OBJECTIVE. Our goal was to determine whether, for preterm newborn infants with respiratory disease, inhaled nitric oxide reduced the rates of death, bronchopulmonary dysplasia, intracranial hemorrhage, or neurodevelopmental disability.
METHODS. We searched Medline, Embase, Healthstar, and the Cochrane Central Register of Controlled Trials using the search terms “nitric oxide,” “clinical trial,” and “newborn” and covering 1985–2006. We also searched abstracts of the Pediatric Academic Societies.
RESULTS. Eleven randomized, controlled trials of inhaled nitric oxide therapy for preterm infants were found. The trials were grouped into 3 categories according to the entry criteria, that is, entry in the first 3 days of life on the basis of oxygenation criteria (early rescue), enrollment after 3 days on the basis of elevated risk of bronchopulmonary dysplasia, and routine use for intubated preterm infants. Early rescue treatment based on oxygenation criteria did not seem to affect mortality or bronchopulmonary dysplasia rates. Routine use for intubated preterm infants showed a barely significant reduction in the incidence of the combined outcome of death or bronchopulmonary dysplasia (relative risk [RR]: 0.91 [95% confidence limits (CLs): 0.84, 0.99]). Later treatment based on the risk of bronchopulmonary dysplasia showed no significant effect on this outcome. Early rescue treatment showed a trend toward increased incidence of severe intracranial hemorrhage, whereas routine use for intubated preterm infants seemed to show a reduction in the incidence of either severe intracranial hemorrhage or periventricular leukomalacia (RR: 0.70 [95% CLs: 0.53, 0.91]).
CONCLUSIONS. Inhaled nitric oxide as rescue therapy for very ill preterm infants undergoing ventilation does not seem to be effective and may increase severe intracranial hemorrhage. Later use of inhaled nitric oxide to prevent bronchopulmonary dysplasia does not seem to be effective. Early routine use of inhaled nitric oxide for mildly sick, preterm infants seems to decrease the risk of serious brain injury and may improve rates of survival without bronchopulmonary dysplasia.
Since the introduction of surfactant therapy, mortality rates for preterm infants have decreased significantly.1 Occasionally infants do not experience adequate improvement in oxygenation after surfactant treatment, however, and other complications of prematurity continue to cause substantial long-term disability.2
Premature animals with models of hyaline membrane disease have elevated pulmonary vascular resistance, and both pulmonary artery pressure and oxygenation may be improved with inhaled nitric oxide (iNO).3,4 Preterm infants with respiratory failure also have increased pulmonary artery pressure.5 This is rarely sufficient to cause reversal of ductal shunts, however, and the hemodynamic profile thus differs from that of term neonates with severe pulmonary hypertension. In term neonates with hypoxic respiratory failure, iNO decreases the requirement for extracorporeal membrane oxygenation without decreasing mortality rates.6 However, because of different pathophysiologic features, different entry criteria, and different outcomes, the results for term infants cannot be extrapolated to preterm infants.
Preterm infants are at risk of long-term pulmonary disability attributable to bronchopulmonary dysplasia (BPD). If iNO therapy leads to a decrease in required ventilatory support, then reductions in lung injury and the frequency of BPD may follow. BPD is important because it leads to chronic medical illness and rehospitalization and is associated with neurodevelopmental impairment.7,8 However, nitric oxide has both prooxidant and antioxidant activities and can potentially worsen lung injury.9 Therefore, the effects of iNO therapy on developing lungs must be evaluated carefully before the introduction of such therapy into clinical practice.
Of particular concern for preterm infants is the fact that iNO affects coagulation.10,11 Preterm infants are at high risk of developing intracranial hemorrhage, which has substantial effects on long-term developmental outcomes. Therefore, it is important that iNO be evaluated for its effect on intracranial hemorrhage in preterm infants.
The few case reports and case series published before randomized, controlled trials were conducted demonstrated that premature infants with severe respiratory failure that did not respond to full conventional management, including surfactant therapy and high-frequency ventilation, might experience improved oxygenation with iNO.12,13 In those reports, death and intracranial hemorrhage were frequent.
The objective of this study was to determine whether, in preterm newborn infants with respiratory disease, treatment with iNO improves oxygenation and reduces the rates of death, BPD, intracranial hemorrhage or other serious brain injury, and adverse, long-term, neurodevelopmental outcomes. This is an edited version of an updated systematic review to be published in the Cochrane Database of Systematic Reviews.
