INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE CONCEPT, that particles deposited on the conducting airways are removed by mucociliary clearance within 24 h after deposition, has been an accepted tenet in our understanding of particle deposition and clearance (1, 20, 26, 32, 52). Recent evidence, however, suggests that not all particles deposited in the conducting airways are cleared rapidly by mucociliary action. A series of human studies indicates prolonged retention of particles deposited on the airway epithelium (18, 45, 54, 56). These studies are based on a special aerosol delivery technique, the inhalation of a radiolabeled aerosol bolus into shallow lung depths, which delivers aerosol particles only to airways but not to the peripheral lung. However, in these studies, there was no direct visualization of the particles retained in airways. Studies using experimental animals have shown that particles could be ingested by airway macrophages and thus retained for >24 h (15, 16). Furthermore, small fractions of particles found in the airways outside of macrophages many weeks after inhalation have been reported (36, 57). A potential mechanism of airway retention of particles is suggested by in vitro studies in which particles deposited on top of a mucous gel covered with a surfactant film are likely to be pulled into the gel phase due to gradients of surface tension forces at the interfacial layer (13). The question of whether particles are rapidly and completely cleared from conducting airways or whether there is substantial long-term retention of particles in the airways has implications for a number of human diseases, most notably bronchogenic lung cancer (4, 66).

The aerosol bolus inhalation technique was introduced by Altshuler et al. (2) and has later been used successfully to test the first-in-last-out principle in respiratory physiology in humans (18, 45, 54) and dogs (50). An air volume inhaled at the beginning of a breath penetrates deeply into the lungs and is recovered late during exhalation, whereas an air volume inhaled at the end of a breath penetrates only slightly into the lungs and is exhaled first. It is therefore expected that particles deposited from an aerosol bolus penetrating into a volumetric lung depth (VLD) of the respiratory system that is less than the dead space volume are located in conducting airways. However, because of convective air mixing, the particles are not confined only to the bolus air while the bolus is transported through the respiratory system. The particles may be convectively transported into air adjacent to the bolus air, resulting in a more dispersed bolus on exhalation, with particles distributed in a much larger volume in the exhaled air than in the inhaled air (18).

Therefore, convective air mixing transports particles deeper into the lungs than convective bulk transport so that the true sites of particle deposition can be different from those anticipated. It is therefore conceivable that long-term particle retention observed in the human lung after shallow aerosol bolus inhalation, and interpreted as prolonged retention of particles in ciliated airways, is actually retention of particles in nonciliated peripheral air spaces.

In this study, shallow aerosol bolus inhalation is used in dogs with radiolabeled or fluorescence-labeled monodisperse aerosols to both replicate the human studies noted above and to define, by subsequent microscopic analysis, the anatomic location of retained particles in the lungs. Because single-breath washout curves are the same in canine and human subjects, it was assumed that dogs would be well suited to this study (25, 28, 49, 51, 67). If a pattern of delayed clearance could be demonstrated with shallow bolus methodology in dogs, it would then be possible to analyze the location of the retained particles morphometrically and to identify the spatial distribution of retained particles in all anatomic sites, including different sizes of airways and alveoli.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Experimental Methodology Considerations

To carry out these experiments, a sufficiently high number of radiolabeled, red fluorescence, and green fluorescence, 2-µm monodisperse particles were delivered in shallow boluses to permit their quantification anatomically at sites of retention. (The term shallow bolus inhalation will be used for a fractional volume containing aerosol particles inhaled at the end of the tidal volume.) Radiolabeled particles provided retention kinetics and gross distribution, whereas red and green fluorescent particles delivered 96 and 24 h before euthanasia yielded the anatomic site of retention. Two-micrometer-sized particles were used to minimize impaction and provide sufficient settling during the breath hold. A deep aerosol bolus inhalation using blue fluorescence-labeled 1-µm monodisperse particles served as a marker for the peripheral lung to further control the applied morphometric analysis. One-micrometer-sized particles were used for the deep bolus to ensure negligible settling and impaction in airways while ensuring adequate settling in alveoli. (The term deep bolus will be used to indicate a fractional volume containing aerosol particles inhaled at the beginning of the tidal volume.) Definition of deposition and retention utilized external gamma radiation emission detection and imaging techniques (10, 21, 23, 33), macroscopic radiation distribution methodology at necropsy (59-62), and microscopic quantitative morphology for the ultimate analyses of particle localization.

