (Glaxo Welcome, UK). The Rotahaler was placed into a moulded mouthpiece attached to the TSI, twisted and the powder was then dispersed by drawing 4 L of air during each measurement (4 s at 60 L/min). Each part (inhaler, stage one and stage two) was then rinsed with 40% methanol (in water) and the rinsed liquid was collected and diluted to an appropriate volume (100 mL). The drug concentration was then determined by a validated high performance liquid chromatographic assay. Five replicates of TSI measurement were performed for each formulation, and the TSI measurements were conducted immediately after removal of the formulations from desiccators. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
1018
DAS ET AL.
The total amount of drug collected from inhaler, stage one (S1) and stage two (S2) was regarded as the recovered dose (RD), and was calculated as a percentage of total dose. The emitted dose (ED) was the amount of drug that was deposited in stage one and stage two and was calculated as the percentage of RD. The FPF was defined as the percentage of RD that was deposited in stage two. ED ¼
ðS1 þ S2 Þ 100 RD
FPF ¼
S2 100 RD
High Performance Liquid Chromatography (HPLC) Assay SX recovered by TSI was analysed by HPLC using a C18 column (5 mm, 4.6 mm 150 mm, Apollo Alltech Associates Inc., Deerfield, IL) and a UV detector (Shimadzu Diode Array Detector, SPD M10A VP, Kyoto, Japan) at a wavelength of 252.2 nm. A freshly prepared mixture of methanol and 0.2% (w/v) ammonium acetate solution in water (55:45, pH 6.9), filtered and degassed using 0.45 mm membrane filter (Millipore, County Cork, Ireland) was used as the mobile phase. A flow rate of 1.0 mL/min at ambient temperature was generated by a HPLC pump (Shimadzu LC-10 AJ, Kyoto, Japan) using an injection volume of 15 mL (Shimadzu Autosampler, SIL-10AD VP, Kyoto, Japan). The retention time of SX was approximately 4 min. The peak area was recorded by integration (Shimadzu Class VP, Kyoto, Japan). A HPLC calibration curve was prepared using five replicates of four concentrations between 0.4 and 10 mg/mL, and the linear regression was performed using Sigmaplot (Jandel Scientific, Chicago, IL). The precision of each assay was determined by analysing five replicates from each of two standard SX solutions (two concentrations: low 1.0 mg/mL and high 4.0 mg/mL). The regression coefficient (R2) was 1.000 and no significant deviation from zero intercept was observed ( p > 0.05). The accuracy ranged from 99.4% to 101.3% and the precision ranged from 0.6% to 1.1%. Scanning Electron Microscopy (SEM) A small amount of powder was sprinkled and glued on a metal sample plates. The samples were gold plated with a sputter coater (BAL-TEC SCD 005, Tokyo, Japan) using an electrical potential JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
of 2.0 kV at 25 mA for 10 min. The surface morphology of the particles was examined at several magnifications under JOEL JSM 6000F scanning electron microscope at 15 kV (JOEL, Tokyo, Japan). Aerodynamic Particle Size Analysis by Aerosizer The aerodynamic particle size of the powders was measured by an Aerosizer (Amherst Process Instruments, Inc., Hadley, MA) using a dry powder dispersion system (aerodisperser). Approximately 5 mg of powder was placed in the sample cup of the aerodisperser. Particle size measurement was carried out at a medium feed rate and a sample run time of 300 s. Different shear pressures (3.4, 10.3, 20.7, 27.6 kPa) were used to understand the effect of increasing shear pressure on the agglomerate strength of the powders. The particle size of the mixture was analysed using API Aerosizer software (LD Version 7.04). The average particle size distribution was determined from five replicates of each sample.
Statistical Analysis All data were subjected to One Way Analysis of Variance (ANOVA) (SPSS, Chicago, IL, Version 15.0). Probability ( p) values of 0.05 when analysed by post hoc multiple comparisons were considered as statistically significant.
