- Email: [email protected]
Agglomerate properties and dispersibility changes of salmeterol xinafoate from powders for inhalation after storage at high relative humidity
European Journal of Pharmaceutical Sciences 37 (2009) 442–450
Contents lists available at ScienceDirect
European Journal of Pharmaceutical Sciences journal homepage: www.elsevier.com/locate/ejps
Agglomerate properties and dispersibility changes of salmeterol xinafoate from powders for inhalation after storage at high relative humidity Shyamal Das a , Ian Larson a , Paul Young b , Peter Stewart a,∗ a b
Drug Delivery, Disposition and Dynamics, Monash Institute of Pharmaceutical Sciences, Monash University (Parkville Campus), 381 Royal Parade, Victoria 3052, Australia Advanced Drug Delivery Group, Faculty of Pharmacy, University of Sydney, NSW 2006, Australia
a r t i c l e
i n f o
Article history: Received 7 November 2008 Received in revised form 27 March 2009 Accepted 30 March 2009 Available online 8 April 2009 Keywords: Salmeterol xinafoate Agglomerate Particle size distribution Storage Relative humidity Dry powder inhaler
a b s t r a c t Purpose: This study investigated changes in agglomeration and the mechanism of dispersibility decrease of salmeterol xinafoate (SX) from SX–lactose mixtures for inhalation after storage at 75% RH for 3 months. Methods: The dispersibility, PSD and in situ PSD of aerosol plumes of SX alone and SX–coarse lactose (CL) mixtures containing 0, 5, 10 and 20% micronized lactose (ML) before and after storage were determined by a Next Generation Impactor (NGI), a Mastersizer 2000 and a Spraytec, respectively. Results: The PSD of ML increased after storage at 75% RH, but dispersibility of SX using the stored ML increased. After storage, the %SX of the mixture containing 20% ML (M20F) significantly increased (P < 0.05) in the throat and mouthpiece, preseparator and stage 1 of NGI, while it significantly decreased in the remaining stages (P < 0.05). In situ analysis of aerosol plumes of M20F supported this result with an increased presence of particles of 4–25 m and a decreased respirable particle distribution of <4 m after storage. Conclusions: The decreased dispersibility of M20F after storage was due to the formation of less dispersible agglomerates, probably occurring through enhanced capillary interaction and/or solid bridging of ML, entrapping and preventing the release of SX particles. Crown Copyright © 2009 Published by Elsevier B.V. All rights reserved.
1. Introduction Changes in dispersibility of dry powder inhalers during storage at high relative humidity (RH) are often encountered by pharmaceutical manufacturers (Borgstrom et al., 2005; Young et al., 2007). The dispersibility of powders is influenced by interparticulate forces (Louey and Stewart, 2002) which can be modified by environmental RH (Podczeck et al., 1996, 1997a,b; Young et al., 2003a, 2004). High RH may increase interparticulate forces due to increased capillary interactions resulting in the formation of larger agglomerates (Hogg, 1989) that are less breakable (Boerefijin et al., 1998). Micronized lactose monohydrate has been found to dissolve under the influence of high RH (75% RH or greater) (Podczeck et al., 1997a). Thus, the liquid bridges which are formed at these high RH conditions are often followed by solid bridges due to recrystallization of the dissolved lactose (Padmadisastra et al., 1994a,b). Many investigations on the effect of storage RH on dispersibility (Braun et al., 1996; Hindle and Makinen, 1996; Young et al., 2003b; Lida et al., 2004; Young and Price, 2004; Borgstrom et al., 2005; Young et al., 2007; Zeng et al., 2007) were limited to either drug
∗ Corresponding author. Tel.: +61 3 9903 9517; fax: +61 3 9903 9583. E-mail address: [email protected] (P. Stewart).
alone formulations or binary mixtures of drugs. These studies did not include dry powder inhaler formulations that contained ternary components (Staniforth, 1996; Zeng et al., 1996), although the addition of ternary components, for example, micronized lactose, has been shown to improve dispersibility (Lucas et al., 1998; Louey and Stewart, 2002; Adi et al., 2006). Moreover, in most studies, formulations were exposed to storage RH for only short periods of time (maximum 7 days). A recent extended study on different types of dry powder formulations (Das et al., 2009) has reported that the salmeterol xinafoate (SX) dispersibility was influenced predominately by the presence of micronized lactose in the formulation and the storage relative humidity. A significant decrease in fine particle fraction (FPF) of SX after storage at 75% RH was observed in the ternary formulation containing 20% ML within 4 weeks and the decrease reached lower static levels within 3 months. Calculations revealed that increased capillary interactions were likely to occur between lactose particles and that deagglomeration required increased shear pressures after storage at 75% RH. The extent of particulate interactions increased, but the study did not identify the mechanism of decreased dispersibility. For example, the study outcomes could be consistent with increased lactose–lactose interactions resulting in (a) increase in the particle size distribution (PSD) of lactose or in (b) the formation of strong lactose agglomerates entrapping the SX and
0928-0987/$ – see front matter. Crown Copyright © 2009 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.ejps.2009.03.016
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
decreasing its dispersibility. An increase in the size distribution of lactose will influence dispersibility in different ways. Small changes in particle size in the cohesive range, essentially less than about 10 m, have been shown to influence dispersibility of SX (Adi et al., 2006). However, significant agglomeration of micronized lactose, producing stronger free flowing agglomerates, will likely decrease SX dispersibility due to the capacity of the agglomerate to act as a secondary carrier and strongly bind the SX to its surface. Decreased dispersibility has been observed in formulations containing ternary lactose acting as secondary carriers (Adi et al., 2006). Thus, particle size changes may potentially result in either decreased or increased dispersibility of drugs. However, the entrapment of a drug within a lactose agglomerate network will decrease dispersibility. An understanding of the mechanism is essential to establish the body of knowledge to enable quality by design and to ensure that appropriate strategies are employed in the development of powders for inhalation. The purpose of this study was to identify the mechanism of dispersibility change during storage at high RH of an SX formulation containing Inhalac 120 as the coarse lactose carrier and micronized lactose monohydrate as the fine added excipient. The study used a combination of particle sizing approaches to achieve the outcome. 2. Materials and methods 2.1. Materials Inhalation grade micronized salmeterol xinafoate (SX) (batch no., B068803, Glaxo SmithKline, Ware, UK) was used as model drug. Coarse ␣-lactose monohydrate, CL (Inhalac ® 120, Meggle AG, Wasserburg, Germany) was used in “as supplied” form as the coarse carrier. Micronized ␣-lactose monohydrate (Lactose New Zealand, Hawera, New Zealand) was prepared using a fluid energy mill (Ktron Soder, NJ, USA) as described in a previous study (Adi et al., 2006). Methanol (HPLC grade, Merck KGaA, Darmstardt, Germany), milli-Q grade water (Millipore Corporation, Melsheim, France) and ammonium acetate (BDH laboratories, Victoria, Australia) were used for HPLC analysis. 2.2. Methods 2.2.1. Preparation of powder formulations SX (2.5%)–CL mixtures containing 0, 5, 10 and 20% micronized lactose (M0F, M5F, M10F and M20F) were prepared according to a previously validated mixing method (Alway et al., 1996; Liu and Stewart, 1998). The batch size for these formulations was 5 g. The micronized SX or SX and ML were placed between equal amounts of CL in a test tube. Placing three ceramic beads (approximately 10 mm diameter) inside, the test tube was stoppered and inverted several times to prevent the drug from sticking to the sides of the test tube which was followed by vigorous shaking for 5 min by hand. A ball-milling effect for breaking up agglomerates was provided by the ceramic beads. 2.2.2. Homogeneity of powder mixtures The homogeneity of mixtures was determined to ensure that the mixing procedure produced homogenous and consistent powder mixtures that approached target SX content with low variability. Twenty samples (20 ± 0.5 mg each) were accurately weighed into suitable volumetric flasks and dissolved in 40% (v/v) methanol/water (HPLC grade) and the amount of SX was determined by a validated UV assay. The average drug content and the variability between samples, expressed by the coefficient of variation (CV), were used to quantify the homogeneity of each powder mixture. A mixture with mean drug content within 95–105% of the
443