The protocol for the systematic review was approved by the Cochrane Neonatal Review Group. The criteria for considering studies for this review were as follows: the studies were randomized or quasi-randomized clinical trials, and the participants were preterm infants (<35 weeks’ gestation) with respiratory failure after adequate treatment with surfactant. The intervention was iNO therapy, compared with placebo or control treatment, in addition to conventional treatment, for respiratory failure. The outcome measures in which we were interested were death before hospital discharge, BPD (oxygen dependence at corrected age of 36 weeks), death or BPD (at corrected age of 36 weeks), intracranial hemorrhage or severe intracranial hemorrhage, periventricular leukomalacia, neurodevelopmental disability (proportion of survivors with neurologic abnormalities sufficient to affect quality of life, developmental index >2 SDs below the mean, using a validated scale at >12 months of age, or mean developmental index among survivors at >12 months of age), severe retinopathy of prematurity, and oxygenation within 2 hours after therapy.
The following search strategy was used for identification of studies. A Medline search was performed by using PubMed and the search terms “nitric oxide” and “newborn.” This search was limited to clinical trials and was updated most recently in November 2006. The years searched were 1985 to the present. The abstracts of the annual Pediatric Academic Societies meetings were also searched from 1995 to 2006. In addition, the standard methods of the Cochrane Neonatal Review Group were used to find any additional references; this involved searches of Embase, Healthstar, and the Cochrane Central Register of Controlled Trials, all last updated in November 2006. Each identified trial was assessed for methodologic quality with respect to (1) masking of allocation, (2) masking of intervention, (3) completeness of follow-up monitoring, and (4) masking of outcome assessment.
For categorical outcomes, typical estimates for relative risk (RR) and risk difference (RD) were calculated by using RevMan software (Nordic Cochrane Centre, Cochrane Collaboration, 2003, Copenhagen, Denmark), and 95% confidence limits (CLs) were used. A fixed-effect model was assumed. Continuous outcomes were analyzed by using weighted mean differences, also assuming a fixed-effect model. Heterogeneity was evaluated by using the I2 statistic, which is reported whenever the result was >50%; a χ2 test for heterogeneity was also performed, and results are reported for P < .05. Sensitivity analyses were performed within groups of trials when the trials were of very different design or quality. Results both with and without inclusion of particular trials are presented.
Description of Studies
We found 11 published, randomized, controlled trials of iNO, compared with control treatment, for preterm infants.14–24 Abbreviated details are found in Table 1. All studies used an intent-to-treat approach to analysis.
The entry criteria for the 11 studies were quite dissimilar, with enrollment in the first 2 days of life for most of the infants in 7 of the trials, all of which required significantly impaired oxygenation for trial enrollment.14–20 Two of the remaining 4 trials had enrollment after the first 3 days of life for infants with elevated risks of developing BPD.23,24 The remaining trials, by Schreiber et al21 and Kinsella et al,22 enrolled infants in the first 2 days of life, and any preterm infant undergoing ventilation was considered eligible. The implications for clinical practice are clearly quite different for these 3 groups of studies, which were therefore analyzed separately.
Studies were eligible to be considered in the first group if they included acutely ill, preterm infants undergoing ventilation, most of whom were enrolled within the first 3 days of life, and the infants needed to satisfy a criterion regarding severity of illness; this group is termed “early rescue treatment.” The average oxygenation indices (OIs), when available, were as follows: median of 32 for the Neonatal Ventilation With Inhaled Nitric Oxide Versus Ventilatory Support Without Inhaled Nitric Oxide for Severe Respiratory Failure (INNOVO) trial,16 23 for the trial by Van Meurs et al,17 18 to 20 for the Franco-Belgium trial,18 12 to 15 for the trial by Hascoet et al,15 and 10.8 to 11.9 for the trial by Srisuparp et al.20
Studies were included in the second group if they enrolled preterm infants who were >3 days of age and qualified because of an elevated risk for BPD. There were 2 such trials. Ballard et al24 reported a median respiratory severity score (cm H2O; which is not directly convertible to the OI because it does not take into account the Pao2; few eligible infants would have had arterial lines), calculated as the fraction of inspired oxygen (Fio2) multiplied by the mean airway pressure (in centimeters of water), of 3.5 in both the treated and control infants, which suggests mild illness.