RESULTS AND DISCUSSION Surface Moisture of CL and ML during Storage at Different RH In order to assess the kinetics of surface moisture uptake for both CL and ML, the surface moisture was monitored by a thermo gravimetric method during storage at three different RH (45%, 75% and 95% RH) for 2 months (Fig. 1). The rate of surface moisture uptake and the equilibrium surface moisture content increased with increasing RH for both CL and ML. Moreover, at any specific RH, both the rate of surface moisture uptake and the equilibrium surface moisture content were higher for ML than for CL. For example, the surface moisture of CL increased to the plateau level of 0.25% w/w in 3 weeks at 75% RH and to 0.12% w/w in a month at 45% RH. In DOI 10.1002/jps
INFLUENCE OF RH ON DISPERSION OF SX POWDERS
1019
Powder Formulations
Figure 1. Monitoring of surface moisture (by a thermo gravimetric analysis) of (A) coarse lactose, CL and (B) micronised lactose, ML during storage at three different RHs (45%, 75% and 95%) for 60 days.
contrast for ML, plateau levels of 0.36% w/w and 0.22% w/w were achieved in only 2 weeks at 75% RH and 3 weeks at 45% RH, respectively. The equilibrium moisture contents found for both CL and ML were consistent with previous reports by Callahan et al.19 who reported a lactose monohydrate equilibrium moisture content of 0.2% in the range of 11–93% RH. The time to reach equilibrium for CL was consistent with the report of Berard et al.20 who concluded from adsorption kinetic studies of lactose monohydrate that at least 21 days was required to establish the equilibrium. The extent of water accumulation and condensation depends on surface properties such as the presence of amorphous domains, surface roughness and surface area.3 The higher surface moisture content of ML compared with CL was expected, but the increase was not proportional to the anticipated increase in surface area determined on a particle size basis; this may have been associated with the reduction in expected surface area due to agglomeration of the micronised lactose. DOI 10.1002/jps
Inhalation grade SX particles form cohesive agglomerates because of their small size and predominant cohesion interactions (Fig. 2A). Coarse carriers such as lactose monohydrate are commonly added to DPI formulations to improve flow properties and increase bulk. The SEM of the SX–CL mixture (M0F) revealed that the majority of the SX particles, existing as either individual particles or agglomerates, were adhered to the CL (Fig. 2B). In SX–CL mixtures containing additional ML, agglomerates were observed both on and off the surface of CL (Fig. 2C and D). These agglomerates are likely to consist of interacting particles of SX and ML13 and will vary in the composition of SX and ML because of complex interparticulate interactions occurring among different particles during powder mixing. The homogeneity and consistency in the in vitro TSI deposition of each batch were determined to ensure that the mixing procedure produced powder mixtures approaching target SX content with low variability and reproducible SX dispersion behaviour. The SX content ranged from 99.5% to 100.7% and the coefficient of variation (CV) was below 1.9%. The results indicated that the homogeneity of powder mixtures was excellent, being well within the specifications applied in this study.18 In order to study intra-batch dispersion variability, the ED and the SX content of the stage one and two of the TSI from a single batch of M0F were compared using five replicates. The ED and the SX content of the stage one and two of the TSI were 80.3 0.5%, 74.8 0.5% and 5.3 0.4%, respectively, and showed no significant differences ( p > 0.05). In order to determine inter-batch variability, the in vitro deposition of five batches of M0F were compared. The ED and the SX content of the stage one and two of the TSI were 79.9 0.8%, 74.2 0.7% and 5.4 0.6%, respectively, and showed no significant differences ( p > 0.05). No significant intra- or interbatch variability in ED, and SX content of stage one and two of the TSI was seen for M20F when tested using the same procedure. Homogeneity studies together with intra- and inter-batch dispersion comparisons confirmed that the powder mixing process used in this research produced homogeneous mixtures with reproducible in vitro aerosol deposition. The ED and FPF for all mixtures were consistent with earlier findings from our laboratory.13 Before storage at specific RH, the ED of SX–CL JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
1020
DAS ET AL.
Figure 2. Scanning electron micrographs (SEM) of (A) Salmeterol xinafoate magnification (2306), and SX–CL mixtures containing 0%, 10% and 20% micronised lactose, (B) M0F, magnification (800), (C) M10F, magnification (412) and (D) M20F, magnification (601) at 10 kV, respectively.