theoretical value and a CV <5% was regarded to have an acceptable degree of homogeneity. These specifications indicated that 95% of samples would fall within 10% of the mean (Crooks and Ho, 1976). 2.2.3. Storage conditions A saturated solution of sodium chloride (NaCl) (BDH laboratories, Melbourne, Victoria, Australia) was used to produce 75% RH in sealed containers (Callahan et al., 1982). Five formulations: SX alone (D) and four SX–CL mixtures containing 0, 5, 10 and 20% ML (M0F, M5F, M10F and M20F, respectively) and CL and ML were stored in open pan as bulk powder at 75% RH for 3 months. Over the study period, the RH and temperature, monitored by a thermohygrometer (Shinyei TRH-CZ, Osaka, Japan), were observed to be 75 ± 2% and 25 ± 2 ◦ C, respectively. 2.2.4. Dispersibility using a Next Generation Impactor (NGI) In vitro dispersibility of the five powder formulations before and after storage (at 75% RH for 3 months) was carried out using a NGI according to the method described in the British Pharmacopoeia for DPIs (Appendix XXI F, http://www.pharmacopeia.co.uk). Using a Rotary vein pump and a solenoid valve timer, the in vitro measurements were carried out at 60 l/min. The pump was set using a calibrated flow meter (TSI instruments Ltd., Buckinghamshire, UK). The collection cups were coated with silicone oil before each measurement to eliminate particle bounce. The preseparator was filled with 15 ml of purified water. Accurately weighed 20 ± 0.5 mg of powder or 0.50 ± 0.03 mg of pure drug was loaded in to a size 3 gelatine capsule (Capsugel, Sydney, Australia), which was placed into a Rotahaler (Glaxo SmithKline, Melbourne, Australia). The Rotahaler was connected to a mouthpiece adaptor that was inserted into a United States Pharmacopoeia (USP) throat which was connected to the NGI. The dispersibility was carried out at 60 l/min for 4 s. The RH and temperature during experiment was 45% RH and 25 ◦ C, respectively. The NGI experiments were conducted rapidly (<5 min) to minimise any effects of RH change. Samples from the 11 parts (capsule + rotahaler, throat + mouthpiece, preseparator, stages 1–7 and micro-orifice collector, MOC) were collected in separate volumetric flasks using 40% (v/v) methanol/water to rinse. Each formulation was tested in triplicate. The amount of SX in each part was determined using a validated HPLC assay. The aerodynamic cut-off diameters at a flow rate of 60 l/min for preseparator, stage 1, stage 2, stage 3, stage 4, stage 5, stage 6 and stage 7 were 12.41 m, 8.06 m, 4.46 m, 2.82 m, 1.66 m, 0.94 m, 0.55 m and 0.34 m, respectively (Marple et al., 2003a,b). The amount of drug recovered from all stages of NGI, preseparator, throat, device and capsule was defined as the recovered dose (RD). The emitted dose (ED) was the percent of recovered dose that was collected from all sections except device and capsule. The amount of SX in each stage was then calculated as a percent of RD. The % fraction of RD that was collected in stage 3 to MOC was regarded as FPF. 2.2.5. In vitro drug dispersibility by twin stage impinger (TSI) In vitro aerosol deposition was determined using a twin stage impinger (TSI, Apparatus, A; British Pharmacopoeia, 2000) (Copley Scientific Ltd., Nottingham, UK) and a Rotahaler (Glaxo SmithKline, Melbourne, Australia). A vacuum pump (Model OD 5/2, Dynavac Engineering, Melbourne, Australia) fitted to the mouthpiece of TSI was used to create the airflow and the flow rate was adjusted to 60 l/min before each measurement. A thermohygrometer (Shinyei TRH-CZ, Osaka, Japan) was used to measure the surrounding temperature (25 ± 2 ◦ C) and RH (45 ± 2% RH). 7.0 and 30.0 ml of 40% (v/v) methanol (HPLC grade) were placed in the stage 1 and 2 of the TSI, respectively. Size 3 hard gelatine capsules (Capsugel, Sydney, Australia), loaded with the powder formulations (20 ± 0.5 mg)
444
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
or drug (0.50 ± 0.03 mg), were inserted into a Rotahaler. The Rotahaler was placed into a moulded mouthpiece attached to the TSI. The Rotahaler was twisted and the powder was then dispersed by drawing 4 l of air during each measurement (4 s at 60 l/min). After rinsing each part (inhaler, stage 1 and 2) with 40% (v/v) methanol (in water), the rinsed liquid was collected and diluted to an appropriate volume (100 ml). A validated high performance liquid chromatographic assay was used to determine the drug concentration. TSI measurements were conducted immediately after removal of the formulations from desiccators and for each formulation, five replicates were performed. The percentage of total dose that deposited in the inhaler, stage 1 (S1 ) and stage 2 (S2 ), was the recovered dose (RD). The amount of drug that was deposited in stage 1 and 2 was the emitted dose (ED) and it was calculated as the percentage of RD. The FPF was defined as the percentage of recovered dose that was deposited in stage 2. ED =
FPF =
(S1 + S2 ) × 100 RD S2 × 100 RD
2.2.9. In situ particle size distribution measurement using a Spraytec In situ PSD measurements of DPI mixtures were performed using a Spraytec (Malvern Instruments Limited, Worcestershire, UK). For each DPI mixture, the particle size distribution (PSD) was measured with the inhalation cell attachment at a flow rate of 60 l/min. The flow rate was controlled using a Critical Flow Controller Model TPK 2000 & Flow meter model DFM 2000 (Copley Scientific Limited, Nottingham, UK). All measurements were conducted on five replicates at room temperature (25 ◦ C) and 45% RH. The measurements were performed for 4 s with triggering level of 50, noise level of 0 and background level of 100. The particle size of the mixture was analysed using Spraytec version 3.0 (Malvern Instruments Limited, Worcestershire, UK). 2.3. Statistical analysis All data were subjected to One Way Analysis of Variance (ANOVA) (SPSS, Version 15.0, Chicago, IL, USA). Probability (P) values of ≤0.05 when analysed by post hoc multiple comparisons were considered as statistically significant. 3. Results and discussion
2.2.6. High performance liquid chromatography (HPLC) assay SX recovered by twin stage impinger (TSI) and Next Generation Impactor was analysed by HPLC using a C18 column (5 m, 4.6 mm × 150 mm, Apollo, Alltech associates, Deerfield, IL, USA) and a UV detector (Shimadzu Diode Array Detector, SPD M10A VP, Kyoto, Japan) at a wavelength of 252.2 nm. A mixture of methanol and 0.2% (w/v) ammonium acetate solution in water (55:45, pH ∼ 6.9) that was freshly prepared, filtered and degassed using 0.45 m membrane filter (Millipore, County Cork, Ireland) was used as the mobile phase. A volume of 15 l was injected (Shimadzu Autosampler, SIL10AD VP, Kyoto, Japan) at ambient temperature at a flow rate of 1.0 ml/min that was generated by a HPLC pump (Shimadzu LC10 AJ, 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 g/ml and 10 g/ml, and the linear regression was performed using Sigmaplot (Jandel Scientific, Chicago, IL, USA). The precision of each assay was determined by analysing five replicates from each of two standard SX solutions (two concentrations: low 1.0 g/ml and high 4.0 g/ml).
3.1. Particle size distributions of formulation ingredients The particle size distributions of salmeterol xinafoate (SX), micronized lactose (ML) and coarse lactose (CL) were determined using a laser diffraction method with a Malvern Mastersizer 2000. The average of at least three replicates at each pressure was calculated.
2.2.7. Particle size analysis by using a Mastersizer 2000 The particle size of the powders was determined by laser scattering using the dry dispersion technique in a Malvern Mastersizer 2000 (Malvern Instruments Limited, Worcestershire, UK). A Scirocco cell and Scirocco 2000 dry powder feeder were used. Particle measurement was carried out at four different pressures (0.5 bar, 1.0 bar, 2.0 bar and 3.0 bar). The particle size of the powders was analysed using Malvern Mastersizer software (version 5.22). The average PSD was measured from three replicates of each sample. 2.2.8. Scanning electron microscopy (SEM) Powder was sprinkled and glued on metal sample plates. Using an electrical potential of 2.0 kV at 25 mA for 10 min, the powder was gold plated with a sputter coater (BAL-TEC SCD 005, Tokyo, Japan). The surface morphology of the particles was examined at several magnifications under JOEL JSM 6000F scanning electron microscope at 15 kV (JOEL, Tokyo, Japan).
Fig. 1. Particle size distributions (PSDs) of (A) micronized lactose (ML) and salmeterol xinafoate (SX) determined at 1.0 bar pressure and (B) coarse lactose (CL) determined at 0.5 bar and 2.0 bar pressures using Mastersizer 2000 employing dry dispersion technique (n = 3, data are represented as mean ± S.D.).
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
At 0.5 bar pressure, both SX and ML showed multimodal distributions with large variability indicating the existence of agglomerates. However, at 1.0 bar pressure, both ML and SX were fully dispersed and showed a monomodal distribution in the range of 0.1–11.0 m (Fig. 1A). The volume mean diameter (VMD) of SX was 1.9 m and 90% particles were <4.1 m. Almost all particles were <8.7 m. The PSD of ML were similar to SX, having 50 and 90% particles <2.4 m and 5.5 m, respectively. Increasing shear pressures did not change the PSD of SX and ML. A monomodal distribution around 100 m was observed for CL at 0.5 bar (Fig. 1B). The VMD of coarse lactose was 126.7 m. However, when the PSD was determined at 1.0 bar pressure, a tri-modal distribution was observed for CL (Fig. 1B) with modes at <7 m, 7–30 m and around 100 m. The frequency of the two smaller sized modes (<7 m and 7–30 m) was small but increased as the pressure was increased. The appearance of these smaller modes at 1.0 bar (and greater) shear pressure could be due to the dissociation of any fines that were present in CL or comminution of CL at high shear pressure. 3.2. Powder formulations 3.2.1. Morphology of powder formulations Inhalation grade SX particles usually form agglomerates (Fig. 2A). High cohesiveness of these micronized particles reduces
445
flow properties and dispersibility (Ganderton and Kassem, 1992). In order to improve flow properties and bulk, coarse carriers such as lactose monohydrate (Fig. 2B) are usually added to SX formulations. In SX–CL binary mixture, the majority of the SX particles, either individual particles or agglomerates, remained adhered to the coarse lactose surface (Fig. 2C). In SX–CL mixtures containing additional ML, agglomerates were observed both on and off the surface of coarse lactose (Fig. 2D). These agglomerates were likely to consist of SX and/or ML particles and the composition and arrangement of SX and ML could vary between mixtures containing different concentrations of ML due to complex interparticulate interactions occurring among different particles during powder mixing (Adi et al., 2006; Das et al., 2009). The size of the agglomerates also varied (Fig. 2D). 3.2.2. Homogeneity of powder formulations 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 (Crooks and Ho, 1976). For the determination of intra-batch variability, the in vitro TSI depositions of five replicates of a single batch of M20F were compared. The ED and the SX content of the stage 1 and 2 of the TSI were 81.1 ± 0.5%, 76.7 ± 0.6% and 11.3 ± 0.5%, respectively, and showed no significant differences (P > 0.05). For the determination of inter-batch variabil-
Fig. 2. Scanning electron micrographs of (A) salmeterol xinafoate, SX (magnification: 3000×), (B) coarse lactose, CL (magnification: 686×), (C) mixture containing 0% additional micronized lactose, M0F (magnification: 529×) and (D) mixture containing 20% micronized lactose, M20F (magnification: 412×).
