INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Dextran is an oligosaccharide being considered for use in the treatment of cystic fibrosis (CF) because of the therapeutic potential it has demonstrated in animal and in vitro experiments. For example, Feng and coworkers (1) show that dextran exhibits significant mucolytic activity in vitro in CF sputum, while Feng and colleagues (2) show enhanced mucociliary clearance rates when aerosolized dextran is delivered in dogs and suggest that dextran reduces cross-linkage bonding in the mucus, leading to reduced mucous viscoelastic modulus. In addition, Barghouthi and colleagues (3) find that dextran interferes with bacterial adhesion to epithelial cells, and Bryan and coworkers (4) find that aerosolized dextran prevents Pseudomonas aeruginosa pneumonia and death in neonatal mice. For these reasons, dextran may be useful in the treatment of the respiratory aspects of cystic fibrosis, where impaired mucociliary clearance is a major cause of the bacterial infections that lead to much of the morbidity associated with this disease.

However, before the therapeutic benefits of dextran can be examined in human clinical trials, an appropriate delivery system is required. Because dextran appears to demonstrate efficacy in the above-mentioned models owing to local activity in the airway surface fluid, the localized delivery of dextran to airway surfaces is expected to be an effective choice of delivery route. A standard approach for such delivery is the use of an inhaled aerosol. Because nebulizers are uniquely suited to deliver the relatively large quantities of aerosolized dextran that are expected to be needed for in vivo efficacy, nebulization of aqueous dextran is a logical choice of delivery method.

In the present work, we examine the potential of several nebulizer systems to deliver dextran as an inhaled aerosol. The data we present are used to suggest an optimized delivery system that is expected to successfully deliver appropriate amounts of dextran in planned clinical trials.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Dextran 4000 (nominal molar mass, 4,000 g/mol) supplied by Dextran Products (Scarborough, ON, Canada) was prepared in isotonic saline (0.9%, wt/vol) and pH adjusted to 7.4 by addition of 0.1 N NaOH.

Surface tension measurements of the dextran-saline solutions were made by axisymmetric drop shape analysis (5) at room temperature (23 ± 1° C). Measurements of viscosity were made with a rotating spindle viscometer (Brookfield Engineering Laboratories, Stoughton, MA) at 23 ± 1° C room temperature and at 18 ± 1° C, which is the range of temperatures measured during nebulization of dextran with jet nebulizers. Measurements of surface tension were done in triplicate. Ten measurements of viscosity were made at dextran concentrations of 50 and 100 mg/ml, while 15 were made at concentrations of 200 and 400 mg/ml.

Nebulizers were filled with either 2.5 or 4 ml of dextran solutions with differing dextran concentrations. Four different nebulizer types were used. These included two of the most commonly used nebulizers in CF therapy (6), as well as two more recently developed nebulizers that are known to outperform many nebulizers with other formulations (7, 8). The nebulizers included three jet nebulizers: the LC STAR (Pari, Starnberg, Germany), the T-Updraft II (model 1732; Hudson RCI, Temecula, CA), and the Acorn II (Marquest Medical Products, Englewood, CO), in addition to one type of ultrasonic nebulizer, the Sonix 2000 (model 86111; Medix, Catthorpe, UK). A Pulmo-Aide compressor (model 5650D; DeVilbiss, Somerset, PA) was used to supply compressed air to the nebulizers. Flow rates of air supplied by the compressor were measured with two of each of the jet nebulizer types using a dry gas meter (DTM-115; American Meter Division, Horsham, PA). All nebulizers were operated under measured room conditions of 50 ± 10% RH and 22 ± 2° C. Unless otherwise stated, 5 units of each nebulizer type were tested under each set of experimental conditions, and manufacturer lot 1352 of dextran was used. Nebulization was stopped when there was a pause of more than 15 s without aerosol production.

