Hypoxic Pulmonary Vasoconstriction in Nonventilated Lung Areas Contributes to Differences in Hemodynamic and Gas Exchange Responses to Inhalation of Nitric Oxide 

After obtaining approval from the local ethics committee and informed consent from patients' next of kin, 11 consecutive patients with severe ARDS treated by venovenous extracorporeal lung assist were studied. Table 1summarizes patients' clinical characteristics. For extracorporeal lung assist, the inferior vena cava was cannulated percutaneously with a heparin-coated 25-French catheter (Medtronic, Dusseldorf, Germany) via a femoral vein. The superior vena cava was cannulated with a 23-French catheter via the right internal jugular vein. Blood was drained from the inferior vena cava into a reservoir and returned to the superior vena cava through two heparin-coated hollow fiber oxygenators (Maxima; Medtronic, Dusseldorf, Germany) by two roller pumps (Stockert, Munich, Germany) in parallel. A continuous flow of oxygen was delivered to the gas port of the oxygenator. Whereas extracorporeal oxygen supply to the patient depends on extracorporeal blood flow, extracorporeal elimination of carbon dioxide depends mainly on the ventilation:perfusion ratio of the membrane oxygenators. To prevent recirculation in the extracorporeal circuit, the tip of the drainage catheter was placed approximately 5 cm below the diaphragm.

Patients were included in the study when arterial oxygen tension (Pa^O2) was 60–65 mmHg at an extracorporeal blood flow of 2–3 l/min and a fraction of inspired oxygen (FI^O2) of 0.4. All patients had a radial artery catheter and fiberoptic pulmonary artery catheter (Opticath P7110-EP8; Abbott, Wiesbaden, Germany) in place, the latter for continuous measurement of mixed venous oxygen saturation (SV with bar^O2).

The patients were sedated with flunitrazepam and paralyzed with pancuronium. The lungs were ventilated with pressure control (peak airway pressure, 25–38 cm H²O) and a positive end-expiratory pressure of 6–12 cm H²O (Servo 900C; Siemens Elema, Lund, Sweden). Respiratory rates ranged from 7 to 10 breaths per minute. During the study period, FI sub O²was maintained at 1. Pv with bar^O2was aimed at 45, 55, 65, and 85 mmHg in random order by changing extracorporeal blood flow. Extracorporeal gas flow was adjusted to keep mixed venous carbon dioxide tension (Pv^CO2) unchanged. At each level of Pv with bar^Osub 2 hemodynamic and gas exchange variables were measured before, during, and after inhalation of NO at a concentration of 15 ppm. Each test period lasted 20 min, and measurements wee made at the end of each period.

Heart rate; systolic, diastolic and mean arterial pressure; systolic, diastolic and mean pulmonary artery pressure (PAP); and pulmonary artery wedge pressure (PAWP) were recorded on a monitoring system (Sirecust 1281; Siemens, Erlangen, Germany). All pressure measurements were made at end-expiration with the patient supine and the transducers (Medex Novotrans II MX 860, Hilliard, OH) zeroed to ambient pressure at the mid-axillary line. Cardiac output was determined as the mean of three thermodilution measurements randomly made during the respiratory cycle using 10 ml saline at room temperature and a cardiac output computer (Sirecust 1281). Blood gas tensions, hemoglobin oxygen saturations (Sa^O2and Sv with bar^O2), and total hemoglobin concentration were determined in arterial and to keep mixed venous blood samples (ABL 510; Radiometer, Copenhagen, Denmark). Cardiac index, Q with dot S /Q with dot T, oxygen transfer by the patients' lungs (V with dot^O2LUNG), and pulmonary vascular resistance index (PVRI) were calculated using standard formulas.

Nitric Oxide was administered as described previously. [7] Nitric oxide and NO²concentrations were monitored continously by electrochemical sensors (GS 8641 NO and GS 8650 NO²; Bieler & Lang, Achern, Germany). Methemoglobin was measured photometrically.

Data are expressed as means +/- SEM. Data before, during, and after NO inhalation at different levels of Pv with bar^O2were analyzed using two-way analysis of variance for repeated measures followed by a Student-Newman-Keuls test for multiple comparisons. Nitric oxide-induced changes in PAP-PAWP, PVRI, Q with dot^S/Q with dot^T, Pa^O2, and Sa^O2at the four levels of Pv with bar sub O²were compared using a one-way analysis of variance for repeated measures followed by a Dunnett test for multiple comparisons. A probability value less than 0.05 was considered significant.

