Nitric oxide (NO) is administered frequently in patients with acute respiratory distress syndrome (ARDS) and pulmonary hypertension. The efficacy of this therapy over several days is not well known. The authors first determined the consistency of the response to repeated administration of NO and then the baseline variables that were associated with improvement in patients with severe ARDS.
In a prospective trial, 32 mechanically ventilated patients with severe ARDS received 10 parts per million NO by inhalation. In 22 of these patients, its effect was tested repeatedly (up to four times) in several days. Improvement was defined as an increase >10% in the ratio of pressure of oxygen in arterial blood (P(aO2)) to the inspiratory pressure of oxygen (FIO2) from baseline. Patients showing such an improvement were maintained on NO inhalation.
Twelve of the 22 patients (54%) showed a clinically significant and reproducible increase in the P(aO2)/FIO2 ratio with NO, from 74 +/- 30 mmHg (mean +/- SD) to 95 +/- 41 mmHg (P < 0.001). In three patients (14%), P(aO2) did not improve, even with multiple exposures. In seven patients (32%), an inconsistent response was seen on different days. Mean pulmonary artery pressure decreased for the entire group from 34 +/- 10 mmHg to 29 +/- 9 mmHg (P < 0.01), but this decrease did not correlate with the increase in P(aO2) in individual patients. The baseline P(aO2)/FIO2 ratio and mixed venous oxygenation (P(vO2)) were significantly lower, and the venous admixture was greater in patients showing beneficial effects of NO inhalation on P(aO2).
Repeated NO inhalation caused a consistent improvement in P(aO2) in about one half of these patients with severe ARDS; no significant benefit or inconsistent effects on pulmonary gas exchange were noted in the others. These findings could be related to the complexity of the mechanisms regulating the vasomotor changes in this syndrome. Severe baseline hypoxemia may be associated with a more favorable effect of NO on P(aO2).
NITRIC oxide (NO) is produced by endothelial cells and participates in the regulation of local vasomotor tone by relaxing vascular smooth muscle cells. [1,2]In acute respiratory failure and in pulmonary artery hypertension, an absolute or relative (or both) deficit of NO or an unbalanced state between vasoconstriction and vasodilation of pulmonary vessels may occur. Thus therapy with NO given by inhalation has been used to treat severe pulmonary hypertension as well as arterial hypoxemia in acute respiratory distress syndrome (ARDS). The rationale for NO therapy is based on its selective vasodilatory effect in ventilated lung regions, thereby improving the overall match of ventilation and perfusion. Studies of limited numbers of patients reported that inhalation of NO at doses of 5–40 parts per million (ppm) improves arterial oxygenation and selectively decreases pulmonary artery pressure in patients with ARDS, and it remains efficient over several days of administration. [4,5]
However, inconsistent effects of NO inhalation on both gas exchange and pulmonary artery pressure have been suggested in severe ARDS [6,7]and septic ARDS. Worsening of gas exchange has been shown in chronic obstructive pulmonary disease. The purpose of the present study therefore was to determine, first, the consistency of response to repeated exposure to inhaled NO in patients with severe ARDS, defined by an improvement in systemic oxygenation of more than 10%. In addition, we examined the association between baselines arterial or mixed (or both) venous oxygen tension and the response to NO.
Materials and Methods
The study was approved by the Committee for Ethics in Human Research of our institution. Thirty-two patients with severe ARDS and associated pulmonary arterial hypertension were given NO by inhalation. ARDS was defined according to the recent American-European consensus conference on ARDS, including acute respiratory insufficiency requiring mechanical ventilation, an arterial oxygen partial pressure (PaO2) to inspired oxygen fraction (FIO2) ratio (PaO2/FIO2) < 200 mmHg, requirement of a positive end-expiratory pressure of >or= to 5 cm H2O, bilateral pulmonary infiltrates evident on a chest radiograph, a pulmonary artery occlusion pressure <or= to 18 mmHg, and a lung injury severity score according to Murray et al. of >or= to 2.5. Pulmonary hypertension was defined as a mean pulmonary artery pressure (MPAP) > 25 mmHg.
