Nitric Oxide Modulation of Pulmonary Blood Flow Distribution in Lobar Hypoxia 

The study was approved by the Animal Research Ethics Committee of Uppsala University.

Twenty-one pigs of Swedish country breed, weighing 22–35 kg, were used in the study. The animals were premedicated with pentobarbital (20 mg *symbol* kg¹) and 0.5 mg atropine intraperitoneally. Anesthesia was induced 30 min later with 180–360 mg intravenous pentobarbital and was maintained with a continuous infusion of chlormetiazole [16](800–1,600 mg *symbol* h¹) and pancuronium bromide (4–8 mg *symbol* h¹). After induction the animals were placed in the supine position for the remainder of the study.

An intravenous injection of 0.5 mg fentanyl was given before surgery (see below), and additional doses of 0.2–0.3 mg were given as necessary. The level of anesthesia was considered appropriate when the surgical procedure did not cause any changes in heart rate or blood pressure. Isotonic saline, 10 ml *symbol* kg¹*symbol* h¹, was given for hydration.

A tracheotomy was performed and a cuffed endotracheal tube with an inner diameter of 6.0 mm was inserted. During the preparation, adequate ventilation was supplied by a servomechanism-controlled ventilator (Servo 900 C, Siemens Elema, Lund, Sweden). The respiratory frequency was maintained at 20 breaths/min, and the inspired fraction of oxygen was 0.4. Another cuffed endotracheal tube (4.5 mm internal diameter) was introduced through the tracheal stoma to the left of the main tube. An introducer inside this smaller tube enabled positioning in the left lower lobe (LLL) bronchus. By inflating the cuff, the LLL was isolated and could be ventilated independently. The position of the tube could be checked and adjusted at the site of the thoracotomy so that the LLL was inflated with the smaller tube but not when ventilation was provided by the main tube. This arrangement was tested by submerging the tube leading to the LLL under water and checking for air leakage when the remaining lung regions were ventilated. It also was possible to ensure that the left middle and upper lobes and the right lung were ventilated by the main tube. The LLL was ventilated with an additional volume-controlled ventilator (Servo 900 C. Siemens Elema) synchronized with the main ventilator. Thirty percent of the total minute ventilation was administered to the LLL, which corresponded to the LLL's weight as a percentage of the total weight of both lungs.* After thoracotomy, both ventilators were adjusted to give a positive end-expiratory pressure of 5 cmH²O. The inspired fraction of oxygen was increased to 0.8, and respiratory frequency was maintained at 20 breaths/min with tidal volumes adjusted to obtain an arterial carbon dioxide tension (Pa^COsub 2) of 37–47 mmHg. At the beginning of each measurement period (see Experimental Protocol, below) the lungs were ventilated three times with two times normal tidal volume to prevent atelectasis.

A mixture of 200 ppm nitric oxide in pure nitrogen was administered with volumetrically calibrated flowmeters (AGA, Lidingo, Sweden) and was connected to the low-flow inlet of the ventilator with a Y-piece. The oxygen cell of the ventilator was used to measure the inspired fraction of oxygen. The inspired gas passed through a canister containing soda lime to absorb any nitrogen dioxide. The concentrations of inspired nitric oxide and nitrogen dioxide were measured close to the endotracheal tubes by means of chemiluminescence (9841 NOx, Measurement Controls, Englewood, CO). Inspired nitrogen dioxide was less than 0.5 ppm.

An ear vein was cannulated for inducing and maintaining anesthesia. A 7-French triple-lumen balloon-tipped catheter (Swan-Ganz) was introduced via the right external jugular vein to the pulmonary artery for blood sampling, pressure recording, and cardiac output (Q with dot^T) measurements. A large-bore catheter was inserted into the contralateral jugular vein for infusion, with its tip in the superior caval vein. The right carotid artery was cannulated to measure arterial blood pressure and to sample blood gases. The catheters were connected to appropriate pressure transducers (Sorenson Transpac, Abbott Critical Care Systems, North Chicago, IL).

