INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

During strenuous exercise in normal animals, intravascular liquid rapidly filters into the lung parenchyma in large amounts, yet the lung does not normally develop overt pulmonary edema (1). This homeostasis occurs despite evidence in sheep (2, 3), humans (4, 5), and horses (6, 7) that pulmonary capillary hydrostatic pressure rises above edemagenic levels, > 30 mm Hg, during strenuous exercise. Clearance of transvascular filtrate, assessed via lung lymphatic flow, rises 7- to 10-fold in sheep during strenuous exercise (2, 3, 8). The pattern of the increase in lymph flow is remarkable; within seconds of the onset of running, lung lymph flow rises to a new level that is sustained for the duration of exercise and then rapidly drops back to resting levels at cessation of exercise (2). This pattern shows that mechanisms of augmented lymph clearance are quickly and effectively brought into play during exercise, and the lung does not normally accumulate excess liquid during exercise.

The handling of filtered liquid during passive pulmonary congestion is very different from exercise. When pulmonary capillary pressure is elevated by mechanical obstruction of the mitral valve, lung lymph flow increases slowly over a period of hours, even when capillary pressure is raised to edemagenic levels (9, 10). Lung lymph-to-plasma protein concentration ratio (L/P ratio) decreases steadily, but does not achieve its lowest washdown value until hours of continuous capillary hypertension. After cessation of capillary hypertension, lymph flow and L/P ratio are restored to normal over a period of several hours (10).

The purpose of this study was to reconcile the apparent paradox that at similar intravascular pressures, lymph flow and hydrostatic washdown of interstitial protein are very different between exercise and passive pulmonary capillary hypertension. One difference between exercise and passive congestion is hyperpnea. In humans, ventilation may rise from about 5 L/ min at rest to 160 L/min with strenuous exercise. We investigated the effect of hyperpnea alone and in concert with left atrial hypertension on lymph flow and protein and compared the results with that of similar levels of ventilation and microvascular pressure produced during exercise (3).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animal Preparation

Seven adult sheep underwent bilateral thoracotomies for placement of vascular and lung lymph catheters, as previously described (2, 3, 12, 13). Through a right thoracotomy, the caudal mediastinal lymph node was isolated, and the efferent duct was cannulated with a thin silastic catheter. Through a second right thoracotomy, the caudal portion of the lymph node was doubly ligated and bisected, and any diaphragmatic communications were severed. Through a left thoracotomy, silastic catheters were placed directly into the main pulmonary artery and left atrium, respectively. An ultrasonic flow probe (Transonic Systems, Ithaca, NY) was placed around the main pulmonary artery to measure cardiac output. Through a small right neck incision, the carotid artery was cannulated, and a No. 8 Cordis catheter sheath introducer was sutured into the jugular vein (Cordis Labs, Miami, FL). All catheters were exteriorized and sewn to the skin of the chest wall. Animals were allowed to recover in pens for 4-7 d with free access to water and feed.

Measurements

Mean pulmonary arterial (), left atrial (), and systemic arterial () pressures were measured continuously, using calibrated pressure transducers (Hewlett-Packard, Andover, MA). The zero reference for pressures was set near the level of the left atrium, and the level was kept constant during all experiments in each sheep. Cardiac output was also recorded continuously by the implanted ultrasonic flow probe, which was connected to a Transonic T101 blood flowmeter. Flowmeter values were calibrated to the cardiac output measured by thermodilution (model 9520A; Edwards, Santa Ana, CA). After placing a 7.5F Swan-Ganz catheter in the pulmonary artery via the jugular vein, we introduced administered isoproterenol in each animal to increase cardiac output. Isoproterenol (Elkins-Sinn, Cherry Hill, NH) 10 mg/250 ml in 0.9% saline was delivered intravenously via an infusion pump to produce outputs similar to those of exercise. Flow probe values were compared to those obtained by thermodilution at several points, and a regression line was made to normalize ultrasonic flow values for each sheep (3, 12, 15).