446
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
ity, the in vitro deposition of five batches of M20F were compared. The ED and the SX content of the stage 1 and 2 of the TSI were 80.7 ± 0.5%, 74.8 ± 0.6% and 11.4 ± 0.9%, respectively, and showed no significant differences (P > 0.05). No significant intra- or interbatch variability in ED, and SX content of stage 1 and 2 of the TSI was seen for M0F when tested using the same procedure. Homogeneity studies together with intra-batch and inter-batch dispersibility comparisons confirmed that the powder mixing process used in this research produced homogeneous mixtures with reproducible in vitro aerosol deposition. 3.3. In vitro dispersibility using a Next Generation Impactor (NGI) In order to examine the dispersibility changes, five SX formulations: SX alone (D) and SX–CL mixtures containing 0%, 5%, 10% and 20% micronized lactose (M0F, M5F, M10F and M20F, respectively), before and after storage at 75% RH for 3 months, were dispersed in triplicate at an airflow of 60 l/min using a Next Generation Impactor (NGI). Before storage, the emitted dose (ED) of SX alone formulation (D) was 43.5 ± 1.3%, in comparison to 55.5 ± 6.1% for the SX–coarse lactose mixture (M0F) (Fig. 3A). When micronized lactose was added to the formulation, the ED decreased and the lowest ED (35.9 ± 4.1%) was observed with the mixture containing 20% micronized lactose, M20F. After storage, the ED of M20F sharply increased from 35.9 ± 4.1% to 53.5 ± 1.1%. The ED of the mixture containing 10% micronized lactose (M10F) also significantly increased from 43.8 ± 1.8% to 51.7 ± 1.7%. Consolidation of the loose agglomerates after storage at the high RH into a more structurally permanent
form due to enhanced capillary interactions/solid bridging possibly developed into a more flowable formulation, resulting in greater capsule emptying and decreased adhesion of free SX on the capsule wall (Adi et al., 2006). No significant difference (P < 0.05) in ED was observed for the other formulations (D, M0F and M5F). No significant differences (P > 0.05) were observed in the percent of SX that remained in (a) the throat and mouthpiece, (b) the preseparator, (c) stage 1 (8.06–12.41 m), (d) stage 2 (4.46–8.06 m), or (e) stage 3–7 including micro-orifice collector (FPF, <4.46 m) for any of the four formulations (D, M0F, M5F, M10F) before and after storage (data not shown). For M20F after storage, the %SX significantly increased from 16.8 ± 1.6% to 21.0 ± 1.8% in throat and mouthpiece and from 11.4 ± 1.5% to 21.3 ± 1.9% in preseparator. The %SX for M20F also significantly increased from 1.5 ± 0.1% to 2.0 ± 0.1% in stage 1 (P < 0.05) (Fig. 3B). In contrast, the %SX collected in stage 2 (4.46–8.06 m) significantly decreased (P = 0.04) after storage while the %SX decrease in stage 3 to MOC (<4.46 m fraction) was marginally significant (P = 0.047) at the 0.05 level. A decrease in %SX occurred in all of the stages which constituted the FPF (stage 3 to MOC). The volume mean diameter (VMD) of SX was 1.9 m, with 90% particles <4.1 m, and almost all particles <8.7 m (Fig. 1A). In the M20F mixture, SX may exist as individual particles, SX agglomerates and/or associated with lactose as mixed SX–ML agglomerates of interactive units with CL. The changes shown in Fig. 3B might result from changes in particulate interactions due to increased capillary interaction and/or solid bridging due to surface moisture at 75% RH. These changes indicated that a decrease in the concentration of fully dispersed particles of SX or small agglomerates <8.06 m and an increase in the concentration of agglomerates above 8.06 m occurred after storage at 75% RH. Such changes could involve interactions between any of the particulate species but a previous study by Das et al. (2009) indicated that lactose–lactose interactions were likely to be more influential in modifying particulate behaviour. Thus, SX–SX and SX–lactose interactions were less likely to be responsible for the increased agglomeration observed, although they could not be dismissed. If the behaviour shown in Fig. 3B was related to increased lactose–lactose interactions, two mechanisms remain possible: (a) modifications of the lactose particle size resulting from increased interaction between lactose particles in the high RH conditions and producing dispersibility changes in a similar manner those occurring in a previous study (Adi et al., 2006) or (b) increased lactose–lactose interactions within mixed agglomerates of ML and SX leading to structural changes in the mixed agglomerates, trapping SX and decreasing its dispersibility. These mechanisms were tested in the following studies. 3.4. Particle size distribution changes of salmeterol xinafoate, coarse lactose and micronized lactose after storage at 75% RH for 3 months
Fig. 3. Comparison of (A) emitted dose (ED) of SX of five formulations: SX alone and SX–coarse lactose mixtures containing 0%, 5%, 10% and 20% micronized lactose, ML (D, M0F, M5F, M10F and M20F, respectively), (B) the %SX fraction at throat and mouthpiece, preseparator, stage 1 (8.06–12.41 m), stage 2 (4.46–8.06 m) and stage 3–7 including micro-orifice collector (MOC) (<4.46 m) of mixture containing 20% micronized lactose (M20F) before and after storage (at 75% RH for 3 months) determined using a Next Generation Impactor (NGI) at an airflow of 60 l/min [the cut-off diameter for relevant stage was shown in parentheses], (n = 3, data are represented as mean ± S.D.).
In order to determine if any of the components of the mixture interacted to produce a permanent change in particle sizes, after storage at 75% RH for 3 months, the PSD of the stored samples of SX, CL and ML were determined using a laser diffraction method with a Malvern Mastersizer 2000. For each sample (SX, CL or ML), the PSDs obtained before and after storage were compared. The average of at least three replicates at each pressure was determined. At 1.0 bar and greater shear pressures, the PSD of SX before and after storage overlapped within the error margin indicating no change in particle size (distributions not shown). This indicated that any interaction between particles of SX before and after stor-
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
447
Fig. 5. Fine particle fraction (FPF) of (A) mixture containing 20% micronized lactose (M20F) before storage, (B) M20F after storage at 75% for 3 months (C) M20F before storage that was prepared using ML stored at 75% RH for 3 months (n = 5, data are represented as mean ± S.D.).
3.5. Effect of stored micronized lactose on SX dispersibility from M20F
Fig. 4. Comparison of (A) particle size distributions (PSDs) at 1.0 bar pressure and (B) volume mean diameter (VMD) at three different pressures (1.0 bar, 2.0 bar and 3.0 bar) of micronized lactose (ML) before and after storage (at 75% RH for 3 months) determined using a Mastersizer 2000 employing dry dispersion technique (n = 3, data are represented as mean ± S.D.).
age were similar and capillary interactions and/or solid bridging that might have occurred during storage at 75% RH did not lead to any particle size change. Similarly, no change in PSD of CL was observed before and after storage at all shear pressures. The distributions were identical with those shown in Fig. 1 for both SX and CL. However, the PSD of ML increased after storage and the volume mean diameter (VMD) of ML increased from 2.4 ± 0.2% m to 3.1± 0.1% m at 1.0 bar pressure (Fig. 4A). SEM images were not helpful in identifying the causes and evidence of solid bridging could not be seen. The increased particle size was an indication of strong interactions between small ML particles. The increase in particle size was found at all shear pressures indicating that very strong interactions either from increased capillary cohesion or solid bridging were very likely (Fig. 4B). This finding is also supported by Bridson et al. who found an increase in the 5th percentile of the PSD of lactose after 1–2 weeks storage at 70% RH (Bridson et al., 2007). Previous studies showed that the dispersibility of SX is dependent on the particle size of added micronized lactose (Adi et al., 2006). Higher FPF for SX mixtures containing 10–20% added micronized lactose was observed when the VMD of the added micronized lactose was increased from 3.0 m to 7.9 m. The change was attributed to a change in the agglomerate structure with a resultant decrease in agglomerate strength. It was uncertain whether the change in particle size observed in this current study (i.e., from 2.4 ± 0.2% m to 3.1± 0.1% m) was sufficient to change the dispersibility of the mixture. Therefore, the effect of using ML (that had been stored at high RH) on the FPF of freshly prepared M20F was tested.
This mixture was prepared using ML stored at 75% RH for 3 months but using CL and SX which were not stored at high RH. The FPF was determined using a twin stage impinger (TSI). The FPF of M20F before storage was 11.3 ± 1.2%. After storage of the mixture at 75% RH for 3 months, the FPF decreased to 8.2 ± 0.6% (Fig. 5). The FPF of the freshly prepared M20F using ML stored at 75% RH increased significantly to 19.4 ± 2.3% (Fig. 5). The increased FPF of freshly prepared M20F using stored ML was consistent with that observed earlier by Adi et al. (2006); however, the extent of increase was surprising given the relatively small change in particle size (i.e., from 2.4 ± 0.2% m before storage to 3.1 ± 0.1% m after storage). Any discussion of the reasons for the marked increase can only be speculative. Given that the extent of dispersibility would be related to the structure of the SX–ML agglomerates (Adi et al., 2006), the shift of the ML PSD to a distribution with a higher mean particle size and a smaller span (decreased from 1.67 ± 0.01 to 1.45 ± 0.01 after storage) might have influenced the agglomerate structure and strength. This part of the study was not pursued further. However, the important outcome was that the increase in particle size of ML in the formulation alone increased SX dispersibility and was not consistent with the decreased SX dispersibility observed in the M20F formulation during storage. Thus, the mechanism of reduced SX dispersibility was not related to increased lactose–lactose interactions producing larger PSD of ML. 3.6. Particle size distribution changes of mixtures containing micronized lactose after storage at 75% RH for 3 months The PSD of the mixtures containing micronized lactose (M5F, M10F and M20F) before and after storage at 75% RH for 3 months were determined using a laser diffraction method with a Malvern Mastersizer 2000 using a dry dispersion technique. The average of at least three replicates conducted at 1.0 bar shear pressure was determined. Each SX–CL mixture containing additional ML (M5F, M10F and M20F) showed tri-modal distributions. The PSD of M5F and M10F are shown in Fig. 6A. The chemical composition of the particles in each mode was not able to be identified. However, while PSD and SEM gave some guidance, the true compositions of the modes could only be speculated. The smallest mode, in the range of <7.0 m, could be predominantly dispersed SX particles, dis-
448
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
and subsequent particle size increase seen when micronized lactose was stored alone. 3.7. Particle size distribution of the aerosol plume before and after storage of the formulations In order to understand the deagglomeration pattern that would be encountered during inhalation, particle size measurements of the aerosol clouds of all five formulations (D, M0F, M5F, M10F and M20F) before and after storage at 75% RH for 3 months were conducted using an in situ particle size analyser (Spraytec) with the Rotahaler as the dispersion device. Each powder was dispersed five times at a flow rate of 60 l/min and the averaged PSD before and after storage was determined. All mixtures produced a bi-modal distribution of powders in the aerosol plume (Fig. 7A). The frequency of the first mode increased with increasing concentration of ML, which was consistent with the PSD observed using Mastersizer 2000. The mode at the lower sizes was broad (3–30 m) and could represent SX and ML individual particles, small agglomerates of SX and ML and SX–ML mixed agglomerates while the second mode represented CL carrier, SX–CL and SX–ML–CL interactive units. SX alone showed a broader distribution ranging from 1.0 m to 1000 m indicating the existence of a wide distribution of individual particles and agglomerates (data not shown). No change was observed in the PSD of SX alone and SX mixtures after storage containing 0% and 5% micronized lactose. The mixture containing 10% ML showed some evidence of the reduced rate of deagglomeration of smaller agglomerates after storage. A clear difference was observed for the mixture containing 20% ML (Fig. 7B). Though there was a large standard deviation associated
Fig. 6. (A) Particle size distributions (PSDs) of mixtures containing 5% and 10% micronized lactose (M5F and M10F, respectively) (B) comparison of PSD of the mixture containing 20% micronized lactose (M20F) before and after storage (at 75% RH for 3 months) determined using a Mastersizer 2000 at 1.0 bar pressure and (C) PSD of M20F (shown in 6B) in the range of <40 m (n = 3, data are represented as mean ± S.D.).
persed ML particles and small agglomerates of SX, ML or SX–ML. The second mode, in the range of 10–30 m, was likely to consist of agglomerates (SX, ML or SX–ML) and the third mode at 95–105 m could be CL, ML–CL, or SX–ML–CL interactive units. The second mode (in the range of 10–30 m) increased with increasing concentration of ML in mixtures (Fig. 6A). The PSD for M5F and M10F before and after storage overlapped within the error margin (data not shown). On the other hand, the PSD of M20F before and after storage differed. After storage, the frequency of particles <7.0 m was lower and the frequency in the size range of 10–30 m was higher than the mixture prior to storage (Fig. 6B and C). The reduced deagglomeration of 10–30 m size agglomerates after storage was assumed to be reason for the decrease in <7.0 m fraction (Fig. 6C). In the mixture containing 20% ML, SX and ML particles formed agglomerates with a greater propensity of ML–ML contact. The formation of stronger agglomerates, due to strong cohesion and/or solid bridging at ML–ML contact during storage at 75% RH (Das et al., 2009), might have caused the SX particles to be mechanically trapped inside the agglomerates, which ultimately resulted in decreased dispersibility after storage. The formation of stronger SX–ML agglomerates during storage at 75% RH was consistent with the increasing interaction between micronized lactose particles
Fig. 7. Comparison of particle size distributions (PSDs) of (A) mixtures containing 0%, 5% and 10% micronized lactose (M0F, M5F and M10F, respectively) before storage, (B) mixture containing 20% micronized lactose (M20F) before and after storage (at 75% RH for 3 months) determined using a Spraytec at an airflow of 60 l/min (n = 5, data are represented as mean ± S.D.).
