© 2003 American Thoracic Society
A Randomized Trial of Inhaled Nitric Oxide to Prevent Ischemia–Reperfusion Injury after Lung TransplantationToronto Lung Transplant Program, Toronto General Hospital, University of Toronto; and Department of Clinical Epidemiology and Biostatistics, McMaster University, Hamilton, Ontario, Canada Correspondence and requests for reprints should be addressed to J. T. Granton, M.D., FRCPC, Toronto General Hospital, 200 Elizabeth Street, EN 10-220, Toronto, ON, M5G 2C4 Canada. E-mail: john.granton{at}uhn.on.ca
Inhalation of nitric oxide (NO) has been advocated as a method to prevent ischemia–reperfusion injury after lung transplantation. We enrolled 84 patients into a concealed, randomized, placebo-controlled trial to evaluate the effect of inhaled NO (20 ppm NO or nitrogen) initiated 10 minutes after reperfusion on outcomes after lung transplantation. The groups (n = 42) were balanced with respect to age, sex, lung disease, procedure, and total ischemic times. PaO2/FIO2 ratios were similar on admission to the intensive care unit (ICU) (NO 361 ± 134; control patients 357 ± 132), and over the duration of the study. There were no differences in hemodynamics between the two groups. Severe reperfusion injury (PaO2/FIO2 < 150) was present at the time of admission to the ICU in 14.6% NO patients versus 9.5% of control patients (p = 0.48). The groups had similar median times to first successful trial of unassisted breathing (25 vs. 27 hours; p = 0.76), successful extubation (32 vs. 34 hours; p = 0.65), ICU discharge (3.0 days for both groups), and hospital discharge (27 vs. 29 days; p = 0.563). Five NO versus six control patients died during their hospital stay. Adjusting for age, sex, lung disease etiology, presence of pulmonary hypertension, and total ischemic time did not alter these results. In conclusion, we did not detect a significant effect of inhaled NO administered 10 minutes after reperfusion on physiologic variables or outcomes in lung transplant patients.
Key Words: nitric oxide • lung transplantation • mechanical ventilation • critical care
Lung transplantation is an accepted therapy for end-stage pulmonary disease, offering the possibility of longer life and improved quality of life. Despite advances in surgical techniques and lung preservation, ischemia–reperfusion (IR) injury remains a common and disturbing complication. Endothelial dysfunction, capillary leak, and an intense neutrophilic inflammatory reaction in the lung parenchyma characterize IR injury. Clinically, this is recognized as a progressive deterioration in gas exchange, opacification of the chest X-ray, and increased pulmonary vascular resistance within 24 hours of transplantation. The development of IR injury is associated with a higher mortality rate, prolonged duration of mechanical ventilation, and attendant increase in intensive care and hospital length of stay (1). Importantly, the development of IR injury has also been correlated with the later development of obliterative bronchiolitis—a common and dreaded cause of late lung dysfunction and death (2, 3). Although the precise pathophysiological process has not been fully elucidated, there are several mechanisms identified that have been implicated in the development of IR injury. Central to the development of IR injury is adhesion and eventual sequestration of activated leukocytes through disrupted endothelium and attendant cytotoxic disruption of lung parenchyma (4). Complement activation, platelet activation, increased endothelin-1 expression, release of reactive oxygen species, and release of inflammatory cytokines have also been implicated (5–7). Indeed, therapeutic strategies aimed at disrupting a variety of these pathways have been shown to attenuate IR injury in animal models of lung transplantation. However, once IR injury occurs, treatment remains supportive, with mechanical ventilation, and in severe circumstances extracorporeal support. Nitric oxide (NO) possesses several properties that have made it attractive to use in the setting of IR injury. The most commonly exploited biological property of NO is its vasodilatory effect. Inhalation of NO causes regional vasodilation leading to a reduction in pulmonary vascular pressures. Through regional vasodilatation, inhaled NO may lead to an improvement in gas exchange via an improvement in In addition to its role as a vasodilator, more recent interest has focused on the potential use of NO to prevent IR injury. Two experimental observations support the notion that the earlier use of inhaled NO at the time of ischemia or reperfusion may attenuate IR injury. First, is the observation that there is a relative NO deficiency that