Special emphasis was also placed on the technique of lung tissue fixation in these studies. Because liquid fixation by either intratracheal instillation or intravascular perfusion had the possibility of displacing particles from their location of retention, physical fixation methods were employed in which the lungs were microwave-dried during inflation at total lung capacity (59-62). In morphometric analysis, stratified random sampling was used with a microdissection technique for analysis of airways > 1 mm in diameter as well as stratified random sampling of sections of peripheral lung tissue to distinguish particle location between airways < 1 mm in diameter and alveoli.

Design of the Study

View larger version (90K): [in this window] [in a new window]   Fig. 1.   A: flow diagram of 48-h, radiolabeled-particle, shallow bolus clearance study. B: flow diagram of comprehensive, 96-h, shallow and deep bolus deposition and clearance study. 99mTc-PSL, technetium 99m-labeled polystyrene particles; SPECT, single-photon-emission computed tomography.

The second and main part took place several months later. A schematic flow chart (Fig. 1B) is provided to clarify this design. Four days (96 h) before euthanasia, each dog inhaled shallow boluses of red fluorescent PSL. Twenty-four hours before sacrifice, each dog inhaled shallow boluses of green fluorescent PSL and radiolabeled ^99mTc-PSL. At the same time, blue fluorescent 1-µm PSL were delivered via deep aerosol bolus inhalation. Lung retention of ^99mTc-PSL from the time of inhalation until sacrifice 24 h later was followed. For the first 2-3 h, retention was determined from the gross particle distribution as imaged by a gamma camera and later by external gamma counting by using a more sensitive gamma spectroscopy lung counter (10, 21, 23).

Experimental Setup for Aerosol Bolus Delivery

The bolus valve system consisted of four electropneumatically driven valves supplying two pathways of ventilation (Fig. 2). One pathway was used to maintain normal ventilation, and the other for delivery of a 30-cm³ bolus. The aerosol bolus volume was inserted into the inhaled air at a preselected volume by computer-controlled switching of the valve system. To minimize aerosol particle deposition and subsequent reentrainment of the deposit during the next firing of the valve system, the aerosol valves were constructed with an actuated center-bore rod that provided an undisturbed aerosol flow without edges and bends.

View larger version (59K): [in this window] [in a new window]   Fig. 2.   Schematic diagram of experimental aerosol generation system and experimental system for measurement of pulmonary volumes and bolus delivery of labeled particles to canine lungs.

Aerosol Generation

3

3

To define particle deposition and convective mixing in the dogs (51), an aerosol of di-(2)-ethylhexyl sebacate (DEHS) was generated from a condensation aerosol generator (Mage generator, Lavoro e Ambiente, Bologna, Italy) and delivered as a 30-cm³ bolus without a breath hold in the breathing maneuver. These particles had an aerodynamic diameter of 2.5 µm and were used with photometry measurements.

Animals and Anesthesia

Each dog was anesthetized and paralyzed by intravenously delivered pentobarbitone (20 mg body wt) and alcuronium (0.2 mg/kg body wt) and maintained in a supine position. Intubation was via the oropharynx with a low-pressure cuffed endotracheal tube with an inner diameter of either 9 or 10 mm depending on the dog's tracheal size. The endotracheal tube was placed carefully such that it penetrated 6 cm into the trachea. The tube was connected to a servo ventilator as described earlier (50). The bolus valve system and the aerosol photometer were placed between the servo ventilator and the tracheal tube. Tidal volume and breathing frequency of ventilation were adjusted for each dog such that physiological values of the monitored end-expiratory carbon dioxide and oxygen levels were maintained. Similarly, oxygen saturation, measured throughout each experiment via pulse oximetry (Pulsox-Probe SP7, Minolta, Osaka, Japan) at the shaved tip of the tail, was maintained at >93%.

After the ^99mTc-PSL aerosol bolus delivery, anesthesia was maintained for about three more hours until the SPECT (single-photon-emission computed tomography) scan (see Clearance Kinetics Using Radioactive Test Particles) was finished. During that time, ventilation of the dog was maintained by a pressure-driven ventilator (Bird ventilator, Bird, Palm Springs, CA). After the SPECT scan, the dog was allowed to recover, and ventilation was maintained until the dog was breathing spontaneously, so that the endotracheal tube could then be removed.

Determination of Lung and Airway Volumes

Series dead space volumes for He and SF₆ were determined by the single-breath washout technique (11). After equilibration of the lung with 1% He and 1% SF6 in air, a standardized single-breath maneuver was performed. After passive expiration, a test-gas-free air volume was inspired up to 80% of TLC at a flow rate of 300 cm³/s. Without a postinspiratory pause, expiration was performed to well below FRC at the same flow rate. The slope of phase III of the expirogram was measured by least squares regression analysis and was used to determine the series dead space volume. The volume of phase I was also determined from the expirograms.