Patients were monitored for clinical reasons with a central venous catheter, a radial arterial catheter, and, at least at the first exposure to NO, a pulmonary artery catheter to assess pulmonary hemodynamics, cardiac output measurement by thermodilution, and for mixed venous blood gas (PvO2) analysis. All patient lungs were mechanically ventilated using Hamilton Veolar (Rhazuns, Switzerland), Drager Evita (Lubeck, Germany), or Siemens 900 C (Siemens Elema, Solna, Sweden) ventilators. Tidal volume was set to limit peak inspiratory airway pressure < 50 cm H2O, and minute ventilation was adapted for an arterial pH >or= to 7.30. The minimal FIO2was chosen to obtain a PaO2between 55 and 65 mmHg and the positive end-expiratory pressure level to achieve optimal compliance. Nitric oxide was delivered from a tank containing a mixture of 400 ppm NO in nitrogen (AGA, Pratteln, Switzerland). Nitric oxide was administered continuously through a T-piece inserted proximal to the endotracheal tube, and its gas flow was adjusted to minute ventilation to obtain an average concentration of 10 ppm. Nitric oxide and NO2concentrations at the level of the endotracheal tube and expiratory limb of the breathing circuit, respectively, were assessed by a chemiluminescence analyzer (Eco-Physics System, Zurich, Switzerland). Arterial methemoglobin concentrations were measured daily.
When a patient was deemed eligible for NO inhalation, baseline data were collected. Nitric oxide was then given to achieve a mean concentration of 10 ppm and all measurements were repeated 30 min later. A test result was considered positive when the PaO2/FIO2ratio, corrected for NO gas flow, increased by > 10%. In patients who responded, NO was continued. In those who did not respond, NO was discontinued.
To test the consistency of NO response, 22 patients, who still met the definition of ARDS, were studied again 48–72 h later. In patients who had continued to receive NO during the interim, NO inhalation was halted for a minimum of 2 h before data were collected and a second exposure to NO was studied. This sequence was repeated up to day 10. The maximum number of tests were four in any individual. Twenty-two patients had more than one test: nine had two, eight had three, and five had four tests at the defined time intervals. Ten patients had one NO test only, and they are only included for the analysis of association between baseline gas exchange values and the response to NO.
The responses to the first and subsequent inhalations of NO were compared for blood gases, Qs/Qt, arterial oxygen content and transport, cardiac output, and pulmonary artery and systemic pressure changes. Values are expressed as means +/- SD. To account for interdependence of observations made in the same patient, standard errors were estimated using methods for cluster sampling as implemented on Epi Info 6. Patients were assigned to different groups according to their response to NO inhalation, for PaO2/FIO sub 2: R = responder (patient with consistently positive tests); NR = nonresponder (patient with consistently negative tests), and IR = inconsistent response (patient with both types of response during repeated trials). Baseline variables analyzed for association with the response to NO included PaO2, PaO2/FIO2, measured oxygen saturation (SaO2)(ABL 520 system, Copenhagen, Denmark), PvO2, Qs/Qt, PaCO2, MPAP, and pulmonary vascular resistance.
These values were compared with values obtained during NO inhalation in individual patients and baseline observations of the responders were compared with baseline data of nonresponders using simple regression. A P value < 0.05 was considered significant.
Between October 1992 and June 1996, 32 consecutive patients with severe ARDS were prospectively included in this study. Fifteen of these were referred from other hospitals for ARDS management. The delay between onset of ARDS and first exposure to NO averaged 4.4 +/- 5.6 days (range, 0–17 days).
Seventy-two NO inhalation tests were performed, and pulmonary hemodynamics and cardiac output were obtained for 66 of these. Ten patients were tested only once, because they died of multiple-organ failure (n = 8) or were transferred to another intensive care unit (n = 2) within 48 h after the first NO trial. The overall mean FIO2for all tests was 0.78, and the average FIO2at initial tests was 0.87 (median, 0.97; range, 0.55–1.0).
(Table 1) shows patient age, diagnosis, APACHE II and lung injury scores, baseline PaO2/FIO2, and outcome.
Consistency of Nitric Oxide Effects on Arterial Oxygen Pressure
(Table 2) shows data concerning the repeated effects of NO on oxygenation. Of the 22 patients with more than one test, 3 (14%) did not have an increase in PaO2during NO inhalation; i.e., they had consistently “negative” test results. Seven other patients (32%) had inconsistent responses during 21 observation; i.e., during their clinical course, arterial oxygenation increased on some days but not on others: 52% of their test results were “positive” and 48% were negative. In the last 12 patients (54%), PaO2increased consistently by more than 10% in the 33 tests performed (Figure 1). The PvO2, oxygen saturation, arterial oxygen content, and oxygen transport increased significantly and Qs/Qt decreased during NO inhalation in positive test results for PaO2(Table 3).
