Pressure-flow behavior of pulmonary interstitium.

A method to measure the pressure-flow behavior of the interstitium around large pulmonary vessels is presented. Isolated rabbit lungs were degassed, and the air spaces and vasculature were inflated with a silicon rubber compound. After the rubber had hardened the caudal lobes were sliced into 1-cm-thick slabs. Two chambers were bonded to opposite sides of a slab enclosing a large blood vessel and were filled with saline containing 3 g/dl albumin. The flow through the interstitium surrounding the vessel was measured at a constant driving pressure of 5 cmH2O and at various mean interstitial pressures. Flow decreased with a reduction of mean interstitial pressure and reached a limiting minimum value at approximately -9 cmH2O. The pressure-flow behavior was analyzed under the assumptions that the interstitium is a porous material described by a single permeability constant that increases with hydration and that the expansion of the interstitium with interstitial pressure was due to the elastic response of the surrounding rubber compound. This resulted in an interstitial resistance (reciprocal of permeability constant) of 1.31 +/- 1.03 (SD) cmH2O.h.cm-2 and a ratio of interstitial cuff thickness to vessel radius of 0.022 +/- 0.007 (SD), n = 11. The phenomenon of flow limitation was demonstrated by holding the upstream pressure constant at 15 cmH2O and measuring the flow while the downstream pressure was reduced. The flow was limited at downstream pressures below -10 cmH2O.

Alveolar liquid pressure measured in the intact rabbit chest by micropuncture.

Previous measurements in isolated lung showed that alveolar liquid pressure was near the pleural pressure at a lung volume near functional residual capacity (FRC). In this study we verified that alveolar liquid pressure in vivo was similar to that of the isolated lung. In anesthetized paralyzed rabbits (3-4 kg, n = 9) ventilated with 100% O2 in the left lateral position, we made a pleural window between the fifth and sixth ribs near midchest by removing tissue down to the parietal pleura. Window height was 6 cm above the base of the lung. During apnea, alveolar liquid and pleural pressures were measured by puncturing through the pleural window with micropipettes connected to a servo-nulling pressure-measuring system. Pressures were measured at airway pressures of 0 (FRC) and 10 cmH2O both in vivo and postmortem. In vivo, alveolar liquid and pleural pressures relative to ambient pressure averaged -2.3 +/- 1.4 (SD) and -1.8 +/- 0.9 cmH2O at FRC and increased to 3.3 +/- 1.8 and 1.8 +/- 1.6 cmH2O after inflation to an airway pressure of 10 cmH2O, respectively. Similar values were obtained postmortem. These results were similar to previous measurements in the isolated lung.

Stress wave velocity measured in intact pig lungs with cross-spectral analysis.

In anesthetized pigs (25-40 kg), we generated stress waves in the lung by rapid deflation of an esophageal balloon. The source distortion was measured by an accelerometer (1 g wt) bonded to the balloon. Stress waves were detected by three accelerometers bonded to intercostal muscle and to the skin near midchest. The distance between the source and chest receivers were measured radiographically. Cross-spectral analysis was used to calculate transit times. We measured stress wave velocities at airway pressures of 0 (functional residual capacity) and 25 cmH2O. Transpulmonary pressure (Ptp) was measured by an esophageal balloon. In vivo, stress wave velocities increased from 291 +/- 117 (SD) cm/s at 3.0 +/- 0.9 cmH2O Ptp to 573 +/- 73 cm/s at 13.8 +/- 3.5 cmH2O Ptp (n = 6). These velocities agreed with longitudinal wave velocities measured in isolated sheep lungs and predictions based on the elastic moduli of lung parenchyma. Post-mortem edema was induced by intratracheal instillation of 200 ml of saline, resulting in a wet-to-dry weight ratio of 7.7 +/- 1.4 (n = 5). At 15 cmH2O Ptp, stress wave velocities decreased from 565 +/- 155 cm/s before edema to 445 +/- 130 cm/s after edema. This decrease correlated well with predictions based on the increased lung density, as dictated by elasticity theory.

Effect of ventilation frequency and tidal volume on pleural space thickness in rabbits.