mixture, M0F (80.0 3.1%) was higher than the other formulation used in this study. As the concentration of ML in the mixtures increased, the ED decreased, for example, M5F (72.5 3.6%), M10F (68.6 2.9%), M20F (57.6 1.5%) and SX alone formulation (60.4 4%). The minimum ED, shown by the mixture containing 20% micronised lactose, was not significantly different from that of drug alone formulation ( p > 0.05). The addition of the CL to SX decreased dispersion performance with FPF of the SX–CL mixture (5.5 0.7%) decreasing from 10.3 1.01% for the SX alone formulation (Fig. 3B). The addition of ML increased dispersion performance with the FPF significantly increased to 8.1 0.8% for M5F ( p < 0.05) and to 9.1 0.8% for M10F ( p < 0.05). The highest FPF was shown by mixture M20F (11.3 0.6%) though it was not significantly different from the drug only formulation ( p > 0.05). JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
Dispersion Mechanism of Powder Formulations In order to understand the reasons for potential changes observed during storage, the mechanisms of dispersion associated with the mixtures used in this study need to be defined. Micronised powders will usually agglomerate due to the dominance of the interparticulate interaction forces over the gravitational detachment forces.21 When larger carrier particles (such as CL) are mixed with micronised powders, micronised powder particles can exist both as adhered particles to the carrier surface (either individually or as networks of particles like agglomerates), and as agglomerates existing freely in the powder mixture and separated from the carrier surface. For complex mixtures used in this study containing SX, CL and ML, both SX and ML can (a) adhere to the CL surface, (b) cohere to form SX and ML agglomerates and DOI 10.1002/jps
INFLUENCE OF RH ON DISPERSION OF SX POWDERS
1021
process is facilitated by the incorporation of ML particles into the agglomerate since mixed agglomerates of SX and ML have reduced strength, possibly due to the change in the packing fraction of the particles within the agglomerate.13
Dispersion Changes (ED and FPF) of DPI Formulations during Storage The FPF was determined before storage and at 7, 14, 28, 56, 84, 168, 364 and 564 days after storage. For 75% RH, two additional readings were taken at 24 and 48 h. In this article, data are only shown where significant changes in dispersion performance occurred. ED
Figure 3. (A) Emitted dose (ED) and (B) fine particle fraction (FPF) of SX of five formulations: SX alone (D), and its four mixtures M0F, M5F, M10F and M20F containing 0%, 5%, 10% and 20% micronised lactose, respectively during storage at 75% RH for 18 months.
(c) adhere to each other to form mixed agglomerates with different ratios of SX and ML. In the study of the formulations used in this research, SX dispersion will occur due to (a) particle detachment from the CL surface and (b) deagglomeration. SX alone mixtures will disperse through deagglomeration of the SX agglomerates. The major mechanisms for dispersion for mixtures containing SX and CL only (where the SX particles are essentially located at the CL surface as individual particles or agglomerates—see Fig. 2B) are SX detachment from the CL surface and SX deagglomeration from the agglomerates on the surface. These mixtures tend to perform poorly due to the relatively strong SX attachment to the CL carrier particles.13 However, when ML is incorporated into the SX– CL mixtures, previous studies13 have demonstrated that deagglomeration of the SX–ML agglomerates is the dominant mechanism of SX dispersion in the powder mixtures. The deagglomeration DOI 10.1002/jps
During storage at 75% RH (Fig. 3A), the ED of mixture M20F sharply increased from 57.6 3.1% to 77.3 4% within a week and did not significantly change over the remaining period. The ED for the mixture M10F gradually increased from 69.1 2.8% before storage to 76.3 3.7% at the end of the study. These two points were statistically different ( p < 0.05); however, no significant change was observed in the ED of all the other formulations (D, M0F and M5F) for the whole study period. At 33% and 55% RH, there was no significant change in ED for any formulation during the study period except for the mixture M20F at 55% RH (data are not shown in figures). The ED of M20F significantly increased from 57.6 2.9% to 65.5 3.2% within 3 days ( p < 0.05), and did not change significantly over the remaining period ( p > 0.05). FPF During storage at 75% RH (Fig. 3B), the FPF of mixture M20F significantly decreased from 11.3 0.7% to 7.7 0.6% within 8 weeks ( p ¼ 0.008). The FPF of mixtures M5F and M10F did not change over the first 6 months ( p > 0.05); however, the FPF had significantly decreased at 18 months ( p < 0.05). Other formulations (D and M0F) at 75% RH and all formulations at 33% and 55% RH showed no significant change in FPF ( p > 0.05) for the whole study period. When M20F was stored at 95% RH, the ED increased from 57.6 2.2% to 77.9 3.2% just in 3 days (Fig. 4A). Then, it gradually rose to 87.0 3.4% in 12 months. Accordingly, the FPF JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
1022
DAS ET AL.
