INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Inhaled nitric oxide (NO) has received attention as a selective pulmonary vasodilator improving blood gases in acute lung injury. These actions are due to rapid inactivation by binding with hemoglobin, which limits potential systemic vasodilation, and to delivery to the lung regions with the highest ventilation/perfusion (A/) relationships, which decreases pulmonary shunt (1). In contrast, intravenously administered vasodilators frequently cause systemic hypotension and worsen A/ mismatch because of a uniform decrease in pulmonary vascular tone (1). Thus, inhaled NO has been reported to decrease pulmonary vascular resistance (PVR) in patients with the adult respiratory distress syndrome (ARDS) (2) and in several animal models of acute lung injury (5). In all these studies, the pulmonary vascular effects of inhaled NO have been of variable importance, occasionally nonexistent in some subjects, and not clearly related to associated improvement in gas exchange. We wondered, therefore, whether these variable effects of inhaled NO might be related to differences in site of action.

Inhaled NO has been reported to decrease precapillary resistance in isolated perfused lungs injured by oxidants (8) or by electrolysis (9). We are not aware of studies on the site of action of inhaled NO and in clinical or intact animal acute lung injury. In isolated-perfused lungs of various species after different vasoconstrictor stimuli, the site of action of inhaled NO has been reported to be predominantly precapillary (9), at large and small arteries and small veins (10), at small arteries and veins (11), or equally distributed to all segmental resistances (12).

We investigated the site of pulmonary vasodilation by inhaled NO in acute lung injury induced by the injection of 100 µm glass beads in dogs. This model is characterized by lung edema, with little increase in pulmonary capillary filtration pressure, severe pulmonary hypertension caused mainly by partially reversible precapillary resistance, and arterial hypoxemia caused by an increased pulmonary shunt (13, 14). We also studied the effects of additional cyclooxygenase inhibition, which decreases pulmonary shunt in oleic acid lung injury (15) and in patients with ARDS (16), probably through an enhancement of hypoxic pulmonary vasoconstriction. An increase in pulmonary vascular tone may augment the improvement of gas exchange induced by inhaled NO in acute lung injury (4, 17).

In the present experiments, PVR was partitioned in arterial and (capillary + venous) segments by effective capillary pressure (Pc') measurements using the analysis of the pressure decay curve after arterial balloon occlusion (18). Pulmonary vascular pressures were compared at several levels of flow to discriminate between active, tone-dependent, and passive, flow-dependent, changes. Gas exchange variables were also compared at constant flow to avoid flow-dependent changes in A/ matching.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Preparation

Thirteen mongrel dogs (mean weight, 27 kg; range, 20 to 35 kg) were anesthetized intravenously with pentobarbital sodium (25 mg/kg), paralyzed intravenously with pancuronium bromide (0.2 mg/kg), and ventilated with a servo-ventilator (Elema 900 B; Siemens Elema, Sölna, Sweden) via a cuffed endotracheal tube. The inspired fraction of O₂ was 0.4, the respiratory rate was 12 breaths/min, and the tidal volume was 10 to 15 ml/kg adjusted to obtain an arterial PCO₂ between 35 and 45 mm Hg. Pentobarbital (2 mg/kg) and pancuronium (0.2 mg/ kg) were repeated hourly to maintain anaesthesia and to prevent spontaneous respiratory efforts. Sodium bicarbonate was given as required to maintain arterial pH above 7.30. Temperature was maintained at 37 to 38° C by use of an electrical heating blanket. The dogs were lying supine. The experiments were conducted in accordance with the Guiding Principles in the Care and the Use of Animals as approved by the Council of the American Physiological Society.