On the day of experiment, a No. 10 Shiley cuffed tracheostomy tube (OEM Medical, Richmond, VA) was inserted. Respiratory flow was measured using a heated Fleisch #3 pneumotachograph (Rudolph Co., Omaha, NE) connected to the tracheostomy tube, and tidal volume was obtained by electrical integration of the flow signal (Validyne, Inc., Northridge, CA) (14). Vascular pressures, cardiac output, airflow, and tidal volume (VT) were continuously and simultaneously recorded on a multichannel recorder (MT95000; Astro-Med, Inc., West Warwick, RI). Ventilation was calculated as VT × rate over a minute of steady-state breathing. Arterial blood pH, PCO₂, and PO₂ were analyzed. We estimated pulmonary microvascular pressure (Pmv) as Pmv = Pla + 0.4 × (PpaPla) (16).

Lung lymph was collected continuously and was decanted at 5-min intervals to measure flow and protein concentration. Simultaneously, heparinized blood samples were drawn from the carotid artery catheter. Total protein concentration of blood plasma and lung lymph was measured by the modified Biuret method to calculate lung L/P ratio (2, 9, 10).

Experimental Protocols

Sheep stood and exercised on a standard treadmill (Quinton Instruments Co., Seattle, WA) (2, 3, 12, 13). In each experiment, we continuously measured pulmonary hemodynamics, cardiac output, and lung lymph flow. We measured minute ventilation, tidal volume, and respiratory frequency continuously in experiments 1-3, and every 30 min in experiment 4. We collected lung lymph every 10 min at baseline and established a stable baseline for all values for over 1 h of observation. Experimental protocols were as follows:

1. Exercise: Sheep exercised at a constant rate of 2.8 mph for 20 min.

2. Hyperpnea plus increased left atrial pressure (DS +  Pla) at rest: First, we added 120-180 ml (150 ml in five sheep, 120 ml in one sheep, and 180 ml in one sheep) of tubing to create sufficient dead space (DS) at rest to increase ventilation to the same level observed during exercise. After 20 min of measurements during constant hyperpnea, we raised Pla to increase Pmv to the level that occurred in that sheep during exercise.

3. Increased Pla plus hyperpnea ( Pla + DS): First, we raised Pla to increase Pmv to the same level as during exercise and held it constant for 20 min. We then added DS to cause the same increase in minute ventilation seen with exercise. When we added DS in this and in the above protocol, we drew arterial blood for pH and gas analysis every 5 min. If the Pa_O2 decreased below 120 mm Hg, we gave oxygen at 0.5 L/min into the flowby tubing to prevent hypoxemia.

4. Increased Pmv ( Pla alone): To observe the effects of a prolonged increase in Pmv alone, we raised Pla for over 2 h on a separate day to raise Pmv to the same level as with exercise.

5. In three sheep, another control study was done to see the independent effects of increased cardiac output, increased ventilation, and a combination of these perturbations on lung lymph filtration. We increased cardiac output with an intravenous infusion of isoproterenol (2.2-3.2 µg/min). In addition, we observed the effect of increased DS alone on lymph flow for 60 min. In these experiments, increased cardiac output and ventilation were matched to the values obtained during exercise.

Lymph Transit Measurement

We used one sheep to measure the actual transit time from the vascular bed to the lymphatic trunk. We also wanted to show that the lymph increase was due to increased filtration and clearance rather than mobilization of basal interstitial and lymphatic filtrate. On three separate days we injected ^99mTc-labeled 4.0-kD dextrans through the jugular vein catheter with the sheep at rest, during steady-state exercise at 2.8 mph, during left atrial hypertension (Pmv = 15 cm H₂O) and during hyperpnea at rest (20 L/min), followed by the addition of left atrial hypertension. Lung lymph was continuously collected and routed directly to a 2-in. NaI crystal scintillation probe.