Aerosol Bolus Inhalation

Aerosol deposition, convective mixing (bolus dispersion), and residual particles present in exhaled volumes at end expiration were determined from aerosol photometry in front of the endotracheal tube (51).

Shallow bolus inhalation. The red 2.5-µm PSL, the green 2.2-µm PSL, and the radioactively labeled 2.4-µm PSL particles were delivered to a shallow lung depth (shallow bolus), which was scaled to the individual airway volume of the dog, i.e., to the volume of phase I of the He-expirogram. The volumetric front of the shallow bolus was varied among the different dogs of the study, ranging from 45 to 85% of phase I. Although essentially no particles were detectable by photometry at the end of the expiratory volume of each breath, three subsequent breaths of clean air were given, and then the next bolus was delivered at the exact same volume in the breathing maneuver. Each dog usually inhaled 50 boluses within 25-30 min.

Deep bolus inhalation. The 1-µm blue PSL particles were delivered to the lung periphery. The front depth of the bolus was set to three times the series dead space of He (deep bolus), corresponding to a VLD of 350-500 cm³. Inspiration was again performed to 80% of TLC at a flow rate of 300 cm³/s. For this bolus there was a 10-s breath hold before exhalation. After three subsequent breaths of clean air, the next bolus was delivered. Fifty aerosol boluses during 20-25 min were delivered for this test aerosol, as was done for the others.

Clearance Kinetics Using Radioactive Test Particles

After the SPECT measurements, total retention measurements were continued with a gamma spectroscopic lung counter (10) because it had 30-fold higher sensitivity than the gamma camera. The collimated detectors of the lung counter were positioned so that ^99mTc-PSL lung retention was measured without significant interference from ^99mTc-PSL either accumulated at the cuff of the endotracheal tube or swallowed into the gastrointestinal tract once the tube was removed. The ^99mTc-PSL activity A(t) was corrected for decay, background, and sensitivity of the two gamma counters. Activity data were normalized to the initially deposited activity A(0). Lung retention data were fitted by the following function of two exponential terms to distinguish between fast and slow clearance

(1)

where L_i (i = 1,2) are the fast and slow cleared fractions with (L₁ + L₂) = 1, and _i (i = 1,2) are the clearance rates. The initial long-term retention, L₂, is referred to as the A-value (54).

Morphometric Analysis

For tissue sampling, the dried lung was sliced base to apex, after a random start, in alternating 10- and 1-mm-thick slices.

Measurement of 24- and 96-h particle retention in airways > 1 mm and mixed parenchyma. Morphometric analysis is shown schematically in Fig. 3. The 1-mm slices were used to measure 24- and 96-h retention in the parenchyma and airways of known size. By using a dissection microscope, the slices were microdissected into four compartments: airways with a diameter > 5 mm, 2.5- to 5-mm airways, 1- to 2.5-mm airways, with the fourth compartment (mixed parenchyma) being the remaining tissue composed of airways <1 mm and parenchyma. Because of the extensiveness of this dissection, only the tissue of every third slice was sampled, resulting in five slices to be dissected and analyzed. To reduce the risk of dislodging particles during dissection, several blades per slice were used, and forceps were frequently cleaned by using an ultrasonic bath. Two methods were used to measure retention in the four separate compartments of each slice. First, the radioactivity from retained ^99mTc-PSL particles in each compartment was counted in a well counter. The retained fraction in each compartment was calculated as the ^99mTc-activity of PSL found in that compartment normalized by the total ^99mTc-activity found in all 1-mm slices.

View larger version (71K): [in this window] [in a new window]   Fig. 3.   Schematic diagram of morphometric studies to quantify particles in small airways and alveoli.

(2)

Measurement of 24- and 96-h particle retention in the parenchyma and airways < 1 mm. Morphometric analysis is shown schematically in Fig. 3. The 10-mm lung slices were used to determine 24- and 96-h retention in the smallest structures of interest: the parenchyma and small airways with diameters < 1 mm. The number of particles in each slice was estimated from radioactive counts from 24-h retained ^99mTc-PSL. The regional distribution of the 24-h retained ^99mTc-PSL should closely match the distribution of the 24-h fluorescent 2-µm PSL, and possibly the 96-h fluorescent 2-µm PSL, because all these particles are of comparable size and were delivered to the lungs in the same manner. Thus those slices with more radioactive particles per unit volume were assumed to also have more fluorescent particles per unit volume. Radioactivity of ^99mTc in each 10-mm slice was expressed by the evenness index (EI). To correct for differences in slice volume, the radioactivity A_i per unit volume V_i of slice i was normalized to the mean radioactivity (A_i)/(V_i) per unit volume of the whole lung according to the following equation

(3)

Thus a slice with an EI = 1 would have the average amount of activity for the whole lung (60).