Increases in extracorporeal blood flow increased Pv with bar sub O²without changes in Pv with bar^CO2and cardiac index (Table 2, Figure 1). The increase in Pv with bar^O2was associated with a decrease in mean PAP, pulmonary perfusion pressure (PAP-PAWP), and PVRI (P < 0.05;Table 2, Figure 2(A and B)), and with an increase in intrapulmonary right-to-left shunt (P < 0.05;Table 2, Figure 2(C)) and Sa^O2(P < 0.05;Table 2). Inhaled NO decreased PAP-PAWP, PVRI, and Q with dot^S/Q with dot^T, and increased Sa^O2, arterial oxygen content, and Pa^O2(Figure 3(A-C), Figure 4(A-C)). The NO-induced decrease in PAP-PAWP and PVRI was smaller at high compared with low Pv with bar^CO2levels (Figure 3(A and B)). In contrast, the decrease in Q with dot^S/Q with dot^Twas independent of the underlying Pv with bar^Osub 2 (Figure 3(C)). Except for the second level of Pv with bar^Osub 2 (i.e., mean 54 mmHg), inhaled NO increased Pv with bar^O2from all other baseline Pv with bar^O2levels (Table 2). The increase in Sa^O2and arterial oxygen content was smaller (P <0.05;Figure 4(A and 4B)), and the increase in Pa^O2was greater (P < 0.05;Figure 4(C)) at high compared to low Pv with bar^O2. Values before and after NO were not different. Methemoglobin concentration never exceeded 1.4%, and NO²concentration did not exceed 1 ppm.

The main result of this study is that hypoxic pulmonary vasoconstriction in nonventilated lung areas is an important factor in determining the hemodynamic and respiratory response to NO inhalation in patients with ARDS. Further, the increase in Pa^O2during NO breathing is influenced by Pv with bar^O2.

Hypoxic pulmonary vasoconstriction increases pulmonary artery pressure and improves V with dot^A/Q with dot matching by diverting blood flow toward ventilated alveoli in the diseased lung. [8] Usually alveolar tension is the primary determinant of hypoxic pulmonary vasoconstriction. [3] However, in nonventilated alveoli Pv with bar^O2becomes the determining factor for HPV. [3] In late ARDS, as in our patients, the pulmonary V with dot^A/Q with dot pattern is characterized by nonventilated and well-ventilated lung units rather than by a broad spectrum of V with dot^A/Q with dot imbalances [4,9,10] During ventilation at 1.0, alveolar P^O2in ventilated lung areas is high enough to inhibit HPV even in alveoli with low V with dot^A/Q with dot ratios. [11] Moreover, at a FI^O2of 1.0 alveoli with low V with dot^A/Q with dot ratios tend to collapse and become atelectatic. [12] Thus, at FI^O21.0 HPV is present only in nonventilated lung areas, and a decrease in PAP and PVRI in response to an increased Pv with bar^O2reflects attenuation of HPV in nonventilated lung areas.

In acute lung injury and ARDS, hypoxic pulmonary vasoconstriction is modulated by several factors. In animals with oleic acid-induced lung injury, pretreatment with indomethacin or aspirin restored the blunted HPV, suggesting that a vasodilating prostaglandin may interfere with HPV. [13] In isolated rat lungs, endothelium-derived relaxing factor attenuates the pressure response to alveolar hypoxia. [14] Increased left atrial pressure has been reported to inhibit HPV in piglets [15] but not in dogs. [16] A high PAP may abolish HPV. [17] Endotoxin, [18] hydrogen peroxide, [19] and platelet-activating factor [20] reduce, whereas metabolic acidosis enhances the pulmonary pressure response to hypoxia in animals. [21] Changes in P^CO2and acid-base status modulate pulmonary vascular tone in animals [22–24] and in humans. [24,25] Because cardiac index, Pv with bar^Osub 2, and acid-base status were kept constant in our patients, the observed changes in pulmonary perfusion pressure, PVRI and Q with dot^S/Q with dot^T(Figure 2and Table 2) must be the result of changes in Pv with bar^O2. Rossaint et al. [26] did not observe a significant change in PAP in patients with ARDS treated by extracorporeal lung assist when Pv with bar^O2was decreased. However, the decrease in Pv with bar^O2was only 20 mmHg, whereas in our patients Pv with bar^O2varied by 37 mmHg (between 47 and 84 mmHg). Moreover, their patients had been ventilated for a longer period, and Pa^O2/FI^O2was higher than in our patients (309 vs. 101 mmHg) at a similar level of Pv with bar^O2(approximately 55 mmHg).