Each nebulizer was tested using procedures described previously by Finlay and coworkers (7), and outlined briefly here. The aerosol path and measurement procedure were designed so that the measured aerosol properties were unaffected by the measurement process. For the LC STAR and Sonix 2000 nebulizers, this involved connecting a breath simulator to the mouthpiece of the nebulizer and collection of "inhaled" aerosol onto a spun polypropylene filter made from 3M Filtrete (Marquest No. 303; Marquest Medical Products). For the Acorn II and T-Updraft II nebulizers, amounts of inhaled aerosol were inferred directly from total amounts of dextran nebulized, using the inhalation:exhalation ratios in the breathing patterns given below. For all nebulizers, initial and final amounts of dextran were determined from pre- and postnebulization measurements of concentration, using freezing point osmometry (Precision Systems, Natick, MA), and of mass.

The breath simulator consisted of an in-house computer-controlled piston and was used with the following tidal breathing patterns. An adult cystic fibrosis pattern was developed on the basis of data in the literature on breathing patterns of patients with cystic fibrosis (9). This pattern consisted of a tidal volume of 0.62 L, inhalation:exhalation time ratio of 1:1.3, 18 breaths/min, and sine wave shapes for the inhalation and exhalation cycles. Two pediatric breathing patterns were also tested with the LC STAR nebulizer. These simulated a 4-yr-old child (square wave, tidal volume of 0.23 L, inhalation flow rate of 11 L/min), and an 8-yr-old child (square wave, tidal volume of 0.32 L, inhalation flow rate of 13 L/min), based on data in the literature (15- 17). A second adult breathing pattern consisting of a square wave with tidal volume of 0.75 L, 12 breaths/min (inhalation flow rate of 18 L/ min) was used for interlot testing of dextran with the LC STAR nebulizer and dextran manufacturer lots 1352, 2750, 2758, and 2764.

Droplet size measurements of the aerosols exiting the nebulizers were done using phase Doppler anemometry (Dantec Electronics, Mahwah, NJ). For the two vented nebulizers (LC STAR and Sonix 2000), droplet size measurements were made during the simulated tidal breathing by using an enclosed, in-line optically clear sizing region consisting of a cylinder with flat optical windows, as described by Prokop and coworkers (18). For the two unvented nebulizers (Acorn II and T-Updraft II), droplet size measurements were made using the procedure of Stapleton and colleagues (19) by removing the T-mouthpieces and measuring the aerosol exiting the nebulizer prior to exposure to any ambient air.

Estimates of regional lung deposition of the aerosol inhaled from the LC STAR were made using a mathematical lung deposition model similar to that described by Finlay and colleagues (7), which compares well with in vivo  scintigraphic measurements on normal subjects (20, 21). In this model, a symmetrically branching model of the lung is used based on the analysis of morphometric data given by Phillips and coworkers (22) for the conducting airways and by Haefeli-Bleuer and Weibel (23) for the alveolar region. Denoting the trachea as generation 0, the conducting airways of this lung model occupy generations 0-14, and the alveolar region occupies generations 15-23. The lung model was scaled to give pediatric lung models for children, 4 and 8 yr old, using the procedures of Hofmann and colleagues (24). With this procedure, the 4-yr-old lung model has one less generation of alveoli than the adult and 8-yr-old lung models. The diameters and lengths of the generations of the different lung models are given in Table 1. Lung volumes of the three models are 3,000 ml for the adult, 1,135 ml for the 8-yr-old, and 529 ml for the 4-yr-old. Amounts of aerosol depositing in each generation of these lungs with the above-described breathing patterns were estimated using a one-dimensional Lagrangian approach and the equations of Chan and Lippmann (25) for inertial impaction, those of Pich (26) and Heyder and Gebhart (27) for sedimentation, and those of Gormley and Kennedy (28) for diffusion. Deposition in the mouth-throat was estimated using the equations of Rudolf and colleagues (29). Hygroscopic effects were not included because of the high aerosol mass fraction produced by the LC STAR, which is sufficient to eliminate such effects. Indeed, the parameter  proposed by Finlay (30) to estimate the importance of such effects was well within the range where hygroscopic effects are negligible (  3), and simulations done with inclusion of two-way coupled hygroscopic effects differed by < 3% in predictions of regional dosages from values obtained with such effects neglected.

