Advancement of a Positive-Pressure Dry Powder Inhaler for Children: Use of a Vertical Aerosolization Chamber and Three-Dimensional Rod Array Interface

Abstract

Purpose:

Available dry powder inhalers (DPIs) have very poor lung delivery efficiencies in children. The objective of this study was to advance and experimentally test a positive-pressure air-jet DPI for children based on the use of a vertical aerosolization chamber and new patient interfaces that contain a three-dimensional (3D) rod array structure.

Methods:

Aerosolization performance of different air-jet DPI designs was first evaluated based on a 10 mg powder fill mass of a spray-dried excipient enhanced growth (EEG) formulation. Devices were actuated with positive pressure using flow rate (10-20 L/min) and inhaled volume (750 ml) conditions consistent with a 5-year-old child. Devices with best performance were connected to different mouthpiece designs to determine the effect on aerosolization and tested for aerosol penetration through a realistic pediatric in vitro mouth-throat model.

Results:

Use of the new vertical aerosolization chamber resulted in high quality aerosol formation. Inclusion of a 3D rod array structure in the mouthpiece further reduced aerosol size by approximately 20% compared to conditions without a rod array, and effectively dissipated the turbulent jet leaving the device. Best case device and mouthpiece combinations produced <2% mouth-throat depositional loss and >70% lung delivery efficiency based on loaded dose.

Conclusions:

In conclusion, use of a 3D rod array in the MP of a positive-pressure air-jet DPI was found to reduce aerosol size by 20%, not significantly increase MP depositional loss, reduce mouth-throat deposition by 6.4-fold and enable lung delivery efficiency as high as 73.4% of loaded dose based on pediatric test conditions.

INTRODUCTION

Dry powder aerosols offer a number of advantages compared with other forms of respiratory drug delivery, including rapid administration of high dose medications, coordination of dose delivery with inhalation, and stable formulations (1–4). Despite these advantages, dry powder aerosols are currently underutilized in children due to a number of real and perceived challenges. As with adults, the lung delivery efficiency of dry powder aerosols in pediatric subjects is known to be low, due to high extrathoracic depositional loss and often poor device emptying (5–8). Across a number of current dry powder inhalers (DPIs), pediatric lung delivery efficiencies are in a range of 5-30% of the nominal dose, based on in vivo (6) and realistic in vitro experiments (5, 7, 8). Potentially more significant than the low dose delivery efficiency is the high intersubject variability associated with pediatric DPI use. For example, Lindert et al. (7) reported a 2-fold difference in lung dose from a DPI based on inhalation waveform conditions even without varying the airway anatomy. Devadason et al. (6) reported coefficients of variation as high as 75% for pediatric DPI use by 3 to 5-year-olds. Children under 6 years old often cannot generate sufficient inhalation flow to operate passive DPIs, even for devices with low to medium resistances (3, 9, 10). A recent study by Lexmond et al. (11) in children ~5-12 years old with an oral DPI simulator reported that 90% of inhalations were associated with oral obstruction, arising from the tongue and cheeks during negative pressure inhalation against the variable resistance device. Lexmond et al. (11) also observed that for 5-year-old subjects, only half of the trained children could successfully perform a passive DPI inhalation maneuver. Furthermore, between 10 and 44% of the trained pediatric subjects exhaled into the DPI simulator, which is expected to adversely affect aerosolization of the powder due to moisture exposure.

Positive-pressure or inline DPIs implement a gas source external to the user to generate the aerosol and potentially deliver a full inhalation (12–19). Gas sources may include an air-filled syringe, ventilation bag or compressed gas reservoir. Depending on the volume of gas used, these DPIs can be classified as operating with high (≥ 200 mL) or low (< 200 mL) actuation air-volume (AAV).

Based on the use of an external gas source that operates independent of the subject, positive-pressure DPIs have a number of potential advantages for the administration of powder aerosols to children. First, positive pressure devices are actuated with a predefined correct and reproducible air volume and flow rate. This approach enables correct aerosolization of the powder, which can increase device emptying, lung delivery efficiency and reduce intersubject variability. Use of a predefined air volume may also enable effective use of DPIs in children younger than 6 years old. Secondly, positive-pressure devices operated with high AAVs may ensure that the aerosol is delivered with a full inhalation breath. This approach may help to better open the lung airways and improve lung distribution of the aerosol, which may be especially important when disease is present. Thirdly, operation with positive pressure may prevent collapse or occlusion of the mouth-throat region associated with negative pressures, as observed by Lexmond et al. (11). This expansion of the extrathoracic airways may improve rather than hamper lung aerosol delivery in human subjects compared with in vitro models. Finally, the positive pressure device forms a sealed airflow passage, which prevents exhalation through the device and may help to guide a breath-hold, when desired.