The thickness of the pleural space was measured by fluorescence video-microscopy during mechanical ventilation in anesthetized paralyzed rabbits. A transparent parietal pleural window was made in the fourth or sixth intercostal space near midchest by dissection of intercostal muscle and endothoracic fascia. Fluorescence-labeled (fluorescein isothiocyanate) dextran solution (1 ml) was injected into the pleural space via a rib capsule and allowed to mix with the pleural liquid. With the rabbit in the left lateral decubitus position and the pleural window superior, the light emitted from the pleural liquid through the pleural window was measured through the videomicroscope. Both ventilation frequency and tidal volume were varied. Pleural space thickness was determined by in vitro calibration of the pleural liquid at the end of the experiment. At a frequency of 40 breaths/min and a tidal volume of 20 ml, pleural space thickness averaged 35 +/- 15 (SD) microns (n = 7). When frequency was reduced to 8 breaths/min, this value was reduced by 40% to 22 +/- 11 microns. A reduction in tidal volume from 20 to 6 ml at a frequency of 40 breaths/min produced a similar reduction in pleural space thickness. During apnea, pleural space thickness averaged 11 +/- 3 microns. Cardiogenic motion had no measurable effect on pleural space thickness. The increased pleural space thickness with ventilation might serve to reduce the power dissipated due to sliding of the lung relative to the chest wall. Results support the concept of lubrication as the primary function of the pleural space.

Effects of electric charge on hydraulic conductivity of pulmonary interstitium.

The hydraulic conductivity of pulmonary interstitium was measured in a short isolated segment of interstitium surrounding a large pulmonary artery (1-3 mm diam) of the rabbit. The flow rate of the following solutions was measured sequentially: normal saline, polycation protamine sulfate (0.08 mg/ml), cationic dextran (0.1 or 1.5%) or anionic dextran (0.1 or 1.5%), and hyaluronidase (testes, 0.02%) solution. The pH of all solutions was adjusted to 7.35-7.40. The ratios of the flow of protamine sulfate and cationic dextran to that of saline averaged 2.3 +/- 0.92 (SD, n = 7) and 3.0 +/- 1.2 (n = 6), respectively. The anionic dextran-to-saline flow ratio averaged 0.72 +/- 0.28 (n = 13). Flow increased in the presence of positively charged molecules and decreased in the presence of negatively charged molecules. At a lower pH of 5.0-6.0, only 0.1% cationic dextran had an effect on interstitial conductivity. Thus pulmonary interstitium at physiological pH has the properties of a negatively charged membrane. The increased interstitial conductivity caused by the positively charged molecules was not observed after treatment with hyaluronidase. These effects of electric charge on interstitial conductivity were partly attributed to the presence in the interstitium of negatively charged hyaluronan.

Effect of lung inflation on regional lung expansion in supine and prone rabbits.

At functional residual capacity, lung expansion is more uniform in the prone position than in the supine position. We examined the effect of positive airway pressure (Paw) on this position-dependent difference in lung expansion. In supine and prone rabbits postmortem, we measured alveolar size through dependent and nondependent pleural windows via videomicroscopy at Paw of 0 (functional residual capacity), 7, and 15 cmH2O. After the chest was opened, alveolar size was measured in the isolated lung at several transpulmonary pressures (Ptp) on lung deflation. Alveolar mean linear intercept (Lm) was measured from the video images taken in situ. This was compared with those measured in the isolated lung to determine Ptp in situ. In the supine position, the vertical Ptp gradient increased from 0.52 cmH2O/cm at 0 cmH2O Paw to 0.90 cmH2O/cm at 15 cmH2O Paw, while the vertical gradient in Lm decreased from 2.17 to 0.80 microns/cm. In the prone position, the vertical Ptp gradient increased from 0.06 cmH2O/cm at 0 cmH2O Paw to 0.35 cmH2O/cm at 15 cmH2O Paw, but there was no change in the vertical Lm gradient. In anesthetized paralyzed rabbits in supine and prone positions, we measured pleural liquid pressure directly at 0, 7, and 15 cmH2O Paw with dependent and nondependent rib capsules. Vertical Ptp gradients measured with rib capsules were similar to those estimated from the alveolar size measurements. Lung inflation during mechanical ventilation may reduce the vertical nonuniformities in lung expansion observed in the supine position, thereby improving gas exchange and the distribution of ventilation.

Pleural pressure distribution and its relationship to lung volume and interstitial pressure.