By means of a left-sided lateral thoracotomy, the artery to the LLL was identified and freed from surrounding tissues. The artery was enclosed in an ultrasonic flow probe, which was connected to a flowmeter (probe 6 SB and flowmeter T208, Transonic, Ithaca, NY) for continuous measurement of blood flow to the LLL (Q with dot^LLL).

Mean arterial pressure (MAP), mean pulmonary arterial pressure (MPAP), and pulmonary capillary wedge pressure (PCWP) were recorded (7010 monitor, Marquette Electronics, Milwaukee, WI). Pressures were averaged over the entire respiratory cycle, and the midthorax was used as a zero reference level.

Q with dot^Twas measured by means of thermodilution: 10 ml isotonic saline at room temperature was injected as a bolus, and the Q with dot^Twas computed (7010, Marquette). Three injections were given for each measurement, and the mean was calculated. The injections were evenly distributed over the respiratory cycle.

Mixed venous and arterial blood samples were collected for blood gas analysis (ABL 3, Radiometer, Copenhagen, Denmark) and measurement of oxygen saturation and hemoglobin concentration (OSM 3, Radiometer). Mean airway pressure and expired minute volume were recorded from both ventilators on each measurement occasion.

The protocol is shown in Figure 1. After preparation, 30 min was allowed before a set of baseline measurements was made. Each set of measurements consisted of MAP, MPAP, heart rate, PCWP, Q with dot^T, Q with dot^LLL, arterial and mixed venous blood gas tensions, arterial and mixed venous blood oxygen saturations, and airway pressures. After baseline measurements had been recorded, the LLL was ventilated with a hypoxic gas mixture for the remainder of the study (5% O², 5% CO², and 90% N²; AGA). This gas mixture was used to obtain a zero gas exchange in the LLL, by establishing approximately the same oxygen and carbon dioxide tensions in the alveoli as in the mixed venous blood. After 30 min of LLL hypoxia, a new set of measurements was performed, and subsequently the animals were divided into two groups.

In the first group (n = 9) 40 ppm nitric oxide was administered according to one of the following procedures:(1) to the hypoxic LLL, (2) to the hyperoxic right lung and upper and middle lobes of the left lung, or (3) to both lungs. Procedure 1 was performed first in four pigs and procedure 2 first in the other five; when comparing these two subgroups no differences were found. Procedure 3 always took place at the end of the experiment. Each phase lasted 30 min and was terminated with measurements.

In the second group (n = 6) an intravenous dose of N^omega-nitro-L-arginine methyl ester (L-NAME) was infused (30 mg/kg). The L-NAME was dissolved in 20 ml isotonic saline immediately before the infusion, which was given over a period of 10 min. New measurements were obtained 5 and 20 min after the infusion was terminated. Thereafter 40 ppm nitric oxide was administered to (1) the hypoxic LLL or (2) the hyperoxic lung regions. Again, each phase lasted 30 min and was terminated with measurements.

In six additional pigs, the same experimental protocol as described above was performed. After measurements at baseline and during lobar hypoxia, nitric oxide was administered to the hypoxic lobe at doses of 5, 10, 20, 40, 80, and 160 ppm, and measurements were performed after 30 min at each dose.

At the end of the experiment, the animals were killed with an intravenous injection of KCl, and the lungs were removed for inspection.

The fractional Q with dot^LLLwas calculated as Q with dot^LLL/Q with dot^T. PVR was calculated according to the formula [17](MPAP -PCWP/Q with dot^T) x 80 (where pressures are expressed in millimeters mercury, Q with dot^Tin liters per minute, and PVR in dynes *symbol* seconds *symbol* centimeters [5]). The vascular resistance of the LLL was calculated as (MPAP - PCWP/Q with dot sub LLL) x 80 and that of the hyperoxic lung regions was calculated as (MPAP - PCWP/Q with dot^T- Q with dot^LLL) x 80 and expressed in the same units as PVR.