hydrophobic powder having a contact angle of 67.78 with water.25 Previous studies have demonstrated that capillary forces do not develop in SX even at RH of 75%.26 Lactose monohydrate is hydrophilic.25 During storage, CL and ML were found to gain only a small amount of water, around 0.1% w/w and 0.2% w/w, respectively, at a RH of 45% (Fig. 1). Capillary forces in lactose monohydrate have not been detected at less than 55% RH.26 Therefore, the contribution of capillary forces due to lactose at 33% RH was unlikely to be significant in the mixtures studied (M0F, M5F, M10F and M20F). Electrical forces such as contact potential and Coulombic forces might be generated during powder handling and contribute to interparticulate interactions of both SX and lactose at this low RH. Charge decay would be expected to occur, but any change in the extent of electrostatic particle interactions did not manifest itself in any observable changes in ED and FPF for the formulations during storage at this RH. Figure 4. (A) Emitted dose (ED) and (B) fine particle fraction (FPF) of SX of mixture containing 20% micronised lactose (M20F), during storage at 95% RH for 12 months.
significantly declined ( p ¼ 0.001) from 11.3 0.7% to 4.9 0.5% within the first 8 weeks of storage (Fig. 4B). The FPF gradually decreased to 3.4 0.3% over the remaining storage period.
Possible Particle Interactions at Low to Medium Environmental RH Conditions That Lead to No Significant Change in Dispersion RH 33% During storage at 33% RH, there was no significant change in dispersion of SX for any formulation. This means that during storage there has been little change in the magnitude of interaction between particles in these mixtures. The obvious force components likely to change during storage are associated with capillary and electrostatic interactions. Though not commonplace, there are a few examples of formation of capillary forces in some systems at low RH.22–24 The emergence and extent of capillary forces mainly depend on material properties. For example, capillary forces are almost negligible for hydrophobic materials, while they can be considerable for hydrophilic substances. SX is a JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
RH 55% At 55% RH, there was no significant change in ED for any formulation during the study period, except for an increase for the mixture M20F. The FPF of all formulations did not significantly change for the whole study period. At 55% RH, a balance between diminishing electrostatic forces and emerging capillary forces was expected. The outcome of storage of the SX formulations was consistent with an earlier report26 which observed only a minimal increase in auto adhesion force in lactose monohydrate up to a RH of 55%. A small increase in ED of M20F within the first 3 days could be due to slightly higher amount of moisture associated with higher percentage of ML facilitating increased agglomeration. However, this did not have any effect on FPF of M20F.
Possible Particle Interactions at High Environmental RH Conditions That Lead to Change in Dispersion RH 75% There was no significant change ( p > 0.05) in both ED and FPF during storage of the SX alone formulation at 75% RH (Fig. 3). This could be attributed to the very low likelihood of capillary interaction at this RH by hydrophobic SX.27 This outcome was supported by the earlier findings that auto adhesion forces of SX were less sensitive to storage humidity and the absence of capillary DOI 10.1002/jps
INFLUENCE OF RH ON DISPERSION OF SX POWDERS
forces in SX when stored at 75% RH or greater for 72 h.26 Jashnani et al.28 also concluded that hydrophobic powders were less sensitive to high storage humidity than hydrophilic powders during a study of the emptying and aerosolisation properties of DPIs. The FPF and ED of SX–CL mixture with no additional micronised lactose (M0F) did not significantly change at 75% RH for 18 months (Fig. 3). In general, capillary forces are not important at a RH less than 50%; however, they start to dominate interparticulate interactions at a RH greater than 60%.3 Capillary forces were observed by other scientists to contribute to adhesion forces between lactose monohydrate and SX when stored at a RH of 75% or 95%, and an equilibrium value of the capillary force was reached within 72 h.29 Thus, capillary interactions between the SX and CL were likely to increase the total interparticulate interaction during storage at 75% RH. However, any increase in interaction that might have occurred did not lead to ED and FPF change during storage. Since the dominant mechanism of dispersion for mixtures of SX and CL at the SX concentration used in this study has been shown to be deagglomeration of the surface agglomerates of SX14 and since capillary interactions within SX particles were less likely to be affected by storage at 75% RH, changes in adhesion that might occur due to any capillary interaction between SX and CL may be less important in the overall dispersion process. At 75% RH, no significant change was observed in the ED of mixture containing 5% micronised lactose (M5F) for the whole study period (Fig. 3). The ED for mixture containing 10% micronised lactose (M10F) gradually increased throughout the study period. On the other hand, the ED of M20F sharply increased from 57.6 3.1% to 77.3 4% within a week (Fig. 3A) though it did not significantly change ( p > 0.05) over the remaining period. The ED increase of M20F and M10F during storage at 75% RH could be due to increased agglomeration caused by increased capillary interactions between the micronised powders leading to better flow properties and entrainment of the mixtures. Significant difference ( p < 0.05) in dispersion behaviour demonstrated by the changes in FPF was observed during storage when micronised lactose (ML) was added to the formulation. The onset of the decrease in FPF during storage depended on the concentration of additional ML. For example, a significant decrease ( p > 0.05) in the FPF for DOI 10.1002/jps