A thermodilution balloon-tipped pulmonary artery catheter (Model 131H-7F; Baxter Edwards, Irvine, CA) was inserted via the right external jugular vein and positioned by means of pressure monitoring in a branch of the pulmonary artery for measurements of pulmonary artery pressure (Ppa), pulmonary artery occluded pressure (Ppao), right atrial pressure (PRA) and central temperature, estimation of effective pulmonary capillary pressure (Pc') from the pressure decay curve after balloon occlusion, and mixed venous blood sampling. A polyethylene catheter was inserted in the abdominal aorta via the right femoral artery for systemic blood pressure (Psa) measurements and arterial blood sampling. A balloon catheter (Percor Stat-DL 10.5 F; Datascope, Paramus, NJ) was advanced into the inferior vena cava through a right femoral venotomy. Inflation of this balloon produced a titratable decrease in cardiac output () by reducing venous return. A large-bore polyethylene cannula was inserted into the left femoral artery and vein to act as an arteriovenous fistula. Closing this fistula resulted in a decrease in by an average of 0.4 L · min¹ · m². Thrombus formation along the catheters was prevented by 100 U/kg of sodium heparin administered intravenously just before the insertion.

Measurements

Pulmonary and systemic arterial pressures were measured using Gould Statham P50 transducers (Gould Inc., Oxnard, CA). The pressure transducers were zero referenced at midchest, and vascular pressures were measured at end-expiration. Heart rate (HR) was determined from a continuously monitored electrocardiographic lead. was measured by thermodilution using injections of 10 ml of 0.9% sodium chloride at 0° C, a computer (9520-A; Edwards Laboratories, Santa Ana, CA), and an automated pneumatic pump electronically synchronized on the ventilatory cycle, and it was calculated as the mean of three determinations. Arterial and mixed venous blood gases were measured immediately after drawing the samples by an automated analyzer (ABL 2; Radiometer, Copenhagen, Denmark) and corrected for temperature.

Venous admixture (VA/T), percent of total blood flow, was calculated as (capillary O₂ content  arterial O₂ content)/(capillary O₂ content  mixed venous O₂ content), where capillary O₂ content is estimated with the alveolar PO₂ and O₂ saturations determined from a nomogram (19). Intrapulmonary shunt was measured using a tracer gas of low solubility, sulfur hexafluoride (SF₆), and the standard shunt equation (20). Starting 30 min before the first measurements, a constant infusion of SF₆ was administered (5 ml · min¹) in the left external jugular vein. Samples of 10 ml of arterial and mixed venous blood were simultaneously withdrawn, equilibrated with 35 ml of N₂ in a heated bath for 45 min, and analyzed for SF₆ by an electron capture detector (5890 A gas chromatograph; Hewlett-Packard, Palo Alto, CA).

The pulmonary vascular pressure signals were sampled at 200 Hz using an analog/digital converter (RTI 800; Analog Device, Norwood, MA) and stored and analyzed on a personal computer. Pc' was computed in triplicate from the Ppa decay curves after inflation of the balloon of the pulmonary artery catheter. For this measurement the dog was disconnected from the ventilator for a period of 8 s. Time zero was defined as the time when pulmonary arterial pressure began to deviate from the normal wave. This instant was chosen because in an intact animal (or clinical) setting, pressure is the only signal usually available for this purpose. A monoexponential curve was fitted to a set of data between 0.2 and 2.0 s after the occlusion, adjusted for Ppao, and extrapolated back toward time 0 + 150 ms. This time lag after occlusion was chosen because it has been shown to correspond to the delay from the beginning of occlusion to zero flow reached in the capillaries (21). A typical Ppa decay curve with analysis for Pc' is shown in Figure 1.

View larger version (27K): [in this window] [in a new window]   Figure 1.   Typical pulmonary artery pressure decay curve with measurements of effective pulmonary capillary pressure (Pc') and pulmonary artery occluded pressure (Ppao) after inflation of the balloon of the pulmonary artery catheter. Exponential fitting and Pc' 150 ms after balloon inflation are automatically calculated using original software.