Inhaled NO administered to patients with ARDS produces vasodilation in ventilated lung areas and increases blood flow to those regions. [2] As a result, intrapulmonary shunt and pulmonary artery pressure decrease. Changes in Q with dot^S/Q with dot^Tdo not correlate with changes in PAP or PVRI. [1,27–29] Pulmonary perfusion pressure and pulmonary vascular resistance are determined by cardiac output and vasoconstriction in ventilated and in nonventilated lung areas. At constant cardiac output, the pulmonary vascular bed may be considered a resistor consisting of two circuits with different resistances in parallel. When vascular resistance in nonventilated lung areas is high, total pulmonary vascular resistance and pulmonary perfusion pressure depend increasingly on vascular resistance in ventilated lung areas, and total PVR will finally approach vascular resistance of ventilated lung areas. In that case, a decrease in vascular resistance of ventilated lung areas by inhaled NO will result in a relatively large decrease in total PVRI and PAP-PAWP. In contrast, when vascular resistance in the nonventilated lung is low, the percentage of NO-induced decrease in PVRI and pulmonary perfusion pressure will be relatively small. Findings in our patients would support such underlying mechanisms. At low Pv with bar^O2, PVRI and PAP-PAWP were high (suggesting high vascular resistance in nonventilated lung areas) and subsequent NO inhalation caused a relatively large decrease in these parameters (Figure 2(A and B), Figure 3(A and B)). Consequently, when Pv with bar^O2was high (suggesting low vascular resistance in nonventilated lung areas), inhaled NO had no significant effect on total PVRI and PAP-PAWP (Figure 2(A and B), Figure 3(A and B)).

Several investigators have shown that in ARDS the NO-induced decrease in PVR is more pronounced when baseline PVR is high. [1,27,28,30] This effect can be attributed, at least in part, to the extent of hypoxic vasoconstriction in nonventilated lung areas. Inhaled NO dilates only preconstricted pulmonary vessels and does not affect pulmonary vessels with normal tone. [31]

If vascular tone in ventilated lung areas is high, total PVRI and the NO-induced change in total PVRI will also be high. In that case, the change in Q with dot^S/Q with dot^Tshould correlate with the change in PVRI. In patients with ARDS, however, the changes in PVRI and PAP do not correlate with changes in Q with dot^S/Q with dot^Tduring NO inhalation. [1,27–29]

In our patients, Q with dot^S/Q with dot^Tincreased with rising Pv with bar^O2levels. Administration of NO decreased Q with dot S /Q with dot T consistently by 4% regardless of the magnitude of (Figure 2(C) and Figure 3(C)). Blood flow to nonventilated lung areas depends on the size of the injured lung segment, the hypoxic vasoconstriction, and the total flow. [32–35] Because cardiac index did not change and because it is unlikely that the size of injured lung segments changed during the study period, the increase in Q with dot S /Q with dot T with increasing Pv with bar^O2must be attributed to the inhibition of HPV at higher levels. The increase in Pv with bar^Osub 2 coincided with the decrease in pulmonary perfusion pressure at high Pv with bar^O2. The finding that intrapulmonary shunt increased with increasing Pv with bar^O2is consistent with similar observations in animal models of ARDS [36] and in segmental atelectasis. [37] Viewing the pulmonary vascular bed as a resistor consisting of two circuits with different resistances in parallel (i.e., ventilated and nonventilated lung areas), resistance of nonventilated areas has little effect on redistribution of flow toward ventilated lung areas when resistance of those areas decreases. Thus vascular resistance in nonventilated lung areas which is modulated by Pv with bar^O2, has a marked influence on the NO-induced reduction in pulmonary perfusion pressure but a negligible one on the decrease in intrapulmonary shunt over a wide range of vascular resistances.