develops during ischemia and subsequent reperfusion. During reperfusion, there is a dramatic decrease in endogenous NO and cyclic guanosine monophosphate levels in the donor lung (12). Our group and others have also demonstrated that the expression of endothelial NO synthase decreases in lung allografts (13). It has been speculated that this deficiency of NO contributes to IR injury. This speculation is supported by the observation that supplementation of NO during either ischemia or reperfusion using NO analogs or donors, inhaled NO, or even endothelial NO synthase gene transfer can improve experimental graft function (14–28). The beneficial effect of NO appears to be related to factors other than its effects on pulmonary vascular tone (29). Rather it is NO's anti-inflammatory properties and ability to potentially target several pathways involved in IR injury (vide infra) that make it therapeutically appealing. NO interferes with neutrophil adherence to endothelium, inhibits platelet aggregation, and inhibits expression of several inflammatory mediators including chemokines, interleukins, endothelin-1, and adhesion molecules (5, 6, 30–32). These anti-inflammatory properties in addition to the observed reduction in endogenous NO production after lung reperfusion have made the rationale for the use of NO to prevent IR injury compelling. Two uncontrolled clinical studies suggest that NO may prevent IR injury (17, 33). However, to date no prospective randomized controlled study has been undertaken to definitively prove this notion. The need for such a study is heightened not only by the expense of NO therapy but also concerns that NO administered during reperfusion may be injurious (34). In this regard, levels of inducible NO synthase derived from leukocytes and alveolar macrophages have been correlated with an increase in nitrite and nitrate levels, and in turn, the severity of acute lung rejection (35–37). Furthermore, inhaled NO has potential proinflammatory properties that could augment IR injury (38–40). NO in the presence of oxygen forms potent oxidizing agents including nitrogen dioxide and peroxynitrite potentially leading to epithelial injury, surfactant disruption, and membrane lipid peroxidation. In this study, we specifically sought to determine the effects of NO administered in the operating room at the time of reperfusion. However, while developing this clinical trial, two experimental observations suggested that delaying NO administration by 10 to 15 minutes after reperfusion may be superior to administration during reperfusion (28, 41). A study by Eppinger and colleagues demonstrated worse initial lung function in animals that received NO during reperfusion compared with those in whom NO administration was delayed by 10 minutes. On the basis of these observations, we conducted a randomized controlled study to evaluate the effects of NO versus placebo administered 10 minutes after the onset of reperfusion on the development of IR injury.
From December 1997 to June 2000, we invited all adults on the Toronto Lung Transplant Program waiting list to participate in this study. We randomly assigned patients to inhaled NO or placebo therapy using opaque sealed envelopes. Pre-existing pulmonary hypertension has been associated with the subsequent development of IR injury, consequently patients were stratified by initial systolic pulmonary artery pressure measurements obtained from a pulmonary artery catheter, using a threshold of 60 mmHg (1).
Patients, investigators, and all caregivers other than respiratory therapists were blinded to the nature of the study gas. NO and placebo gases were stored in identical cylinders containing a compressed mixture of NO in nitrogen gas in a concentration of 1,000 ppm (PraxAir), and pure compressed nitrogen gas, respectively. Respiratory therapists administered both gases in the operating room and the intensive care unit (ICU) using the NOdomo (Drager) apparatus and an EVITA (Drager) ventilator. The NOdomo apparatus is a commercially available NO administration device that allows for the delivery of NO (over a range of 1–99 ppm) proportional to flow rate independent of the ventilatory mode used. The NOdomo apparatus also allowed for the continuous measurement of NO (0–100 ppm, resolution of 1 ppm) and nitrogen dioxide (0–50 ppm, resolution of 0.1 ppm) concentrations. Respiratory therapists measured NO concentrations with each change in dose or