(4)

Estimate of PSL Surface Density in Small-Airways Compartment and the Parenchyma

Estimate of PSL surface density in small-airways compartment. To estimate the surface area of the small airways between 0.3- and 1-mm diameter, we assume that all airways are cylinders of circular cross section. If a cylinder is cut arbitrarily through its jacket, the shortest elliptical axis is the diameter d_a of the cylinder. Orthogonal to d_a is the elliptical axis a_a. If  is the angle at which the cylinder is tilted against the plain perpendicular to the cutting plane of the section, then a_a becomes

(5)

Similarly the length l_a of the jacket of the tilted cylinder is related to the analyzed depth h = 0.145 µm of the section

(6)

One then can show that the surface area M_a of the jacket of the tilted cylinder is

(7)

in which U_a is the perimeter of the circular cross section of the cylinder U_a =  d_a. Substituting Eqs. 5 and 6 into 7, the surface area of the cylinder jacket becomes

(8)

The surface area of the small airways is overestimated because it includes the blood vessels. In beagle dogs (63), the volume density of blood vessels vs. that of bronchi including nonrespiratory bronchioli was determined to be 6.9 ± 1.9 and 5.3 ± 0.8%, respectively, of the lung volume. Presuming a similar distribution of blood vessels and bronchi-to-bronchioli ratio over the whole size range and no PSL retention in blood vessels, the number of tubes analyzed between diameters of 0.3 and 1.0 mm represents 57% blood vessels and 43% nonrespiratory bronchioli.

Estimate of PSL surface density in parenchyma. The alveolar volume of each section V_as was determined from the planar area of the field of view S_fov, the number of fields of view n_fov per section, the analyzed depth h minus the volume of all tubes  V_t in this section

(9)

Then, the alveolar volumes of all sections were summed. Because we have not estimated the alveolar surface density in this study, the mean value of the alveolar surface density SD_a = 600 cm²/cm³ was taken from Refs. 12 and 64. The alveolar surface area SA of each section is then the product of SD_a and the analyzed volume V_as. The PSL surface density in alveoli was obtained by the total number of PSL observed in the alveolar region of all sections divided by the total alveolar surface area measured SA of all the sections.

Statistical Analysis

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Aerosol Bolus Inhalation

The fraction of particles present in expiratory volume after after VLD was 2.4 × 10⁵ ± 3.4 × 10⁴, which is the mean fraction obtained from all eight dogs. This low fraction indicates the noise level of the aerosol photometer and the fact that no particles were contained in this part of the expired breath after the 20-s breath hold.

Tracheal Deposition at the End of the Endotracheal Tube

Particle Clearance Kinetics

View larger version (43K): [in this window] [in a new window]   Fig. 4.   Two kinetics of 99mTc-PSL clearance of dogs 1-4 determined 2 mo apart. A: 48-h clearance study. B: 24-h clearance study. Both sets of data were fitted with a function of 2 exponential terms. Initial measurements during first 2 h were performed by using gamma camera; the rest of the time, retention was followed by lung counter.

Overall, the clearance patterns in both studies were quite similar. Dog 3, in the 24-h study, had the most shallowly delivered bolus, and therefore, the retained fraction was distinctively the lowest. Slow particle clearance in all eight dogs decreased with a median half-life of 63 h, with an interquartile range of 33 (25% quartile) and 158 h (75% quartile). Although the half-life of slow particle clearance increased slightly with the A-value, there was no significant correlation. There was also no significant correlation between the VLD and the half-life of slow particle clearance.

Analysis of all 16 clearance studies on the 8 dogs showed that the A-value increased with increasing bolus penetration. In Fig. 5, all the A-values are plotted against VLD expressed as a fraction of the series dead space volume of He(R). There is a significant linear correlation (r = 0.82; P < 0.05) described by the following equation

(10)

In this relationship, an absolute difference in the front depths of ~20 cm³ predicts the A-values to range from near 0 to 0.5. In Table 2, the A-values and the retentions of ^99mTc-PSL 24 h after inhalation are given for the four dogs before morphometric analysis. The 24-h retention is presented as fractions of both the deposited particles (^99mTc-PSL activity) and the inhaled particles (determined by aerosol photometry during ventilation). Although the fraction of deposited particles varies between 24 and 50% of the inhaled PSL, the 24-h retention relative to the inhaled aerosol is 1-3% in three dogs, but dog 2 shows a value of 20%.