Protocol

As soon as steady-state conditions (stable HR, Psa, and Ppa for 20 min) were ensured, a baseline at first P/ plot was generated from Ppa, Pc', and Ppao determinations with open arteriovenous fistula (1 point), after closing the fistula (1 point), and then after stepwise inflations of the inferior vena cava balloon with occluded fistula (3 points). The average time to generate a P/ plot was 30 min. Embolization with 100 µm glass beads (Sigma, St. Louis, MO) was carried out progressively for 30 min until mean Ppa reached 40 mm Hg. This pulmonary hypertension stabilized with a Ppa around 30 mm Hg during a subsequent 30-min period, after which a second P/ plot was constructed. Thereafter in eight of the dogs, inhaled NO was instituted through a catheter inserted into the tracheal tube with the flow adjusted to produce an inspired concentration of 80 ppm. This concentration was chosen as the upper limit of a range previously reported to be well tolerated in intact animals without systemic hemodynamic effects or increased circulating methemoglobin (1). NO was supplied from a pure NO source tank (Air Liquide, Haren, Belgium) and delivered into the tracheal tube with a syringe pump. The inspired concentration of NO was verified by the chemiluminescence after calibration against standard NO concentrations (Model 42; Thermo-Environmental, Franklin, MA). Inspired or expired NO₂ remained below 1 ppm.

After 10 min at steady state, a third 5-point P/ plot was constructed. Thereafter, inhaled NO was stopped, the eight treated dogs received 1 g aspirin intravenously, and, after 60 min, a fourth P/ plot was constructed. Inhaled NO 80 ppm was resumed in the eight dogs, and a last 5-point P/ plot was obtained after 10 min at steady state. Five dogs randomly chosen were taken as controls. In this control group, P/ plots were constructed at 60, 100, 190, and 230 min after embolization.

Blood gases were measured at the highest and at the lowest of each P/ plot. S/T and blood gases were measured at an intermediate around 2.5 L · min¹ · m².

Mathematical Modeling of the Arterial Occlusion Data

To ensure that the analysis of the arterial occlusion data chosen in the present experiments gave a good estimate of Pc', the reference electrical analog model of the pulmonary vascular bed of a single compartment with a large capillary compliance between arterial and venous resistances (18) was modified to provide for the parallel structure of the pulmonary vascular bed at the level of small arteries occluded by the 100 µm glass beads (Figure 2). The equations predicting the transient potentials at the site of the occlusion (corresponding to Ppa) and at the occluded and nonoccluded small arteries (corresponding to Pc') were adapted accordingly, and they are presented in detail in the Appendix . As illustrated in Figure 3, an adequate monoexponential fitting between 200 ms and 2 s was predicted at baseline and when the fraction of obstructed small pulmonary arteries was smaller than 0.5. When the fraction of obstructed small pulmonary arteries was above 0.5, Pc' given by the model was slightly higher than the computed value. Previous modeling confronted with hemodynamic and angiographic measurements has indicated that, in acute microembolic pulmonary hypertension induced by the injection of 500 µm glass beads in dogs, a Ppa stabilized around 30 mm Hg at uncontrolled , as in the present experiments, corresponds to a 50 to 60% obstruction of the pulmonary arterial bed (22).

View larger version (10K): [in this window] [in a new window]   Figure 2.   Electrical analog model for the pulmonary vascular bed. R1, R2, and R4 figure the resistances of the pulmonary arterial tree; R3, R5, and R6 figure the resistances of the venous tree; and C2 and C3 figure the compliance of the capillary sheet. F is the fraction of embolized arterioles when the switch SW2 is open. R2, R3, and C2 can also be computed as a fraction of R4, R5, and C3, respectively. G figures a current generator with a constant value before and after embolization equal to i (constant flow conditions). Vi, VO, VA, and VC represent the potential at the input of the pulmonary vascular bed (equivalent to Ppa), at the site of occlusion, and at two capillary levels (equivalent to Pc'), respectively. The opening of SW1 allows the simulation of arterial occlusion.

View larger version (21K): [in this window] [in a new window]   Figure 3.   Transient potentials recorded at the site of occlusion (VO) and at two capillary levels (VA and VC) before embolization (top panel ), after 33% embolization (middle panel ), and after 66% embolization (bottom panel ). In addition a monoexponential fitting between 200 ms and 2 s, as used for fitting the pressure transient data after pulmonary arterial occlusion in the dogs, was added to allow comparison with the simulated data (VO). Electrical potentials (mV) are equivalent to pressure (mm Hg) in hydraulic circuits.