The crystal was encased in a lead cylinder and the silastic cannula was routed through the cylinder and shielded from the atmosphere and the sheep with a fitted lead cover. The signal was multiplied by a Tennelec 15.5 photomultiplier, conditioned by a Tennelec TC247 amplifier and read by a pulse-light analyzer (PCA-2000; The Nucleus, Oak Ridge, TN). Data were displayed as the sum of every 60 s of radioactivity. A Fortran 77 program was written to correct counts for background noise, and the results were displayed by Excel graphics software.

As a check for the on-line measure of lymph radioactivity, blood and lymph samples were collected every 5 min after intravenous injection of dextran radioisotope. Samples were weighed and centrifuged at 900 × g for 10 min to remove cells. A known volume of plasma or lymph was placed in a scintillation vial with 3 ml of ethanol, followed by centrifugation at 3,600 × g for 10 min and decanting into a second scintillation vial for count. Gamma radioactivity was measured in both precipitate and liquid in a well-type Packard autogamma scintillation spectrometer. Counts were normalized to plasma and lymph weight.

Radiolabeling of Dextran

Dextrans (40 kD, range 37-43 kD) (Sigma Co., St. Louis, MO) were labeled with ^99mTc in a manner similar to that described by Zanelli and coworkers (17, 18). Initially 20 mg of stannous chloride dihydrate (SnCl₂H₂O) was dissolved in 1 ml of 12 N HCl and allowed to stand at room temperature until clear (20-30 min). During this time 100 ml of sterile water was degassed by passage through a 0.45-µm nylon filter using a vacuum and collection in a sterile container. Twenty milliliters of the water was then transferred into a 30-ml sterile vial, and a sufficient amount of paraaminobenzoic acid (PABA) was added to obtain a concentration of 2 mg/ml. A 10% solution of the 6-kD or 40-kD dextrans was prepared using a portion of the degassed water. A sufficient volume of the 10% solution was added to the 30-ml vial to obtain a concentration of 1 mg/ml, 0.1 ml of the SnCl₂. H₂O was then added to the dextran solution and mixed under N₂. The final solution was transferred in 1-ml aliquots to 10-ml glass vials, stoppered under N₂, and frozen at 10° C. These "kits" were lyophilized, stoppered under vacuum, and stored at 4° C. Labeling was performed by adding 5 ml of sterile citrate-buffered saline to the lyophilized kit and dissolving the dextrans. Approximately 1-2 mCi of ^99mTc in a volume of 0.2-0.4 ml was added to the vial and allowed to incubate at room temperature for 10 min. Labeling efficiency was determined by instant thin-layer chromatography on silica gel and developed with methyl ethyl ketone. The labeled dextrans remain at the origin and free pertechnetate migrates to the solvent font.

Statistics

Data were transcribed to Quattro-Pro (Borland Co., Scotts Valley, CA), transferred and analyzed by NCSS Statistical Systems (Kaysville, UT) and Minitab Statistical Software 11.21 (State College, PA). Data are presented as means ± SEM. Comparison within group changes from baseline to experimental values for pulmonary hemodynamics and ventilation were made by the Wilcoxon rank sum test. Lung lymph data were analyzed by one-way analysis of variance, and tests for difference between baseline and experimental values were by Duncan's new multiple range test. Significant differences were assumed at p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pulmonary hemodynamic changes in protocols 1-4 are summarized in Table 1. Pulmonary artery pressure rose significantly during exercise at 2.8 mph with no significant change in left atrial pressure. Cardiac output rose rapidly to about 70% over baseline, remained stable during exercise, and decreased rapidly back to baseline at cessation. There were no significant changes in pulmonary artery pressure, left atrial pressure, or cardiac output resulting from dead space alone.

In  Pla experiments, left atrial pressure was mechanically increased to the desired level; pulmonary artery pressure rose as expected and cardiac output decreased by about 10%. Microvascular pressure estimated by the Gaar equation (1) increased significantly and remained constant during exercise and with left atrial hypertension.