Estimates of the average concentration of dextran in the mucus immediately after completion of aerosol delivery were made by combining the above-described deposition estimates with estimates of mucous volume within each airway generation, assuming uniform disposition of depositing dextran and mucus within each generation. Mucous volume in each lung generation was determined from mucous thickness estimates with the airway lengths and diameters given in Table 1. Mucous thickness was estimated using mass conservation and a model of generational mucous velocities for which values of the tracheal mucous velocity and the mucous volume production rate were specified. For the adult model, tracheal mucous velocities of 5 and 15 mm/min and a mucous production rate of 10 ml/d were used, which are representative of values in the literature (31). For the two pediatric ages, tracheal mucous velocities of 5 and 15 mm/min were used with mucous production rates obtained by scaling the adult value by body weight (giving 2.6 and 4 ml/d for the 4- and 8-yr-old child models, respectively). Mucous velocities were estimated using a procedure similar to that described by Cuddihy and Yeh (34), Yu and colleagues (35), and Yu (36). With this procedure, the mucociliary clearance of the tracheobronchial region is treated as a series of "escalators." Using a deposition model to estimate amounts depositing in each generation, the mucous velocities, assumed constant in each generation, are estimated in order to match the clearance rates of in vivo deposition and clearance studies. Here we used our above-described regional deposition model to estimate generational deposition of the various radiolabeled aerosols inhaled in the in vivo study of Stahlhofen and coworkers (37). Generational mucous velocities were then determined in order to match the average in vivo clearance rates given by Stahlhofen and coworkers (37). It is possible to estimate the time needed to clear the kth generation, for example, by taking the deposited amount predicted by the deposition model for this generation and using the clearance curve to obtain the time needed to clear the kth fast-cleared fraction corresponding to the same amount. The generational mucous velocity is then estimated as U_k = L_k/_k, where L_k is the length of the kth generation and _k is the time needed to clear this generation. The mucous thickness follows by calculating the annular cross-section that corresponds to the same mucous volume flow rate as prescribed at the trachea. The mucous thickness of the most distal tracheobronchial generation (generation 14) was set equal to that estimated for generation 13.

Statistical analysis of the results was done using ANOVA (SYSTAT; SYSTAT, Evanston, IL), with Tukey highest significant differences (HSD) multiple means tests. Differences were deemed significant at a p value of 0.01, unless otherwise stated.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Values of viscosity and surface tension for different dextran concentrations are shown in Figures 1 and 2. Surface tension varies only marginally with dextran concentration (varying by < 6%) and differs little from that of water (73 mN/m), although these differences are significant (p < 0.01). The viscosity of the dextran solutions shown in Figure 2 is significantly greater than that of water (1.0 mPa · s). Viscosity varies dramatically with concentration over the range tested, and these variations are significant (p < 0.01). Viscosity is also significantly higher at 18° C than at 23° C (Student t test: p < 0.01).

View larger version (18K): [in this window] [in a new window]   Figure 1.   Surface tension of solutions of dextran (lot 1352) in isotonic saline at various dextran concentrations. Error bars are not shown since they are smaller in size than the symbols.

View larger version (20K): [in this window] [in a new window]   Figure 2.   Viscosity of solutions of dextran (lot 1352) is shown at various dextran concentrations. Error bars represent the standard deviation.

Flow rates from the compressor were 5.6 L/min for the Acorn II, 5.7 L/min for the T-Updraft II, and 4.7 L/min for the LC STAR.

Amounts of dextran inhaled, nebulization times, aerosol mass median diameter (MMD), and geometric standard deviation (GSD) are shown at various dextran concentrations in Table 2 for a nebulizer volume fill of 2.5 ml with the adult CF breathing pattern. Similar data are shown in Table 3 for a nebulizer volume fill of 4 ml. Data for the Sonix 2000 is not included in Table 3 because two of the Sonix nebulizers became inoperational after nebulization of 200 mg/ml with the 2.5-ml volume fill, and further testing with these units was not done to avoid possible breakdown of additional units at these high dextran concentrations.

