Incoming Flow Turbulence Reduction Increases the Eciency of Aerosolized Medication Delivery to 3D Oral Mucosal Tissue Models.

Background: Inhalable medication devices on the market deliver aerosolized drugs in a turbulent ow, which creates signicant obstacles for reaching the lower lungs. The complexity of the interaction of external turbulent ow from an inhalation device with complex anatomy of the upper airways, including the oropharyngeal cavity, makes the delity of aerosolized medication delivery to the lungs extremely low. Unpredictable outcomes, waste, and side effects result from unintended upper airway drug deposition. Methods: Here we compared the eciency of aerosolized medication (uticasone) delivery via a novel Flow Modication Device (ModiFlow) to that of a Standard Spacer (SS) device. The ability of ModiFlow to minimize the turbulence in the ow was assessed preliminarily by measuring the length of the Laminar Outow using video recording. Oral mucosal 3D tissue culture (SkinAxis) was used as a target for delivery, and was placed either “well within” the range of the length of the oro-pharyngeal cavity (5cm) or “well outside” of it (20cm) from exit points of each device. The eciency of uticasone delivery to the surface of tissue cultures was quantied by mass spectrometry. Results: The results of the study demonstrated a statistically signicant advantage of ModiFlow over a Standard Spacer in delivering aerosolized uticasone to target tissue at both distances. The difference in the eciency of delivery between the two spacers was more pronounced at a longer distance. Conclusions: Lamination of the outow using internal septi helps improve the delivery of aerosolized medication to target tissue despite an increased inner surface area of the spacer device, which also indicates lesser deposition of the medication on the walls of the test tube. This suggests that the use of ModiFlow will potentially result in the more ecient delivery of aerosolized medications to the lungs with lesser deposition in the oral cavity and fewer side effects. Each tissue sample was processed by: adding 50 ml 0.1% formic acid and 200ml of methanol to a culture plate, scrapping with pipette tips, and transferring to an Eppendorf tube. The plate was washed sequentially with 200ml 0.2% formic acid and 100ml methanol and the washes combined with the initial extract. Extracts were sonicated for 1 min and centrifuged for 5 min at 25000 x g. Supernatants were diluted 10-fold using 50% methanol/0.1% formic acid before analysis by LC-MS. HPLC–MS experiments were performed using a ThermoFisher Velos LTQ Orbitrap Pro mass spectrometer interfaced with a Dionex U3000 chromatography system. Samples (5 µL) were injected in microliter pick up mode and separated on a reverse-phase column (Discovery BIO Wide Pore C18, 5cm x 2.1mm, Supelco Analytical). Chromatography was conducted at a ow rate of 200µl/min using a gradient formed with an aqueous solution of 0.2 % acetic acid (solvent A) and methanol (solvent B) as follows: 60% B (1 min), 60-90% B (linear increase in 3 min), 90% B for 1 min, 90- 60% B (linear decrease in 0.1 min), and equilibration at 60% B (3 min). The column temperature was maintained at 45°C. MS acquisition parameters were as follows: the electrospray ion source was operated in positive ion mode (ESI+). The positively charged uticasone (m/z= 501.3) was isolated in the ion trap with an isolation window of 3 m/z and fragmented with CID with a relative collision energy of 25% and activation time of 10 milliseconds. Fragments were detected using the ion trap and the 303.15 m/z fragment used for quantication. A standard curve consisting of dilutions of uticasone in methanol (0.01 ng/ml to 100mg/ml) was analyzed in parallel with samples. Peak areas of the 501-301