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
with the PSD of the stored formulations, the frequency of particles in the VMD range of 4–25 m increased after storage and the frequency <4 m VMD slightly decreased. For example, the percentage of particles <3.98 m size for M20F decreased from 7.1 ± 1.2% to 5.4 ± 1.8%. After storage the dispersibility of individual particles of SX in the aerosol plume decreased. Given that the major mechanism for SX dispersibility is deagglomeration of SX–ML mixed agglomerates (Adi et al., 2006), the influence of the high humidity conditions on these agglomerates is evident with a clearly decreased capacity of the agglomerates to release SX. This finding is in agreement with those observed in the NGI and Mastersizer 2000 study and supports the presence of agglomerates that were less likely to disperse and release SX. 4. Conclusions This study has identified the mechanism by which mixtures of SX containing high concentrations of micronized lactose (i.e. 20%) decrease in vitro deposition after storage at 75% RH. The %SX of an SX mixture containing 20% micronized lactose significantly increased in the throat and mouthpiece, preseparator and stage 1 of the NGI (P < 0.05) after storage. In contrast, the %SX for the same formulation significantly decreased (P < 0.05) in remaining stages (stage 2 to micro-orifice collector, MOC) of NGI after storage. The ED of this formulation also increased after storage indicating greater capsule emptying through better flow of the formulation or less SX adhesion to the capsule wall. While the particle size of micronized lactose increased when stored at 75% RH for 3 months, the FPF of freshly prepared SX mixture containing 20% of the stored micronized lactose increased, indicating that the particle size changes to the micronized lactose alone were not responsible for the decreased in vitro performance of this mixture after storage. The extent of increase caused by a small change in the PSD (i.e. from 2.4 m to 3.1 m and a decreased span) was surprising and may have been related to changes in agglomerate structure due to particle size changes (Adi et al., 2006). This was not pursued further as it was not core to understanding the mechanism. Particle size data of both the SX mixtures at various shear pressures and of the aerosol plume after dispersion from a model device showed an increased presence of agglomerates in the range 4–25 m and decreased dispersed particles <4 m. SX and the coarse lactose carrier showed no change in PSD after storage and thus any increased interaction between these mixture components was not the cause of the increased agglomeration and decreased dispersibility. In addition, significant changes in PSD only occurred in mixtures containing higher concentrations of micronized lactose (clearly seen when 20% was used and some indication when 10% was used). Thus, agglomerate formation was related to the presence of micronized lactose. The mechanisms of dispersibility of SX mixtures containing micronized lactose (Islam et al., 2004; Adi et al., 2006) have been identified. In these mixtures where there were relatively high concentrations of cohesive powders, the major mechanism of dispersibility was identified as the deagglomeration of mixed agglomerates of SX and micronized lactose. The mechanism of reduced dispersibility during storage, therefore, must be related to increased cohesion within the SX–micronized lactose agglomerates due to increased capillary interactions or solid bridging (Das et al., 2009) producing stronger agglomerates that did not disperse. This outcome is consistent with the particle sizing results of this project. This study is important as it addressed a practical industrial problem encountered by the dry powder inhaler formulation industry. The study has revealed the mechanism of dispersibility change during storage and thus this body of knowledge provides guidance and direction to mixture design and formulation strategies.
449
A laboratory scale mixing process was employed using small batches (5 g) like many of the studies in the literature. The batch size and type of mixer will influence the magnitude of shear generated during mixing and thus will influence particle contacts and agglomerate structure. The outcomes of this research will be related to the mixing process; however, these principles will apply to other mixing conditions where agglomerates are produced. Careful interpretation of the data is therefore required to apply these outcomes in industry applications. Acknowledgements Shyamal Das would like to acknowledge the scholarship support through the Faculty of Pharmacy, Monash University. Shyamal Das also acknowledges Dr Handoko Adi and Dr Herbert Chiou for their support during the use of NGI and Mastersizer 2000, and Dr Joseph A Nicolazzo for his help in statistical analysis. References 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. Alway, B., Sangchantra, R., Stewart, P., 1996. Modelling the dissolution of diazepam in lactose interactive mixtures. Int. J. Pharm. 130, 213–224. Boerefijin, R., Ning, Z., Ghadiri, M., 1998. Disintegration of weak lactose agglomerates for inhalation applications. Int. J. Pharm. 172, 199–209. 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. Braun, M., Oschmann, R., Schmidt, P., 1996. Influence of excipients and storage humidity on the deposition of disodium cromoglycate (DSCG) in the twin impinger. Int. J. Pharm. 135, 53–62. Bridson, R., Robbins, P., Chen, Y., Westerman, D., Gillham, C., Roche, T., Seville, J., 2007. The effects of high shear blending on lactose monohydrate. Int. J. Pharm. 339, 84–90. Callahan, J., Cleary, G., Elefant, M., Kaplan, G., Kensler, T., Nash, R., 1982. Equilibrium moisture content of pharmaceutical excipients. Drug Dev. Ind. Pharm. 8, 355–369. Crooks, M., Ho, R., 1976. Ordered mixing in direct compression of tablets. Powder Technol. 14, 161–167. Das, S., Larson, I., Young, P., Stewart, P., 2009. Influence of storage relative humidity on the dispersion of salmeterol xinafoate powders for inhalation. J. Pharm. Sci. 98, 1015–1027. Ganderton, D., Kassem, N., 1992. Dry powder inhalers. In: Ganderton, D. (Ed.), Advances in Pharmaceutical Sciences, vol. 6. Academic Press, London, UK, pp. 165–191. Hindle, M., Makinen, G., 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. Hogg, R., 1989. Role of colloid and interface science in agglomeration. In: Cross, M., Oliver, R. (Eds.), Proceedings of the 5th International Symposium on Agglomeration. The Institute of Chemical Engineers. Rugby, Warwickshire. Islam, N., Stewart, P., Larson, I., Hartley, P., 2004. Lactose modification by decantation: are drug–fine lactose ratios the key to better dispersion of salmeterol xinafoate from lactose-interactive mixtures? Pharm. Res. 21, 492– 499. 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. Liu, J., Stewart, P., 1998. De-aggregation during the dissolution of benzodiazepines in interactive mixtures. J. Pharm. Sci. 87, 1632–1638. Louey, M., Stewart, P., 2002. Particle interactions involved in aerosol dispersion of ternary interactive mixtures. Pharm. Res. 19, 1524–1531. Lucas, P., Anderson, K., Staniforth, J., 1998. Protein deposition from dry powder inhalers: fine particle multiplets as performance modifiers. Pharm. Res. 15, 562– 569. Marple, V., Olson, B., Mitchell, J., Murray, S., Hudson-Curtis, B., 2003a. Next generation pharmaceutical impactor (a new impactor for pharmaceutical inhaler testing). Part II. Archival calibration. J. Aerosol Med. 16, 301–324. Marple, V., Roberts, D., Romay, F., Miller, N., Truman, K., Van, O., Olsson, B., Holroyd, M., Mitchell, J., Hochrainer, D., 2003b. Next generation pharmaceutical impactor (a new impactor for pharmaceutical inhaler testing). Part I. Design. J. Aerosol Med. 16, 283–299. Padmadisastra, Y., Kennedy, R., Stewart, P., 1994a. Influence of carrier moisture adsorption capacity on the degree of adhesion of interactive mixtures. Int. J. Pharm. 104, R1–R4. Padmadisastra, Y., Kennedy, R., Stewart, P., 1994b. Solid bridge formation in sulphonamide-Emdex interactive systems. Int. J. Pharm. 112, 55–63.
450
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
Podczeck, F., Newton, J., James, M., 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. Podczeck, F., Newton, J., James, M., 1997a. Influence of relative humidity of storage air on the adhesion and autoadhesion of micronized particles to particulate and compacted powder surfaces. J. Colloid Int. Sci. 187, 484–491. Podczeck, F., Newton, J., James, M., 1997b. 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. Staniforth, J., 1996. Improvement in dry powder inhaler performance. In: Proc. Drug Del. Lungs Conf. VII, pp. 86–89. Young, P., Price, R., 2004. The influence of humidity on the aerosolization of micronized and SEDS produced salbutamol sulphate. Eur. J. Pharm. Sci. 22, 235–240. Young, P., Sung, A., Traini, D., Kwok, P., Chiou, H., Chan, H., 2007. Influence of humidity on the electrostatic charge and aerosol performance of dry powder inhaler carrier based systems. Pharm. Res. 24, 963–970.
Young, P., Price, R., Tobyn, M., Buttrum, M., Dey, F., 2003a. Investigation into the effect of humidity on drug–drug interactions using the atomic force microscope. J. Pharm. Sci. 92, 815–822. Young, P., Price, R., Tobyn, M., Buttrum, M., Dey, F., 2003b. Effect of humidity on aerosolization of micronized drugs. Drug Dev. Ind. Pharm. 29, 959–966. Young, P., Price, R., Tobyn, M., Buttrum, M., Dey, F., 2004. The influence of relative humidity on the cohesion properties of micronized drugs used in inhalation therapy. J. Pharm. Sci. 93, 53–761. Zeng, X., Tee, S., Martin, G., Marriott, C., 1996. Effect of mixing procedure and particle size distribution of carrier particles on the deposition of salbutamol sulphate from dry powder inhaler formulations. In: Proc. Drug Del. Lungs Conf. VII, pp. 40–43. Zeng, X., MacRitchie, H., Marriott, C., Martin, G., 2007. Humidity-induced changes of the aerodynamic properties of dry powder aerosol formulations containing different carriers. Int. J. Pharm. 333, 45–55.