At each dextran concentration and nebulizer fill volume in Tables 2 and 3, the nebulization times, MMD, and GSD of the LC STAR are less than for the other nebulizers. These differences are significant (p < 0.01).

Average nebulization times for all devices at both volume fills are significantly longer (p < 0.01) at 400 mg/ml than at either 100 or 200 mg/ml, but do not differ significantly between 100 and 200 mg/ml. Nebulization times are also significantly longer (p < 0.01) for all delivery systems at all concentrations when using the 4-ml volume fill compared with the 2.5-ml volume fill.

Mass median diameters and geometric standard deviations were not significantly affected by either volume fill or dextran concentration except for the LC STAR, where MMDs were significantly smaller at 400 mg/ml (p < 0.01), and the T-Updraft II, where GSDs were significantly larger at 400 mg/ml with a 4-ml volume fill.

The amount of dextran inhaled as a percentage of that placed in the delivery device decreased significantly between 200 and 400 mg/ml (p < 0.05), but did not differ significantly between 100 and 200 mg/ml except for the Sonix 2000 (where significantly less was delivered at 200 mg/ml, and data were not taken at 400 mg/ml). At each dextran concentration, the percentages of inhaled dextran for the three jet nebulizers were not significantly affected by volume fill. Although significant differences exist in amounts of inhaled dextran between different nebulizer types, these differences are not significant among the three jet nebulizer types for a 2.5-ml volume fill at dextran concentrations of 100 and 200 mg/ml.

Interlot variations in the amounts of dextran inhaled (0.75-L tidal volume, 18-L/min inhalation/exhalation flow rate) with four different dextran lots are shown in Figure 3 for the LC STAR, using a 2.5-ml volume fill and various dextran concentrations. Interlot differences are not significant (p > 0.01) at each concentration  200 mg/ml shown in Figure 3, but are significant at 300 mg/ml.

View larger version (19K): [in this window] [in a new window]   Figure 3.   Amount of dextran inhaled (0.75-L tidal volume, square wave 18 L/min inhalation/exhalation flow rate) as aerosol with the LC STAR for four different dextran lots at various dextran concentrations. Error bars show the standard deviation (n = 5).

Results from further testing of LC STAR nebulizers with the 4-yr-old and 8-yr-old breathing patterns are shown in Table 4 (using dextran lot 1352) at a dextran concentration of 200 mg/ml. Data from Table 2 for the adult case are repeated here for the reader's convenience. Mass median diameters did not differ significantly between the three ages (p > 0.05). Nebulization times are significantly longer for the 4-yr-old breathing pattern (p < 0.05), but did not differ significantly between the two older age breathing patterns. Amounts of inhaled dextran increased significantly with age (p < 0.01).

Table 4 also shows amounts of dextran predicted to deposit in the different regions of the mathematical lung models. Estimated amounts depositing in the alveolar and tracheobronchial regions of the mathematical lung models varied significantly with age (p < 0.01), but amounts in the extrathoracic (mouth-throat) region did not (p > 0.05).

Concentrations of dextran in the mucus immediately after aerosol exposure as estimated with our mucous model are shown in Figure 4 for the two assumed tracheal mucous velocities (5 and 15 mm/min) and the assumed mucous production rates given earlier in METHODS. Error bars are not shown, because they are smaller than the symbol sizes in Figure 4. The estimated mucous concentrations increase significantly with age and decrease significantly with mucous velocity (p < 0.01). Mucous concentrations also decreased in nearly direct proportion with mucous production rate. For example, increasing mucous production rate by a factor of 10 reduced dextran concentrations on average by a factor of 9.6 in the adult model and 9.3 in the 8-yr-old model.

