Meeting Report: Summary of Abbreviated Impactor Measurement Workshop Organized By the European Pharmaceutical Aerosol Group

Abstract

This year, the Drugs Delivery to the Lungs 21 conference broke new ground with the first half-day devoted to a workshop focusing on experimental aspects of the abbreviated impactor measurement concept. The workshop had the following objectives: to define what further needs to be done experimentally to establish abbreviated impactor measurement; and to identify the pathway towards adoption of existing methods into the pharmacopeias, as the next step towards what is hoped will eventually be acceptance by the key regulatory agencies in Europe, Canada and the USA.

Seven speakers presented different aspects associated with the implementation of abbreviated impactor measurement (AIM)-based measurement apparatuses in their laboratories as new techniques for potential application in early stage product screening, as well as later in the product quality control role.

A report of the panel discussion that followed will be presented as a poster at Respiratory Drug Delivery – Europe 2011, as part of the focus on AIM combined with effective data analysis that will be featured at this meeting, and also in a satellite symposium on AIM – effective data analysis, being prepared by the International Pharmaceutical Consortium on Regulation and Science in conjunction with the Respiratory Drug Delivery Organization.

Workshop highlights

Mitchell and Copley provided an overview of the various studies undertaken at Trudell Medical International since 2007 to develop and validate abbreviated systems based on the Andersen eight-stage Cascade Impactor (ACI) [1], primarily for use in characterizing aerodynamic size-relevant metrics of aerosols from pressurized metered-dose inhalers (pMDIs). The AIM concept was developed out of the need for the more rapid assessment of aerodynamic, size-pertinent properties of aerosols from oral inhaled products. Initial work was based on the Copley Fast Screening Andersen Impactor, revealing that additional care in coating the collection surfaces with an adhesive substance was needed to prevent bias caused by particle bounce and re-entrainment, which are exacerbated by removing intermediate stages from the ACI [2]. In a follow-on study [3], a similar in-house abbreviated impactor (T-Fast Screening Andersen) was slightly modified by incorporating stage zero of the ACI without a collection plate above the first impaction stage in order to provide comparable conditions to the ACI when sampling aerosols containing a low-volatile excipient, in this case ethanol. In 2009, an experiment designed to quantify the precision and accuracy of two different abbreviated ACI systems was undertaken on behalf of the International Pharmaceutical Consortium on Regulation and Science. This successful study confirmed that an abbreviated system, optimized for product quality control by measuring just two subfractions with the boundary set at 2.1 µm (AIM-QC system), had comparable precision to the ACI [4]. An alternative abbreviated system (AIM-pHRT) that measured three subfractions – coarse particles >4.7 µm, fine particles <4.7 µm and extra-fine particles <1.1 µm – was also comparable with the ACI for fine and coarse fractions. However, a glass fiber filter saturated with Brij-35 surfactant was needed to eliminate particle bounce on the second impaction stage where the extrafine fraction is determined [5].

Russell-Graham and colleagues presented an assessment of the two-stage fast-screening impactor (FSI; MSP Corp., MN, USA) with the cut-point set to 5 µm [6]. The FSI was developed from the preseparator of the next-generation pharmaceutical impactor (NGI). In their previous work [7], they had shown fair agreement comparing FSI-measured fine-particle mass for a commercially available dry-powder inhaler (DPI) tested at 70 l/min with that from the NGI after coating the base of the FSI as well as the collection stage with silicone fluid to minimize particle bounce and re-entrainment. The present study extended their original work to include four different DPI products, and in all instances the FSI appeared to slightly, but significantly, overestimate the fine-particle fraction (FPF_<5µm). In contrast, mass recovery from both full-resolution and abbreviated impactors was excellent, indicating minimal internal losses with either system. The difference between FSI and NGI data was most marked for the DPI having the highest flow-rate dependence on emitted fine-particle mass, where the estimate of the FPF_FSI<5µm was 30% of label claim, compared with 16% of label claim for FPF_NGI<5µm. The cause of the discrepancy was attributed to the smaller internal volume of the FSI that may have created different pressure drop-time characteristics during the initial stages of sampling when the powder is dispersed from the inhaler.

Tservistas et al. also used an FSI, but in their case they applied their abbreviated system to the investigation of aerosols comprising aqueous droplets of salbutamol emitted from an e-FLOW* vibrating membrane nebulizer (PARI GmbH, Starnberg, Germany) [8]. They intentionally plugged three of the six nozzles of the impaction stage in the FSI in alternating order, so that the system could be operated at 15 l/min, which is a recommended flow rate for sampling from nebulizing systems [9]. They also compared the FSI operated at ambient room temperature (22°C) and cooled to 18°C to reduce the risk of heat transfer-related evaporation [10]. They found excellent agreement between the FSI and NGI for measures of FPF_<5µm, whether or not the FSI was cooled, and also with the same metric determined by a laser diffractometer (Mastersizer-X, Malvern Instruments, UK). Although not as rapid as laser diffractometry (2.5 h to process six replicate tests), the FSI (12.5 h – operated at ambient room temperature; 14.5 h – cooled), was approximately twice as rapid as the NGI (28 h), mainly due to the reduced number of samples that needed to be manipulated during the course of recovery/assay of the API.