We initiated therapy with the study gas 10 minutes after the start of reperfusion for each lung. The starting dose and minimum dose of study gas in the operating room was 20 ppm. Surgeons were allowed to increase the dose of the study drug at their discretion. Unblinded NO or unplanned cardiopulmonary bypass (CPB) could be used (but was not required) for any one of the following three conditions: (1) severe hypoxemia (FIO2 1.0, PEEP On transfer to the ICU, clinicians stabilized patients on FIO2 1.0 and PEEP (minimum of 5 cm H2O), and then recorded initial gas exchange and hemodynamic data. In the ICU, oxygenation, mechanical ventilation, and titration of the study gas were managed according to protocol. Briefly, respiratory therapists titrated FIO2, PEEP, and the study gas, maintaining PaO2 in the range of 75 to 100 mm Hg and SpO2 between 92 and 96%. They reduced FIO2 first, and then adjusted PEEP. The initial and all subsequent PEEP settings were based on the FIO2 being used according to a FIO2/PEEP chart (Table E1 in the online supplement). The lowest possible FIO2/PEEP combination was used at any time. Once the FIO2 was 0.4 and PEEP less than or equal to 7.5 cm H2O, the study gas was titrated by decrements of 10 ppm every 30 minutes. The study gas was not titrated to less than 10 ppm for the first 6 hours postoperatively. It was felt that mandating the use of NO for a longer time period would interfere with attempts to liberalize ventilation and delay time to extubation. Once the study drug was at 10 ppm for 30 minutes (and 6 hours had elapsed since admission to the ICU), the study drug was reduced to 5 ppm. To minimize the risk of rebound pulmonary hypertension or hypoxemia, subsequent reductions occurred by 1 ppm every 10 minutes. If gas exchange deteriorated during titration, the titration was halted and the dose returned to the previous level. A negative response was defined as any one of a fall in saturation more than or equal to 4%, or a rise in mean pulmonary arterial pressure by more than or equal to 25%. A second, third, and fourth attempt was made after waiting 4, 8, and 24 hours, respectively, from the previous attempt at titration. If after the first 96 hours in the ICU, a patient was still receiving study gas, physicians could wean it at their discretion. If the patient deteriorated within 48 hours after the study gas was discontinued, the study gas was restarted in a blinded fashion at 20 ppm. Unblinded NO could be used in the ICU if all of the following four conditions were met: (1) a FIO2 of 1.0 beyond 6 hours or FIO2 of 0.8 or higher beyond 24 hours, (2) titration of study gas without improvement, (3) trial of paralysis, (4) trial of PEEP up to 15 cm H2O, if tolerated without hemodynamic compromise. Unblinded NO was administered from a separate cylinder. PG therapy could be used when a study patient met criteria for unblinded NO. We used pressure control ventilation with VTs in the range of 6 to 12 ml/kg and plateau pressures up to 35 cm H2O (including PEEP), unless pH fell below 7.2. Weaning from mechanical ventilation began once the patient met all of the following criteria; (1) the patient was awake and alert, and exhibited inspiratory efforts, (2) the FIO2 less than 0.4 for more than 2 hours or FIO2 equal to 0.5 for more than 24 hours, and PEEP less than or equal to 7.5 cm H2O, (3) the VT greater than or equal to 4 ml/kg ideal body weight, (4) not receiving vasopressors (less than or equal to 5 µg/kg/minute of dopamine was allowed), and (5) there were no anticipated surgical procedures for the next 24 hours. From pressure control mode, patients underwent a 2-hour trial of unassisted breathing using T-piece or CPAP mode. Using predefined criteria, if they were able to tolerate the 2-hour trial they were extubated. If they did not tolerate the 2-hour T-piece trial, we implemented a standardized pressure support wean. A further T-piece trial of unassisted breathing was performed only once they were receiving a pressure support of less than 8 cm H2O for more than 2 hours. We recorded ventilation and hemodynamic variables at 8-hour intervals for 4 days, and daily at 8:00 A.M., thereafter. Daily chest radiographs were evaluated by one of us (J.T.G.) in a blinded fashion on a daily basis. The radiograph was scored as demonstrating the presence of a focal, diffuse, or perihilar opacity. Met-hemoglobin levels were measured every 6 hours for 24 hours, every 12 hours for the second 24 hours, and daily, thereafter, during administration of study gas. To ensure blinding, only respiratory therapists had access to met-hemoglobin levels. If a met-hemoglobin level was documented above 5%, the protocol mandated the administration of methylene blue (1 mg/kg intravenously over 15 minutes) and a reduction in the NO level by 10 ppm. This was repeated if the met-hemoglobin level remained above 5% 1 hour later. During this study, no patient had a met-hemoglobin level greater than 2%.