Farkas et al. (12) recently reported the development of a positive-pressure pediatric DPI that implements an air-jet approach for aerosolizing a powder formulation. As described in previous studies (13–15, 20), the air-jet DPI implements a small diameter inlet airflow passage, aerosolization chamber, and small diameter outlet aerosol flow passage. Positive-pressure gas passes through the inlet airflow passage and forms a high speed turbulent jet within the aerosolization chamber (21). Secondary flow velocities formed by the high speed jet are used to initially fluidize the powder. As the fluidized powder enters the high speed jet region, additional powder deaggregation occurs (21). The small diameter outlet orifice serves to both help form the secondary velocities and allow passage of sufficiently deaggregated particles out of the aerosolization chamber. Using this approach, AAVs of 10 ml and lower have been shown to effectively aerosolize 10 mg powder masses (13–15, 20). For pediatric drug delivery, Farkas et al. (12) developed a positive-pressure pediatric air-jet DPI that was operated with a ventilation bag or compressed gas supply of 750 ml of air in order to aerosolize the powder and provide a full inhaled breath for a 5-year-old child. Using a highly dispersible spray-dried formulation (22), the best case pediatric air-jet DPI produced an aerosol mean mass median aerodynamic diameter (MMAD) <1.75 μm and a fine particle fraction (<5μm) ≥90% based on emitted dose. Actuation with the ventilation bag enabled lung delivery efficiency through the nasal and oral interfaces to a tracheal filter of 60% or greater, based on loaded dose. In both oral and nose-to-lung administrations, extrathoracic depositional losses were <10% (12).

While the pediatric air-jet DPI developed by Farkas et al. (12) performed well, it is expected that further design improvements are possible. One potential limitation of the previous device is the use of a horizontal aerosolization chamber with high AAV. It was determined experimentally that device performance was improved when the inlet jet did not impinge on the initial powder bed (13) (thereby allowing secondary velocities to initially fluidize the powder (23)). Computational fluid dynamics (CFD) studies (21, 23, 24) have indicated that excessive turbulence in the region of initial powder formation increases the size of the aerosol provided by the air-jet design. One feature controlling the level of turbulence on the initial powder bed is the distance between the jet and powder. Orienting the aerosolization chamber vertically enables more control of this distance. Vertical orientation may also enable loading larger powder masses while maintaining aerosolization performance.

A second limitation of the previous pediatric air-jet DPI was the generation of a high speed air-jet that forms in the outlet aerosol flow passage and then exits the device. It is well known that turbulent jets from DPIs and other inhalers can significantly increase downstream depositional losses in patient interfaces and extrathoracic airways (25–31). As a result, in the study of Farkas et al. (12), mouthpiece (MP) depositional loss was ~5% and mouth-throat (MT) depositional loss was ~7% of loaded dose, which are higher values than expected for the aerosol size produced. To further improve performance of the pediatric air-jet DPI, a method to disperse the high speed jet exiting the inhaler without incurring additional particle loss is needed.

Computational fluid dynamics (CFD) simulations of the air-jet DPI have recently explored different patient interface designs to better dissipate the turbulent jet exiting the device. Designs tested include rapid and stepped expansions, streamlined bodies and three-dimensional (3D) rod arrays (32). Of these designs, placing a 3D rod array across the air-jet outlet flow passage was found to be most effective. As previously described, the 3D rod array structure increases turbulent energy in the small eddies, thereby improving powder dispersion (33). Bass and Longest (32) demonstrated that this structure could also be used to diffuse a turbulent jet from a DPI, with minimal depositional loss. While interesting, the CFD-predicted results of coupling an air-jet DPI with a 3D rod array have not been experientially tested.

The objective of this study is to advance and experimentally test a positive-pressure air-jet DPI for children based on the use of a vertical capsule chamber and 3D rod array patient interface. The DPI is operated for conditions consistent with aerosol delivery to a 5-year-old child, which include an approximately 750 ml AAV delivered at a positive-pressure of 6 kPa, as generated from a compressed air source or ventilation bag. Conditions for a 5-year-old child were selected based on poor DPI usage and performance at this age with passive devices in the study of Lexmond et al. (11). For the vertical aerosolization chamber design, different inlet and outlet flow passage diameters are considered based on the CFD study of Bass et al. (24) with a 10 mg powder fill mass of excipient enhanced growth formulation. A colliding inlet jet design is also considered. Patient interface MP designs include a rapid expansion and three additional designs containing 3D rod arrays in a 3-4-3 pattern. In vitro experiments are initially conducted to test aerosol performance of each DPI assessed at the outlet flow passage. Best performing devices are then connected to different MPs and retested. Finally, the best performing device and MP combinations are evaluated in terms of aerosol delivery through a pediatric (5-year-old) MT model and to a tracheal filter (estimated lung delivery). As with previous pediatric DPI studies, high efficiency aerosolization targets include a mouthpiece emitted dose (ED) ≥ 80% of the loaded dose, a mean mass median aerodynamic diameter (MMAD) < 1.75 μm (and preferably ≤ 1.5 μm), as well as fine particle fraction (FPF) of ED < 5 μm (FPF_<5μm/ED) > 90% and FPF_<1μm/ED > 20%. Based on design improvements, air-jet and MP combinations are sought that can delivery >70% of the loaded drug mass through the pediatric MT model and to the tracheal filter, with <5% MT depositional loss.

MATERIALS AND METHODS

Materials and Powder Formulation

Albuterol Sulfate (AS) USP was purchased from Spectrum Chemicals (Gardena, CA) and Pearlitol® PF-Mannitol was donated from Roquette Pharma (Lestrem, France). Poloxamer 188 (Leutrol F68) was donated from BASF Corporation (Florham Park, NJ). L-leucine and all other reagents were purchased from Sigma Chemical Co. (St. Louis, MO).