1023
mixtures containing 5% and 10% additional micronised lactose (M5F and M10F) was first observed at 18 months storage while that for M20F was observed at 4 weeks (Fig. 3B). The FPF of SX for M20F significantly decreased ( p < 0.05) from 11.3 0.7% to 8.2 0.6% within the first 4 weeks and continued to decrease to 7.7 0.6% until week 8 of storage, and then it did not significantly change ( p > 0.05) for the remaining storage period (Fig. 3B). It is interesting to note that the FPF of M20F after 8 weeks storage at 75% RH was not significantly different ( p > 0.05) from the FPF of mixture M5F before storage. RH 95% When M20F was stored at 95% RH, the ED significantly increased in just 3 days (Fig. 4A) while the FPF significantly declined ( p ¼ 0.001) from 11.3 0.7% to 4.9 0.5% within the first 8 weeks of storage (Fig. 4B), which is almost a 57% reduction from the initial value. The FPF gradually decreased to 3.4 0.3% over the remaining storage period. At 95% RH, surface moisture of ML and CL increased to 0.45% (w/w) and 0.3% (w/w) in 1 and 2 weeks, respectively. A more pronounced decreasing effect was observed more quickly at 95% RH than it was at 75% RH because of higher capillary forces followed by solid bridging.
Proposed Mechanisms for the Change in Dispersion of the SX Mixtures during Storage at High RH The critical parameters in the decrease in dispersion of the SX mixtures during storage were (a) the presence of ML in sufficient concentrations (clearly evident at 20%) and (b) the presence of sufficient atmospheric moisture during storage (75% RH and greater). The dispersion of SX from mixtures containing micronised lactose has been shown to be mainly controlled by the deagglomeration of mixed agglomerates of SX and ML. The presence of ML was well known to increase dispersion performance due to its influence on the agglomerate structure,13 and its contribution to decreased dispersion during storage at RH of 75% and greater was also likely to be associated with its influence on agglomerate structure. Micronised lactose was seen to absorb moisture during storage at high RH (Fig. 1B); therefore, capillary interactions will contribute particularly to greater cohesion between ML particles and also to greater adhesion between JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
1024
DAS ET AL.
ML and SX. In order to understand the extent of capillary interaction influence within the agglomerate, capillary forces ( Fc) were theoretically calculated using the equation mentioned by Zimon.3 Fc ¼ 2prg cos u where, g is surface tension, r is the radius of particle in contact, and u is contact angle. For the contact between two particles, capillary forces ( Fc) can be calculated extending the above equation to the following: Fc ¼ 2pr1 g cos u1 þ 2pgr2 cos u2 where, g is surface tension, r1, r2 are the radii of particles in contact, and u1, u2 are contact angles of the contacting particles. The surface tension of water is 0.0729 N/m3, while the radii for SX and micronised lactose particles are 1.10 and 1.55 mm, respectively. The contact angle of water for SX is 67.78. Assuming the contact angle for lactose is zero, the calculated capillary forces for SX–SX, SX–ML and ML–ML contacts were 0.382, 0.836 and 1.42 mN, respectively. Capillary forces have fivefold higher effect on the adhesion forces for ML–ML contacts compared with that of the SX–SX contact. In order to understand the relative abundance of different types of contacts, a very simple randomised distribution matrix of SX and ML within the mixed agglomerate was designed. In mixtures containing 5%, 10% and 20% ML, the ratio between SX and ML are 1:2, 1:4 and 1:8, respectively. A hypothetical randomised distribution of SX and ML in the agglomerate (Fig. 5) shows that the abundance of ML–ML contact increased while SX–SX contact or SX–ML contact decreased with increasing concentration of micronised lactose in mixtures. For example, the hypothetical number of ML–ML contacts in the randomly distributed agglomerate of the M5F was 88, whereas 186 contacts occurred in the agglomerates of M20F. Moreover, there was no SX–SX contact in M20F, although six SX–SX contacts were present in M5F. Therefore, during storage at high humidity conditions, the extent of ML–ML contact and the magnitude of capillary forces increased with the increased concentration of ML in mixtures. These additional capillary forces would increase the extent of particle interactions within the agglomerates, increasing the agglomerate strength and decreasing the FPF during storage. Although the randomisation process can be JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
Figure 5. Random distribution of micronised lactose (ML) and salmeterol xinafoate (SX) in mixtures of 2.5% SX with (A) 5%, (B) 10% and (C) 20% micronised lactose.