Statistical Analysis

Results are expressed as means ± SE. Inspection of the individual P/ plots showed them to be essentially rectilinear, and thus a linear least-squares regression analysis was used to compute slopes and extrapolated pressure intercepts for each of them. To obtain composite Ppa/, and Pc'/ plots for each group of dogs, pressures interpolated from the regression analysis of individual dogs were averaged at 0.5 L · min¹ · m² intervals of from 2 to 4 L · min¹ · m². The blood gases and hemodynamic data were analyzed by a repeated-measures analysis of variance. When the F-ratio of the analysis of variance reached a p < 0.05 critical level, specific comparisons were made using modified t tests, that is, t tests computed with the residual variance of the ANOVA (23).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Effects of Microembolism

Injection of 100 µm glass beads increased Ppa and Pc' with no effect on , HR, Psa, Ppao, and PRA (Tables 1 and 3). In the control group, at uncontrolled flow decreased at 100 min, and Ppa and Pc' were the highest at 60 min and stabilized at a slightly lower level thereafter (Table 1). Gas exchange deteriorated, with mainly a decrease in arterial PO₂, and increases in AaPO₂, VA/T, and S/T (Tables 2 and 4). These changes remained stable in the control group (Table 2). Both Ppa/ and Pc'/ plots were shifted to higher pressures, and these changes appeared stable over time in the control group (Figures 4 and 5). The arterial component of PVR, defined as (Ppa  Pc')/(Ppa  Ppao) at the intermediate of 3 L · min¹ · m², decreased from 77 ± 5 to 66 ± 5% in the control group, and from 77 ± 4 to 63 ± 4% in the treatment group (p < 0.05 and p < 0.001, respectively) (Figure 6).

View larger version (18K): [in this window] [in a new window]   Figure 4.   Composite plots of mean pulmonary arterial pressure (Ppa) minus pulmonary artery occluded pressure (Ppao) versus flow () in the control group (left panel ) (n = 5) and in the treated group (right panel ) (n = 8). Embolism T + 60 min is shown by the dashed line; Embolism T + 100 min (NO 80 ppm in the treated group) is shown by the dotted line; Embolism T + 190 min (ASA 1 g in the treated group) is shown by the dotted chain line; Embolism T + 230 min (ASA 1 g + NO 80 ppm in the treated group) is shown by the long dash line. *p < 0.05, compared with embolism at the same .

View larger version (18K): [in this window] [in a new window]   Figure 5.   Composite plots of effective pulmonary capillary pressure (Pc') minus pulmonary artery occluded pressure (Ppao) versus flow () in the control group (left panel ) (n = 5) and in the treated group (right panel ) (n = 8). See legend to Figure 4 for indications of the various embolisms. *p < 0.05, compared with baseline at the same .

View larger version (20K): [in this window] [in a new window]   Figure 6.   Pressure drop at a constant flow of 3 L · min1 · m2 across the arterial (open columns) and the capillary + venous (hatched columns) segments in the control group (left panel ) (n = 5) and in the treated group (right panel ) (n = 8). Numbers inside the arterial columns are the percentages of pressure drops across the arterial segment (arterial component of pulmonary vascular resistance).

Inhaled NO

Inhaled NO had no effect on hemodynamics (Table 3), indices of gas exchange (Table 4), or Pc'/ plots, but it shifted Ppa/ plots to lower pressures at high (Figures 4 and 5). Inhaled NO did not affect the arterial component of PVR (65 ± 4 versus 63 ± 4%, p = NS) (Figure 6).

Aspirin

Aspirin has no effect on hemodynamics (Table 3) or indices of gas exchange (Table 4), but it shifted Ppa/ plots to higher pressures (Figure 4). Aspirin increased the arterial component of PVR (72 ± 4 versus 63 ± 4%, p < 0.001) (Figure 6).