Our primary endpoint was the duration of mechanical ventilation from the time of admission to ICU to first successful extubation. We took time to the event (liberation from the ventilator) into account by conducting a survival analysis using Cox proportional hazards regression analysis and censoring all deaths at the time the last patient was liberated from the ventilator. We conducted similar analyses for other comparisons (i.e., duration of ICU stay and hospitalization). Additional endpoints included unplanned use of CPB, use of unblinded NO, and the incidence of severe IR injury (PaO2/FIO2 < 150) which we analyzed using
To determine the number of patients needed in the trial, we made the following assumptions. First, we assumed that the use of inhaled NO would reduce the duration of mechanical ventilation by 50%. In a survey of 38 lung transplant patients over a 14-month period before initiation of our study, the mean duration of mechanical ventilation was 201 hours with a SD of 271 hours. Second, on the basis of previous studies, it was assumed that the standardization of mechanical ventilation and weaning would result in a 30% reduction in variation in duration of mechanical ventilation to 200 and 100 hours in the control and NO groups, respectively (44). Assuming an 80% power and with an An independent data and safety monitor reviewed all outcomes and adverse events.
Eighty-four transplant recipients participated, and two declined. The two study groups were balanced with respect to age, sex, and etiology of lung disease (Table E2 in the online supplement). In the operating room, the groups were similar with respect to operative procedures and ischemic times (Table E3 in the online supplement). The control group had a marginally higher frequency of unplanned CPB that did not achieve statistical significance. In contrast, the NO group had a slightly higher frequency of unblinded NO administered in the operating room (not statistically significant). The most common reason for the use of unblinded NO use was hemodynamic instability followed by hypoxemia. On admission to the ICU, acute physiology and chronic health evaluation II scores were balanced, as were gas exchange and hemodynamic variables (Table 1) . Only six and four patients in the NO and control groups had a PaO2/FIO2 ratio less than 150 on admission to the ICU.
The evolution of oxygenation remained similar in the two groups while they were mechanically ventilated (Figures 1A–1C) . There was no significant difference between groups at any time point when one considered either PaO2/FIO2 ratio, oxygenation index, or level of inspired oxygen. On examining Figure 1A, there appeared to be a separation between the two groups after 60 hours (2.5 days), suggesting that NO may have had a late beneficial effect on oxygenation. However, this represented only 19 study patients and did not achieve statistical significance. The oxygenation index is perhaps a better indicator of the degree of derangement in oxygenation as it incorporates mean airway pressure. The separation between the two groups in PaO2/FIO2 at 60 hours was not seen in the oxygenation index (Figure 1B), suggesting that NO did not have a benefit on oxygenation. There was no difference in the appearance of focal, diffuse, or perihilar opacities on the chest radiograph between the two groups. Two NO patients and one control patient received unblinded NO therapy in the ICU for refractory hypoxemia. Two patients in each group received PG infusions.
The two groups were comparable with respect to median times to the first successful trial of unassisted breathing, successful extubation, discharge from the ICU, and discharge from the hospital (Table 2) . There was no difference between groups with respect to the evolution of organ dysfunction (using the multiple organ dysfunction score) over the first 4 days (p = 0.42), the first week (p = 0.71), or the first 4 weeks (p = 0.49) (data not shown). The overall mean difference between groups was also nonsignificant in each of these analyses (p = 0.72, p = 0.55, and p = 0.70 for the group main effect at 4 days, 1 week, and 4 weeks, respectively). There were also no differences with respect to the evolution of any of the five components of the multiple organ dysfunction score (cardiac, respiratory, gastrointestinal, renal, or hematologic components) and no overall group mean differences. No differences were found between groups in either systemic or pulmonary vascular hemodynamics (data not shown).
The 30-day operative mortality was 9.5% in the NO group and 11.9% in the control group (Table 2). Eleven patients died in hospital: five NO and six control patients. One patient died of acute myocardial infarction during surgery after initiation of NO therapy. Three NO patients died in the ICU (of IR injury, pneumonia, and sepsis) while still receiving mechanical ventilation. One additional NO patient died of pulmonary embolism in the ICU after extubation. In the control group, three patients died in the ICU before extubation (two with multiple organ failure complicating sepsis, and one of congestive heart failure). Three control patients died after discharge from the ICU. The median duration of blinded therapy was similar in the groups (NO: 10.5 hours, range 6.5–73.0 hours; control group: 9.6 hours, range 7.8–150.7 hours; p = 0.66). The results are unchanged if one considers the duration of both blinded study drug (NO or placebo) and unblinded NO use together. The median duration of all inhaled therapy was 13.1 hours (range 6.5–172 hours) in the NO group and 10.4 hours (range 7.8–150.7 hours) in the control group. No patient in either the NO or control group had a met-hemoglobin level greater than 2% at any time during the study. There were no adverse effects during the study that were felt to be directly attributable to the use or withdrawal of the study medication. Before performing the primary analysis of the data, we identified age, sex, reason for transplantation, and ischemic time as factors that may have influenced the efficacy of NO in this trial. However, reanalysis of the data after adjusting for these variables did not alter any of these results.