View larger version (23K): [in this window] [in a new window]   Figure 4.   Estimated dextran concentration in the mucus immediately after nebulization for the different generations of the conducting airways with the three mathematical lung models and LC STAR nebulizer, using 2.5 ml of 200-mg/ml dextran (lot 1352) in isotonic saline placed in the nebulizer. Data for two assumed tracheal mucous velocities (5 and 15 mm/min) are shown.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our results show that as dextran concentrations increase above 200 mg/ml, nebulization is adversely affected. In particular, nebulization times increase dramatically (e.g., almost doubling between 200 and 400 mg/ml for the T-Updraft II and a 4-ml volume fill), nebulizer efficiencies decrease, where nebulizer efficiency is defined as the amount of inhaled dextran as a percentage of that placed in the nebulizer (e.g., the Acorn II delivers almost four times as much dextran at 200 mg/ml compared with 400 mg/ml), and interlot variations in dextran formulation begin to cause significant interlot differences in amounts of inhaled dextran. For these reasons, the use of dextran concentrations greater than ~ 200 mg/ml for nebulizer delivery does not appear to be warranted.

The dramatic adverse changes in nebulizer behavior that occur at dextran concentrations greater than 200 mg/ml appear to be the result of the rapid increase in viscosity of the dextran solutions that occurs at these concentrations, since surface tension varies only slightly with dextran concentration, while the viscosity nearly triples from 200 to 400 mg/ml to a value that is six to seven times that of water, as seen in Figures 1 and 2. Failure of ultrasonic nebulizers to operate with such viscous liquids has been observed by other authors (see Reference 38), and the increase in viscosity at the higher concentrations is the probable explanation for the failure of the Sonix nebulizers.

Viscous effects also appear to be responsible for the differences in amounts of dextran inhaled from the LC STAR at the three different ages. In particular, the design of this breath- enhanced nebulizer is such that the lower minute volumes at the younger ages lead to lower aerosol output rates. This causes nebulization times to increase at the younger ages. This in turn leads to enhancement of the concentration effect owing to evaporation seen with jet nebulizers (39), so that higher dextran concentrations occur over more of the nebulization time with the younger ages. This results in more viscous droplets because of the effect of dextran concentration on viscosity (see Figure 2), leading to increased holdup of drops on the nebulizer walls and baffles due to decreased ability of droplets to flow back into the nebulizer solution. Increased holdup of dextran on the interior walls of the device results in increased wastage of dextran at the end of nebulization, and less delivered as aerosol, as indeed seen in Table 4. This explanation also implies that the absence of increases in viscosity with increases in concentration should remove this effect. This is observed to be the case, since data we obtained using isotonic saline alone (n = 3) showed no significant decreases (p > 0.01) in amounts of inhaled salt between the adult CF breathing pattern and either child breathing pattern with the LC STAR and a 2.5-ml volume fill. Indeed, average inhaled amounts with the LC STAR were significantly larger for isotonic saline alone than with dextran (as a percentage of that placed in the nebulizer, averaging  30% with isotonic saline alone), also because of the described viscous effects.

A similar effect involving increased holdup due to viscosity can be used to explain the lack of expected improvement in nebulizer efficiency with 4- versus 2.5-ml volume fills. Because nebulization times are considerably longer ( 16 min) when volume fills of 4 ml were used instead of 2.5 ml, and because nebulizer efficiency was not significantly different between the two volume fills, a volume fill of 2.5 ml is preferable in order to give reasonable expectations of patient compliance.

Because the LC STAR was the only nebulizer that gave nebulization times  12 min at all concentrations up to 200 mg/ml with a 2.5-ml volume fill, and since none of the other nebulizers delivered significantly more inhaled dextran than the LC STAR with this volume fill at concentrations up to 200 mg/ml, the LC STAR is the logical choice of delivery system for the dextran formulations considered here.