View larger version (18K): [in this window] [in a new window]   Fig. 5.   A-values of all 16 clearance studies obtained from 8 dogs plotted as a function of ratio of volumetric lung depth and anatomic dead space. There is a significant linear correlation (r = 0.82).

Particle Distribution After 2- to 3-h Retention Using SPECT

View larger version (44K): [in this window] [in a new window]   Fig. 6.   SPECT scan 3 h after inhalation. A: series of transversal (apex-to-base) and coronal (dorsal-to-ventral) slices of 99mTc-PSL retention in lungs of dog 4 around level of carina. Slices were calculated at a distance of 2 mm apart. Retention of 99mTc-PSL is overlaid in color scale over a ventilation scan in gray scale performed earlier. B: series of sagital (right-to-left lateral) slices of 99mTc-PSL retention in lungs of dog 4 and planar image from dorsal view of 99mTc-PSL deposit in lungs and PSL accumulation in trachea at site of ciliar obstruction because of inflation of cuff of endotracheal tube. Planar image was taken at beginning of SPECT scan 2.5 h after end of inhalation.

Distribution of Retained ^99mTc-PSL on a Macroscopic Level in Lung Slices

View larger version (53K): [in this window] [in a new window]   Fig. 7.   A: distribution of evenness indexes (EI) of all slices i (EIi) according to Eq. 3 as means ± SE of all 4 dogs. Count of slices is arranged such that slice 10 is located on 1st bifurcation for each dog. Count of slices starts at lung base. B: histogram of all EIi of all 4 dogs given as means ± SE.

Figure 7B shows the frequency histogram of EI_i as function of values of EI observed. It showed marked nonuniformity with a large number of slices with low EI. Similarly, only a few of the slices had a high EI, indicating high ^99mTc-PSL retention.

Retained Particle Fractions in Airways and the Alveolar Region

View larger version (42K): [in this window] [in a new window]   Fig. 8.   Particle retention 24 and 96 h after shallow bolus inhalation. Green PSL and 99mTc-PSL, 24-h retention; red PSL, 96-h retention after shallow aerosol bolus inhalation. Top: fractions of 99mTc-PSL, green PSL, and red PSL found in various compartments of large airways and combined compartment of small airways <1 mm plus alveolar structures, as determined from 1-mm lung slices. Bottom: fractions of retained PSL in 0.3- to 1-mm airways and in alveoli, as determined from 200-µm sections of 1-cm3 blocks of 10-mm lung slices. Each fraction is mean ± SE obtained from all 4 dogs.

The mixed parenchymal compartment (small airways <1 mm and alveolar structures) contained ~90% of the retained PSL at both 24 and 96 h. This mixed parenchymal fraction was further resolved by the analysis of 200-µm sections into the small-airway compartment with diameters between 0.3 and 1 mm and the alveolar compartment, which is shown in Fig. 8, bottom. About 55% of either retention, 24 or 96 h, was found in the alveolar compartment, whereas <35% was found in the small-airway compartment. There is no statistically significant difference (P > 0.05) between the particles retained for 24 h and those retained for 96 h in these two compartments, indicating that clearance in the intervening time between these measurements is quite slow. It also needs to be emphasized that there was no statistically significant difference in the particle fractions observed in the various airway compartments and the alveolar compartment between dog 2, which had retained 20% of the inhaled particles at euthanasia after 24 h, and the mean fractions in the various compartments found in the other three dogs, which had retained only 1-3% of the inhaled particles at euthanasia.

Figure 9 shows the 24-h retention data obtained from deep aerosol bolus inhalations by using 1-µm PSL particles, which served as a positive control for deposition and 24-h retention in the deep lung. Data were analyzed in the same manner as described above for the shallow bolus inhalations. As expected, ~96% of the retained PSL were found in the alveolar compartment. Interestingly, a very small particle fraction (0.08%) was found in the compartment of airways with diameters between 0.3- and 1-mm, whereas fractions of 3.1, 0.8, and 0.4% were found in the compartments of airways with diameters of 1-2.5 mm, 2.5-5 mm and >5 mm, respectively.

View larger version (62K): [in this window] [in a new window]   Fig. 9.   Blue fluorescent PSL retention 24 h after deep bolus inhalation. PSL fractions found in various compartments of large airways were determined from 1-mm lung slices. Fractions of retained PSL in 0.3- to 1-mm airways and in alveoli were determined from 200-µm sections of 1-cm3 blocks of 10-mm lung slices. Mean values ± SE from 4 dogs. Note value of "alveoli" bar is 0.956 ± 0.0004.