This is the first randomized controlled trial evaluating inhaled NO after lung transplantation. On the basis of compelling studies demonstrating the ability of NO to preserve lung function after IR injury, we anticipated that NO would attenuate lung damage and preserve graft function in patients after lung transplantation. However, in this controlled study, we did not detect a significant effect of inhaled NO at a dose of 20 ppm administered 10 minutes after reperfusion on physiologic variables or clinically important outcomes. To date, only case series or studies using historical control patients have evaluated the potential effects of NO in preventing IR injury. Ardehali and colleagues initiated inhaled NO during reperfusion in 28 consecutive patients, and transiently withdrew NO after 6 and 12 hours postoperatively (33). The incidence of severe IR injury was similar to ours (18 vs. 20% in our study population). However, the observed mortality (20%) was slightly higher than ours (13%). Interestingly, when inhaled NO was transiently withdrawn, they observed a significant deterioration in gas exchange and pulmonary artery pressures. On the basis of the observed deterioration in gas exchange during withdrawal of NO, they concluded that inhaled NO improves lung function and reduces pulmonary artery pressure in those patients who develop IR injury. Alternatively, these physiologic changes may simply have represented a rebound complication of acute NO withdrawal. Their study design could not address the ability of NO to prevent IR injury. We did not observe any differences in gas exchange or pulmonary artery pressures between groups in our randomized trial. Transient worsening in oxygenation and pulmonary arterial pressures were not seen in our study during weaning of the study drug. This may have related to the gradual withdrawal of NO in our trial. More recently, Thabut and colleagues reported 23 consecutive patients who received inhaled NO (10 ppm) and pentoxifylline at the time of reperfusion. To determine if NO conferred any protection to the development of IR injury, they compared their outcome retrospectively with historical unmatched control patients. The control patients represented 23 consecutive patients transplanted just before the use of NO and pentoxifylline (patients transplanted from 1994 to 1998—Control Group 1) and 95 patients representing all the patients of the series who did not receive NO–pentoxifylline (presumably those transplanted before 1994—Control Group 2). They noted that pulmonary edema, was reduced by half, and that oxygenation was better in the NO–pentoxifylline group. The duration of mechanical ventilation was 2.1 ± 2.4 days in the NO–pentoxifylline group versus 7 ± 9 days in Control Group 1 (p = 0.02) and 6 ± 7 days in Control Group 2 (p = 0.01). The 2-month mortality rate was 4.3% in the NO–pentoxifylline group versus 26% (p = 0.04) and 21% (p = 0.07) in Control Groups 1 and 2, respectively. It is difficult to make comparisons between these two studies. The dose of NO was similar to that used in our study, however, we did not concomitantly use pentoxifylline. Although provocative, their study is open to criticism as it used retrospective control patients, and the intervention was neither blinded nor randomly assigned. Other aspects of clinical care were also not controlled to the same degree as in our trial. In addition, the duration of mechanical ventilation and mortality rates in the control groups were appreciably higher than our study. Consequently, the observed benefit of the intervention may have been exaggerated and more related to the severity of illness in their control population. It is important to emphasize that if NO was acting to prevent reperfusion, to the extent that it has been reported in preclinical and the uncontrolled clinical trials done to date, we might have seen a difference in the development of IR injury, oxygenation, and or pulmonary hemodynamics between groups. However, this was not observed. The incidence of severe IR injury (22% in the NO group vs. 19% in the control group), oxygenation, and hemodynamics were similar between the two groups. Although it did not achieve statistical significance, it is interesting to note that the NO group in our trial had a lower incidence of requirement for unplanned CPB than the placebo group. In our center, CPB is used for all patients with significant pulmonary hypertension. For all other patients, the operations are done as bilateral sequential lung transplants with the use of CPB only when needed. The usual time in which CPB is required in this setting occurs after the implantation of the first lung, due to intolerable dysfunction of the newly implanted lung—manifest as hypotension, hypoxemia, or suprasystemic pulmonary arterial pressures. The lower frequency of CPB could imply that NO provided better hemodynamic stability and oxygenation during surgery. In contrast, the observation that the NO group received more unblinded NO in the operating room argues against this potential benefit. However, the discordant use of CPB may indicate a