Background
In recent years the targeted aerosol delivery to the respiratory tract has rapidly gained interest as the preferred route for the treatment of lung diseases locally, as well as accessing the systemic circulation.
Many therapeutic aerosols containing large molecules -proteins, hormones, nucleic acids, chemotherapeutic drugs [1,2], have shown a promise as agents for gene therapy, antiviral therapy for in uenza and measles, insulin and vaccine delivery [3,4].
Our increasing understanding of medication receptor distribution throughout the lungs further emphasizes the need for delivering different aerosolized medications to very speci c areas of the respiratory tract [5]. In recent years more than 100 different medications have been brought to the market in the aerosolized form [6]. However, the e ciency of aerosolized medication delivery to the lungs remains extremely low due to signi cant variability in parameters within the upper respiratory tract and the design of the delivery devices. Consequently, up to 77% of the medication can deposit in the upper airways [7], most notably in the oral and pharyngeal areas, thus not only reducing e ciency and lowering the predictability of the outcomes but also causing side effects. With the entrance of newer inhalable medications to the market, new delivery standards are being explored [8,9].
The emergence of turbulent ow in different segments of the respiratory tract can have a signi cant impact on the deposition and absorption of the aerosols. Although largely overlooked, for many aerosolized medications this may be a factor of critical importance. The presence of turbulence can increase the localized shear wall stress [10], create signi cant resistance to the ow, and reduce farther propagation of the aerosolized substances. To our knowledge, no signi cance has ever been given to turbulence emerging from the delivery device itself as a major obstacle to the effective delivery of aerosols to the lungs. Lamination of the device out ow with a reduction of turbulence to optimize its interaction with oropharyngeal inspiratory air ow may have a positive impact on the e ciency of aerosol delivery to the lungs. Even with the use of spacers, despite the variations of "external" parameters to improve their performance, except for volume, the overall e ciency of aerosol delivery to the lower lungs remains poor [11][12][13]. The inner space of a spacer device, where the ow of aerosolized medication occurs, offers signi cant opportunity for manipulating the ow [14]. Most, if not all of the existing spacers on the market lack any internal structures, and the exiting ow of aerosol is highly turbulent.
A device that can reduce the turbulence in the aerosolized medication out ow may prove to be more e cient in delivering the medication to the lower lungs and reducing its deposition in the upper airways.
ModiFlow is a novel spacer-like delivery device speci cally designed to create a Laminar Out ow of aerosolized medication (Flow Modi cation Device, G. Greg Haroutunian, MD, US Patent # 8,371,291 B2).
Modeling of the air ow and particle deposition rates in the airways is often done with the use of Computational Flow Dynamics (CFD) methods [15][16][17]. However, the extreme complexity of parameters and factors to be accounted for naturally limit the capabilities of these methods. More direct methods of assessing the medication deposition and absorption rates could be obtained on actual 3D tissue samples. In recent years progress has been made in constructing 3D tissue models and using them for various research and clinical applications [18][19][20][21]. This allowed for departure from the use of animal tissue models, and for obtaining data in realistic physiologic conditions [21][22][23][24][25][26][27].

Aerosol drug delivery devices
In this study, we compared the performance of the ModiFlow (MF) to an idealized Standard Spacer (SS). For standardization purposes, a hollow cylindrical tube of identical length and inner diameter represented both devices, with the only difference being that MF had speci cally designed inner septal structures, which SS didn't.
Since most, if not all spacer devices on the market have a cylindrical shape, and no internal structures, our Standard Spacer model was considered to be a fair representation of spacers on the market. In addition, several studies have demonstrated that with the exception of volume, all other structural modi cations to the spacers, such as valves and masks, have not made a signi cant difference in their performance. Regardless, our goal was to minimize any structural variability between the tubes for the purpose of controlling all parameters, and leaving only one main difference -the presence of inner septal structures in MF, and the absence thereof in SS.
In accordance with the above, for this study the parameters for ModiFlow were selected as follows: Total length -100mm, Inner Diameter -38mm, Septi -62mm, 34mm, 24mm. The calculated Inner Surface Area (ISA) was 164.9 sq. cm, of which 45.6 sq. cm is accounted for Septi only, and 119.3 sq. cm for inner walls. As a SS, a hollow cylindrical tube of the same Length (100mm) and Inner Diameter (38mm) as in MF was used, with the calculated ISA of 119.3 sq. cm.
To compare the e ciencies of aerosol delivery for ModiFlow vs a Standard Spacer, 4 different Test Tubes (TT) have been devised, all of which had a similar structure as depicted in Figure 1. Two of them had a length of 15cm, and two -30cm. All four TT's were of cylindrical shape, and of the same inner diameter (38 mm). In the TT of each length, either ModiFlow ( Figure 1