criticised as it did not consider the polydispersity of SX and ML particles and variation in particle shape, this simplistic approach was still useful to view the propensity of different contact. Moisture also has some plasticising effect and will affect the interaction in other ways.25 Surfaces of ML are softened after absorbing moisture at high RH and contact area is increased through particle deformation. Therefore, in addition to the contribution of the increased capillary interactions, higher intermolecular interactions between the micronised particles of the agglomerate could occur and lead to greater interaction and stronger agglomerates with a resultant decreased dispersion performance. In addition to increased capillary and intermolecular interactions during storage, slight dissolution at the surface of micronised particles was likely to occur in presence of absorbed moisture. The temperature of storage fluctuated between 23 and 278C during study period and was conducive to the formation of solid bridges formed through the crystallisation of lactose during temperature variations. Solid bridges formation is not uncommon in pharmaceutical systems. Lactose monohydrate was found to dissolve under the influence of high RH.29 Liquid bridges are often followed by solid bridges due to crystallisation to adjust evaporation of moisture from liquid bridges.30 Weak solid bridges of lactose after recrystallisation were also claimed earlier by other scientists.21 Solid bridges have been also observed in model sulphathiazole–Emdex DOI 10.1002/jps
INFLUENCE OF RH ON DISPERSION OF SX POWDERS
and succinyl sulphathiazole–Emdex1 mixtures during storage of dry powders over 7 days.31 Thus, during storage at high humidity, interparticulate interactions within the agglomerate increased through possible increased capillary and intermolecular interaction forces and solid bridging. These magnitudes of the interactions were greater for ML–ML contacts particularly as the concentration of ML increased within the mixtures and the SX particles within the agglomerates were encaged within the agglomerates by the micronised lactose matrix. Thus dispersion performance decreased during storage. Comparison of Agglomerate Strengths of Mixtures before and after Storage In order to support the above predicted mechanism, the aerodynamic particle sizes of mixtures containing 0%, 5%, 10% and 20% micronised lactose before and after storage at 75% RH for 3 months were determined using an Aerosizer. Each mixture was dispersed at four different shear pressures (3.4, 10.3, 20.7 and 27.6 kPa) to develop particle size-shear pressure profiles to provide an understanding of the extent and propensity of the mixtures to deagglomerate.13 No significant difference in the extent or rate of deagglomeration was observed for the SX–CL mixture (M0F) before and after storage (Fig. 6A).
1025
The aerodynamic particle size was around 100 mm at 3.4 kPa. This size was dominated by the particle size of CL as the mixture contained 97.5% CL. The aerodynamic particle sizes slightly decreased with increasing shear pressures both before and after storage, possibly because of dispersion of drug particles/agglomerates on the surface of CL. It should be noted that when the CL alone was stored under these conditions, the particle sizeshear pressure profile was not significantly different before and after storage ( p > 0.05) and that there was no significant change in particle size between the shear pressures ( p > 0.05). This indicated that no comminution of the carrier occurred. The slightly higher aerodynamic particle size of the mixture containing 5% micronised lactose (M5F) was an indication of increased agglomerate formation in M5F because of the presence of micronised lactose. However, no significant difference in the extent and propensity of deagglomeration was observed before and after storage. This finding supports the proposed mechanisms discussed above that the particle interactions in M5F were not sufficient to bring about any significant changes in interparticulate interactions after storage (Fig. 6B). The aerodynamic particle size of mixture containing 10% micronised lactose (M10F) was higher after storage than before at the lowest shear pressure (3.4 kPa) indicating that agglomerates were not
Figure 6. (A–D) Comparison between before and after storage mixtures containing 0%, 5%, 10% and 20% micronised lactose, respectively. DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
1026
DAS ET AL.
as dispersible at this low shear pressure after storage (Fig. 6C). However, there was no significant difference in deagglomeration at higher shear pressures (10.3, 20.7 and 27.6 kPa) between before and after storage ( p > 0.05). In contrast, the particle size-shear pressure profile of the mixture containing 20% micronised lactose (M20F) was higher after storage than before at all shear pressures (Fig. 6D). There was no significant difference between particle sizes at 20.7 kPa and at 27.6 kPa after storage ( p > 0.05). Thus, the extent of deagglomeration was less after storage, although the propensity to deagglomerate over the shear pressure range was similar. This is a clear indication of the formation of stronger agglomerate in M20F after storage at 75% RH and the outcomes of this Aerosizer study provided strong support to the mechanistic proposals derived from hypothetical distribution matrix and capillary force calculation.