potential utility for prophylactic NO that requires further investigation. We conducted this study in a single institution. Consecutive patients were enrolled in the study, and the characteristics of our recipients and donors are similar to other centers (45). Consequently, we feel that the results of this study are applicable to other transplant centers. We chose to evaluate factors (selected a priori) that may have had an effect on the efficacy of NO in this study (1). We repeated our analysis of all outcome variables after adjusting for age, sex, lung disease etiology, presence of pulmonary hypertension, and total ischemic time. None of these variables had a significant effect on patient outcome. The final donor PaO2 in our study averaged 435 mm Hg. In addition, our cold ischemic times were relatively short (3.7 hours). Consequently, we cannot comment if NO would have had any benefit for more marginal lungs with longer ischemic times. Most of our patients underwent bilateral lung transplantation. We had insufficient numbers of patients to reliably compare differences between patients undergoing double versus single lung transplantation. Our study did not specifically address the role for therapeutic inhaled NO for established IR injury. However, approximately 20% of patients in each group met the criteria for severe IR injury (PaO2/FIO2 < 150) over the first 48 hours in the ICU. Therefore, comparison to an earlier clinical study of inhaled NO for the treatment of severe IR injury is appropriate (46). Date and colleagues compared 15 patients who received inhaled NO (20–60 ppm) with 17 historical control patients. Patients receiving NO had a shorter duration of mechanical ventilation (12 vs. 17 days), fewer airway complications (0 versus 4), and lower mortality (7% [1/15] vs. 24% [4/17]). These results did not achieve statistical significance, however, they did generate tremendous interest in NO after lung transplantation. In our study, the clinical course of severe IR injury (defined as a PaO2/FIO2 < 150) was identical for patients who were randomly assigned to NO or to placebo therapy. The results of our study are surprising as they contrast sharply with several preclinical studies of inhaled NO in animal models of lung IR injury. Indeed, the benefits of NO have consistently been demonstrated across species, using a variety of cold and warm ischemic periods, in a variety of models of lung ischemia and reperfusion with either whole blood or buffer-perfused isolated lung preparations or transplanted in vivo models (15, 22–25, 28, 41, 47–49). In contrast to its use for established hypoxic respiratory failure, the reported ability of NO to attenuate IR injury appears to be independent of its vasodilatory properties. NO is thought to preserve lung function by reducing neutrophil adhesion, and sequestration, and subsequent oxidant injury. In this regard, NO administration utilizing NO donors coadministered with lung preservation solution has also been shown to be protective. Minamoto and colleagues demonstrated that nitroglycerin given during flush and preservation of rat lungs led to an improvement in subsequent animal survival and function compared with either control group or a delayed flush with nitroglycerin. Interestingly, they also demonstrated that the benefits of nitroglycerin were associated with a reduction in graft endothelin-1 expression. Fugino and colleagues also demonstrated that NO inhalation at the time of lung harvest in dogs improved recipient oxygenation, and reduced pulmonary vascular resistance. For practical reasons, we were not able to administer NO during lung harvest. Consequently, we chose to evaluate the effects of NO administration during the reperfusion phase. Clearly, however, the timing of NO administration may be important. One potential criticism of our study design is that we administered NO 10 minutes after the onset of reperfusion. It could be argued that this delay may have reduced the efficacy of NO. We chose to administer NO 10 minutes after reperfusion on the basis of observations from a study by Eppinger and colleagues. In their study in a rat model of lung ischemia, they demonstrated worse early (30 minutes after reperfusion) lung function when NO was administered at the time of reperfusion compared with administration 10 minutes after reperfusion (28). They also demonstrated that the early dysfunction may have been due to oxidant injury as the effect was attenuated with the use of superoxide dismutase. At 4 hours, there were no differences in lung function between the immediate and delayed groups of animals. These experimental observations of early lung dysfunction with the use of NO during reperfusion supported concerns we had at the time of our study that NO, in the presence of high concentrations of oxygen, may lead to the generation of reactive oxygen species and consequent lung injury. At the time of the study protocol development, we were also reassured that a 10-minute delay in NO administration would not miss the window of effect by the observations from the Paris-Sud University Lung Transplant