3D Oral Epidermal tissues
To better approximate physiological conditions of drug delivery, we measured aerosolized drug deposition on a SkinAxis model of the oral mucosal tissues. Normal human Gingival keratinocytes (SkinAxis) were cultured on specially prepared cell culture inserts using serum-free medium and differentiated in vitro using proprietary SkinAxis' cell culture technology to form multilayered, highly differentiated models of the human gingival phenotypes (Figure 3, and www.skinaxis.com). SkinAxis oral epidermal tissue models are highly reproducible and exhibit in vivo-like morphological and growth characteristics. The differentiated tissue was inserted at the end of the spacer, as described above and to quantify drug deposition tissues were processed for Mass Spectrometry.
Fluticasone extraction from oral tissues and quanti cation.
Each tissue sample was processed by: adding 50 ml 0.1% formic acid and 200ml of methanol to a culture plate, scrapping with pipette tips, and transferring to an Eppendorf tube. The plate was washed sequentially with 200ml 0.2% formic acid and 100ml methanol and the washes combined with the initial extract. Extracts were sonicated for 1 min and centrifuged for 5 min at 25000 x g. Supernatants were diluted 10-fold using 50% methanol/0.1% formic acid before analysis by LC-MS.
HPLC-MS experiments were performed using a ThermoFisher Velos LTQ Orbitrap Pro mass spectrometer interfaced with a Dionex U3000 chromatography system. Samples (5 µL

Statistical analysis
For each of the two spacers, independent t-tests were used to test the effects of varying drug deposition in different experimental settings. P-value < 0.05 was considered statistically signi cant.

Results
Comparing the e cacy of drug delivery of Standard Spacer to ModiFlow.
The geometry of ModiFlow inner space is structured using uniquely designed longitudinal septi of different lengths that produce a laminar (non-turbulent) ow, which persists after exiting the device (82-179 mm in one particular model).
Subdivision of the main ow into two sub-ows iterated multiple times (a fractal tree), helps to disrupt the cycle of growth of the lateral force, preventing the emergence of a high Reynolds number turbulence. This tree-like branching pattern generated by iteratively applying a set of simple rules is pervasive in biological networks and is also utilized to model lung function and the bronchial tree [28,29]. In ModiFlow the septi serve to prevent the emergence of turbulence by periodically sub-dividing the ow, and with it, dividing the lateral force that grows in an iterative side-to-side movement, and would eventually break the ow into a turbulent one. At the same time, they reduce the mixing of sub-ows, with their larger surface area they lter out larger size droplets that tend to deposit on the upper portions of the respiratory tract mucosa, and also remove the tangential (lateral, non-parallel) parts of the ow as well. As a result, the aerosol ow emerges from the device more laminar, coherent, unidirectional, and with more uniform particle size. The lengths of the septi and the total length of the ModiFlow tube are selected to be proportional to sequential Fibonacci numbers (these are numbers in a sequence in which each number is a sum of two preceding ones), assuring that their ratios are close to "golden ratio" = 1.618 (it is known that if a Fibonacci number is divided by its immediate predecessor, the resulting quotient approximates 1.618). There is some evidence that this particular ratio is effective in keeping the ow laminar.
This improved laminar ow can have potentially signi cant implications: 1) There is a higher likelihood of such a laminated ow to overcome the oro-pharyngeal barrier without depositing most of the particles there, a larger portion of the ow will reach trachea and main bronchi; 2) The higher degree of lamination of the ow will skew the distribution of aerosol particles pattern towards deposition in lower parts of the respiratory tract, including ner bronchioli and alveoli; 3) The manipulation of the geometry of ModiFlow could allow a more targeted deposition of particles in speci c areas of respiratory tract depending on the need and type of medication being delivered and the distribution map of its receptors in the airways; 4) Reduction of side effects due to reduction of deposition in the mouth, throat, trachea and upper bronchi; 5) Improved/more e cient delivery may overall reduce the cost of expensive medications.
To assess the ability of MF and SS to laminate a ow, multiple measurements of the Laminar Out ow (the portion of the exit ow that remains continuously laminar) were taken using video-recording of a smoke ow via both spacers. For ModiFlow the measurements indicated that the ow remained completely laminar in the range from 82 to 129 mm, transitioned to non-laminar between 82 and 203 mm, and was completely non-laminar above 179-203 mm from the exit point ( Figure 2). This was signi cantly more extended laminar ow than for Standard Spacer could be achieved, which remained in the ranges of less than 30 mm. This data was only obtained and used for the purpose to demonstrate the signi cantly better ability of the ModiFlow to laminate the ow. No attempts were ever made to correlate these measurements with the outcomes of the study.
The tissue samples were inserted at the distal end of each Test Tube (TT) through an adaptor (see Figure  1, membrane holder), perpendicular to the axis of the TT (axis of ow), opposite to the end where the medication dispenser was inserted. ModiFlow or Standard Spacer were incorporated ush with the proximal end of the TT, and oriented in such a way that the ow is directed towards the tissue sample ( Figure 1). The positioning of the tissue sample perpendicular to the axis of ow at different distances allows for the assessment of aerosol ow (delivery) through a particular cross-section of the TT. Placing the tissue samples on the lateral walls of the TT would make it very di cult to discern any correlation between the aerosol ow and its deposition on the tissue sample, having to take into account too many variables.
Experiments were conducted to maximally mimic the real-life application of aerosol oral delivery of Fluticasone. After shaking for 5 seconds and spraying away once to assure proper functioning, the Fluticasone Metered Dose Inhaler was inserted into the medication pump adapter of the Test Tube containing the target tissue and was activated twice with a 3-minute interval by pushing the top of the medication canister all the way down. This delivered a total dose of 440 µg (220 µg per spray) in accordance with standard dosing recommendations. Because of the hermetic closure of Test Tubes, and of the same standard amount of Fluticasone delivered into each TT, it was possible to not only measure the amount of Fluticasone delivered along the axis of ow to the tissue samples at different distances but to also indirectly assess the amount of Fluticasone deposited on the lateral walls of each TT. This could be achieved by subtracting the amount of Fluticasone deposited on each tissue sample from the total dose of 440 µg sprayed in each TT.
The aerosol droplet sizes and the ow velocities were controlled by using the same Fluticasone MDI in all Test Tubes.
To minimize possible electro-static in uences, all test tubes and spacers were pre-washed with standard dishwasher detergent.
The results presented in Figure 4 indicates that ModiFlow signi cantly increased drug delivery to the tissue as compared to the Standard Spacer. ModiFlow delivered on average 20.19 μg and 8.9 μg of drug/tissue at 5 cm and 20 cm distance, while Standard Spacer delivered 7.8 μg and 3.05 μg/tissue respectively (Figure 4), strongly suggesting a more e cient Fluticasone delivery to mucosal tissue by ModiFlow.