Contents lists available at ScienceDirect
European Journal of Pharmaceutical Sciences journal homepage: www.elsevier.com/locate/ejps
Agglomerate properties and dispersibility changes of salmeterol xinafoate from powders for inhalation after storage at high relative humidity Shyamal Das a , Ian Larson a , Paul Young b , Peter Stewart a,∗ a b
Drug Delivery, Disposition and Dynamics, Monash Institute of Pharmaceutical Sciences, Monash University (Parkville Campus), 381 Royal Parade, Victoria 3052, Australia Advanced Drug Delivery Group, Faculty of Pharmacy, University of Sydney, NSW 2006, Australia
a r t i c l e
i n f o
Article history: Received 7 November 2008 Received in revised form 27 March 2009 Accepted 30 March 2009 Available online 8 April 2009 Keywords: Salmeterol xinafoate Agglomerate Particle size distribution Storage Relative humidity Dry powder inhaler
a b s t r a c t Purpose: This study investigated changes in agglomeration and the mechanism of dispersibility decrease of salmeterol xinafoate (SX) from SX–lactose mixtures for inhalation after storage at 75% RH for 3 months. Methods: The dispersibility, PSD and in situ PSD of aerosol plumes of SX alone and SX–coarse lactose (CL) mixtures containing 0, 5, 10 and 20% micronized lactose (ML) before and after storage were determined by a Next Generation Impactor (NGI), a Mastersizer 2000 and a Spraytec, respectively. Results: The PSD of ML increased after storage at 75% RH, but dispersibility of SX using the stored ML increased. After storage, the %SX of the mixture containing 20% ML (M20F) significantly increased (P < 0.05) in the throat and mouthpiece, preseparator and stage 1 of NGI, while it significantly decreased in the remaining stages (P < 0.05). In situ analysis of aerosol plumes of M20F supported this result with an increased presence of particles of 4–25 m and a decreased respirable particle distribution of <4 m after storage. Conclusions: The decreased dispersibility of M20F after storage was due to the formation of less dispersible agglomerates, probably occurring through enhanced capillary interaction and/or solid bridging of ML, entrapping and preventing the release of SX particles. Crown Copyright © 2009 Published by Elsevier B.V. All rights reserved.
1. Introduction Changes in dispersibility of dry powder inhalers during storage at high relative humidity (RH) are often encountered by pharmaceutical manufacturers (Borgstrom et al., 2005; Young et al., 2007). The dispersibility of powders is influenced by interparticulate forces (Louey and Stewart, 2002) which can be modified by environmental RH (Podczeck et al., 1996, 1997a,b; Young et al., 2003a, 2004). High RH may increase interparticulate forces due to increased capillary interactions resulting in the formation of larger agglomerates (Hogg, 1989) that are less breakable (Boerefijin et al., 1998). Micronized lactose monohydrate has been found to dissolve under the influence of high RH (75% RH or greater) (Podczeck et al., 1997a). Thus, the liquid bridges which are formed at these high RH conditions are often followed by solid bridges due to recrystallization of the dissolved lactose (Padmadisastra et al., 1994a,b). Many investigations on the effect of storage RH on dispersibility (Braun et al., 1996; Hindle and Makinen, 1996; Young et al., 2003b; Lida et al., 2004; Young and Price, 2004; Borgstrom et al., 2005; Young et al., 2007; Zeng et al., 2007) were limited to either drug
∗ Corresponding author. Tel.: +61 3 9903 9517; fax: +61 3 9903 9583. E-mail address: [email protected] (P. Stewart).
alone formulations or binary mixtures of drugs. These studies did not include dry powder inhaler formulations that contained ternary components (Staniforth, 1996; Zeng et al., 1996), although the addition of ternary components, for example, micronized lactose, has been shown to improve dispersibility (Lucas et al., 1998; Louey and Stewart, 2002; Adi et al., 2006). Moreover, in most studies, formulations were exposed to storage RH for only short periods of time (maximum 7 days). A recent extended study on different types of dry powder formulations (Das et al., 2009) has reported that the salmeterol xinafoate (SX) dispersibility was influenced predominately by the presence of micronized lactose in the formulation and the storage relative humidity. A significant decrease in fine particle fraction (FPF) of SX after storage at 75% RH was observed in the ternary formulation containing 20% ML within 4 weeks and the decrease reached lower static levels within 3 months. Calculations revealed that increased capillary interactions were likely to occur between lactose particles and that deagglomeration required increased shear pressures after storage at 75% RH. The extent of particulate interactions increased, but the study did not identify the mechanism of decreased dispersibility. For example, the study outcomes could be consistent with increased lactose–lactose interactions resulting in (a) increase in the particle size distribution (PSD) of lactose or in (b) the formation of strong lactose agglomerates entrapping the SX and
0928-0987/$ – see front matter. Crown Copyright © 2009 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.ejps.2009.03.016
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
decreasing its dispersibility. An increase in the size distribution of lactose will influence dispersibility in different ways. Small changes in particle size in the cohesive range, essentially less than about 10 m, have been shown to influence dispersibility of SX (Adi et al., 2006). However, significant agglomeration of micronized lactose, producing stronger free flowing agglomerates, will likely decrease SX dispersibility due to the capacity of the agglomerate to act as a secondary carrier and strongly bind the SX to its surface. Decreased dispersibility has been observed in formulations containing ternary lactose acting as secondary carriers (Adi et al., 2006). Thus, particle size changes may potentially result in either decreased or increased dispersibility of drugs. However, the entrapment of a drug within a lactose agglomerate network will decrease dispersibility. An understanding of the mechanism is essential to establish the body of knowledge to enable quality by design and to ensure that appropriate strategies are employed in the development of powders for inhalation. The purpose of this study was to identify the mechanism of dispersibility change during storage at high RH of an SX formulation containing Inhalac 120 as the coarse lactose carrier and micronized lactose monohydrate as the fine added excipient. The study used a combination of particle sizing approaches to achieve the outcome. 2. Materials and methods 2.1. Materials Inhalation grade micronized salmeterol xinafoate (SX) (batch no., B068803, Glaxo SmithKline, Ware, UK) was used as model drug. Coarse ␣-lactose monohydrate, CL (Inhalac ® 120, Meggle AG, Wasserburg, Germany) was used in “as supplied” form as the coarse carrier. Micronized ␣-lactose monohydrate (Lactose New Zealand, Hawera, New Zealand) was prepared using a fluid energy mill (Ktron Soder, NJ, USA) as described in a previous study (Adi et al., 2006). Methanol (HPLC grade, Merck KGaA, Darmstardt, Germany), milli-Q grade water (Millipore Corporation, Melsheim, France) and ammonium acetate (BDH laboratories, Victoria, Australia) were used for HPLC analysis. 2.2. Methods 2.2.1. Preparation of powder formulations SX (2.5%)–CL mixtures containing 0, 5, 10 and 20% micronized lactose (M0F, M5F, M10F and M20F) were prepared according to a previously validated mixing method (Alway et al., 1996; Liu and Stewart, 1998). The batch size for these formulations was 5 g. The micronized SX or SX and ML were placed between equal amounts of CL in a test tube. Placing three ceramic beads (approximately 10 mm diameter) inside, the test tube was stoppered and inverted several times to prevent the drug from sticking to the sides of the test tube which was followed by vigorous shaking for 5 min by hand. A ball-milling effect for breaking up agglomerates was provided by the ceramic beads. 2.2.2. Homogeneity of powder mixtures The homogeneity of mixtures was determined to ensure that the mixing procedure produced homogenous and consistent powder mixtures that approached target SX content with low variability. Twenty samples (20 ± 0.5 mg each) were accurately weighed into suitable volumetric flasks and dissolved in 40% (v/v) methanol/water (HPLC grade) and the amount of SX was determined by a validated UV assay. The average drug content and the variability between samples, expressed by the coefficient of variation (CV), were used to quantify the homogeneity of each powder mixture. A mixture with mean drug content within 95–105% of the
443