Two batches of spray dried albuterol sulfate (AS; as a model inhaled drug) excipient enhanced growth (EEG) formulation were produced using the optimized method described by Son, Longest and Hindle (22) using a Büchi Nano Spray Dryer B-90 HP (Büchi Laboratory-Techniques, Flawil, Switzerland). While AS delivery is not the intended final application of the pediatric air-jet DPI system, it was implemented in this study as a model inhaled medication that is readily available and can be directly quantified. The EEG powder formulation contained a 30:48:20:2% w/w ratio of AS, mannitol, L-leucine, and Poloxamer 188. The primary particle size of the spray-dried powder was initially determined using a laser diffraction method with a Sympatec ASPIROS dry dispersing unit and HELOS laser diffraction sensor (Sympatec GmbH, Clausthal-Zellerfeld, Germany). A pressure drop of 4 bar (400 kPa) was used to disperse a small amount of powder for primary particle size analysis. The geometric diameters given by the laser diffraction system were converted to aerodynamic diameters using the theoretical particle density of 1.39 g/cm³, resulting in a primary particle MMAD of 1.18-1.20 μm.

Air-Jet DPI System Overview

An overview of the pediatric air-jet DPI system connected to a pediatric MT model and tracheal filter is provided in Figure 1. Figure 1b illustrates the inner flow pathway of the air-jet DPI connected to the device mouthpiece (MP). As described, the air-jet DPI contains a small diameter inlet flow passage, aerosolization chamber, and small diameter outlet flow passage. The diameters of the flow passages control the strength of the high speed jet of air within the aerosolization chamber and the release characteristics of the aerosol from the device. All designs included a vertical aerosolization chamber consistent with volume and shape of a Size 0 capsule. The outlet flow passage is constructed with a stainless steel hollow capillary that has been shown to produce minimal depositional internal loss. The DPI outlet flow passage creates a high speed air jet that induces unnecessarily high depositional losses in the patient interface and extrathoracic airways. As suggested by Bass and Longest (32), most MPs considered in this study pass this high speed air jet leaving the DPI and containing the aerosol through a 3D rod array, with the rods in a parallel staggered arrangement. The purpose of the 3D rod array is two-fold. First, Longest et al. (33) previously demonstrated that the 3D rod array was the most effective of the structures analyzed at deaggregating an aerosol with a given amount of input energy and with minimal depositional loss. Secondly, Bass and Longest (32) demonstrated that the 3D rod array effectively dissipated a turbulent jet minimizing downstream deposition including low depositional loss on the rods. Bass and Longest (32) also demonstrated that widening the MP in the vicinity of the rods was required to avoid depositional loss on the MP side walls. This study first explores the performance of different air-jet DPI designs based on varying the inlet and outlet flow passage diameters and configurations. For selected air-jet DPI designs, different MPs are considered both without and with 3D rod arrays. Finally, the best performing air-jet DPI systems are connected to a pediatric MT geometry to estimate the lung delivery efficiency of the inhalers.

Figure 1.

Images showing (a) schematic of the setup for testing oral administration from the air-jet DPI to a tracheal filter through a pediatric MT model and (b) detailed view of the air-jet DPI with important components labeled.

Pediatric Mouth-Throat (MT) Model

The pediatric MT model was based on the VCU Medium Adult MT geometry (www.rddonline.com) with an applied geometric scale factor. The value of the scale factor was 0.75 in order to match the adult outlet tracheal diameter to an estimated tracheal diameter for a 5-year-old child, based on the study of Phalen et al. (34). The pediatric MT oral opening was defined to be a 1.7 by 2.2 cm ellipse. The specified oral airway opening was smoothly blended with the rest of the scaled MT model. The resulting pediatric MT model volume and surface area through the upper tracheal region were 26.2 cm³ and 63.9 cm², respectively. The interior airway surface and shell geometry of the pediatric MT model are shown in Figures 2a and b. To verify the use of the 0.75 scale factor, dimensional comparisons were made with other studies. Distance from the MT inlet to the vertical pharynx midline was 54 mm, which is smaller than with the Golshahi and Finlay (35) pediatric MT model (62.7 mm), and the Wachtel et al.(36) data (58.4 mm; adjusted to the midline of the pharynx) and associated pediatric MT geometry (5). Cheng et al. (37, 38) showed that MT deposition correlates with minimal glottic diameter, which was also supported by Xi and Longest (39). The pediatric MT model had a minimum glottic diameter of 6.3 mm, compared with a value of approximately 6.2 mm in the Golshahi and Finlay (35) pediatric MT model. As a result, it is expected that pediatric MT deposition will be similar to other pediatric airway geometries.

Figure 2.

Isometric view of the pediatric (5-year-old child) MT model including (a) inner airway surface and (b) hollow geometry used for in vitro testing.

Positive Pressure Gas Delivery Conditions

Gas delivery conditions through the inhalers were set to provide the aerosol and a full inhalation breath to a 5-year-old child. Based on ICRP (40) estimates, the vital capacity of a 5-year-old child is approximately 1 L, and the inhaled volume was set at 75% of this value, or 750 ml. An inhaled air volume greater than the 500 ml limit suggested by Lexmond et al. (11) was viewed as acceptable considering that it was delivered with positive pressure and not as a result of the child’s effort breathing against a resistance. ICRP (40) estimates of tracheal gas flow rates for a 5-year-old child are ~10 L/min (LPM) at rest and 20 LPM during light exercise. As a result, 10-20 LPM was used as an appropriate range of flow rates for administering the aerosol and inhalation breath to the in vitro model.