Rubertsson et al. [18] have shown a difference between the Q with dot^Tof the pig in the main pulmonary artery when measured by thermodilution (Q with dot^Tthermo) or by ultrasonic flow probes (Q with dot^Tprobe), thermodilution giving higher values. The equation Q with dot^Tthermo = Q with dot^Tprobe x 1.26 + 0.38 (r = 0.98, P < 0.01) provided good correlation. The Q with dot^Tmeasured by thermodilution was therefore recalculated according to the formula of Rubertsson et al. [18] The absence of a correction would not affect the relation between Q with dot^Tand Q with dot^LLL.

All data in the tables are given as means plus/minus SEM. Comparisons were made by analysis of variance for repeated measurements. In all analysis of variance tests, Fisher's least significance method was used. Differences were considered significant at P < 0.05.

Only Q with dot^LLLis given as mean plus/minus SEM in the text; for most other data, see Table 1and Table 2and Figure 2, Figure 3, Figure 4. All differences described are statistically significant (P < 0.05), unless otherwise stated. The data obtained for baseline and LLL hypoxia in Results and in Figure 2apply to both groups together; data for the separate groups are given in Table 1and Table 2.

At baseline, the Q with dot^T, systemic pressure, MPAP, and PCWP demonstrated similar values to ones previously reported for mechanically ventilated pigs with healthy lungs. [19] The Q with dot sub LLL was 23 plus/minus 2.1% of the total Q with dot^T, a value that is in accordance with a previous report. [19] However, Q with dot^LLL/Q with dot^Twas on average lower in the pigs that later received L-NAME (group 2; mean 18 plus/minus 2.6%) than in those who did not (group 1; mean 27 plus/minus 2.4%). Arterial oxygenation was good: mean arterial oxygen tension (Pa^O2) was greater than 400 mmHg. The animals were normocapnic with a Pa^CO2averaging 42 mmHg.

LLL hypoxia caused an increase in MPAP of 4 mmHg from a mean baseline value of 17 mmHg. The Q with dot^Tincreased slightly by an average of 0.5 1 *symbol* min sup -1. The increase was only significant in the pigs that later received L-NAME (group 2). The Q with dot^LLL/Q with dot^Twas markedly reduced (to a mean 6 plus/minus 1% for both groups). No changes in PVR were seen. The vascular resistance in the hypoxic LLL was markedly increased, while that in the hyperoxic regions remained essentially unaltered. The Pa^O2decreased to an average of 275 mmHg and the Pa^CO2increased by 5 mmHg in the second group.

The infusion of L-NAME caused a prompt and persistent effect on the systemic and pulmonary circulation. Recordings obtained 5 and 20 min after infusion did not differ. The data presented are from the 20- min recordings. These showed significant increases in MAP and MPAP, and also in the PCWP. PVR increased by a factor of 3. The Q with dot^Twas decreased when compared with baseline values and when compared with LLL hypoxia results. The Q with dot^LLL/Q with dot^Tdecreased in all six pigs from a mean of 4.2 plus/minus 0.7% to 2.2 plus/minus 0.8%, but this reduction was not clearly significant by analysis of variance that also included the large change in Q with dot^LLL/Q with dot^Tby hypoxia per se (see Figure 3). At the same time there was a significant improvement of arterial oxygenation.

The delivery of 40 ppm nitric oxide to LLL did not affect systemic pressure or MPAP nor Q with dot^T. The vascular resistance in the hypoxic LLL decreased by approximately 75%. Q with dot^LLL/Q with dot^Tincreased during hypoxia from 7 plus/minus 1.4% to 18 plus/minus 2.7%, but the latter value was still significantly reduced, to 66% of baseline. The increase in Q with dot^LLL/Q with dot^Twas accompanied by a reduction in Pa^O2.