Particle Density in Small Airways and Parenchyma

2

2

View larger version (33K): [in this window] [in a new window]   Fig. 10.   PSL density in small airways and alveoli, i.e., no. of PSL per unit surface area, 24 and 96 h after shallow bolus inhalation. For comparison, PSL densities 24 h after deep bolus inhalation are shown. Densities are presented as means ± SD of all 4 dogs.

In alveoli, the 1-µm PSL density after deep bolus inhalation was almost two orders of magnitude higher than after shallow bolus inhalation (green and red PSL) because of the much higher deposition probability in the alveolar region during deep bolus inhalation. Very few blue particles were found in small airways (diameter 0.3-1 mm) after deep aerosol bolus inhalation. However, because the alveolar surface area is about two orders of magnitude larger than the surface area of small airways, the PSL density in small airways after deep bolus inhalation is on the same order of magnitude as the alveolar PSL density.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

This study clearly shows that shallow aerosol bolus inhalation into a VLD less than the series dead space resulted in predominant airway deposition of particles. A fraction of 0.5-0.95 of the deposited particles was cleared within 24 h after shallow bolus inhalation by mucociliary action. The retention pattern 2.5 h after inhalation was irregularly shaped, usually located centrally at the level of the carina but occasionally reaching as far as the perimeter of the lungs. The remaining particle fraction of 0.05-0.5 was cleared slowly in all dogs after 24 h. This slowly cleared fraction increased linearly with increasing bolus front depth. Of the slowly cleared fraction, ~40% retained at 24 and 96 h were found in airways. Most of the particles retained in the airways were found in the small-airways compartment with diameters of 0.3-1 mm. Although there was a tendency of fewer particles in the large airways (>2.5 mm) after 96 h compared with 24-h data, there was no significant difference between 24- or 96-h retention. Therefore, we conclude that slow clearance from the airways represents clearance from small airways. More than 50% of the retained particles (2.5-25% of the particles deposited at t = 0) were in alveoli at 24 and 96 h.

This study directly examined the question of airway retention of particles. It was designed to assess particle retention in airways of all sizes and used the shallow aerosol bolus approach as a means to preferentially deposit particles in a very central location. As such, it is also a test of the capability for shallow boluses to achieve this pattern of deposition, as well as a test of our ability to predict the anatomic and/or functional distributions of a labeled aerosol inhaled into a volume less than the anatomic dead space. Thus the design of this experiment was crucial. The choice of the experimental animal rested on a number of considerations. Small animals, such as rodents, were not feasible because of the need for reliable external imaging of the deposition pattern, limitations on usable volumes with the bolus delivery system, and lung geometry. Larger animals, such as canines, had advantages on the basis of the size of their respiratory tract. Anatomic and geometric design considerations of the respiratory tract in terms of similarities to and differences from humans were also considered. Single-breath washout curves in human and canine subjects are very similar and indeed are functionally the same (25, 28, 67). Similarly, in bolus studies, it appears that bolus penetration is into the same anatomic regions (48-51). However, one known difference is that measurements of shallow bolus dispersion in the anesthetized and ventilated dog show a smaller degree of convective mixing than do similar studies in humans (49, 51). Thus in the canine lungs the site of bolus deposition should be more related to bulk transport and less influenced by convective mixing, which would carry particles deeper into the lungs. Finally, experiments carried out in canine lung casts had precisely defined bolus dispersion and effect of flow rate on bolus penetration (9, 44).

The degree of ventilation nonuniformity is an important issue in these studies both in terms of differences between humans and animals as well as in the capability of bolus methodology to confine inhaled particles to the conducting airways. A major confounding effect is that ventilation nonuniformities can transport particles to lung depths deeper than the VLD, where they will readily deposit due to smaller dimensions in the deep lung. One transport mechanism operates when the lung fills unevenly due to nonuniformities in compliance (35). Another possible transport mechanism is the nonslug inspiratory flow profiles that arise from the repeated division of the gas flow at bifurcations (47, 65). In theory, the more uniform the lung ventilation and the more symmetric the lung, the better the ability of the bolus method in confining the inhaled particles to the desired VLD within the conducting airways. A symmetric lung has dichotomous branching equal-caliper airways, a geometry that has an equal path length (and volume) from the larynx to each airway of the same generation. Because these ideal geometric conditions are not present in the human and in the monopodial canine airways (17, 19, 69, 72), the lungs of both species do exhibit nonuniform ventilation (25, 28, 67). Thus some particles may reach alveolar air spaces, although the VLD is smaller than the anatomic dead space volume of the airways. Because we directly quantified the location of particles, the important fact is the observed slowly cleared fraction in airways and not so much this alveolar penetration of particles, which may be specific to the canine species, and/or due to specific differences in the experimental protocol between the human and the canine bolus inhalation.