In this study, a custom-built automated air source was used to actuate the air-jet DPIs and provide a simulated full inhalation to the in vitro model. The automated air source consisted of a pressure regulator, solenoid valve and microprocessor controlled timer. Push button actuation of the device provided constant pressure application for a defined time period, resulting in a square waveform flow profile. Farkas et al. (12) previously demonstrated that air-jet performance was not significantly different between ventilation bag and automated gas source actuation, and that an adult can generate a 6 kPa pressure source with one hand operation of a small ventilation bag. Therefore, in this study the pressure supplied by the automated gas source at the air-jet DPI inlet was 6 kPa. For each device, depending on resistance, actuation timing was set to deliver an inhaled air volume of 750 ml. The combination of the device resistance and actuation time then determined the flow rate through each device within the range of 10-20 LPM. As an initial guide, a flow rate of 15 LPM (250 ml/s) requires 3 seconds to deliver the full 750 ml inhalation.

Flow rate measurements were taken using a mass flow meter (Sensirion EM1, Sensirion AG, Stafa, Switzerland) positioned between the solenoid valve and the device inlet for each device in order to set the delivery time for a 750 ml delivery volume. First, the steady state device pressure was set to 6 kPa, then the flow rate was read from the flow meter software (Sensiview, Sensirion AG). The delivery time was then calculated using this flow rate and a volume of 750 ml. After device testing, the MPs were connected to the devices and the flow rate was measured again to ensure that the measured flow rate did not change. Flow rates and delivery times are given in Table 1.

Table 1:

Measured flow rate values for devices at 6 kPa pressure drop. Delivery time is calculated using the measured flow rates and a delivery volume of 750 ml. Inlet and outlet flow passage diameters of each device are also provided.

Device	Flow Rate (LPM)	Time (s)	Inlet/Outlet Diameter (mm)
Device 1	16.3	2.76	1.83/2.39
Device 2	28.1	1.60	3.00/2.69
Device 3	9.7	4.64	1.40/2.39
Device 4	12.1	3.72	1.29(x2)/2.39

Device and Interface Designs

Considering the air-jet DPI base unit, Device 1 (Figure 3a) was designed with inlet and outlet flow passage diameters of 1.83 mm and 2.39 mm, respectively, which were selected as the best case from the previous study of Farkas et al. (41) with a horizontal aerosolization chamber. Device 2 (Figure 3b) was designed based on CFD results from Bass, Farkas and Longest (24) and had an inlet of 3.00 mm and an outlet of 2.69 mm. Device 3 (Figure 3c) had the same outlet diameter as Device 1 (2.39 mm), but included a smaller 1.4 mm inlet to reduce the flow rate through the device. Device 4 (Figure 3d) had two 1.29 mm inlets (to match the cross-sectional area of the 1.83 mm inlet from Device 1) that formed colliding jets within the aerosolization chamber, and then maintained the 2.39 mm outlet diameter used in Devices 1 and 3.

Figure 3.

Cross-sectional views of the air-jet DPI including: (a) side-view of Device 1, (b) side-view of Device 2, (c) side-view of Device 3, and (d) top-view of Device 4.

Mouthpieces are connected to the outlet flow passage of the air-jet DPI and offer the potential for enhanced performance. Mouthpiece designs were generally based on the previous CFD concept study of Bass and Longest (32). Mouthpieces were also selected to determine the effect of a simple rapid expansion from the air-jet outlet flow passage vs. a rapid expansion and 3D rod array. Mouthpiece 1 (Figure 4a) did not contain a rod array and implemented a rapid expansion from the outlet flow passage to a straight elliptical section with a width of 18 mm and a height of 13 mm. Based on the study of Bass and Longest (32), Mouthpiece 2 (Figure 4b) included a 3D rod array in a 3-4-3 pattern and moved the walls of the MP away from the rod array in the direction of intended jet dispersion. In the vicinity of the rod array, MP 2 then had an elliptical cross section with dimensions of 36×13 mm, which was tapered to a MP outlet cross section of 18×13 mm. The 3-4-3 rod array was made from ten 0.5 mm diameter rods and was positioned 1.25 mm away from the outlet flow passage. Each row of rods was spaced 1.75 mm apart (center to center distance) and was staggered 0.5 mm from the previous row, so that there is no flow path straight through the rod array. While the Bass and Longest (32) study implemented the rod array in the mouthpiece design, it was not optimized in that study. The rod array design was based on previous optimization work for both active and passive DPIs (17, 42, 43). Mouthpieces 3 (Figure 4c) and 4 (Figure 4d) had a flow path with an initial elliptical shape of 36×26 mm that tapered to the same size outlet as the other MPs (18×13 mm).

Figure 4.

Top and side cross-sectional views of Device 3 connected to (a) MP 1, (b) MP 2, (c) MP 3, and (d) MP 4.

The air-jet DPIs and MPs were created using Autodesk Inventor and exported as .STL files to be prototyped. Next, they were built using stereolithography (SLA) in Accura ClearVue by 3D Systems On Demand Manufacturing (3D Systems Inc., Rock Hill, SC). The capillaries used in the air-jet DPIs were custom cut from lengths of stainless steel (SAE 304) capillary tubing. O-rings were used as needed to ensure that all components were joined with airtight seals that could withstand the expected operating pressures.