The in vitro experiments of Barghouthi and coworkers (3) suggest that dextran mucous concentrations of approximately  40 mg/ml are optimal in reducing bacterial adhesion, although significant reductions in adhesion are observed at dextran concentrations as low as a few milligrams per milliliter. Furthermore, the in vitro studies of Feng and colleagues (1) indicate that effective mucolysis (significant reduction in mucous viscoelastic modulus) occurs in the concentration range 4-40 mg/ml. Using the LC STAR nebulizer at dextran concentrations of 200 mg/ml and a 2.5-ml volume fill, the results from our mathematical models (Figure 4) show that mucous concentrations  40 mg/ml occur throughout most of the tracheobronchial airways in our 4-yr-old lung model. Dextran mucous concentrations less than 40 mg/ml are predicted in our 8-yr-old and adult lung models with this delivery system, although concentrations  4 mg/ml are predicted throughout much of the conducting airways, so that some antiadhesive and mucolytic effects might still be expected at these two older ages. Increased mucous production rates would be expected to lead to decreased mucosal dextran concentrations, as indeed is observed in the results with our model, but even a factor of 10 increase in mucous production rate gives concentrations  4 mg/ml throughout much of the 4-yr-old lung model. If higher concentrations than those obtained with a 2.5-ml volume fill are desired, our results with 4-ml fills indicate that the use of larger volume fills should result in nearly directly proportional larger mucosal concentrations, provided the longer nebulization times are acceptable.

It should be noted that our mucous model provides an estimate of mucous concentrations only immediately after aerosol treatment, since it does not include the effect of clearance or absorption on dextran mucous concentrations. However, neglecting clearance and absorption is expected to be reasonable in estimating initial dextran mucous concentrations, since the time constants for clearance and absorption (32, 40) of the dextran used here (nominal molar mass, 4,000 g/mol) are expected to be much longer than the aerosol delivery times (10 min) for the LC STAR nebulizer with which this model was used. The model may also be useful in aiding delivery system design for other new therapeutic agents, such as antibiotics, where efficacy occurs via immediate onset of local activity in the airway surface fluid.

Because our lung models do not include any morphological or physiological changes in the lung associated with the presence of cystic fibrosis, our model calculations must be interpreted with due care. In particular, increased mucous production rates and slower mucus transport rates will decrease mucosal dextran concentrations, as is clearly seen from our results at the lower tracheal mucous velocities and higher mucous production rates given in Results. However, our results do provide indications that the amounts of dextran delivered to the airways using the LC STAR nebulizer/Pulmo-Aide compressor with dextran at 200 mg/ml and a 2.5-ml volume fill are enough that this delivery system is appropriate for consideration in future clinical trials with patients whose tracheal mucous velocities and mucous production rates are similar to those considered here. Because of concerns that this delivery system may give optimal dextran concentrations only in younger children, consideration of delivery systems that yield higher concentrations of dextran in the mucus (without increasing treatment times) would be worthwhile. Note, however, that this delivery system is already delivering 20-30 mg of dextran to the conducting airways according to our lung models (and total inhaled amounts of dextran near 100 mg). These are doses that are beyond the capabilities of many aerosol delivery systems (such as typical dry powder or pressurized metered dose inhalers).

Summary

The ability of nebulizers to deliver aqueous dextran to the lung as an inhaled aerosol was examined. Experimental methods were used to characterize the aerosols inhaled during simulated tidal breathing. Nebulization deteriorated significantly at dextran concentrations greater than 200 mg/ml, resulting in excessive nebulization times and reductions in the amount of inhaled dextran caused by the high viscosity of these dextran solutions. Volume fills of 4 ml were not associated with higher nebulizer efficiency compared with 2.5 ml, but instead gave excessive nebulization times. At 200 mg/ml and a 2.5-ml volume fill, the Pari LC STAR had significantly shorter nebulization times and smaller droplet sizes than the other delivery systems, while delivering similar amounts of dextran.

Mathematical lung deposition models were combined with mucus thickness models and the above-described experimental measurements of inhaled aerosols to estimate mucosal dextran concentrations with the Pari LC STAR nebulizer for a 4-yr-old child, an 8-yr-old child, and an adult. Mucosal dextran concentrations are predicted to be  40 mg/ml with this delivery system in our 4-yr-old model. Such concentrations are associated with optimal positive effects in previous in vitro bacterial adhesion experiments. In our 8-yr-old and adult models, significantly lower mucosal concentrations are predicted, although these values are  4 mg/ml over much of the airways and are still associated with significant positive effect in vitro. Whether these results translate into in vivo efficacy with this delivery system remains for future clinical trials to determine.