The mechanics of the pleural space has long been controversial. We summarize recent research pertaining to pleural mechanics within the following conceptual framework, which is still not universally accepted. Pleural pressure, the force acting to inflate the lung within the thorax, is generated by the opposing elastic recoils of the lung and chest wall and the forces generated by respiratory muscles. The spatial variation of pleural pressure is a result of complex force interactions among the lung and other structures that make up the thorax. Gravity contributes one of the forces that act on these structures, and regional lung expansion and pleural pressure distribution change with changes in body orientation. Forces are transmitted directly between the chest wall and the lung through a very thin but continuous pleural liquid space. The pressure in pleural liquid equals the pressure acting to expand the lung. Pleural liquid is not in hydrostatic equilibrium, and viscous flow of pleural liquid is driven by the combined effect of the gravitational force acting on the liquid and the pressure distribution imposed by the surrounding structures. The dynamics of pleural liquid are considered an integral part of a continual microvascular filtration into the pleural space. Similar concepts apply to the pulmonary interstitium. Regional differences in lung volume expansion also result in regional differences in interstitial pressure within the lung parenchyma and thus affect regional lung fluid filtration.

Pleural space thickness in situ by light microscopy in five mammalian species.

The thickness of the pleural space was measured by a focusing method using a light microscope (X157, 2.5-micron depth of focus). In anesthetized animals, thin transparent parietal pleural windows were made by dissection of intercostal muscle. Multiple postmortem measurements were made of the combined thickness of the pleural space and the window by focusing in sequence on the lung surface and on 1- to 2-micron tantulum particles sprayed on the window. The window thickness was measured after creating a pneumothorax and retracting the lungs. In supine rabbits the pleural space measured at various heights on the costal surface was of uniform thickness (16 micron) except for a thicker region (62 micron) located within 3 mm of the most dependent part of the lung. The thicker region reverted to the uniform thickness after it was placed in a nondependent position by inverting the animal from the supine to prone position, indicating fluid drainage by gravity. In the prone position near midchest, pleural space thickness (t) averaged 6.9 micron in the mouse, 10.2 in the rat, 17.2 in the rabbit, 18.3 in the cat, and 23.6 in the dog. Animals of larger body mass (M, kg) had a wider pleural space: t = 13.1 X M0.20. There was no contact between the two pleurae, indicating that fluid lubrication facilitates sliding between the lung and chest wall. Based on the t vs. M relationship and estimates of the viscous flow of pleural liquid, pleural fluid exchange rate would be proportional to body mass and the work of sliding as a fraction of the work of breathing would be smaller in larger animals.

Pulmonary blood flow vs. gas volume at various perfusion pressures in rabbit lung.

To obtain a detailed description of the dependence of pulmonary blood flow on changes in lung volume, we perfused isolated rabbit lungs with homologous blood at 37 degrees C while controlling vascular pressures during lung deflation. We set pulmonary arterial pressure (Ppa) and pulmonary venous pressure (Ppv) to constant values relative to alveolar pressure (Palv) to keep the effective driving pressure for flow constant during lung deflation from total lung capacity (TLC) to 25% TLC. The shapes of the flow vs. lung volume curves were dependent on the levels of Ppa-Palv and Ppv-Palv at which they were obtained. When Ppv greater than Palv throughout the lung (zone 3 conditions), flow increased as the lungs were deflated from TLC, independent of the level of Ppa-Palv. When Ppv less than Palv (zone 2 conditions) and Ppa-Palv was moderately high, flow increased as the lungs were deflated from 100 to approximately 50% TLC, then decreased at lower lung volumes. When Ppa - Palv was less than 10 cmH2O in zone 2 conditions, flow decreased monotonically during deflation from TLC. We concluded that the dependence of blood flow on lung volume is complex, which may be a reflection of the nonlinear pressure-diameter properties of pulmonary vessels.

Effect of airway and left atrial pressures on microcirculation of newborn lungs.

To determine the effect of lung inflation and left atrial pressure on the hydrostatic pressure gradient for fluid flux across 20- to 60-microns-diam venules, we isolated and perfused the lungs from newborn rabbits, 7-14 days old. We used the micropuncture technique to measure venular pressures in some lungs and perivenular interstitial pressures in other lungs. For all lungs, we first measured venular or interstitial pressures at a constant airway pressure of 5 or 15 cmH2O with left atrial pressure greater than airway pressure (zone 3). For most lungs, we continued to measure venular or interstitial pressures as we lowered left atrial pressure below airway pressure (zone 2). Next, we inflated some lungs to whichever airway pressure had not been previously used, either 5 or 15 cmH2O, and repeated venular or interstitial pressures under one or both zonal conditions. We found that at constant blood flow a reduction of left atrial pressure below airway pressure always resulted in a reduction in venular pressure at both 5 and 15 cmH2O airway pressures. This suggests that the site of flow limitation in zone 2 was located upstream of venules. When left atrial pressure was constant relative to airway pressure, the transvascular gradient (venular-interstitial pressures) was greater at 15 cmH2O airway pressure than at 5 cmH2O airway pressure. These findings suggest that in newborn lungs edema formation would increase at high airway pressures only if left atrial pressure is elevated above airway pressure to maintain zone 3 conditions.