Evaluation of Air-Jet DPI Performance and Lung Delivery

Experiments in this study were performed using 10 mg of EEG-AS powder formulation, which was manually weighed and placed in a removable section of the aerosolization chamber. To load the air-jet DPI, the removable section with loaded powder was connected to the device forming the complete aerosolization chamber. The air-jet DPIs were actuated one time per each powder loading using the automated gas source, which provided 750 ml of positive pressure gas through the device at an inlet pressure of 6 kPa. For experiments characterizing the air-jet DPIs without the attached MPs, the devices were attached to an adapter which was then connected to the inlet of the next generation impactor (NGI). This adapter, shown in Figure 5a, allowed the air-jet DPI to be positioned the same distance from the NGI inlet each time and enabled make-up air to be pulled into the NGI to achieve a steady state NGI flow rate of 45 LPM. Characterizing the performance of the MPs was done using a similar adapter, shown in Figure 5b, which connected to the outlet of the MPs to the NGI inlet.

Figure 5.

Experimental setup for testing aerosol performance of the air-jet DPI operated with compressed air for (a) device testing without the MP and (b) device testing with the MP.

The best-case device/MP combinations were chosen to test for extrathoracic losses in the pediatric MT model. The MT model used in this study was built using SLA in Accura ClearVue resin by 3D Systems On Demand Manufacturing. As shown in Figure 6, the air-jet DPI and MP system was connected to the MT geometry. The MT was coated with MOLYKOTE^® 316 silicone spray (Dow Corning, Midland, MI) to minimize particle bounce and re-entrainment, and two filters (Pulmoguard II) were connected in series to the tracheal outlet of the MT model. Aerosol deposition on these filters was used as an approximation for lung delivery. Because the device was operated with positive pressure, the final filter was open to atmospheric conditions, which represents a scenario where the subject does not resist or assist with inhalation of the aerosol.

Figure 6.

Experimental setup used for testing aerosol delivery through the pediatric MT model.

Aerosol Characterization

After aerosolization, drug masses retained in the air-jet DPI and MP, as well as the drug collected on the preseparator, impaction plates and the filter of the NGI were recovered by washing with appropriate volumes of deionized water and quantified by high performance liquid chromatography (HPLC) analysis. The mass of AS retained in the air-jet DPI and MP, determined by HPLC, was expressed as a percentage of the loaded AS dose. The air-jet DPI emitted dose (ED) was calculated by subtracting the mass of AS retained in the device from the loaded AS dose. Likewise, the MP ED was calculated by subtracting the AS retained in the air-jet DPI and the MP from the loaded dose. In addition, the MT model and filter deposition were determined using the same methods. Model and filter deposition were also expressed as a percentage of the loaded dose. Recovery values were expressed as a percentage of the total amount of drug on all components (air-jet DPI, MP, model, and filter) compared to the loaded AS dose.

In order to determine the nominal dose of AS in the EEG-AS formulation, known masses of the formulation were dissolved in 50 ml of water and the mean amount of AS per mg of formulation was determined using HPLC analysis. For each aerosolization experiment, the measured formulation AS content and the mass of formulation loaded into the capsule was used to determine the nominal dose of AS.

The cut-off diameters of each NGI stage at the operating flow rate of 45 LPM were calculated using the formula specified in USP 35 (Chapter 601, Apparatus 5) and were used to calculate MMAD and fine particle fractions of the delivered aerosol. Fine particle fraction of the EEG formulation (FPF_<5μm/ED) and sub-micrometer FPF (FPF_<1μm/ED) were defined as the mass fractions less than 5 μm and 1 μm, respectively, expressed as a percentage of the ED. MMAD, FPF_<5μm/ED and FPF_<1μm/ED were calculated by linear interpolation using a plot of cumulative percentage drug mass vs. cut-off diameter.

Computational Fluid Dynamics (CFD) Illustration of Rod Array Influence on Flow Field

Computational fluid dynamics (CFD) simulations were used in this study to illustrate the effects of the 3D rod array on the flow field to aid in the discussion of the results. Methods were consistent with the previous study of Bass and Longest (32). Briefly, the flow field and particle trajectory models followed our previously established best practices for CFD-based evaluation of pharmaceutical aerosol delivery to the respiratory airways (44, 45). This includes use of the computationally-efficient low-Reynolds number (LRN) k-ω turbulence model (45), and near-wall corrections to account for anisotropic turbulence and particle-wall hydrodynamic interactions (46–48). Mesh independence was established using the Roache method (49) by evaluating velocity magnitude and turbulent kinetic energy on successively refined meshes, as these field quantities have the most influence on the computational particle trajectory models. FLUENT v19.0 (ANSYS Inc., Canonsburg, PA) was used to obtain solutions for all flow and turbulence transport equations, as well as the particle equations of motion, with results successfully validated against experimental deposition data in the patient interface and extrathoracic airways. Further details on the transport equations and particle tracking models are available in other publications (50, 51).

RESULTS

Performance of Air-Jet DPI Base Units

Gas delivery flow rates and delivery times required to administer the 750 ml air volume are reported in Table 1. Aerosolization performance of the air-jet DPI devices without attached MPs are given in Table 2. Devices 1-3 were similar in all areas except FPF_<1μm/ED, where Device 2 had a statistically higher FPF_<1μm/ED. The emptying of Device 4 was highly variable and while the results were statistically similar to Devices 1-3, the average value was much lower than the other options (ED=69.9%) with a much higher standard deviation (SD=14.6%). While MMAD was similar for Devices 1-3, Device 2 was more variable and also had a much higher flow rate at the specific inlet pressure of 6 kPa. Devices 1 and 3 were statistically similar in all measured performance areas and as a result, Devices 1 and 3 were selected for further investigation in the study. The two different devices were retained because it was expected that different flow rates would have an impact on MP performance and MT penetration.