Considering each subsequent NO trial separately for all patients, we observed a constant percentage of positive and negative responses for PaO2to NO, at 69% versus 31%, 73% versus 27%, and 69% versus 31% at the first, second, and third exposures, respectively. The analysis over time of individual responses for PaO2to NO inhalation showed that 81%(13 of 16) of the positive responders at the first trial were consistently positive at the second trial and 88%(seven of eight) continued to have positive responses at the third trial. Among negative responders at first exposure to NO, only 50%(three of six) were consistently negative again at the second trial, and only one patient remained a nonresponder at the third trial. Fifty-seven percent (four of seven) of inconsistent responders showed a positive response at first exposure, and 75%(three of four) of these positive responders turned to negative responses at the second trial, whereas the nonresponders at the initial trial subsequently all became responders.
Relation between Baseline Conditions and Response
Considering all trials of NO administration (n = 72), the baseline PaO2/FIO2ratio and PvO2were lower with positive than in negative test results: PaO2/FIO2was 74 +/- 30 mmHg compared with 97 +/- 42 mmHg (P < 0.01;Figure 2), PvO2was 34 +/- 5 mmHg compared with 39 +/- 7 mmHg (P < 0.05), and oxygen saturation was 79 +/- 12% compared with 86 +/- 8%(P < 0.01). In addition, baseline Qs/Qtwas higher in patients with positive test results, at 54 +/- 14 compared with 42 +/- 11%(P < 0.05).
Mean pulmonary artery pressure and pulmonary vascular resistance decreased significantly during NO inhalation for the ARDS group as a whole (from 34 +/- 10 mmHg to 29 +/- 9 mmHg; P < 0.001, Table 3), but only 6 of 18 (33%) consistently had such decreases by more than 10% when tested on different days. Concordance of a decrease in MPAP with an increase in PaO2was observed in 21 of 32 patients and in 43 of the 66 tests, but no significant correlation (by linear regression analysis) between changes in PaO2and in MPAP nor pulmonary vascular resistance was noted (r =-0.15 and -0.05, respectively).
The hospital mortality rate for our patients with ARDS was 47%(15 of 32), and there was no significant difference in outcome between responders and nonresponders to NO inhalation for PaO2.
Methemoglobin concentration never exceeded 3 micro Meter; that is, it remained < 1% of the respective hemoglobin values. The concentration of NO2measured in the endotracheal tube and in the expiratory limb of the respiratory circuit never exceeded 1 ppm.
This investigation shows that only 54% of patients with severe ARDS have a clinically significant and consistent improvement in arterial oxygenation during NO inhalation. Thirty-two percent of patients showed an inconsistent effect, and in 14% no change or a deterioration of PaO2on different days of their clinical course was seen. Similarly, NO caused a consistent decrease in pulmonary arterial pressure in only one third of the patients studied. Baseline PaO2/FIO2and PvO2were significantly lower, whereas Q sub s /Qtwas higher in patients who increased arterial oxygenation with NO. Some patients maintained positive responses even in the later phases, when FIO2could be decreased to 0.5, whereas in others no effect of NO inhalation was observed even in the acute phase of the disease (i.e., during the first 3 days). Finally, the mortality rate seems not to be related to the response to NO inhalation.
These results are consistent with data from patients with ARDS described in a preliminary report of a multicenter study and observations in other pulmonary diseases. [8,16]Our findings contradict those of a previous investigation that observed a consistently positive response in ARDS. More recently, Rossaint et al. and Kraff et al. reported a significant increase in PaO2during NO inhalation in 40–80% of patients with severe or septic ARDS, with the percentage of positive effects depending on the criteria chosen to define significant changes. Looking at each trial separately in the present study, a positive response in PaO2occurred in 70%, and this percentage remains constant for repeated trials. The inconsistent effect of NO on pulmonary hemodynamics observed in our patients with severe ARDS also contrasts with the consistently successful NO application described in non-ARDS types of pulmonary hypertension, such as persistent pulmonary hypertension in neonates, in children with congenital heart disease, and in patients after mitral valve replacement. [17–20]
The inconsistent or absent responses to NO inhalation observed in many of our patients likely reflect the complexity of pulmonary vascular involvement in this disease. We could speculate that in responder patients a reduction of the endogenous endothelial release of NO could have induced a predominance of hypoxic vasoconstriction, thus increasing pulmonary vascular resistance. Previous studies showed that NO is continuously released to regulate pulmonary vascular tone [21,22]and that endothelium-dependent relaxation is impaired in animals exposed to chronic alveolar hypoxia. 