theoretical value and a CV <5% was regarded to have an acceptable degree of homogeneity. These specifications indicated that 95% of samples would fall within 10% of the mean (Crooks and Ho, 1976). 2.2.3. Storage conditions A saturated solution of sodium chloride (NaCl) (BDH laboratories, Melbourne, Victoria, Australia) was used to produce 75% RH in sealed containers (Callahan et al., 1982). Five formulations: SX alone (D) and four SX–CL mixtures containing 0, 5, 10 and 20% ML (M0F, M5F, M10F and M20F, respectively) and CL and ML were stored in open pan as bulk powder at 75% RH for 3 months. Over the study period, the RH and temperature, monitored by a thermohygrometer (Shinyei TRH-CZ, Osaka, Japan), were observed to be 75 ± 2% and 25 ± 2 ◦ C, respectively. 2.2.4. Dispersibility using a Next Generation Impactor (NGI) In vitro dispersibility of the five powder formulations before and after storage (at 75% RH for 3 months) was carried out using a NGI according to the method described in the British Pharmacopoeia for DPIs (Appendix XXI F, http://www.pharmacopeia.co.uk). Using a Rotary vein pump and a solenoid valve timer, the in vitro measurements were carried out at 60 l/min. The pump was set using a calibrated flow meter (TSI instruments Ltd., Buckinghamshire, UK). The collection cups were coated with silicone oil before each measurement to eliminate particle bounce. The preseparator was filled with 15 ml of purified water. Accurately weighed 20 ± 0.5 mg of powder or 0.50 ± 0.03 mg of pure drug was loaded in to a size 3 gelatine capsule (Capsugel, Sydney, Australia), which was placed into a Rotahaler (Glaxo SmithKline, Melbourne, Australia). The Rotahaler was connected to a mouthpiece adaptor that was inserted into a United States Pharmacopoeia (USP) throat which was connected to the NGI. The dispersibility was carried out at 60 l/min for 4 s. The RH and temperature during experiment was 45% RH and 25 ◦ C, respectively. The NGI experiments were conducted rapidly (<5 min) to minimise any effects of RH change. Samples from the 11 parts (capsule + rotahaler, throat + mouthpiece, preseparator, stages 1–7 and micro-orifice collector, MOC) were collected in separate volumetric flasks using 40% (v/v) methanol/water to rinse. Each formulation was tested in triplicate. The amount of SX in each part was determined using a validated HPLC assay. The aerodynamic cut-off diameters at a flow rate of 60 l/min for preseparator, stage 1, stage 2, stage 3, stage 4, stage 5, stage 6 and stage 7 were 12.41 m, 8.06 m, 4.46 m, 2.82 m, 1.66 m, 0.94 m, 0.55 m and 0.34 m, respectively (Marple et al., 2003a,b). The amount of drug recovered from all stages of NGI, preseparator, throat, device and capsule was defined as the recovered dose (RD). The emitted dose (ED) was the percent of recovered dose that was collected from all sections except device and capsule. The amount of SX in each stage was then calculated as a percent of RD. The % fraction of RD that was collected in stage 3 to MOC was regarded as FPF. 2.2.5. In vitro drug dispersibility by twin stage impinger (TSI) In vitro aerosol deposition was determined using a twin stage impinger (TSI, Apparatus, A; British Pharmacopoeia, 2000) (Copley Scientific Ltd., Nottingham, UK) and a Rotahaler (Glaxo SmithKline, Melbourne, Australia). A vacuum pump (Model OD 5/2, Dynavac Engineering, Melbourne, Australia) fitted to the mouthpiece of TSI was used to create the airflow and the flow rate was adjusted to 60 l/min before each measurement. A thermohygrometer (Shinyei TRH-CZ, Osaka, Japan) was used to measure the surrounding temperature (25 ± 2 ◦ C) and RH (45 ± 2% RH). 7.0 and 30.0 ml of 40% (v/v) methanol (HPLC grade) were placed in the stage 1 and 2 of the TSI, respectively. Size 3 hard gelatine capsules (Capsugel, Sydney, Australia), loaded with the powder formulations (20 ± 0.5 mg)
444
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
or drug (0.50 ± 0.03 mg), were inserted into a Rotahaler. The Rotahaler was placed into a moulded mouthpiece attached to the TSI. The Rotahaler was twisted and the powder was then dispersed by drawing 4 l of air during each measurement (4 s at 60 l/min). After rinsing each part (inhaler, stage 1 and 2) with 40% (v/v) methanol (in water), the rinsed liquid was collected and diluted to an appropriate volume (100 ml). A validated high performance liquid chromatographic assay was used to determine the drug concentration. TSI measurements were conducted immediately after removal of the formulations from desiccators and for each formulation, five replicates were performed. The percentage of total dose that deposited in the inhaler, stage 1 (S1 ) and stage 2 (S2 ), was the recovered dose (RD). The amount of drug that was deposited in stage 1 and 2 was the emitted dose (ED) and it was calculated as the percentage of RD. The FPF was defined as the percentage of recovered dose that was deposited in stage 2. ED =
FPF =
(S1 + S2 ) × 100 RD S2 × 100 RD
2.2.9. In situ particle size distribution measurement using a Spraytec In situ PSD measurements of DPI mixtures were performed using a Spraytec (Malvern Instruments Limited, Worcestershire, UK). For each DPI mixture, the particle size distribution (PSD) was measured with the inhalation cell attachment at a flow rate of 60 l/min. The flow rate was controlled using a Critical Flow Controller Model TPK 2000 & Flow meter model DFM 2000 (Copley Scientific Limited, Nottingham, UK). All measurements were conducted on five replicates at room temperature (25 ◦ C) and 45% RH. The measurements were performed for 4 s with triggering level of 50, noise level of 0 and background level of 100. The particle size of the mixture was analysed using Spraytec version 3.0 (Malvern Instruments Limited, Worcestershire, UK). 2.3. Statistical analysis All data were subjected to One Way Analysis of Variance (ANOVA) (SPSS, Version 15.0, Chicago, IL, USA). Probability (P) values of ≤0.05 when analysed by post hoc multiple comparisons were considered as statistically significant. 3. Results and discussion
2.2.6. High performance liquid chromatography (HPLC) assay SX recovered by twin stage impinger (TSI) and Next Generation Impactor was analysed by HPLC using a C18 column (5 m, 4.6 mm × 150 mm, Apollo, Alltech associates, Deerfield, IL, USA) and a UV detector (Shimadzu Diode Array Detector, SPD M10A VP, Kyoto, Japan) at a wavelength of 252.2 nm. A mixture of methanol and 0.2% (w/v) ammonium acetate solution in water (55:45, pH ∼ 6.9) that was freshly prepared, filtered and degassed using 0.45 m membrane filter (Millipore, County Cork, Ireland) was used as the mobile phase. A volume of 15 l was injected (Shimadzu Autosampler, SIL10AD VP, Kyoto, Japan) at ambient temperature at a flow rate of 1.0 ml/min that was generated by a HPLC pump (Shimadzu LC10 AJ, 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 g/ml and 10 g/ml, and the linear regression was performed using Sigmaplot (Jandel Scientific, Chicago, IL, USA). The precision of each assay was determined by analysing five replicates from each of two standard SX solutions (two concentrations: low 1.0 g/ml and high 4.0 g/ml).
3.1. Particle size distributions of formulation ingredients The particle size distributions of salmeterol xinafoate (SX), micronized lactose (ML) and coarse lactose (CL) were determined using a laser diffraction method with a Malvern Mastersizer 2000. The average of at least three replicates at each pressure was calculated.
2.2.7. Particle size analysis by using a Mastersizer 2000 The particle size of the powders was determined by laser scattering using the dry dispersion technique in a Malvern Mastersizer 2000 (Malvern Instruments Limited, Worcestershire, UK). A Scirocco cell and Scirocco 2000 dry powder feeder were used. Particle measurement was carried out at four different pressures (0.5 bar, 1.0 bar, 2.0 bar and 3.0 bar). The particle size of the powders was analysed using Malvern Mastersizer software (version 5.22). The average PSD was measured from three replicates of each sample. 2.2.8. Scanning electron microscopy (SEM) Powder was sprinkled and glued on metal sample plates. Using an electrical potential of 2.0 kV at 25 mA for 10 min, the powder was gold plated with a sputter coater (BAL-TEC SCD 005, Tokyo, Japan). The surface morphology of the particles was examined at several magnifications under JOEL JSM 6000F scanning electron microscope at 15 kV (JOEL, Tokyo, Japan).
Fig. 1. Particle size distributions (PSDs) of (A) micronized lactose (ML) and salmeterol xinafoate (SX) determined at 1.0 bar pressure and (B) coarse lactose (CL) determined at 0.5 bar and 2.0 bar pressures using Mastersizer 2000 employing dry dispersion technique (n = 3, data are represented as mean ± S.D.).
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
At 0.5 bar pressure, both SX and ML showed multimodal distributions with large variability indicating the existence of agglomerates. However, at 1.0 bar pressure, both ML and SX were fully dispersed and showed a monomodal distribution in the range of 0.1–11.0 m (Fig. 1A). The volume mean diameter (VMD) of SX was 1.9 m and 90% particles were <4.1 m. Almost all particles were <8.7 m. The PSD of ML were similar to SX, having 50 and 90% particles <2.4 m and 5.5 m, respectively. Increasing shear pressures did not change the PSD of SX and ML. A monomodal distribution around 100 m was observed for CL at 0.5 bar (Fig. 1B). The VMD of coarse lactose was 126.7 m. However, when the PSD was determined at 1.0 bar pressure, a tri-modal distribution was observed for CL (Fig. 1B) with modes at <7 m, 7–30 m and around 100 m. The frequency of the two smaller sized modes (<7 m and 7–30 m) was small but increased as the pressure was increased. The appearance of these smaller modes at 1.0 bar (and greater) shear pressure could be due to the dissociation of any fines that were present in CL or comminution of CL at high shear pressure. 3.2. Powder formulations 3.2.1. Morphology of powder formulations Inhalation grade SX particles usually form agglomerates (Fig. 2A). High cohesiveness of these micronized particles reduces
445
flow properties and dispersibility (Ganderton and Kassem, 1992). In order to improve flow properties and bulk, coarse carriers such as lactose monohydrate (Fig. 2B) are usually added to SX formulations. In SX–CL binary mixture, the majority of the SX particles, either individual particles or agglomerates, remained adhered to the coarse lactose surface (Fig. 2C). In SX–CL mixtures containing additional ML, agglomerates were observed both on and off the surface of coarse lactose (Fig. 2D). These agglomerates were likely to consist of SX and/or ML particles and the composition and arrangement of SX and ML could vary between mixtures containing different concentrations of ML due to complex interparticulate interactions occurring among different particles during powder mixing (Adi et al., 2006; Das et al., 2009). The size of the agglomerates also varied (Fig. 2D). 3.2.2. Homogeneity of powder formulations 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 (Crooks and Ho, 1976). For the determination of intra-batch variability, the in vitro TSI depositions of five replicates of a single batch of M20F were compared. The ED and the SX content of the stage 1 and 2 of the TSI were 81.1 ± 0.5%, 76.7 ± 0.6% and 11.3 ± 0.5%, respectively, and showed no significant differences (P > 0.05). For the determination of inter-batch variabil-
Fig. 2. Scanning electron micrographs of (A) salmeterol xinafoate, SX (magnification: 3000×), (B) coarse lactose, CL (magnification: 686×), (C) mixture containing 0% additional micronized lactose, M0F (magnification: 529×) and (D) mixture containing 20% micronized lactose, M20F (magnification: 412×).