Table 2:

Aerosolization performance of air-jet DPI Devices 1 through 4.

Mean aerosol characteristics with standard deviations (SD) shown in parenthesis [n=3].

Description	Device 1	Device 2	Device 3	Device 4
ED (%)	82.8 (2.1)	86.8 (2.7)	85.3 (0.7)	69.9 (14.6)
Device (%)	17.2 (2.1)	13.4 (2.7)	14.7 (0.7)	30.1 (14.6)
FPF<5μm/ED (%)*	82.1 (3.0)	78.3 (1.7)	79.1 (0.4)	90.1 (0.6)**
FPF<1μm/ED (%)*	22.4 (3.0)	27.0 (1.2)**	21.7 (1.0)	26.2 (0.7)**
MMAD (μm)*	1.71 (0.06)	1.85 (0.14)	1.76 (0.03)	1.56 (0.01)***

^*p<0.05 significant effect of Device on FPF_<5μm/ED, FPF_<1μm/ED, and MMAD (one-way ANOVA).

^**p<0.05 significant difference compared to Device 1 (post-hoc Tukey).

^***p<0.05 significant difference compared to Device 2 (post-hoc Tukey).

Performance of MPs

To determine the best-case device-MP combination, both Device 1 and Device 3 were tested with the four different MP designs. Aerosolization performance for Device 1 with the four different MPs are shown in Table 3. Without the rod array, MP 1 produced a significantly smaller FPF_<5μm/ED and larger MMAD compared with MPs 2-4, which did include 3D rod arrays. All of the MPs produced similar MP ED values between 78% and 82%, as well as similar FPF_<1μm/ED between 17% and 20%.

Table 3:

Aerosolization performance of Device 1 connected to MP 1 through 4.

Mean aerosol characteristics with standard deviations (SD) shown in parenthesis [n=3].

Description	MP 1	MP 2	MP 3	MP 4
MP ED (%)	80.3 (0.9)	78.2 (1.7)	82.3 (1.8)	77.9 (3.3)
Device (%)	12.5 (2.1)	14.1 (1.8)	11.8 (1.6)	14.9 (2.3)
MP (%)	7.3 (1.5)	7.8 (3.3)	5.9 (3.1)	7.2 (2.5)
FPF<5μm/ED (%)*	87.8 (0.8)	96.1 (1.1)**	96.8 (1.0)**	97.0 (0.8)**
FPF<1μm/ED (%)	17.4 (1.2)	19.6 (0.8)	18.6 (1.4)	20.3 (4.2)
MMAD (μm)*	1.85 (0.06)	1.64 (0.04)**	1.67 (0.04)	1.65 (0.11)**

^*p<0.05 significant effect of MP on FPF_<5μm/ED and MMAD (one-way ANOVA).

^**p<0.05 significant difference compared to MP 1 (post-hoc Tukey).

Aerosolization performance of the different MPs connected to Device 3 is reported in Table 4. Significantly smaller FPFs and larger MMAD values were found with MP 1 compared to the other MPs, which all contain 3D rod arrays. Specifically, the 3D rod array reduced the emitted MMAD by approximately 0.3-0.4 μm compared with MP 1 without a 3D rod array. Similar to Device 1 results, MP ED values for all MP designs were similar between 78% and 81%. The two simpler MP designs, MP 2 and 3 were selected for further investigation of lung delivery through a MT model. Since Devices 1 and 3 produced similar results, both devices were also selected for testing lung delivery efficiency.

Table 4:

Aerosolization performance of Device 3 connected to MP 1 through 4.

Mean aerosol characteristics with standard deviations (SD) shown in parenthesis [n=3].

Description	MP 1	MP 2	MP 3	MP 4
MP ED (%)	80.9 (0.5)	79.3 (1.7)	80.4 (0.3)	78.3 (4.6)
Device (%)	13.7 (0.3)	14.2 (2.1)	13.6 (0.3)	16.1 (5.4)
MP (%)	5.4 (0.5)	6.5 (0.3)	6.0 (0.3)	5.7 (1.0)
FPF<5μm/ED (%)*	85.4 (1.0)	96.2 (0.3)**	96.9 (0.5)**	98.0 (0.5)**
FPF<1μm/ED (%)*	17.3 (1.4)	20.1 (0.7)	20.8 (1.1)**	21.0 (1.5)**
MMAD (μm)*	1.98 (0.05)	1.68 (0.02)**	1.67 (0.04)**	1.62 (0.05)**

^*p<0.05 significant effect of MP on FPF_<5μm/ED, FPF_<1μm/ED, and MMAD (one-way ANOVA).

^**p<0.05 significant difference compared to MP 1 (post-hoc Tukey).

Oral Aerosol Delivery

Devices 1 and 3 were both tested in combination with MPs 2 and 3 with the results shown in Table 5. In addition, MP1 was also considered connected to Devices 1 and 3 in order to illustrate the effect of the smaller aerosol produced by the 3D rod array designs (MPs 2 and 3). As expected for both Devices 1 and 3, the smaller aerosols produced by the 3D rod array MPs resulted in a significant reduction in MT deposition compared with MP1. Device 3 showed a decrease in MP loss across all MPs, although this difference did not result in a significant increase in MP ED or filter delivery when used with MPs 2 and 3. Although there was no significant difference between Devices 1 and 3 with MPs 2 and 3, the highest average lung dose achieved was 73.2% with the Device 3-MP 2 combination.

Table 5:

Aerosolization and lung delivery efficiency for oral administration through the 5-year-old pediatric MT geometry.