Indeed, NO only partially reverses hypoxic vasoconstriction in anesthetized dogs. [24,25]This does not necessarily imply that hypoxic vasoconstriction is counterregulated by NO. Other factors than NO are involved in the development of pulmonary hypertension and increased Qs/Qtin ARDS. Mediators such as endothelin, prostaglandin F2alpha, cytokines, and free radicals enhance hypoxic pulmonary vasoconstriction in ARDS, and NO inhalation seems able to reduce the release of oxygen-reactive species by activated leukocytes and cytokines, accounting for the discrepancy between the effect of NO on PaO2and MPAP. Our results are consistent with a multifactorial origin of pulmonary hypertension in ARDS involving factors not reversible by NO-induced dilation. Previous reports have noted that a lower PvO2is associated with a higher degree of hypoxic vasoconstriction. In the present study, lower baseline PaO2and PvO2values were observed in patients showing positive effects of NO on PaO2.
Basal rates of NO production and the degree of its reduction may be different in individual patients with ARDS. In addition, intrapulmonary diffusion of NO to lesser ventilated regions could reverse hypoxic vasoconstriction in these areas, thereby increasing intrapulmonary shunt. We could speculate that, in this case, the effect of NO would be comparable to an intravenous infusion of a pulmonary vasodilator, inducing a decrease in MPAP and arterial oxygenation. 
Barbera et al. reported no improvement in arterial oxygenation with NO in patients with chronic obstructive pulmonary disease in whom hypoxemia is mainly due to pulmonary ventilation-perfusion mismatching and much less to Qs/Qt. In our study, patients with positive responses had significantly higher baseline Qs/Qtvalues, supporting the findings by Barbera et al. ; that is, patients most likely to benefit from NO inhalation seem to be those in whom increased shunt is the principal determinant of hypoxemia. The possibility that NO could increase venous admixture via systemic recirculation cannot be excluded. In the lungs, S-nitrosylation of hemoglobin occurs and NO can enter the systemic circulation by binding to hemoglobin, forming S-nitrohemoglobin, or by binding to proteins containing thiol groups such as albumin. These compounds constitute a NO reservoir, allowing delayed release and possible recirculation in the pulmonary vascular tree. [30–32]No systemic hemodynamic effects are observed despite this perhaps because of a more marked dependence of the pulmonary circulation on NO to regulate its vascular tone compared with other vascular beds. 
The presence (or absence) of a positive effect of NO on gas exchange does not seem to be related to outcome in our patient series. In addition, the clinical effect of the changes in arterial oxygenation observed during NO inhalation remains to be shown. It is recognized that ARDS patients rarely die from hypoxia, and the relevance of a decrease of 5–7 mmHg of MPAP in absence of right ventricular failure is not clear. As suggested by others, the limit for clinical significance could be set at 20% for PaO2. In this case, the proportion of consistently positive responders would decrease to 23% in our series. The interpretation of the effect of NO can also be influenced by spontaneous variability of PaO2in this type of patient. In the present study, this problem was minimized by choosing a 10% cutoff to define an increase.
Our data suggest that even patients who are nonresponders one day may become responders the next day. If the initial NO test does not show an effect, it does not imply that it is not worth testing again, particularly when hypoxemia persists.
It should be emphasized that the slow-response chemiluminescence analyzer used allowed us to estimate mean NO concentration only, but not to determine peak and minimal levels or the importance of changes during the respiratory cycle. Newer systems of delivery result in improved stability of NO concentration during inspiration. Nitric oxide inhalation withdrawal should be done progressively to avoid an important decrease in PaO2, an increase in MPAP, or both.
In conclusion, NO inhalation produces a consistently beneficial effect on PaO2in only 50–60% of patients with severe ARDS. Our data suggest that the decision to start this therapy should be made on a clinical basis and the response assessed repeatedly. Clinicians should be aware that NO inhalation is of uncertain benefit in ARDS; there is no evidence at present for lower mortality and morbidity rates. Because its effects may vary during the clinical course, blood gas changes should be checked frequently. A low baseline PaO2/FIO sub 2 value and high venous admixture are associated with a more favorable reaction to NO inhalation.
The authors thank Dr. Thomas Perneger, Institute of Social and Preventive Medicine, University Hospital of Geneva, for assistance in the statistical analysis.