Aspirin Combined with Inhaled NO

Aspirin combined with inhaled NO had no effect on hemodynamics (Table 3) or gas exchange (Table 4), and it shifted Ppa/ plots back to pretreatment values (Figure 4), with no effect on the partitioning of PVR (64 ± 4 versus 63 ± 4%, p = NS) (Figure 4).

Changes in Cardiac Output

Both Ppa and Pc' decreased with decreasing flow, but without significant change in the partitioning of PVR.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present results show that small glass beads pulmonary embolism in intact dogs is associated with an increase in both arterial and effective capillary resistance, which is partially reversible, mainly on the arterial side, by inhaled NO and by dilating products of the cyclooxygenase pathway of arachidonic acid metabolism. Pulmonary vascular tone does not appear to affect gas exchange in this model of acute lung injury.

The arterial occlusion technique, originally developed in isolated lungs perfused in steady flow (24), has been extensively used in intact animals and in patients for the estimation of capillary pressure as a major determinant of fluid filtration and edema in the lung (18). A variety of electrical analog reference models and methods of analysis of the Ppa decay curve have been reported (18). Gilbert and Hakim (21) recently showed in intact dogs equipped with a balloon-tipped catheter in the left pulmonary artery, a left lower lobe artery flow probe and a laser Doppler flow probe on the surface of the left lower lobe that, after balloon inflation, pulmonary artery pressure and flow decreased simultaneously, lobar arterial flow reached zero after an average of 72 ms, and microvascular flow changed on average 80 ms after lobar flow. On the basis of these observations, these investigators proposed that the best estimate of Pc' would be obtained by a back extrapolation to 152 ms after the initial change in pulmonary artery pressure of a single exponential fitting of the pressure decay curve between 0.2 to 2 s (21). To ensure that the monoexponential fitting used to estimate Pc' remained adequate after embolization, we modified the reference simple electrical analog model of a single compartment with a large capillary compliance between arterial and venous resistances (18) to provide for the parallel structure of the pulmonary vascular bed at the level of small arteries occluded by the 100 µm glass beads. This model predicted that a monoexponential fitting of the Ppa decay curve is a reasonable approximation before as well as after small glass bead embolism. After embolization we still used a time lag of 150 ms after occlusion to estimate Pc'. The rationale behind this was that embolization increased the arterial resistance but also decreased the arterial compliance (22), leading to little change in the time constant of the arterial tree. Using the method proposed by Gilbert and Hakim (21), we found that the arterial component of PVR amounts to about 77%, which is somewhat higher than found in other arterial occlusion studies in intact dogs (25).

In the present experiments, 100 µm glass bead embolism increased Pc' proportionally more than Ppa so that the arterial component of PVR decreased. This observation poses the problem of the anatomic correlate of Pc'. Ehrhart and Hofman (13) previously reported that 100 µm glass bead embolism in isolated dog lung lobes increased Pc' more than Ppa, but hardly changed double occlusion pressure (Pdo) as measured after simultaneous arterial and venous occlusion, and which gave an excellent approximation of capillary pressure at isogravimetry. Because previous studies have shown that more than 90% of beads obstruct vessels of the same diameter (29), the investigators reasoned that the vascular segment represented by Pc'  Pdo was the major site of mechanical obstruction and vasoconstriction, and likely included 100 µm arteries (13). Maarek and colleagues (30) found also that embolization with 100 µm glass beads increased Pc' more than Pdo in isolated perfused dog lungs. Hakim and Kelly (31, 32) applied different techniques, including micropuncture, small retrograde catheter, and arterial and venous occlusion to isolated perfused dog lungs, and they showed that arterial and venous occlusion measures pressures in vessels that are > 50 µm in diameter, and very likely close to 100 µm. Taking into account that in our dogs, the glass beads had diameters ranging from 75 to 150 µm (14), it seems reasonable to assume that the increases in Ppa and in Pc' with a slight decrease in the arterial component of PVR reflects obstruction of small arteries distributed around a diameter of 100 µm at the periphery of the arterial vascular tree.