Mean aerosol characteristics with standard deviations (SD) shown in parenthesis [n=3].

	Device 1			Device 3
Description	MP 1	MP 2	MP 3	MP 1	MP 2	MP 3
Device (%)	14.5 (3.8)	14.1 (0.8)	15.4 (3.0)	15.4 (1.3)	16.8 (1.0)	16.5 (1.1)
MP (%)*	9.4 (0.8)	11.8 (1.0)	9.5 (1.4)	4.7 (0.5)**	5.9 (0.5)**	6.8 (0.8)**
MP ED (%)	76.1 (4.5)	74.1 (0.5)	75.1 (1.8)	79.8 (1.4)	77.3 (1.0)	76.7 (1.7)
MT (%)*	6.2 (0.7)***	4.0 (0.2)**,***	2.5 (0.4)**,***	8.9 (0.3)**	2.9 (0.4)**,***	1.4 (0.5)**,***
Filter (%)*	65.6 (4.0)***	65.9 (3.7)**,***	67.8 (0.9)**,***	66.0 (2.6)**	73.2 (1.4)**,***	70.4 (1.6)**,***
Recovery (%)	95.7 (1.3)	95.8 (3.4)	95.3 (1.3)	95.0 (2.2)	98.8 (2.6)	95.2 (1.0)

^*p<0.05 significant effect of Device-MP combination on MP, Model, and Filter (one-way ANOVA).

^**p<0.05 significant difference compared to Device 1 MP 1 (post-hoc Tukey).

^***p<0.05 significant difference compared to Device 3 MP 1 (post-hoc Tukey).

DISCUSSION

A primary finding of this study was that a vertically oriented aerosolization chamber with the air-jet passing across the upper section of the chamber empties well and forms a high quality aerosol. As described, this orientation of the aerosolization chamber can accommodate higher powder masses and may be easier to load than the horizontal aerosolization chamber considered in Farkas et al. (12). Considering the new air-jet DPI without a MP, Device 3 had the best combined aerosolization performance with an ED of 85.3% and a MMAD of 1.76 μm, which was also statistically similar to Device 1. This performance was also similar to Case 4 considered in Farkas et al. (12) with the air-jet oriented symmetrically along the primary axis of the capsule chamber. While aerosolization performance was similar between the current and previous devices, the vertical aerosolization chamber of the current study provided lower variability in the ED, with a Device 3 ED coefficient of variation of only 0.8%. Importantly, Device 3 of the current study provided similar aerosolization compared with the best case device in the previous study at a lower flow rate of 9.7 LPM, compared with the previous flow rate of 13.7 LPM. This lower flow rate results in a small increase in aerosol delivery time (Table 1), but will improve penetration of the aerosol through the extrathoracic airways.

A second primary finding of this study was that the 3D rod array effectively reduced the aerosol size exiting the MP without significantly increasing the MP deposition fraction. While Devices 1 and 3 were statistically similar, Device 3 provided the best overall transmission of aerosol through the MT geometry and will therefore be the focus of this discussion. Considering Device 3, inclusion of the rod array decreased ex-MP aerosol size by approximately 0.3-0.4 μm. Interestingly, this reduction in size did not result in a net statistically significant increase in MP deposition fraction including any potential aerosol deposition on the rods. As a result, the 3D rod array appears to be an effective structure to further reduce the aerosol size. This result would not have been possible without the CFD study of Bass and Longest (32), which demonstrated that expanding the MP side walls in the vicinity of the rod array and in the direction of jet dispersion was required to maintain low MP deposition. It is also noted that with the rod array arrangement of Bass and Longest (32), the rod array does not need to extend across the entire width of the MP. Instead, a small rod array with a 3-4-3 pattern is used directly at the outlet of the small diameter flow pathway leading out of the air-jet DPI, without the complexity and expense of additional rods.

In addition to significantly reducing the particle size, the 3D rod array also reduces the turbulent jet entering the MP and MT model. Without the rod array, this turbulent jet resulted in ~5% MP deposition and ~10% MT deposition, which are relatively high for the 2 μm aerosol produced by Device 3 and MP 1. This deposition arises from both turbulent dispersion and impaction on the back of the throat, as illustrated in Bass and Longest (32). As described (32), the rod array effectively dissipates the turbulent jet creating nearly uniform flow entering the MT model. This dissipation is illustrated in Figure 7 for MP 1 and MP 3 at a flow rate of 13.3 LPM using the computational fluid dynamics (CFD) methods described in Bass and Longest (32). For illustrative purposes, a flow rate of approximately 13 LPM was selected which is approximately midway between the flow rates of Devices 1 and 3. In Figure 7, iso-surfaces of 10, 15, and 20 m/s are shown (yellow, orange, and red iso-surfaces, respectively) in which the volume inside the respective isosurfaces has a velocity equal to or greater than 10, 15, and 20 m/s. For MP 1 without the 3D rod array, the isosurface of high velocity jet flow extends almost the entire length of the MP in the direction of the MT geometry (Figure 7a). In contrast, the 3D rod array effectively shatters the high velocity isosurface and largely eliminates its presence in the MP. For Device 3, findings of this study illustrate that the presence of the 3D rod array reduces MT deposition from 8.9% (MP 1) to 2.9% and 1.4% with MP 2 and MP3, respectively (Table 5). As a result, the best case combination of Device 3 and MP3 reduced MT deposition loss by a factor as high as 6.4-fold.