446
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
ity, the in vitro deposition of five batches of M20F were compared. The ED and the SX content of the stage 1 and 2 of the TSI were 80.7 ± 0.5%, 74.8 ± 0.6% and 11.4 ± 0.9%, respectively, and showed no significant differences (P > 0.05). No significant intra- or interbatch variability in ED, and SX content of stage 1 and 2 of the TSI was seen for M0F when tested using the same procedure. Homogeneity studies together with intra-batch and inter-batch dispersibility comparisons confirmed that the powder mixing process used in this research produced homogeneous mixtures with reproducible in vitro aerosol deposition. 3.3. In vitro dispersibility using a Next Generation Impactor (NGI) In order to examine the dispersibility changes, five SX formulations: SX alone (D) and SX–CL mixtures containing 0%, 5%, 10% and 20% micronized lactose (M0F, M5F, M10F and M20F, respectively), before and after storage at 75% RH for 3 months, were dispersed in triplicate at an airflow of 60 l/min using a Next Generation Impactor (NGI). Before storage, the emitted dose (ED) of SX alone formulation (D) was 43.5 ± 1.3%, in comparison to 55.5 ± 6.1% for the SX–coarse lactose mixture (M0F) (Fig. 3A). When micronized lactose was added to the formulation, the ED decreased and the lowest ED (35.9 ± 4.1%) was observed with the mixture containing 20% micronized lactose, M20F. After storage, the ED of M20F sharply increased from 35.9 ± 4.1% to 53.5 ± 1.1%. The ED of the mixture containing 10% micronized lactose (M10F) also significantly increased from 43.8 ± 1.8% to 51.7 ± 1.7%. Consolidation of the loose agglomerates after storage at the high RH into a more structurally permanent
form due to enhanced capillary interactions/solid bridging possibly developed into a more flowable formulation, resulting in greater capsule emptying and decreased adhesion of free SX on the capsule wall (Adi et al., 2006). No significant difference (P < 0.05) in ED was observed for the other formulations (D, M0F and M5F). No significant differences (P > 0.05) were observed in the percent of SX that remained in (a) the throat and mouthpiece, (b) the preseparator, (c) stage 1 (8.06–12.41 m), (d) stage 2 (4.46–8.06 m), or (e) stage 3–7 including micro-orifice collector (FPF, <4.46 m) for any of the four formulations (D, M0F, M5F, M10F) before and after storage (data not shown). For M20F after storage, the %SX significantly increased from 16.8 ± 1.6% to 21.0 ± 1.8% in throat and mouthpiece and from 11.4 ± 1.5% to 21.3 ± 1.9% in preseparator. The %SX for M20F also significantly increased from 1.5 ± 0.1% to 2.0 ± 0.1% in stage 1 (P < 0.05) (Fig. 3B). In contrast, the %SX collected in stage 2 (4.46–8.06 m) significantly decreased (P = 0.04) after storage while the %SX decrease in stage 3 to MOC (<4.46 m fraction) was marginally significant (P = 0.047) at the 0.05 level. A decrease in %SX occurred in all of the stages which constituted the FPF (stage 3 to MOC). The volume mean diameter (VMD) of SX was 1.9 m, with 90% particles <4.1 m, and almost all particles <8.7 m (Fig. 1A). In the M20F mixture, SX may exist as individual particles, SX agglomerates and/or associated with lactose as mixed SX–ML agglomerates of interactive units with CL. The changes shown in Fig. 3B might result from changes in particulate interactions due to increased capillary interaction and/or solid bridging due to surface moisture at 75% RH. These changes indicated that a decrease in the concentration of fully dispersed particles of SX or small agglomerates <8.06 m and an increase in the concentration of agglomerates above 8.06 m occurred after storage at 75% RH. Such changes could involve interactions between any of the particulate species but a previous study by Das et al. (2009) indicated that lactose–lactose interactions were likely to be more influential in modifying particulate behaviour. Thus, SX–SX and SX–lactose interactions were less likely to be responsible for the increased agglomeration observed, although they could not be dismissed. If the behaviour shown in Fig. 3B was related to increased lactose–lactose interactions, two mechanisms remain possible: (a) modifications of the lactose particle size resulting from increased interaction between lactose particles in the high RH conditions and producing dispersibility changes in a similar manner those occurring in a previous study (Adi et al., 2006) or (b) increased lactose–lactose interactions within mixed agglomerates of ML and SX leading to structural changes in the mixed agglomerates, trapping SX and decreasing its dispersibility. These mechanisms were tested in the following studies. 3.4. Particle size distribution changes of salmeterol xinafoate, coarse lactose and micronized lactose after storage at 75% RH for 3 months
Fig. 3. Comparison of (A) emitted dose (ED) of SX of five formulations: SX alone and SX–coarse lactose mixtures containing 0%, 5%, 10% and 20% micronized lactose, ML (D, M0F, M5F, M10F and M20F, respectively), (B) the %SX fraction at throat and mouthpiece, preseparator, stage 1 (8.06–12.41 m), stage 2 (4.46–8.06 m) and stage 3–7 including micro-orifice collector (MOC) (<4.46 m) of mixture containing 20% micronized lactose (M20F) before and after storage (at 75% RH for 3 months) determined using a Next Generation Impactor (NGI) at an airflow of 60 l/min [the cut-off diameter for relevant stage was shown in parentheses], (n = 3, data are represented as mean ± S.D.).
In order to determine if any of the components of the mixture interacted to produce a permanent change in particle sizes, after storage at 75% RH for 3 months, the PSD of the stored samples of SX, CL and ML were determined using a laser diffraction method with a Malvern Mastersizer 2000. For each sample (SX, CL or ML), the PSDs obtained before and after storage were compared. The average of at least three replicates at each pressure was determined. At 1.0 bar and greater shear pressures, the PSD of SX before and after storage overlapped within the error margin indicating no change in particle size (distributions not shown). This indicated that any interaction between particles of SX before and after stor-
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
447
Fig. 5. Fine particle fraction (FPF) of (A) mixture containing 20% micronized lactose (M20F) before storage, (B) M20F after storage at 75% for 3 months (C) M20F before storage that was prepared using ML stored at 75% RH for 3 months (n = 5, data are represented as mean ± S.D.).
3.5. Effect of stored micronized lactose on SX dispersibility from M20F
Fig. 4. Comparison of (A) particle size distributions (PSDs) at 1.0 bar pressure and (B) volume mean diameter (VMD) at three different pressures (1.0 bar, 2.0 bar and 3.0 bar) of micronized lactose (ML) before and after storage (at 75% RH for 3 months) determined using a Mastersizer 2000 employing dry dispersion technique (n = 3, data are represented as mean ± S.D.).
age were similar and capillary interactions and/or solid bridging that might have occurred during storage at 75% RH did not lead to any particle size change. Similarly, no change in PSD of CL was observed before and after storage at all shear pressures. The distributions were identical with those shown in Fig. 1 for both SX and CL. However, the PSD of ML increased after storage and the volume mean diameter (VMD) of ML increased from 2.4 ± 0.2% m to 3.1± 0.1% m at 1.0 bar pressure (Fig. 4A). SEM images were not helpful in identifying the causes and evidence of solid bridging could not be seen. The increased particle size was an indication of strong interactions between small ML particles. The increase in particle size was found at all shear pressures indicating that very strong interactions either from increased capillary cohesion or solid bridging were very likely (Fig. 4B). This finding is also supported by Bridson et al. who found an increase in the 5th percentile of the PSD of lactose after 1–2 weeks storage at 70% RH (Bridson et al., 2007). Previous studies showed that the dispersibility of SX is dependent on the particle size of added micronized lactose (Adi et al., 2006). Higher FPF for SX mixtures containing 10–20% added micronized lactose was observed when the VMD of the added micronized lactose was increased from 3.0 m to 7.9 m. The change was attributed to a change in the agglomerate structure with a resultant decrease in agglomerate strength. It was uncertain whether the change in particle size observed in this current study (i.e., from 2.4 ± 0.2% m to 3.1± 0.1% m) was sufficient to change the dispersibility of the mixture. Therefore, the effect of using ML (that had been stored at high RH) on the FPF of freshly prepared M20F was tested.
This mixture was prepared using ML stored at 75% RH for 3 months but using CL and SX which were not stored at high RH. The FPF was determined using a twin stage impinger (TSI). The FPF of M20F before storage was 11.3 ± 1.2%. After storage of the mixture at 75% RH for 3 months, the FPF decreased to 8.2 ± 0.6% (Fig. 5). The FPF of the freshly prepared M20F using ML stored at 75% RH increased significantly to 19.4 ± 2.3% (Fig. 5). The increased FPF of freshly prepared M20F using stored ML was consistent with that observed earlier by Adi et al. (2006); however, the extent of increase was surprising given the relatively small change in particle size (i.e., from 2.4 ± 0.2% m before storage to 3.1 ± 0.1% m after storage). Any discussion of the reasons for the marked increase can only be speculative. Given that the extent of dispersibility would be related to the structure of the SX–ML agglomerates (Adi et al., 2006), the shift of the ML PSD to a distribution with a higher mean particle size and a smaller span (decreased from 1.67 ± 0.01 to 1.45 ± 0.01 after storage) might have influenced the agglomerate structure and strength. This part of the study was not pursued further. However, the important outcome was that the increase in particle size of ML in the formulation alone increased SX dispersibility and was not consistent with the decreased SX dispersibility observed in the M20F formulation during storage. Thus, the mechanism of reduced SX dispersibility was not related to increased lactose–lactose interactions producing larger PSD of ML. 3.6. Particle size distribution changes of mixtures containing micronized lactose after storage at 75% RH for 3 months The PSD of the mixtures containing micronized lactose (M5F, M10F and M20F) before and after storage at 75% RH for 3 months were determined using a laser diffraction method with a Malvern Mastersizer 2000 using a dry dispersion technique. The average of at least three replicates conducted at 1.0 bar shear pressure was determined. Each SX–CL mixture containing additional ML (M5F, M10F and M20F) showed tri-modal distributions. The PSD of M5F and M10F are shown in Fig. 6A. The chemical composition of the particles in each mode was not able to be identified. However, while PSD and SEM gave some guidance, the true compositions of the modes could only be speculated. The smallest mode, in the range of <7.0 m, could be predominantly dispersed SX particles, dis-
448
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
and subsequent particle size increase seen when micronized lactose was stored alone. 3.7. Particle size distribution of the aerosol plume before and after storage of the formulations In order to understand the deagglomeration pattern that would be encountered during inhalation, particle size measurements of the aerosol clouds of all five formulations (D, M0F, M5F, M10F and M20F) before and after storage at 75% RH for 3 months were conducted using an in situ particle size analyser (Spraytec) with the Rotahaler as the dispersion device. Each powder was dispersed five times at a flow rate of 60 l/min and the averaged PSD before and after storage was determined. All mixtures produced a bi-modal distribution of powders in the aerosol plume (Fig. 7A). The frequency of the first mode increased with increasing concentration of ML, which was consistent with the PSD observed using Mastersizer 2000. The mode at the lower sizes was broad (3–30 m) and could represent SX and ML individual particles, small agglomerates of SX and ML and SX–ML mixed agglomerates while the second mode represented CL carrier, SX–CL and SX–ML–CL interactive units. SX alone showed a broader distribution ranging from 1.0 m to 1000 m indicating the existence of a wide distribution of individual particles and agglomerates (data not shown). No change was observed in the PSD of SX alone and SX mixtures after storage containing 0% and 5% micronized lactose. The mixture containing 10% ML showed some evidence of the reduced rate of deagglomeration of smaller agglomerates after storage. A clear difference was observed for the mixture containing 20% ML (Fig. 7B). Though there was a large standard deviation associated
Fig. 6. (A) Particle size distributions (PSDs) of mixtures containing 5% and 10% micronized lactose (M5F and M10F, respectively) (B) comparison of PSD of the mixture containing 20% micronized lactose (M20F) before and after storage (at 75% RH for 3 months) determined using a Mastersizer 2000 at 1.0 bar pressure and (C) PSD of M20F (shown in 6B) in the range of <40 m (n = 3, data are represented as mean ± S.D.).
persed ML particles and small agglomerates of SX, ML or SX–ML. The second mode, in the range of 10–30 m, was likely to consist of agglomerates (SX, ML or SX–ML) and the third mode at 95–105 m could be CL, ML–CL, or SX–ML–CL interactive units. The second mode (in the range of 10–30 m) increased with increasing concentration of ML in mixtures (Fig. 6A). The PSD for M5F and M10F before and after storage overlapped within the error margin (data not shown). On the other hand, the PSD of M20F before and after storage differed. After storage, the frequency of particles <7.0 m was lower and the frequency in the size range of 10–30 m was higher than the mixture prior to storage (Fig. 6B and C). The reduced deagglomeration of 10–30 m size agglomerates after storage was assumed to be reason for the decrease in <7.0 m fraction (Fig. 6C). In the mixture containing 20% ML, SX and ML particles formed agglomerates with a greater propensity of ML–ML contact. The formation of stronger agglomerates, due to strong cohesion and/or solid bridging at ML–ML contact during storage at 75% RH (Das et al., 2009), might have caused the SX particles to be mechanically trapped inside the agglomerates, which ultimately resulted in decreased dispersibility after storage. The formation of stronger SX–ML agglomerates during storage at 75% RH was consistent with the increasing interaction between micronized lactose particles
Fig. 7. Comparison of particle size distributions (PSDs) of (A) mixtures containing 0%, 5% and 10% micronized lactose (M0F, M5F and M10F, respectively) before storage, (B) mixture containing 20% micronized lactose (M20F) before and after storage (at 75% RH for 3 months) determined using a Spraytec at an airflow of 60 l/min (n = 5, data are represented as mean ± S.D.).