After small bead embolism, inhaled NO decreased Ppa, confirming previous reports of partial reversibility by papaverine of pulmonary hypertension in this experimental model (13). Inhaled NO affected, but not significantly, the partitioning of PVR, also suggesting a site of action predominantly at small arteries of the same diameter or slightly larger. Roos and colleagues (11) showed by arterial, venous, and double occlusion that 170 and 670 ppm inhaled NO in endothelin-constricted isolated blood-perfused rat lungs decreased resistance in small arteries and veins, with no effect on larger capacitance arteries and veins (11). Using the same technique, Rimar and Gillis (9) showed in isolated rabbit lungs perfused with Krebs solution containing dextran and indomethacin that 120 ppm inhaled NO affected primarily the difference between Ppa and Pdo after administration of either a thromboxane analogue or electrolysis-induced acute lung injury (9). Using the same technique, Tod and coworkers (10) showed in isolated blood-perfused lamb lungs with hypoxia or thromboxane analogue-induced vasoconstriction that inhaled 45 ppm NO decreased resistance in large and small arteries, and also in small veins but not in large veins. These responses to NO were not affected by cyclooxygenase inhibition by indomethacin or by inhibition of endogenous NO synthase by L-NNA (10). Kavanagh and colleagues (8) showed that 90 to 120 ppm inhaled NO decreased Ppa  Pdo in isolated buffer-perfused rabbit lungs with oxidant-induced lung injury. Lindebord and coworkers (12) reported that 5 to 240 ppm inhaled NO caused a dose-related proportional reduction in all segmental resistances in buffer- or blood-perfused rabbit lungs before and after a thromboxane analogue-induced vasoconstriction. These observations in experimental models with variable increases in Ppa and a wide range of concentrations of NO, and our findings, are compatible with the idea that inhaled NO acts predominantly on resistance vessels, arteries, and/or veins, with increased tone at a short diffusion distance from the alveoli.

In the present experiments, cyclooxygenase inhibition by aspirin prevented the pulmonary vasodilating effects of inhaled NO. In isolated perfused lungs, cyclooxygenase inhibition by indomethacin did not prevent the vasodilating effects of inhaled NO (9, 10). Administration of pulmonary vasoconstrictors such as almitrine in patients with ARDS (4) or a L-arginine analogue in lavage-induced acute lung injury (7) did not prevent pulmonary vasodilation and associated improvement in gas exchange induced by inhaled NO. There is no easy explanation available for these discrepancies, especially since our arterial occlusion measurements suggest the same site of action for endogenous vasodilating products of cyclooxygenase and inhaled NO, that is, at vessels at a diameter equal or slightly higher than the injected 100 µm glass beads.

An increase in flow has been reported to decrease the tone of small pulmonary arteries as a consequence of shear-stress-induced release of endothelium-derived relaxing factors in isolated perfused lungs (32). However, over the range of flows studied in the present experiments, no flow-induced decrease in the arterial component of PVR could be evidenced. Larger changes in flow are probably necessary to release endothelial-derived vasodilating mediators in intact dogs.

In our dogs, venous admixture was slightly higher than true pulmonary shunt measured by the SF₆ method. This is in keeping with studies using the multiple inert gas elimination technique, which established that hypoxemia in ARDS results mainly from perfusion to unventilated lung units, with, in about one third of cases, some additional perfusion to lung units with a low A/ (33). However, neither NO nor aspirin improved gas exchange, which is at variance with several studies on clinical (2, 16) and experimental (5, 7, 15, 17) ARDS. It may be noted that individual responses to inhaled NO may sometimes be variable, or absent. Rossaint and colleagues (34) observed that 17% of patients with ARDS did not response to inhaled NO. Contrary to Putensen and coworkers (17), Romand and colleagues (35) found that NO did not improve gas exchange in dogs with oleic-acid-induced acute lung injury. The reasons for these discrepancies are unclear. Young and colleagues (29) showed that the functional unit for pulmonary gas exchange is perfused by vessels with a diameter less than 150 µm. It may be that pulmonary vascular tone does not improve gas exchange in acute microembolic pulmonary hypertension, or, in some cases or models of acute lung injury, it may be caused by changes occurring mainly at a site upstream from the vessels that perfuse functional units.