The findings of this study suggest that, in healthy large airways, mucociliary clearance quickly and effectively removes almost all particles. However, in healthy small airways, mucociliary clearance is less effective and there are even sites of longer-term retention. Mucus transport velocity is highest in the trachea and decreases with increasing generations of airways (70, 71). Measured mucus transport velocity in the human trachea is ~5 mm/min (37). In contrast, model estimates on the basis of water and ion balances suggest that the mucus transport velocity in terminal bronchioli is only ~10³ mm/min (7). Thus, for a typical terminal bronchioli that is 2 mm long, the calculated particle retention times will be >35 h. Another potential factor leading to longer-term retention is that the ciliated airway epithelium may not be continuous, particularly in small airways. Although there is agreement that the human trachea is covered by a continuous ciliated epithelium, controversy remains how these cells diminish with increasing generations of airways (27). The lack of cilia and the nature of the mucous layer in these areas need further definition. Nevertheless, if particles deposit onto nonciliated areas, it is less likely that they will be cleared by mucociliary transport. In this case, they are likely to be phagocytized by either airway macrophages (14, 15) or epithelial cells themselves (68), resulting in prolonged retention of the particles in the airways.

Another mechanism of airway retention of particles has been proposed by Gehr and co-workers (13, 14), who showed in vitro that particles deposited on top of the mucous gel covered with a bimolecular layer of surfactant are likely to be pulled into the gel phase because of gradients of surface tension forces at the interfacial layer. In addition, findings of microscopic particles among the cilia on the surface of ciliated epithelial cells in hamster airways led to the hypothesis that particles may be moved even further into the sol phase by those gradients of surface tension forces (15, 16). Particles located at the base of cilia appear to be immobilized and are likely to be retained for prolonged times. Also at this location, particles may be subject to phagocytosis by either airway macrophages or epithelial cells. The present study offers some support for these hypotheses in small airways but little support in the largest airways.

Another possible explanation for the observed longer-term retention is that particles observed in small airways were initially deposited in alveoli and then transported to small airways by alveolar macrophages. However, such a mechanism is not consistent with alveolar macrophage particle transport rates from the alveolar region and the small airways to the larynx at 24- and 96-h retention. Particle transport from the alveolar region of the dog's lungs has been estimated to be 0.002 ± 0.0002 d¹ (daily particle fraction normalized to the particles contemporarily retained in the lungs) (22, 53). Multiplying the observed alveolar-retained fraction of 0.55 by this transport rate shows the fraction of particles that could have left the alveolar region was only 0.001 at 24 h (for both shallow and deep boluses) and 0.004 at 96 h (shallow bolus). Even if no particle transport would occur in the proximal airways, these fractions are still negligible compared with the observed 24-h fractions of 0.33 in airways <1 mm in diameter and 0.15 in airways >1 mm in diameter (see Fig. 8). Also, if the PSL retention in <1-mm-diameter airways and in the alveolar region are taken together, the fractions of transitional particles in large airways would increase to 0.002, which is still negligible compared with the observed fraction of 0.15 of retained PSL.

Another approach to the question of airway penetration of particles was used in recent studies (3, 8, 58). These investigations sought to enhance particle deposition in airways proximal to the alveolar region during extremely slow inhalation of radiolabeled aerosols (50 cm³/s airflow) by healthy subjects. An intermediate, slowly cleared particle fraction (half-life 30 d¹) was seen which was not observed after inhalation at normal air flow. This fraction was found between rapid clearance within a few hours and slow clearance (half-life > 100 d¹). The observed particle fraction cleared at an intermediate, slow half-life is considered to represent particle retention in small airways and supports the findings of the human aerosol bolus studies. Our study, which anatomically defined the location of these particles in small airways, is supportive of these conclusions.

The results of this study are in conflict with the concept that, in humans, a bolus delivered to a shallow lung depth deposits predominantly in the airways and not in the peripheral lungs (45, 54). However, there has been no anatomic localization of particles in the human studies. These studies are all based on the ability of subjects to breathe in a defined pattern to ensure the replication of the protocol, the delivery of boluses to a portion of the breath corresponding to volumes predicted to represent conducting airways, and external measurements of radiolabeled particles (5, 45, 54). The study reported herein uses the same type of particles used in some of the human studies (45), so the particles are unlikely to account for any differences between human and canine studies. The differences in morphology between the human and the canine lungs have been discussed, and these are likely to account for similarities rather than differences.