Pleural liquid pressure measured with rib capsules in anesthetized ponies.

Pleural liquid pressure was measured at end expiration in 11 spontaneously breathing anesthetized ponies in the prone and supine positions. A liquid-filled capsule was implanted into a rib to measure pleural liquid pressure with minimal distortion of the pleural space (Wiener-Kronish et al., J. Appl. Physiol. 59: 597-602, 1985). Capsule position relative to lung height was measured from thoracic radiographs taken in each position. In each body position, pleural liquid pressure was most negative in the superior lung regions and least negative in the inferior lung regions. In the supine position, the magnitude of the vertical gradient in pleural liquid pressure was 0.67 cmH2O/cm ht and was not significantly different from 1 cmH2O/cm ht. In the inferior lung regions (less than 50% lung ht), pleural liquid pressure averaged -1.3 cmH2O, indicating a low transpulmonary pressure over the region of the chest where most of the lung mass is located. When animals were in the prone position, the magnitude of the vertical gradient in pleural liquid pressure was 0.14 cmH2O/cm ht and was not statistically different from 0 cmH2O/cm ht. In each body position, mean transpulmonary pressure, measured postmortem, was similar to the estimated magnitude of pleural liquid pressure at 50% lung ht. This suggests that pleural liquid pressure is closely related to pleural surface pressure. These results are consistent with the poor ventilation distribution and reduced lung volumes measured in anesthetized horses in the supine position compared with values measured in horses in the prone position.

Upward flow of pleural liquid near lobar margins due to cardiogenic motion.

In anesthetized, paralyzed, supine rabbits (3-4 kg) during apnea, we injected fluorescent dye or fluorescent microspheres (2 or 6 microns diam) into the dependent pleural space and observed the arrival and movement of the dye or microspheres at superior regions. Injection was through a rib capsule located in the dependent right chest. The dye or microspheres were observed through a pleural window overlying a lobar margin. The vertical distance between the capsule and window was 3-4 cm. The movement of the dye or microspheres was recorded via a fluorescence videomicroscope, and the signals were analyzed for dye transit time and microsphere velocity. The transit time of the dye to traverse the height of the pleural space was calculated from the light intensity vs. time curve. Transit time during apnea averaged 6.0 +/- 3.4 (SD) min (n = 4). Transit time measured after the onset of mechanical ventilation was < 1 min. The direction and speed of a microsphere moving in the relatively thick pleural space adjacent to the lobar margin depended on its distance from the lobar margin. Microspheres moved upward in the pleural space that was in proximity to the lobar margin but downward at farther distances from the lobar margin. Pleural liquid recirculation occurs via the pleural space adjacent to lobar margins.

Effects of age on elastic moduli of human lungs.

The model of the lung as an elastic continuum undergoing small distortions from a uniformly inflated state has been used to describe many lung deformation problems. Lung stress-strain material properties needed for this model are described by two elastic moduli: the bulk modulus, which describes a uniform inflation, and the shear modulus, which describes an isovolume deformation. In this study we measured the bulk modulus and shear modulus of human lungs obtained at autopsy at several fixed transpulmonary pressures (Ptp). The bulk modulus was obtained from small pressure-volume perturbations on different points of the deflation pressure-volume curve. The shear modulus was obtained from indentation tests on the lung surface. The results indicated that, at a constant Ptp, both bulk and shear moduli increased with age, and the increase was greater at higher Ptp values. The micromechanical basis for these changes remains to be elucidated.

Sequence of perivascular liquid accumulation in liquid-inflated dog lung lobes.