Figure 7.

Computational fluid dynamics (CFD) predictions of velocity isosurfaces at a flow rate of 13 LPM with (a) MP 1 and (b) MP 3.

The overall outcome of this study is the development of a pediatric air-jet device that met the objectives of high efficiency performance and lung aerosol delivery. For Device 3 with MP 2 or MP 3, the mouthpiece ED was ~80% whereas FPF_<5μm/ED and FPF_<1μm/ED were >95% and >20%, respectively. MMAD values were also <1.7 μm. Small MMAD alone was insufficient to minimize MT deposition. Instead, the combination of small MMAD and dissipation of the turbulent jet with the 3D rod array reduced MT deposition to <3% and produced lung delivery efficiencies of >70% based on device loaded dose. The best case combination of Device 3 and MP 2 produced a MT depositional loss of only 1.4% and a lung delivery efficiency of 73.2%. These performance metrics were achieved using a 10 mg powder fill mass and a single device actuation. Compared with the previous study of Farkas et al. (12), air-jet DPI advancements reduced MT depositional loss from 6.6% to as low as 1.4% of loaded dose and increased estimated lung delivery from 63.8% to 73.2%.

In addition to high efficiency aerosolization and dispersion of the turbulent jet, the air-jet DPI is able to overcome difficulties of delivering aerosol to the lungs of pediatric patients by using positive pressure to aerosolize the powder and inflate the lungs without relying on the child’s inspiration. High variability was observed between inhalation waveforms of trained children in a lab setting, which can have a negative effect on the performance of passive DPIs (11). In addition, the positive pressure provided by the system developed in this study expands the airways to enable more efficient passage for the aerosol to the lungs, instead of negative pressure produced by the patient’s inhalation, contracting the already small airways.

Limitations of this study include using a single airway geometry, investigating only oral delivery, and testing using one fixed powder mass and formulation for all experiments. Additional airway geometries can be tested using CFD modeling of realistic geometries obtained from CT scans. Once validated with the experimental results from this study, use of a CFD model to assess intersubject variability will allow more efficient testing of a large number of geometries (24, 32). Although this study only investigated oral delivery, our previous study (12) showed similar lung delivery efficiency (within 3% of the loaded dose) with a nasal cannula interface when using a nasal airway model of a five-year-old child. A powder mass of 10 mg was used to assess general inhaler performance; however, modifying the powder for low dose (~ 2 mg) and high dose (~75 mg) applications will likely affect performance. Similarly, different spray-dried therapeutic formulations such as inhaled antibiotics or surfactants will also influence device performance and require additional assessment and optimization.

An additional limitation of this study is the use of a set dose (10 mg powder mass) of AS EEG formulation. AS delivery is not the intended application of the pediatric air-jet DPI system as this medication is relatively inexpensive, has a broad therapeutic index and relatively mild side effects. However, AS was implemented in this study as a model drug that can be readily quantified. Moreover, we have developed highly dispersible optimized EEG formulations of AS in previous studies (22). Intended applications of the pediatric air-jet DPI are the delivery of higher dose inhaled medications where efficacy can be increased with improved lung and deep lung targeting, and where reduced inter- and intra-subject variability is important. Potential candidate medications include inhaled antibiotics, surfactants, clearance agents, insulin, and anti-inflammatories. Expected doses of these medications are in the range of 10-100 mg or more. As a result, future studies should also evaluate the effect of increasing the loaded powder mass on device performance. In a different DPI, we have previously shown that higher loaded doses typically result in higher percentages of dose emission (i.e., higher ED) (52). However, this effect requires future verification in the pediatric air-jet DPI as well as an evaluation of the impact of higher doses on aerosol size.

Conclusions

In conclusion, the pediatric air-jet DPI was shown to achieve high efficiency aerosolization of a spray-dried excipient enhanced growth (EEG) powder formulation and high efficiency lung delivery based on estimates with a realistic pediatric in vitro model. Compared with other published in vitro evaluations with oral inhalers used for children, lung delivery efficiencies can be improved from approximately 5-30% (5, 7, 8) to >70%, or by factors ranging from 2.3 to 14-fold. The vertical aerosolization chamber orientation used in this study maintained air-jet aerosolization performance at a lower flow rate and reduced variability in ED. Use of the 3D rod array as developed by Bass and Longest (32) was highly successful, reducing aerosol size by 0.3 to 0.4 μm (~20% relative difference) and reduced MT depositional loss from 8.9% to 1.4% (6.4-fold reduction). The vertically oriented aerosolization chamber made the device easy to load and the auto-actuator made the device consistent and straight forward to operate. Further improvements in aerosolization chamber emptying can be used to increase device ED, but will require CFD optimization to simultaneously increase ED while maintaining or reducing MMAD, as with Bass et al. (24). Further pre-clinical studies are needed to improve device handling, develop nasal cannula interfaces for high efficiency nose-to-lung aerosol delivery, explore performance with higher dose loadings, and optimize the device for different spray-dried powder formulations and different age groups.

ABBREVIATIONS

3D

three dimensional

AAV

actuation air volume

AS

albuterol sulfate

CFD

computational fluid dynamics

DPI

dry powder inhaler

ED

emitted dose

EEG

excipient enhanced growth

FPF

fine particle fraction

HPLC

high performance liquid chromatography

LPM

Liters per minute

MMAD

mass median aerodynamic diameter

MP

mouthpiece

MT

mouth-throat

NGI

Next Generation Impactor

SD

standard deviation