When 40 ppm nitric oxide was administered to both lungs at the same time during LLL hypoxia, Q with dot^LLLand Pa^O2did not differ from values obtained when nitric oxide was delivered to the LLL alone, nor did the vascular resistance in the LLL alter, but there were reductions in PVR and in the vascular resistance of the hyperoxic regions.

The effect of inhaling nitric oxide for 30 min at stepwise increased concentrations (0, 5, 10, 20, 40, 80, and 160 ppm) to the hypoxic LLL was tested in six pigs. Figure 4shows the results for Q with dot^LLL/Q with dot^T. As small a concentration as 5 ppm nitric oxide produced a Q with dot^LLL/Q with dot^Tratio of 60% of the baseline value. With 40 ppm nitric oxide, the Q with dot^LLL/Q with dot^Tincreased to 83% of baseline, but neither 80 nor 160 ppm nitric oxide led to any further significant increase in the Q with dot^LLL/Q with dot^T, which leveled at 87% of the baseline value. Thus, the results show a dose-response relation with almost complete response at 40 ppm nitric oxide.

When 40 ppm nitric oxide was administered to the hypoxic LLL after the infusion of L-NAME the MPAP diminished markedly, although it was still higher than the baseline value. The vascular resistance in the hypoxic LLL was dramatically reduced and did no longer differ from the baseline value. The vascular resistance in the hyperoxic regions also decreased by approximately 50%. This resulted in an overall decrease in PVR, to less than half the value measured after the infusion of L-NAME during LLL hypoxia. The Q^LLL/Q^Tincreased to 88 plus/minus 9.3% of the baseline value, an increase that, on average, was larger than that registered when nitric oxide was delivered without L-NAME. Thus the Q^LLL/Q^Tno longer differed from its baseline value. The Pa^O2was markedly reduced as a consequence of the redistribution of blood toward the hypoxic lobe.

When nitric oxide was administered to the hyperoxic lung regions while the LLL was still hypoxic, no effects on pulmonary blood flow distribution were seen and there were no changes in MAP or Q^T. The MPAP was slightly lower than during hypoxia without nitric oxide administration. The arterial blood gases remained unchanged, when compared with data obtained before nitric oxide administration.

When nitric oxide was given to the hyperoxic regions after blockade of endogenous nitric oxide production, the MPAP returned to baseline whereas the MAP remained increased and the Q^Tstayed at the same low level as after L-NAME without nitric oxide inhalation. The Q^LLL/Q^Tdiminished to 0.6 plus/minus 0.4%, or 2.5% of the baseline perfusion of the LLL. Although this decrease was not statistically significant the individual data for the six animals in this group showed a decrease in Q^LLL/Q^Tfor five pigs; in the sixth pig Q^LLL/Q^Tremained at zero (see Figure 3). With this reduction of Q^LLL/Q^Tthere was an increase in Pa^O2from 320 mmHg after L-NAME to 435 mmHg. This increase was unexpectedly large and prompted an analysis of the oxygen content of pulmonary blood. The oxygen content of arterial and mixed venous blood and the venous admixture of the hyperoxic lung regions were calculated using standard formulas. [17,20] To make these calculations, the oxygen content of blood from the LLL was assumed to be the same as that of the mixed venous blood. These calculations showed that the oxygen content of arterial blood increased by 2.8 ml Oxygen²*symbol* l sup -1 blood when nitric oxide was added to the hyperoxic lung regions after L-NAME infusion. Blood was transferred at a rate of 50 ml/min from the hypoxic LLL to the hyperoxic regions and thereby increased the oxygen content of the arterial blood by 1 ml Oxygen²*symbol* l sup -1 blood. At the same time, the venous admixture in the hyperoxic lung regions was reduced from 9.4% to 3.4%, causing an increase in the systemic arterial oxygen content by a further 1.8 ml Oxygen²*symbol* l sup -1. These data suggest an improved ventilation-perfusion matching also in the hyperoxic lung regions when they receive nitric oxide after blockade of endogenous nitric oxide production.