Another difference between the study reported herein and human studies is the delivery of the aerosol bolus inhalation. The human subjects were breathing actively and visually guided to achieve a constant flow. The animals were anesthetized, paralyzed, and ventilated with positive pressure in a supine position. This system allowed greater control of breathing maneuvers than could be expected even in highly trained human subjects. There are no proven differences between inspiration by positive pressure and negative pressure in both species (39, 40). In dogs, ventilation by a positive servo ventilator results in a more uniform distribution of the inspired air compared with active breathing in humans (39, 41). For example, the active inspiration in humans is usually maintained by the action of the diaphragm, resulting in a predominant ventilation of the lung base. This is enhanced by the gravity-dependent filling of the lungs in the upright position (6). This implies that a shallow aerosol bolus is more likely to be delivered to the airways and parenchyma of the lung base in an actively breathing human, but it is more uniformly distributed in the airways of a ventilated dog. It is notable that these differences would suggest that particles were less likely to be transported beyond the dead space of conducting airways in the ventilated dog compared with the actively breathing human subject.

Another factor is that, in human studies, shallow bolus inhalations at flow rates of ~250 cm³/s delivered to a lung depth of 50-75 cm³ require that the flow be instantaneously stopped 0.2-0.3 s after bolus delivery. It is likely that this level of control is not achievable by an actively breathing human subject. Although closing the valve of the inhalation apparatus instantaneously stops inhalation at the preselected tidal volume, some degree of negative pressure is still created within the lung before the inspiratory muscles relax. Because in dogs a small variation in volume fraction creates a large difference in the A-value, i.e., a change of 20 cm³ in the VLD increases the slowly cleared fraction (A-value) from 0.05 to 0.5, it is possible in humans that small increases in inhaled volume, due to the short period of negative pressure created at the end of inspiration, caused redistribution of PSL among the anatomic compartments. These phenomena can certainly be excluded in the dog model, where the lungs of the paralyzed animal exactly follow the prescribed volumes of the computer-controlled ventilator system.

The effect of anesthesia on the experiment is potentially problematic. Alveolar deposition and retention fractions suggest that macrophage function was intact, and there is ample evidence in this study to suggest that ciliary function was maintained, as shown by the rapid particle transport to the obstruction point of the trachea caused by the cuffed endothelial tube (Fig. 6). In this study, oxygen saturation was maintained at a physiological level, as were the expired concentrations of oxygen and carbon dioxide. However, there is neither direct nor indirect evidence to determine whether the ability of epithelial cells to take up particles was maintained in these dogs, which could potentially explain, in part, the observed low airway retention in large airways.

Nevertheless, it must be remembered that all dogs in each of the two clearance studies show a slowly cleared particle fraction. In both humans and dogs, particle deposition patterns after shallow bolus inhalation are central and irregular as observed by gamma camera imaging (5). Although in humans the deposited particle fractions are consistently larger in the left lungs at 75% TLC bolus inhalation maneuvers, the ratio of left-to-right deposited fractions varies widely in the eight dogs studied. Furthermore, in humans the slowly cleared particle fraction of A = 0.4 for 2-µm particles is basically independent of the VLD in a range of 50-110 cm³, corresponding to 0.4-0.8 of the anatomic dead space. This is clearly different from the canine lungs, which show a linear correlation between VLD in the same range of the anatomic dead space, and the slowly cleared fraction of A ranging from 0.05 to 0.5.

Because slow clearance of particles from small airways has been shown morphometrically in this study, the relevance of this clearance phenomenon for normal breathing needs to be estimated. For deposition measurements of large aerosol volumes in beagle dogs, using the same aerosol photometry system, Schulz and co-workers (51) found a total deposited fraction 0.36 ± 0.04 of the inhaled monodisperse 2.0-µm DEHS particles after inhalation of a tidal volume corresponding to 75% of TLC. After bolus delivery of the same-size particles to a VLD of 120 and 160 cm³, they observed deposited fractions of 0.05 and 0.11, respectively. On the basis of the results of this study and by using an average cleared fraction of 85% within 24 h found in the present study, a conservative estimate of the fraction of particles deposited and slowly cleared from small airways would be 0.003-0.007 after tidal volume inhalation. Hence, ~0.8-2% of the totally deposited 2-µm particles may be subject to slow clearance from small airways. Because of the similarity of total deposition of 2-µm particles in human and canine lungs, a similar fraction of slowly cleared particles from small airways would be expected from the human lungs.

Conclusion