In this study, Q with dot^S/Q with dot^Tat different Pv with bar^O2levels was calculated using the standard shunt equation. [38] In contrast to the inert gas elimination technique, [39,40] calculated intrapulmonary shunt theoretically could be influenced by the oxygen uptake in the lungs and by the arteriovenous oxygen content difference (av with bar D^O2). Several authors have compared oxygen shunt to inert gas shunt at FI^O21.0. [40–43] Wagner et al. [40] found an excellent correlation between both methods of shunt determination in dog lungs under various experimental conditions (normal, oleic acid injury, pulmonary embolism, and pneumonia). In their animals, intrapulmonary shunt ranged from 0 to 40%. Values of Pv with bar sub O²were not reported. Hlastala et al. [41] found in dog lungs with atelectasis and after oleic acid injury a notable correlation of both methods when the shunt was between 20% and 80%. However, Pv with bar^O2values were not reported. At lower shunt levels, oxygen shunt was systematically higher than inert gas shunt. This difference was attributed to the fact that in contrast to inert gas shunt, oxygen shunt includes shunts from the bronchial circulation. An additional source of error in calculated shunt may come from the properties of oxygen electrode. Sato et al, [42] compared oxygen shunt and inert gas shunt in dog lobes with varying degrees of intrapulmonary shunt, ranging from 0% to 66%. They found a good correlation between both methods, although at low shunt levels (less than 12%), oxygen shunt was higher than inert gas shunt whereas at higher shunt levels oxygen shunt was lower than inert gas shunt. The main reason for these differences was the properties of the electrode. Pv with bar^O2ranged from 22 to 48 mmHg. Breen et al. [43] compared inert gas shunt and oxygen shunt in dog lungs and in the left lower lobe after oleic acid injury of the left lower lobe. Cardiac output was varied by arteriovenous fistulas. Opening the arteriovenous fistulas not only increased cardiac output, but it also increased Pv with bar^O2from 48 to 62 mmHg and decreased the arteriovenous oxygen content difference (av with bar D^O2). Despite the increase in Pv with bar^O2and the decrease in av with bar D^O2calculated shunt and inert gas shunt in left lower lobes and in the whole lung did not differ for a range of inert gas shunts.

Oxygen consumption by lung tissue results in a systematic error in calculated shunt. However, this potential error was not relevant under various experimental conditions. [40–43] When oxygen uptake by the lungs is zero (i.e., all oxygen is transferred by the extracorporeal circuit), the shunt equation [38] should not be used because dividing a term by zero is impossible. In all our patients oxygen uptake by the lungs was greater than zero even at high Pv with bar^O2(Table 2). Thus in our patients calculated shunt represents real shunt, although a small systematic error may be inherent.

The consistent decrease in intrapulmonary shunt during NO administration at each level of Pv with bar^O2coincided with varying increases in arterial oxygen tension (Figure 4(C)). As a consequence of the shape of the oxygen dissociation curve, the decrease in Q with dot^S/Q with dot^Tat low Sv with bar^O2resulted in a relatively large increase in Sa^O2and arterial oxygen content (Figure 4(A and B)) but in a relatively small increase in Pa^O2. In contrast, the same decrease in Q with dot S /Q with dot T at high Sv with bar^O2resulted in a relatively large increase in Pa sub O²with only a small increase in Sa^O2and arterial oxygen content (Figure 4(A and B)). Thus the assessment of the beneficial effect of NO purely on the basis of an increase in Pa^O2may be misleading.

In patients with ARDS, the NO-induced increase in Pa^O2was more pronounced during administration of almitrine, a substance that enhances HPV in nonventilated lung areas. [5,6] Although initial arterial oxygen saturation was higher during almitrine administration, subsequent NO application decreased venous admixture similarly in the presence or absence of almitrine. [5] Thus the larger increase in Pv with bar^O2during NO and almitrine can be attributed to the location of the Pv with bar^O2in the oxygen dissociation curve.

In conclusion, our data suggest that hypoxic pulmonary vasoconstriction in nonventilated lung areas contributes to the hemodynamic and respiratory responses during nitric oxide inhalation. The stronger the HPV in nonventilated lung areas the more pronounced is the NO-induced decrease in PAP-PAWP and PVRI. The increase in Pa^O2during NO breathing is greater at high compared with low mixed venous oxygen tension.