Discussion
The e cient delivery of aerosolized medications to the patient's airways continues to be a challenging task. Various devices such as pumps, spacers, and nebulizers are currently used for that purpose. In all of the devices on the market, the out ow of aerosolized medication is signi cantly turbulent. Turbulent ow is one of the main reasons for the deposition of up to 77% of the medication in the upper airways [7,10].
The interaction of the out ow of a delivery device with the complex geometry of the oropharyngeal segment of the upper airways may further contribute to these ine ciencies. To our knowledge, no speci c measures have been taken so far to modify the out ow of these devices with the intention to reduce the turbulence. Accurately assessing the e ciency of laminated aerosol delivery may yield important results.
It is important to note that all 4 Test Tubes with attached medication pumps constituted a closed system, and the amount of Fluticasone sprayed into each one of them was identical. Since the amount of Fluticasone that reached the cell cultures at the end of Test Tubes was signi cantly higher (2.5-3-fold) with ModiFlow than with Standard Spacer, therefore smaller amounts were deposited elsewhere in Test Tubes with ModiFlow. Also, although both the ModiFlow and the Standard Spacer have the same length and diameter, the Inner Surface Area of the Standard Spacer is 119.3 sq. cm, while that of the ModiFlow is 38% larger -164.9 sq. cm due to the presence of inner septi. This would cause a larger amount of Fluticasone deposition within ModiFlow as compared to the Standard Spacer.
As indicated above, the delivery of the aerosolized Fluticasone as measured by the tissue surface deposition rates was signi cantly more effective via ModiFlow than via Standard Spacer in both distance ranges -short and long (p < 0.01 and p < 0.05 respectively). Interestingly, the comparative e ciency of ModiFlow can be judged to be higher in the long-range than in the short-range if ratios of deposited

Conclusion
Turbulence plays a signi cant role in reducing the e ciency of aerosolized medication delivery to the lungs. In the ModiFlow device the inner space is structured in a manner that helps laminate the out ow and reduce the emergence of turbulence in the oropharyngeal cavity. This study demonstrated that despite the larger inner surface area, ModiFlow increases the e ciency of aerosol delivery to farther distances, and reduce its deposition in anterior/upper airways. This becomes increasingly important with newer aerosolized medications entering the marketplace with high potential for side effects, high cost of medication waste, higher need for e cient, consistent and targeted delivery.