CONCLUSIONS The dispersion performance of SX mixtures for inhalation, shown by changes to the FPF, was influenced predominately by the presence of micronised lactose in the formulation and the RH of the storage condition. A significant decrease in FPF on storage over 18 months was observed in formulations containing 10% and 20% ML when the RH was 75% or greater. All other formulations showed no change in FPF during storage at 75% and all formulations showed no change in FPF at 33% and 55% RH. For mixtures containing high concentrations of micronised powders, such as 2.5% SX–CL mixtures containing 10% and greater ML, previous studies have demonstrated that the dominant dispersion mechanism was deagglomeration of SX–ML agglomerates formed during mixing.13 At 75% RH, capillary forces between particles in the agglomerate contributed to increased particle interactions within the mixture giving rise to stronger agglomerates. In particular, a calculation of the capillary interactions demonstrated that the capillary interaction between ML particles was significantly greater than SX–ML or SX– SX capillary interactions. Thus, the increase in agglomerate strength occurred predominately in mixtures containing high concentrations of ML. The use of a simple hypothetical two-dimensional model of a random distribution of SX and ML particles in an agglomerate demonstrated that JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009
the increased likelihood of ML–ML interactions within the agglomerate as the concentration of ML increased. Therefore, while the presence of moisture in the storage environment increased all capillary interactions within the agglomerate giving rise to stronger agglomerates, the propensity for increased extent and magnitude of ML– ML interactions at high concentrations of ML was likely to result in the development of a ‘‘cage-like’’ structure with the SX being trapped within the micronised lactose matrix. The increased capillary interactions may also have been supplemented with increased intermolecular interactions due to particle deformation at high moisture conditions and the development of solid bridges due to temperature variations. The use of a novel particle size-shear pressure profile generated by the Aerosizer provided confirmation of increased agglomerate strength for mixtures containing high concentrations of ML after storage at high RH. The study was conducted on a laboratory scale using small batches (5 g). The mixing conditions, both in relation to batch size and type of mixer, will influence the magnitude of shear generated during mixing procedures and thus will influence the extent and structure of agglomerates. This research provides an explanation of the changes observed during storage of SX mixtures for inhalation based on the agglomerate characteristics developed during laboratory mixing. These general principles will apply to other mixing conditions where agglomerates are produced, but these laboratory scale findings should be carefully interpreted for industry scale operation.
ACKNOWLEDGMENTS Shyamal Das would like to acknowledge the scholarship support through the Faculty of Pharmacy, Monash University and thank to Dr Handoko Adi and Dr Herbert Chiou for their support during using SEM and Aerosizer. He also acknowledges Dr Richard J. Prankerd and Dr Joseph A. Nicolazzo for their help during TGA use and statistical analysis, respectively.
REFERENCES 1. Borgstrom L, Asking L, Lipniunas P. 2005. An in vivo and in vitro comparison of two powder inhalers following storage at hot/humid conditions. J Aerosol Med 18:304–310. DOI 10.1002/jps
INFLUENCE OF RH ON DISPERSION OF SX POWDERS
2. Stewart PJ. 1986. Particle interaction in pharmaceutical systems. Pharm Int 7:146–149. 3. Zimon AD, editor. 1982. Adhesion in a gas medium. Sources of adhesion. In: Adhesion of dust and powder. 2nd edition. New York: Consultants Bureau. pp 93–144. 4. Podczeck F, Newton JM, James MB. 1996. The influence of constant and changing relative humidity of the air on the autoadhesion force between pharmaceutical powder particles. Int J Pharm 145: 221–229. 5. Young PM, Price R, Tobyn MJ, Buttrum M, Dey F. 2003. Effect of humidity on aerosolization of micronized drugs. Drug Dev Ind Pharm 29:959–966. 6. Hindle M, Makinen GM. 1996. Effects of humidity on the in vitro aerosol performance and aerodynamic size distribution of cromolyn sodium for inhalation. Eur J Pharm Sci 4:S142. 