The major findings in the current study are that the degree of vasoconstriction in a hypoxic lobe could be decreased by inhaling nitric oxide or increased by impairing nitric oxide production with an nitric oxide synthase inhibitor. Moreover, by blocking the overall nitric oxide production by means of the nitric oxide synthase inhibitor and simultaneously administrating nitric oxide to the hypoxic lobe, Q sub LLL was almost completely restored. The delivery of nitric oxide to the hyperoxic lung regions but not the hypoxic lobe during nitric oxide production blockade reduced Q^LLL. This reduction, seen in five of six animals (in the sixth, lobar blood flow already had been eliminated after the administration of L-NAME), was followed by a significant increase in Pa^O2. It can thus be concluded that the degree of hypoxic vasoconstriction was modified by administering nitric oxide or inhibiting the synthesis of nitric oxide.

Therefore, nitric oxide was used as a vasodilator to adjust the relative vascular impedances in the hypoxic or hyperoxic lung regions. This finding raises the question as to what extent constitutive nitric oxide synthesis participates in the hypoxic pulmonary vasoconstrictor response. As recently demonstrated, [5] nitric oxide synthase is found in a variety of cell types in human airway and pulmonary endothelium. However, although endogenous nitric oxide very likely interferes with hypoxic pulmonary vasoconstriction, it should be remembered that hypoxic pulmonary vasoconstriction is thought to be a complex response mechanism in which several biochemical modulators are involved. [21].

Nitric oxide is synthesized from L-arginine with the help of nitric oxide synthase. [22] In the current study, a competitive inhibitor, L-NAME, was used to block nitric oxide production. [23] A dose of 10 mg/kg in rabbits eliminated any measurable nitric oxide in the exhaled air. [4] In pigs a dose of 10 mg/kg was shown to decrease pulmonary artery conductance and increase MPAP, [24] and in sheep a dose of 30 mg/kg was shown to increase PVR and PAP for at least 1 h. [13] To achieve a stable L-NAME effect in the current study, which lasted 90–120 min, 30 mg/kg was administered. This dose caused persistent changes in systemic arterial pressure and Q^T, changes that were not seen in the group that did not receive L-NAME. The decrease in Q^Tmay in itself have affected the distribution of pulmonary blood flow, but because Q^Twas constant once L-NAME had been given, the succeeding experiments should hardly have been biased by the L-NAME dose.

It can be argued that the calculation of PVR, and in particular that of a single lobe, suffers from several uncertainties. Thus, if the perfusion in a part of a lung or a lobe has been altered from zone III to zone II or I conditions, the calculation of vascular resistance will be erroneous. Furthermore, the small flow values registered in a hypoxic lobe will produce a large degree of scatter, as can be seen from the standard deviations under prevailing conditions. The variation in vascular resistance must therefore be interpreted with caution.

There is increasing evidence that the bronchial circulation has an influence on the hypoxic pulmonary vasoconstriction and that the bronchial vessels regulate the tone of the larger pulmonary arteries. [25] However, it seems less likely that inhaled nitric oxide exerts any effect via the bronchial vessels because of the inactivation of nitric oxide in the blood by its binding to hemoglobin. In addition, inhaled nitric oxide to one lung was found to have no measurable effect on the vascular tone of the other lung in anesthetized patients during individual lung ventilation. [26] The extent to which blockade of endogenous nitric oxide production by the infusion of L-NAME affects bronchial circulation and influences pulmonary vascular tone remains to be evaluated.