Svensson and Berg described abbreviating the NGI by modifying the full-resolution impactor [11]. They explored two configurations, assessing their capability for the measurement of a variety of DPI- and pMDI-generated aerosols, testing the DPIs at flow rates from 40 to 77 l/min and the pMDIs at 30 l/min:

▪	In the ‘external filter‘ concept, one small NGI cup was modified and equipped with an outlet tube onto which a filter was mounted. After passing the filter, where the aerosol particles were collected, the air flow was directed back into the NGI air flow passage via a similar outlet tube on the next cup in a downstreamposition;
▪	In the ‘internal filter‘ concept, a glass-fiber filter with a diameter suitable to fit the inner diameter of the nozzles, was placed in the nozzle dwell covering all jet inlets. The top lid of the NGI had to be separated from the seal body to achieve this configuration. The filter was manipulated with a pair of tweezers.

Both methods were applied from stages 3–5 individually of the full-resolution NGI, so that the cut-off size between fine and coarse fractions could be varied from 1.5 µm (stage 5) to 6.4 µm (stage 3). Regardless of inhaler type, either internal- or external-filter methods provided very good agreement with the NGI for the equivalent measures of fine particle mass, but the internal-filter approach was twice as rapid to execute as the external-filter method.

Després-Gnis and Williams used the FSI to investigate a DPI currently in development, making measurements at 35 l/min with one and three actuations/determination to test the sensitivity of the abbreviated method compared with the full-resolution NGI where the API (fluticasone propionate) has to be recovered from many more samples [12]. They found that the FSI and NGI provided broadly equivalent measures of fine-particle and total-emitted mass per actuation, noting that precoating the collection cup of the FSI with an adhesive agent (glycerol) did not have an appreciable effect on fine-particle mass determined with only a single actuation (where the influence of bias from bounce and re-entrainment might have been expected to be most severe). Importantly, they also found comparable measures of total emitted mass/actuation determined by dose uniformity sampling apparatus and the FSI when only one dose was delivered, suggesting the possibility of a combined apparatus for dose uniformity and fine-particle mass, based on the FSI.

Sheng et al. also evaluated the FSI, but this time for the fast screening of formulations during early stage development for potential pMDI- and nebulizer-based delivery of medication [13]. Their pMDI-focused testing evaluated excipient screening, manufacturing-process screening and the effects of flow rate on FSI performance evaluated as FPF. They obtained comparable measures for FPF_<5µm with the NGI in their excipient screening study (sampling at 30 l/min). However, they found FPF_FSI > FPF_ACI in the manufacturing process screening study, where they compared their FSI (FPF_<4.7µm) with an ACI at 28.3 l/min. They attributed the slight bias to the steeper collection efficiency curve of the impaction stage in the FSI compared with that of stage 2 in the ACI. Finally, in their flow rate study comparing the FSI with both NGI and ACI at 30, 45 and 60 l/min, they found that FPF_FSI ≍ FPF_NGI at the lowest flow rate. However, FPF_FSI was larger than the corresponding values from either their NGI or ACI at the two higher flow rates, with smaller divergence between FPF_FSI and FPF_NGI. Further comparisons are needed to confirm if differences in stage size-selectivity is indeed the correct cause of the differences that were observed. They also observed higher FPF_FSI in their DPI-based studies in which measurements were made at each of the same flow rates with the divergences decreasing at the higher flow rates, such that FPF_FSI ≍ FPF_NGI at 45 and 60 l/min. The FSI is better compared with the NGI as the reference full resolution impactor, based on the fact that stage size-selectivity has been optimized for both systems in comparison with the ACI.

Rogueda et al. also provided insight into their comparison of a FSI with NGI for DPIs using carrier-lactose delivery technology in the development phase and also preparations for nebulization [14]. In contrast with previous comparisons, rather than test for statistically significant differences, they proposed a fixed acceptance criterion for in vitro equivalence of FPF_FSI<5µm = FPF_NGI<5µm ± 10%. Unsurprisingly, they established that the FSI required coating the base (developed from the NGI preseparator bottom) with surfactant to mitigate bias from particle bounce for their DPI work at 90 l/min. When this precaution was taken, agreement between the two techniques according to their criterion was generally achieved with mean FPF_FSI marginally greater than mean FPF_NGI, except for one early phase lactose–API blend, having a 12% difference in FPF, which just failed to meet the criterion. Interestingly, in their nebulizer-related work, where the flow rate was reduced to 15 l/min, FPF_FSI<5µm was slightly less than FPF_NGI<5µm, and both products evaluated met their acceptance criterion, whether operated at ambient room temperature (23 ± 2°C) or chilled to 5 ± 2°C to prevent possible bias from heat-transfer related evaporation [10].

The workshop consensus was that AIM-based measurements are both practical and afford significant time savings per measurement compared with full-resolution cascade impactor-derived measurements. Close agreement between FPF measured by the chosen abbreviated systems was readily achievable for measurements involving nebulizers and pMDIs. However, for DPI-based comparisons, there was evidence that the FPF_FSI was detectably higher than the FPF_NGI. Further investigation is required, but the cause may originate with differences in internal dead volume between the full-resolution and abbreviated systems. These differences may become important in aerosolizing the powder at the earliest phase of the measurement process when the flow rate is rapidly increasing from zero at initiation of the test towards its final stable value. Comparisons between the ACI and FSI were less reliable than those between NGI and FSI, possibly due to differences in stage size-selectivity between these systems.

Financial & competing interests disclosure

Steve C Nichols is an independent specialist consultant in OINDPs. Jolyon Mitchell is a full-time employee of Trudell Medical International. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.