The peribronchovascular interstitium of the lung is a potential space that expands in pulmonary edema with the formation of large liquid cuffs. To study the time course of cuff formation we inflated nine isolated dog lung lobes with liquid to total lung capacity, rapidly froze them in liquid N2 after inflation periods of 1-300 min, then photographed 20 blocks of each lobe at X3 magnification. From the photographs we measured the ratio of cuff area to vessel area for arteries and veins of 0.05-8 mm diam. We found that the cuff-to-vessel area ratio attained a maximum value of 3-4, which was independent of vessel size. However, the first cuffs to reach maximum size were those around vessels of 0.1-0.5 mm diam, whereas cuffs around larger vessels filled more slowly. No cuffs were visible around vessels smaller than 0.1 mm diam. After 45 min cuffs had formed around 99% of all vessels larger than 0.5 mm diam but had formed around only 38% of veins and 91% of arteries of smaller diameter. We simulated the observed rate and pattern of cuff growth using electrical analog models. The filling pattern and model analyses suggest that liquid entered the interstitium from an air space site associated with arteries of approximately 0.1-1.0 mm diam, spread to adjacent sites, and eventually reached the lobe hilum. The estimated perivascular interstitial flow resistance decreased approximately 100-fold with cuff expansion.

Microvascular pressures measured by micropipettes in isolated edematous rabbit lungs.

To study the mechanical effects of lung edema on the pulmonary circulation, we determined the longitudinal distribution of vascular resistance in the arteries, veins, and microvessels, and the distribution of blood flow in isolated blood-perfused rabbit lungs with varying degrees of edema. Active vasomotor changes were eliminated by adding papaverine to the perfusate. In three groups of lungs with either minimal [group I, mean wet-to-dry weight ratio (W/D) = 5.3 +/- 0.6 (SD), n = 7], moderate (group II, W/D = 8.5 +/- 1.2, n = 10), or severe (group III, W/D = 9.9 +/- 1.6, n = 5) edema, we measured by direct micropuncture the pressure in subpleural arterioles and venules (20-60 micron diam) and in the interstitium surrounding these vessels. We also measured pulmonary arterial and left atrial pressures and lung blood flow, and in four additional experiments we used radio-labeled microspheres to determine the distribution of blood flow during mild and severe pulmonary edema. In lungs with little or no edema (group I) we found that 33% of total vascular pressure drop was in arteries, 60% was in microvessels, and 7% was in veins. Moderate edema (group II) had no effect on total vascular resistance or on the vascular pressure profile, but severe edema (group III) did increase vascular resistance without changing the longitudinal distribution of vascular resistance in the subpleural microcirculation. Perivascular interstitial pressure relative to pleural pressure increased from 1 cmH2O in group I to 2 in group II to 4 in group III lungs.(ABSTRACT TRUNCATED AT 250 WORDS)

Effect of hydration on lung interstitial permeability response to albumin and hyaluronidase.

Previous studies showed that the flows of albumin and hyaluronidase solutions increased relative to that of saline in isolated segments of rabbit lung interstitium (Lai-Fook et al. J. Appl. Physiol. 67:606-613, 1989). We questioned whether these effects were hydration dependent. In interstitial segments the flows of lactated Ringer, albumin (5 and 10 g/dl), and hyaluronidase (0.02%) solutions were measured at mean interstitial pressures (Pm) between -5 and 15 cmH2O with a constant driving pressure of 5 cmH2O. The albumin-to-Ringer flow ratio increased monotonically from near the viscosity-dependent value (0.75-0.77) at -5 cmH2O Pm to values of 1.6-2.1 at 15 cmH2O Pm. A similar behavior was observed for the flow of the hyaluronidase solution relative to that of Ringer solution. The increased permeability response to albumin was independent of the albumin concentration used. By contrast, the response to hyaluronidase was lower when the interstitium was perfused with the higher concentration albumin solution (10 g/dl) before the flow of hyaluronidase, indicating an inhibitory effect of albumin on the hyaluronidase response. Estimates of interstitial hydration from Pm indicated an increased interstitial permeability (conductivity) to the flows of albumin and hyaluronidase solutions only after interstitial volume had doubled, whereas interstitial permeability was viscosity dependent at normal interstitial hydration.

Effect of vascular volume and edema on wave propagation in canine lungs.

The velocities of longitudinal and transverse stress waves transmitted through inflated lung parenchyma depend on the lung stiffness, as defined by the bulk and shear moduli, and the lung density. We examined the relationship between stress wave velocities and lung density. A saline-filled reservoir was connected to the vessels of caudal dog lobes held inflated at 5 cmH2O transpulmonary pressure, and vascular volume and extravascular lung water were increased in steps by increasing vascular pressure. At each step, we measured the transmitted signals at locations 2 and 7 cm from an impulse surface distortion by means of microphones embedded in the lung surface. Longitudinal and transverse wave velocities were computed by using cross-correlation analysis of microphone signal pairs. Both wave velocities decreased as lung density increased: as a first approximation, wave velocities were inversely proportional to the square root of lung density. This behavior is consistent with the propagation of small-amplitude stress waves through an elastic continuum. Estimated bulk and shear moduli were 26 and 3.5 cmH2O, respectively, and were consistent with results from quasi-static deformation tests.