Inhaled nitric oxide has been shown to cause a dose- dependent decrease in MPAP in various animals and humans, [6,7,9] and a decrease in airway resistance. [27,28] Inhaled concentrations as low as 1–5 ppm have been reported to lower pulmonary artery pressure substantially whereas concentrations as great as 300 ppm or more are required to counter bronchoconstriction. The pig is known to develop a very strong vascular response to hypoxia, [29] which may raise the question as to whether the response to nitric oxide in pigs differs from that in humans. A dose-response study in six pigs was therefore undertaken and showed that 40 ppm nitric oxide exerted an almost full effect and that an increase in inhaled nitric oxide concentration to 160 ppm added little to vasodilatation and redistribution of blood flow to the hypoxic lobe. It can thus be concluded that close to a maximum effect of inhaled nitric oxide was obtained in this study.

Inhaled nitric oxide reduced vasoconstriction in the hypoxic lobe and increased the blood flow but did not reestablish control values. However, when endogenous nitric oxide production was blocked by L-NAME infusion, nitric oxide inhalation to the hypoxic lobe had a more marked vasodilatory effect, so that the vascular resistance and blood flow in the hypoxic lobe were no longer different from control values. It may be argued that the increase in vascular resistance in hyperoxic lung regions caused by L-NAME and previously shown in normoxic and hypoxic rabbits [11] forced the blood flow to the hypoxic lobe and caused a passive vasodilatation by distention and recruitment of the vascular bed. Of interest, the vascular resistance in the hyperoxic lung regions (left upper and middle lobes and right lung), which increased threefold after L-NAME infusion, was reduced to almost half when nitric oxide was administered to the hypoxic lobe. Whether this indicates that a reflex mechanism exists between hypoxic and hyperoxic lung regions or is an effect of nitric oxide that has diffused or circulated in the blood is unclear.

When nitric oxide was inhaled to the hyperoxic lung regions after L-NAME infusion blood flow to the hypoxic lobe was reduced in five pigs and remained at zero in the sixth (see Figure 3). At the same time Pa^O2increased significantly. This method of minimizing shunting by constricting vessels in a shunted area has also been tried in animal models of ARDS. In an ovine model of ARDS, Rovira et al. [13] did not find any improvement in gas exchange by adding intravenous L-NAME to inhaled nitric oxide. In a canine model of ARDS Putensen et al. [14] used another nitric oxide synthase inhibitor, N^G- monomethyl-L-arginine (L-NMMA), and its administration with nitric oxide caused a more pronounced reduction of shunt than did nitric oxide alone. There is evidence that constitutive types of nitric oxide synthase are more sensitive to L-NAME and inducible nitric oxide synthase more sensitive to L-NMMA. [30] Our model used regional hypoxia in a healthy lung, in which constitutive nitric oxide synthase probably dominates, and hence L-NAME had a strong effect. In ARDS there is an increase in inducible nitric oxide synthase, which may explain why L- NMMA had an effect whereas L-NAME did not. In patients with ARDS [15] a peripheral chemoreceptor stimulant, almitrine, was combined with inhaled nitric oxide, and the combination improved Pa^O2without increasing pulmonary artery pressure. The use of a vasoconstrictor may in the future be a way of potentiating the effect of inhaled nitric oxide.

The Pa^O2increase observed when nitric oxide was delivered to the hyperoxic regions after L-NAME infusion was remarkable and cannot be fully explained in terms of blood redistribution away from the hypoxic lobe, although blood flow in this lobe was less than 1% of Q sub T. In fact, it was found that the venous admixture in the hyperoxic regions decreased when nitric oxide was inhaled to these parts of the lungs. This result suggests an improved ventilation-perfusion matching when nitric oxide is inhaled after endogenous nitric oxide production has been blocked.

In conclusion, lobar blood flow during lobar hypoxia could be markedly varied. The combination of an intravenous nitric oxide synthase inhibitor and nitric oxide to the hyperoxic regions reduced lobar blood flow and caused a marked increase in Pa^O2. This method of reducing the shunt could be valuable in the treatment of patients undergoing one-lung ventilation and of patients with acute respiratory failure.

The authors thank Monika Hall, for technical assistance, and Eva-Maria Hedin, for drawing the figures.

*Freden F, Hayberg A, Hedenstierna G: Unpublished data, 1993.