Lung lymph flow increased rapidly to levels about double that of baseline during exercise and then remained constant. Lung lymph flow decreased rapidly to baseline within minutes after cessation of exercise as in previous studies (2, 3). Increased ventilation with dead space alone did not significantly affect lymph flow. During left atrial hypertension lung lymph flow slowly increased by 30% above baseline, lower than was seen during exercise. However, the combination of increased ventilation and left atrial hypertension resulted in a lung lymph flow similar to that of exercise.

L/P protein ratio changes are shown in Figures 1 and 2. L/P decreased rapidly during exercise, reaching statistical difference from baseline at 20 min. At cessation, L/P rose back to the baseline level over the next 20 min. During left atrial hypertension, L/P decreased slowly, achieving significant change at 100 min. Figure 2 demonstrates the change associated with the combination of left atrial hypertension and hyperpnea. The most dramatic change was seen in sheep exposed first to 20 min of left atrial hypertension. The addition of dead space resulted in a rapid reduction in L/P ratio, consistent with facilitated clearance.

View larger version (14K): [in this window] [in a new window]   Figure 1.   Changes in lymph-to-plasma (L/P) protein concentration during exercise (open circles) or left atrial hypertension (closed circles) at similar calculated microvascular pressures in seven sheep. L/P ratio dropped quickly to significant levels within 20 min of exercise and rose back to baseline rapidly after cessation. L/P ratio achieved a significant decrease after 110 min during left atrial hypertension.

View larger version (16K): [in this window] [in a new window]   Figure 2.   Lymph-to-plasma (L/P) protein concentration ratios in paired studies in seven sheep at rest with left atrial hypertension and dead space hyperpnea applied in alternate order. L/P dropped most quickly in animals with left atrial hypertension for 30 min followed by hyperpnea.

The ventilatory effects of each experiment are shown in Table 2. Minute ventilation more than doubled during exercise at 2.8 mph and remained constant for the 20 min duration. Adding dead space at rest raised minute ventilation to levels similar to that with exercise; compared with exercise, the response included a greater increase in tidal volume and a smaller increase in respiratory rate. Left atrial hypertension did not influence tidal volume or respiratory rate. Ventilation during combined left atrial hypertension and increased dead space was slightly but not significantly higher than during exercise. The only effects of the experimental perturbation on blood gas tensions were a small increase in Pa_O2 during dead space breathing in response to the small flow of added O₂, and a mild elevation in Pa_CO2, Table 3. Arterial pH was unaffected.

Figure 3 delineates the relationships of ventilation and lymph flow under the several experimental conditions. Lymph flow is normalized to resting levels for each experimental area. This figure shows that neither microvascular pressure nor hyperpnea alone raises lymph flow to the levels seen during comparable microvascular pressure and minute ventilation of exercise, but the combination of the two yields lymph flows similar to that of exercise, in either order of sequence.

View larger version (16K): [in this window] [in a new window]   Figure 3.   Changes in lung lymph flow and minute ventilation (L/min) during several perturbations. The arrows depict the direction of change from rest to experimental conditions. Exercise (closed squares) was at 2.8 mph for 20 min. PLa (closed circles) was a step increase in microvascular pressure to attain the same level as during exercise. DS (closed triangles) was the addition of airway dead space that raised minute ventilation to a level similar to that of exercise. PLa + DS and DS + PLa are the combination of dead space and increased Pmv in each sequence. In each case the first perturbation was applied for 20 min, followed by the addition of the other.

In protocol 4, cardiac output increased from 4.4 ± 0.5 to 8.7 ± 0.8 with intravenous isoproterenol. Pulmonary arterial pressure increase from 23.6 to 25.3 cm H₂O and left atrial pressure was unchanged. The addition of dead space during isoproterenol infusion increased ventilation from 6.5 ± 1 to 13.0 ± 3.4 L/ min. Normalized lymph flow was 1 at baseline, 1.3 ± 0.1 with isoproterenol, and 1.2 ± 0.1 with isoproterenol plus dead space. Thus, pulmonary blood flow alone did not markedly increase lymph clearance.