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
with the PSD of the stored formulations, the frequency of particles in the VMD range of 4–25 m increased after storage and the frequency <4 m VMD slightly decreased. For example, the percentage of particles <3.98 m size for M20F decreased from 7.1 ± 1.2% to 5.4 ± 1.8%. After storage the dispersibility of individual particles of SX in the aerosol plume decreased. Given that the major mechanism for SX dispersibility is deagglomeration of SX–ML mixed agglomerates (Adi et al., 2006), the influence of the high humidity conditions on these agglomerates is evident with a clearly decreased capacity of the agglomerates to release SX. This finding is in agreement with those observed in the NGI and Mastersizer 2000 study and supports the presence of agglomerates that were less likely to disperse and release SX. 4. Conclusions This study has identified the mechanism by which mixtures of SX containing high concentrations of micronized lactose (i.e. 20%) decrease in vitro deposition after storage at 75% RH. The %SX of an SX mixture containing 20% micronized lactose significantly increased in the throat and mouthpiece, preseparator and stage 1 of the NGI (P < 0.05) after storage. In contrast, the %SX for the same formulation significantly decreased (P < 0.05) in remaining stages (stage 2 to micro-orifice collector, MOC) of NGI after storage. The ED of this formulation also increased after storage indicating greater capsule emptying through better flow of the formulation or less SX adhesion to the capsule wall. While the particle size of micronized lactose increased when stored at 75% RH for 3 months, the FPF of freshly prepared SX mixture containing 20% of the stored micronized lactose increased, indicating that the particle size changes to the micronized lactose alone were not responsible for the decreased in vitro performance of this mixture after storage. The extent of increase caused by a small change in the PSD (i.e. from 2.4 m to 3.1 m and a decreased span) was surprising and may have been related to changes in agglomerate structure due to particle size changes (Adi et al., 2006). This was not pursued further as it was not core to understanding the mechanism. Particle size data of both the SX mixtures at various shear pressures and of the aerosol plume after dispersion from a model device showed an increased presence of agglomerates in the range 4–25 m and decreased dispersed particles <4 m. SX and the coarse lactose carrier showed no change in PSD after storage and thus any increased interaction between these mixture components was not the cause of the increased agglomeration and decreased dispersibility. In addition, significant changes in PSD only occurred in mixtures containing higher concentrations of micronized lactose (clearly seen when 20% was used and some indication when 10% was used). Thus, agglomerate formation was related to the presence of micronized lactose. The mechanisms of dispersibility of SX mixtures containing micronized lactose (Islam et al., 2004; Adi et al., 2006) have been identified. In these mixtures where there were relatively high concentrations of cohesive powders, the major mechanism of dispersibility was identified as the deagglomeration of mixed agglomerates of SX and micronized lactose. The mechanism of reduced dispersibility during storage, therefore, must be related to increased cohesion within the SX–micronized lactose agglomerates due to increased capillary interactions or solid bridging (Das et al., 2009) producing stronger agglomerates that did not disperse. This outcome is consistent with the particle sizing results of this project. This study is important as it addressed a practical industrial problem encountered by the dry powder inhaler formulation industry. The study has revealed the mechanism of dispersibility change during storage and thus this body of knowledge provides guidance and direction to mixture design and formulation strategies.
449
A laboratory scale mixing process was employed using small batches (5 g) like many of the studies in the literature. The batch size and type of mixer will influence the magnitude of shear generated during mixing and thus will influence particle contacts and agglomerate structure. The outcomes of this research will be related to the mixing process; however, these principles will apply to other mixing conditions where agglomerates are produced. Careful interpretation of the data is therefore required to apply these outcomes in industry applications. Acknowledgements Shyamal Das would like to acknowledge the scholarship support through the Faculty of Pharmacy, Monash University. Shyamal Das also acknowledges Dr Handoko Adi and Dr Herbert Chiou for their support during the use of NGI and Mastersizer 2000, and Dr Joseph A Nicolazzo for his help in statistical analysis. References 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. Alway, B., Sangchantra, R., Stewart, P., 1996. Modelling the dissolution of diazepam in lactose interactive mixtures. Int. J. Pharm. 130, 213–224. Boerefijin, R., Ning, Z., Ghadiri, M., 1998. Disintegration of weak lactose agglomerates for inhalation applications. Int. J. Pharm. 172, 199–209. 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. Braun, M., Oschmann, R., Schmidt, P., 1996. Influence of excipients and storage humidity on the deposition of disodium cromoglycate (DSCG) in the twin impinger. Int. J. Pharm. 135, 53–62. Bridson, R., Robbins, P., Chen, Y., Westerman, D., Gillham, C., Roche, T., Seville, J., 2007. The effects of high shear blending on lactose monohydrate. Int. J. Pharm. 339, 84–90. Callahan, J., Cleary, G., Elefant, M., Kaplan, G., Kensler, T., Nash, R., 1982. Equilibrium moisture content of pharmaceutical excipients. Drug Dev. Ind. Pharm. 8, 355–369. Crooks, M., Ho, R., 1976. Ordered mixing in direct compression of tablets. Powder Technol. 14, 161–167. Das, S., Larson, I., Young, P., Stewart, P., 2009. Influence of storage relative humidity on the dispersion of salmeterol xinafoate powders for inhalation. J. Pharm. Sci. 98, 1015–1027. Ganderton, D., Kassem, N., 1992. Dry powder inhalers. In: Ganderton, D. (Ed.), Advances in Pharmaceutical Sciences, vol. 6. Academic Press, London, UK, pp. 165–191. Hindle, M., Makinen, G., 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. Hogg, R., 1989. Role of colloid and interface science in agglomeration. In: Cross, M., Oliver, R. (Eds.), Proceedings of the 5th International Symposium on Agglomeration. The Institute of Chemical Engineers. Rugby, Warwickshire. Islam, N., Stewart, P., Larson, I., Hartley, P., 2004. Lactose modification by decantation: are drug–fine lactose ratios the key to better dispersion of salmeterol xinafoate from lactose-interactive mixtures? Pharm. Res. 21, 492– 499. 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. Liu, J., Stewart, P., 1998. De-aggregation during the dissolution of benzodiazepines in interactive mixtures. J. Pharm. Sci. 87, 1632–1638. Louey, M., Stewart, P., 2002. Particle interactions involved in aerosol dispersion of ternary interactive mixtures. Pharm. Res. 19, 1524–1531. Lucas, P., Anderson, K., Staniforth, J., 1998. Protein deposition from dry powder inhalers: fine particle multiplets as performance modifiers. Pharm. Res. 15, 562– 569. Marple, V., Olson, B., Mitchell, J., Murray, S., Hudson-Curtis, B., 2003a. Next generation pharmaceutical impactor (a new impactor for pharmaceutical inhaler testing). Part II. Archival calibration. J. Aerosol Med. 16, 301–324. Marple, V., Roberts, D., Romay, F., Miller, N., Truman, K., Van, O., Olsson, B., Holroyd, M., Mitchell, J., Hochrainer, D., 2003b. Next generation pharmaceutical impactor (a new impactor for pharmaceutical inhaler testing). Part I. Design. J. Aerosol Med. 16, 283–299. Padmadisastra, Y., Kennedy, R., Stewart, P., 1994a. Influence of carrier moisture adsorption capacity on the degree of adhesion of interactive mixtures. Int. J. Pharm. 104, R1–R4. Padmadisastra, Y., Kennedy, R., Stewart, P., 1994b. Solid bridge formation in sulphonamide-Emdex interactive systems. Int. J. Pharm. 112, 55–63.
450
S. Das et al. / European Journal of Pharmaceutical Sciences 37 (2009) 442–450
Podczeck, F., Newton, J., James, M., 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. Podczeck, F., Newton, J., James, M., 1997a. Influence of relative humidity of storage air on the adhesion and autoadhesion of micronized particles to particulate and compacted powder surfaces. J. Colloid Int. Sci. 187, 484–491. Podczeck, F., Newton, J., James, M., 1997b. 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. Staniforth, J., 1996. Improvement in dry powder inhaler performance. In: Proc. Drug Del. Lungs Conf. VII, pp. 86–89. Young, P., Price, R., 2004. The influence of humidity on the aerosolization of micronized and SEDS produced salbutamol sulphate. Eur. J. Pharm. Sci. 22, 235–240. Young, P., Sung, A., Traini, D., Kwok, P., Chiou, H., Chan, H., 2007. Influence of humidity on the electrostatic charge and aerosol performance of dry powder inhaler carrier based systems. Pharm. Res. 24, 963–970.
Young, P., Price, R., Tobyn, M., Buttrum, M., Dey, F., 2003a. Investigation into the effect of humidity on drug–drug interactions using the atomic force microscope. J. Pharm. Sci. 92, 815–822. Young, P., Price, R., Tobyn, M., Buttrum, M., Dey, F., 2003b. Effect of humidity on aerosolization of micronized drugs. Drug Dev. Ind. Pharm. 29, 959–966. Young, P., Price, R., Tobyn, M., Buttrum, M., Dey, F., 2004. The influence of relative humidity on the cohesion properties of micronized drugs used in inhalation therapy. J. Pharm. Sci. 93, 53–761. Zeng, X., Tee, S., Martin, G., Marriott, C., 1996. Effect of mixing procedure and particle size distribution of carrier particles on the deposition of salbutamol sulphate from dry powder inhaler formulations. In: Proc. Drug Del. Lungs Conf. VII, pp. 40–43. Zeng, X., MacRitchie, H., Marriott, C., Martin, G., 2007. Humidity-induced changes of the aerodynamic properties of dry powder aerosol formulations containing different carriers. Int. J. Pharm. 333, 45–55.