7. Braun MA, Oschmann R, Schmidt PT. 1996. Influence of excipients and storage humidity on the deposition of disodium cromoglycate (DSCG) in the twin impinger. Int J Pharm 135:53–62. 8. Young PM, Price R. 2004. The influence of humidity on the aerosolisation of micronised and SEDS produced salbutamol sulphate. Eur J Pharm Sci 22: 235–240. 9. Lida K, Hayakawa Y, Okamoto H, Danjo K, Luenberger H. 2004. Influence of storage humidity on the in vitro inhalation properties of salbutamol sulphate dry powder with surface covered lactose carrier. Chem Pharm Bull 52:444–446. 10. Zeng XM, MacRitchie HB, Marriott C, Martin GP. 2007. Humidity-induced changes of the aerodynamic properties of dry powder aerosol formulations containing different carriers. Int J Pharm 333:45–55. 11. Harjunen P, Lankinen T, Salonen H, Lehto VP, Jarvinen K. 2003. Effects of carriers and storage of formulation on the lung deposition of a hydrophobic and hydrophilic drug from a DPI. Int J Pharm 263:151–163. 12. Maggi L, Bruni R, Conte U. 1999. Influence of moisture on the performance of a new dry powder inhaler. Int J Pharm 177:83–91. 13. Adi H, Larson I, Chiou H, Young P, Traini D, Stewart P. 2006. Agglomerate strength and dispersion of salmeterol xinafoate from powder mixtures for inhalation. Pharm Res 23:2556–2565. 14. Islam N, Stewart P, Larson I, Hartley P. 2004. Effect of carrier size on the disperson of salmeterol xinafoate from interactive mixtures. J Pharm Sci 93:1030–1038. 15. Louey MD, Stewart PJ. 2002. Particle interactions involved in aerosol dispersion of ternary interactive mixtures. Pharm Res 19:1524–1531. 16. Always B, Sangchantra R, Stewart PJ. 1996. Modelling the dissolution of diazepam in lactose interactive mixtures. Int J Pharm 130:213–224.
DOI 10.1002/jps
1027
17. Liu J, Stewart PJ. 1998. Deaggregation during the dissolution of Benzodiazepines in Interactive mixtures. J Pharm Sci 87:1632–1638. 18. Crooks MJ, Ho R. 1976. Ordered mixing in direct compression of tablets. Powder Technol 14:161– 167. 19. Callahan JC, Cleary GW, Elefant M, Kaplan G, Kensler T, Nash RA. 1982. Equilibrium moisture content of pharmaceutical excipients. Drug Dev Ind Pharm 8:355–369. 20. Berard V, Lesniewska E, Andres C, Pertuy D, Laroche C, Pourcelot Y. 2002. Affinity scale between a carrier and a drug in DPI studied by atomic force microscopy. Int J Pharm 247:127–137. 21. Visser J. 1995. Particle adhesion and removal: A review. Particulate Sci Technol 12:169–196. 22. Eastman T, Zhu DM. 1996. Adhesion forces between surface-modified AFM tips and a mica surface. Langmuir 12:2589–2862. 23. Price R, Young PM, Edge S, Staniforth JN. 2002. The influence of relative humidity on particulate interactions in carrier-based dry powder inhaler formulations. Int J Pharm 246:47–59. 24. Xu L, Lio A, Ogletree DF, Salmeron M. 1998. Wetting and capillary phenomena of water on mica. J Phys Chem B 102:540–548. 25. Podczeck F, Newton JM, James MB. 1997. Influence of relative humidity of storage air on the adhesion and autoadhesion of micronized particles to particulate and compacted powder surfaces. J Colloid Interface Sci 187:484–491. 26. Podczeck F, Newton JM, James MB. 1996. The influence of constant and changing relative humidity of the air on the autoadhesion force between pharmaceutical powder particles. Int J Pharm 145: 221–229. 27. Podczeck F, Newton JM, James MB. 1995. Assessment of adhesion and autoadhesion forces between particles and surfaces. Part II. The investigation of adhesion phenomenoa of salmeterol xinafoate and lactose monohydrate particles in particle-onparticle and particle-on surface contact. J Adhes Sci Technol 9:475–486. 28. Jashnani RN, Byron PR, Dalby RN. 1995. Testing of dry powder aerosol formulations in different environmental conditions. Int J Pharm 113:123–130. 29. Podczeck F, Newton JM, James MB. 1997. Variations in the adhesion force between a drug and a carrier particles as a result of changes in the relative humidity of the air. Int J Pharm 149:151– 160. 30. Padmadisastra Y, Kennedy RA, Stewart PJ. 1994. Influence of carrier moisture adsorption capacity on the degree of adhesion of interactive mixtures. Int J Pharm 104:R1–R4. 31. Padmadisastra Y, Kennedy RA, Stewart PJ. 1994. Solid bridge formation in sulphonamide-Emdex interactive systems. Int J Pharm 112:55–63.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 98, NO. 3, MARCH 2009