Pleural liquid pressure in dogs measured using a rib capsule.

We have developed a minimally invasive method for measuring the hydrostatic pressure in the pleural space liquid. A liquid-filled capsule is bonded into a rib and a small hole is cut in the parietal pleura to allow direct communication between the liquid in the capsule and the pleural space. The pressure can be measured continuously by a strain gauge transducer connected to the capsule. The rib capsule does not distort the pleural space or require removal of intercostal muscle. Pneumothoraces are easily detected when they occur inadvertently on puncturing the parietal pleura. We examined the effect of height on pleural pressure in 15 anesthetized spontaneously breathing dogs. The vertical gradients in pleural pressure were 0.53, 0.42, 0.46, and 0.23 cmH2O/cm height for the head-up, head-down, supine, and prone body positions, respectively. These vertical gradients were much less than the hydrostatic value (1 cmH2O/cm), indicating that the pleural liquid is not in hydrostatic equilibrium. In most body positions the magnitudes of pleural liquid pressure interpolated to midchest level were similar to the mean transpulmonary (surface) pressure determined postmortem. This suggests that pleural liquid pressure is closely related to the lung static recoil.

Microvascular pressures measured by micropuncture in lungs of newborn rabbits.

The purpose of this study was to determine the pattern of vascular pressure drop in newborn lungs and to define the contribution of active vasomotor tone to this longitudinal pressure profile. We isolated and perfused with blood the lungs from 22 rabbit pups, 5-19 days old. We inflated the lungs to a constant airway pressure of 7 cmH2O, and at constant blood flow, we maintained outflow pressure in the circulation greater than airway pressure at the level of micropuncture (zone 3). By the use of glass micropipettes and a servo-nulling device, we measured pressures in small (20-60 micron diam) subpleural arterioles and venules in the lungs of 13 newborn rabbits. We found that 60% of the pressure drop was in arteries, 31% in microvessels of less than 20-60 micron diam, and 9% in veins. In the lungs of an additional nine rabbit pups we measured microvascular pressures before and after the addition to the perfusate of the vasodilator, papaverine hydrochloride. We found that removal of vasomotor tone resulted in a 33% reduction in total lung vascular resistance, which resulted from a decrease in pressure in arterial vessels, with no change in microvascular pressure. These findings indicate that arteries of greater than 60 micron diam constitute the major source of vascular resistance in isolated perfused newborn rabbit lungs.

Liquid thickness vs. vertical pressure gradient in a model of the pleural space.

In recent studies using relatively noninvasive techniques, the vertical gradient in pleural liquid pressure was 0.2-0.5 cmH2O/cm ht, depending on body position, and pleural liquid pressure closely approximated lung recoil (J. Appl. Physiol. 59: 597-602, 1985). We built a model to discover why the vertical gradient in pleural pressure is less than hydrostatic (1 cmH2O/cm). A long rubber balloon of cylindrical shape was inflated in a plastic cylinder. The "pleural" space between the balloon and cylinder was filled with blue-dyed water. With the cylinder vertical, we measured pleural pressure by a transducer through side taps at 2-cm intervals up the cylinder. The pressure was measured with different amounts of water in the pleural space. With a clear separation between the balloon and the container, the vertical gradient in pleural liquid pressure was hydrostatic. As water was withdrawn from the pleural space, the balloon approached the wall of the container. Over an 8-cm-long midregion of the model where the balloon diameter matched the cylinder diameter, the vertical gradient was not hydrostatic and was virtually absent. In this region, the pleural liquid pressure was uniform and equal to the recoil of the balloon. In this section we could not see any pleural space. By scintillation imaging using 99mTc-diethylenetriamine pentaacetic acid in the water, we estimated the thickness of this flat "costal" pleural space to be approximately 20 microns. Radioactive tracer injected at the top of the pleural space appeared by 24 h at the bottom, which indicated a slow drainage of liquid by gravity.(ABSTRACT TRUNCATED AT 250 WORDS)