Appearance of ⁹⁹Tc-labeled Dextran 40 in lung lymph is summarized in Table 4. Mean time to appearance of tracer was 18-21 min at rest and 5-6 min with exercise. Increased ventilation alone or left atrial hypertension did not alter the specific time of appearance, but the combination of hyperpnea and left atrial hypertension reduced the time of appearance to that of exercise.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Lung fluid balance is primarily determined by the net vector between the opposing forces of transvascular hydrostatic pressure and transvascular oncotic pressure, modified by barrier functions of the capillary endothelium and alveolar epithelium (19). Barrier function (permeability) is measured primarily by the ability to sieve proteins and is influenced by water and ion channels (16, 20). Interstitial water and protein, reflecting the balance of these forces and processes, are generally viewed as flowing passively from the perivascular space to the pulmonary lymphatics that join the thoracic duct. Although large lymphatic channels are valved and have pulsatile contractions, the major mechanism of interstitial lymph clearance has been assumed to be the pressure difference between the pulmonary capillaries and the junction of the thoracic duct and jugular vein (16, 19). If this were purely the case, there should be no difference in clearance of lung liquid among various physiological conditions, if the sum of hydrostatic and oncotic pressures differences was the same.

Because the rate of increase in lymph flow is so different between exercise and left atrial hypertension at similar hydrostatic measures, we hypothesized that another mechanism of clearance must come into play. What we found was that lung water clearance was greatly enhanced by hyperpnea to such a degree that it is likely that ventilatory excursions are the major mechanism keeping the lungs dry during exercise. Prior studies of the effect of ventilation on lung water have not generally been done under edemagenic conditions, nor have they reproduced the extreme ventilation of exercise (21). The effect of passive hyperpnea in the resting, unanesthetized sheep has been previously measured to cause a transient 30% increase in lymph flow (22). These data suggest that hyperpnea alone has a small effect on clearing resident interstitial fluid but that hyperpnea does not significantly alter Starling factors responsible for transvascular filtration. Thus, unless there is large transvascular filtration, hyperpnea would not be observed to facilitate clearance. In situations in which hyperpnea is induced by ventilatory pump loading, usually via fixed respiratory resistance or negative inspiratory pressure threshold, lymph flow rises and lymph protein content decreases, consistent with a net increase in transvascular pressure gradient (24).

These data reconcile the puzzling difference in the rate and magnitude of lymph flow between left atrial hypertension and exercise. In acute severe left atrial hypertension (as high as 32 mm Hg) lung lymph flow rises to a peak after only 2-4 h, compared with seconds to minutes during exercise, and the lymph protein washdown may continue for as long as 2-4 h, suggesting a considerable degree of interstitial mixing related to slow clearance (9, 10). In exercise, the lymph protein content decreases over minutes.

It is not clear what aspect of hyperpnea facilitates clearance of filtered fluid during exercise. The lung inflation associated with spontaneous breathing is associated with a small decrease in interstitial pressure that favors transvascular filtration (16, 20). Interstitial volume is variably affected; alveolar septal wall volume may decrease whereas perivascular volume increases via circumferential traction by collagen fibers as the lung expands (16). These pressures and tissue volume changes favor movement of fluid and solute from the alveolar membrane toward bronchial lymphatics. Because bronchial lymphatics are valved (20), the oscillations of lung pressure and volume may create a pumping action that facilitates flow away from alveolar tissue (10, 25, 26). This process may be augmented by the high stretch of exercise ventilation where tidal volume excursions increase.

The effect of lymphatic peristalsis in augmenting lung liquid clearance is unknown. It is known that lung lymphatics have periodic contractions of a magnitude of several mm Hg and that with catecholamine stimulation lymphatic pressure may rise to 20-30 mm Hg, but the effect on bulk flow of fluid is unknown. Presumably, experiments where lymphatic peristalsis is paralyzed might help determine the magnitude of this function on lymph clearance.