Group (41). They demonstrated that in an isolated blood-perfused rat model, there was no difference in lung function when NO inhalation was delayed by 15 minutes versus when administered immediately during reperfusion. We, however, concur with the notion that NO administered during ischemia or during lung harvest may be beneficial and needs to be prospectively evaluated. Another reason for the disparate results of our study and earlier experimental observations is that we may have used an inappropriate dose. We chose not to evaluate different doses of NO as there were insufficient patient numbers to reliably determine a difference between doses. On the basis of experimental observations and our own clinical practice, we felt that a dose of 20 ppm was reasonable. Reviewing earlier animal studies of IR injury, the dose of inhaled NO used ranged from 10 to 80 ppm. In fact, the effects of higher doses of NO have either been less impressive or worse than the effects when lower doses were used (20, 41). Another potential criticism of our study is that it was not completely blinded. Though we were unable to blind the respiratory therapists involved in managing ventilatory care, we did address this limitation by following objective protocols for weaning study gas, FIO2, PEEP, and mechanical support. The similar duration of study gas therapy between the two groups suggests that respiratory therapists followed these protocols. We based our sample size on a projection that NO would lead to a 50% reduction in duration of mechanical ventilation. This assumption is supported in part by the observation by King and colleagues who showed that IR injury was associated with a twofold increase in the duration of mechanical ventilation (1). Randomization, concealment, and protocolized care notwithstanding, it is possible that we simply lacked an adequate sample size to detect a true but small benefit of inhaled NO therapy in our patients. The fact that not all patients develop IR injury after lung transplantation may have amplified the limitation of our sample size. With these criticisms in mind, we have estimated that to detect a more conservative 20% reduction in the duration of mechanical ventilation as a result of prophylactic NO therapy would have required an additional 620 patients in this study. Clearly, a larger multicenter trial would have to be undertaken to evaluate a lower magnitude of effect. In summary, we observed no benefit of inhaled NO administered 10 minutes after reperfusion in the prevention of IR injury, with respect to physiologic or clinically important outcomes. Our findings in 84 patients do not entirely rule out the possibility of important harm or benefit. Our study does not support the earlier observations of uncontrolled clinical trials and preclinical studies that demonstrate the ability of inhaled NO to prevent reperfusion injury.
The authors acknowledge the contributions of the members of the Toronto Lung Transplant Program and the Respiratory Therapy Department at the Toronto General Hospital. They also wish to thank Dr. Gordon Guyatt for his ongoing advice, and review of the manuscript and statistical analysis. The authors are grateful for the contributions of Dr. Desmond Bohn, Chair of our Data Safety Monitoring Committee. They would also like to acknowledge the generous support from Lifetronics Medical Inc., who donated the NOdomo NO delivery system.
Supported in part by the Canadian Cystic Fibrosis Foundation. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Received in original form March 22, 2003; accepted in final form February 6, 2003
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
/
matching. Given that IR injury is characterized by an increase in pulmonary arterial pressures and hypoxemia, inhaled NO has been advocated to support patients with IR injury (
10 cm H2O, SpO2 
90%), (2) hemodynamic instability (systolic blood pressure < 90 mm Hg or mean blood pressure < 60 mm Hg), or (3) suprasystemic pulmonary artery pressures. 
2 tests. To evaluate the evolution of oxygenation, we compared mean PaO2/FIO2 ratios, mean oxygenation index (mean airway pressure x PaO2/FIO2), and mean FIO2 levels over time. For these comparisons, we utilized general linear models with time, group, and patient within group as the main variables. The general linear model is analogous to a repeated measures analysis of variance with factors of time, treatment, and interaction between the two. Indeed, when there are no missing data, the general linear models and analysis of variance yield identical results. The advantage of using general linear models with patient as a factor is that it allows analysis with some missing data or with varying lengths of follow-up (as in the case of this study—owing to differences in duration of ventilation) from individual patients, without the imputation required if we were to use analysis of variance. We used the multiple organ dysfunction score to compare organ dysfunction between groups (
error of 0.05, for a two-tailed test of statistical significance, 39 patients would be required in each group. 