Studies were eligible for the third group if they enrolled infants early in life (<3 days of age) and had no criteria regarding severity of illness, other than being intubated. In the study by Kinsella et al,22 the mean OIs were 5.4 (iNO group) vs 5.8 (control group); in the study by Schreiber et al,21 the median OI was 6.94. This group is termed “early routine use.”
Trials of Early Rescue Treatment With iNO for Preterm Infants
Kinsella et al14 randomly assigned preterm infants if the ratio between Pao2 and alveolar oxygen pressure (calculated as [Fio2 × (atmospheric pressure − 47)] − Paco2) was <0.1 in 2 successive blood gas analyses in the first 7 days of life (expected mortality rate: 50%). The planned sample size was 210 infants. A planned interim analysis, performed after 2.5 years, found no detectable difference in the main outcome (survival to discharge). Forty-eight infants had received iNO, and 32 were control subjects. The study was terminated at that time because the analysis suggested that significant benefit was unlikely to be detected. Baseline characteristics of the groups were similar, apart from a greater number of iNO-treated infants having no intracranial hemorrhage at the start of the study (73% vs 59%). The Pao2/Fio2 values were 42 (SD: 18) for the iNO group and 42 (SD: 16) for the control group.
The Franco-Belgium collaborative NO trial group18 performed a multicenter international trial. All except 1 of the 85 preterm infants (<33 weeks’ gestation) received surfactant, and the majority (75%) were receiving high-frequency ventilation. The majority of the infants were enrolled on the first or second day of life. The study was terminated early because of slowing enrollment after 27 months. After assessment of the primary outcome at 2 hours, treatment with open-label iNO was allowed if the infant's OI exceeded 30. Five of the control infants eventually received iNO. The availability of backup treatment with iNO for control infants limited the ability of the study to address long-term outcomes. All of the baseline characteristics were similar between the groups.
Srisuparp et al20 performed a single-institution study that randomly assigned intubated preterm infants of <2000 g who had received surfactant and had clinical respiratory distress syndrome. OI thresholds, noted in Table 1, allowed random assignment of the smallest infants (≤1000 g) with only mild disease and required increasing OIs for enrollment of infants with increasing birth weight. Allocation was through a “card-picking scheme”; therefore, masking was uncertain. This was a pilot study for the study by Schreiber et al21 and therefore was underpowered. Baseline characteristics were similar between the groups.
Hascoet et al15 conducted a study in 10 European centers. Any intubated preterm infant could be enrolled and assigned randomly, but the randomization was revealed only if the infants developed hypoxic respiratory failure, which led to 61 iNO-treated infants and 84 control infants actually being exposed to the study intervention. If an infant developed refractory hypoxemia at any time, then the case was defined as a failure and iNO was administered according to French Drug Agency recommendations. Refractory hypoxemia before 6 hours of age occurred in 20 infants, who therefore were not entered into the study. An additional 20 iNO-treated infants and 28 control infants received open-label iNO for treatment of refractory hypoxemia, which complicated the analyses of the results with significant contamination of the control group. The initially planned sample size was achieved.
Van Meurs et al17 performed a multicenter study in which initial entry criteria restricted enrollment to infants with gestational ages of <34 weeks and birth weights of 401 to 1500 g who were undergoing assisted ventilation ≥4 hours after surfactant therapy and were considered at high risk because of OIs of ≥10 in 2 consecutive blood gas analyses. This criterion was revised after the initial interim analysis showed an unexpectedly high mortality rate with an OI of ≥5 followed by an OI of ≥7.5 after ≥30 minutes. Although it was initially planned to recruit 440 infants, the study was terminated after two thirds of the infants had been assessed for the primary outcome, because there seemed to be an increase in the severe intracranial hemorrhage rate but no benefit in the primary outcome. By the time the analysis was completed, 420 infants had been enrolled and the study was terminated. Baseline characteristics were similar between the groups.
The INNOVO trial16 was a European multicenter study with masked allocation using a telephone system; treatment assignment was through minimization25 “with a probabilistic element,” rather than strict randomization. Eligibility criteria were “severe respiratory failure requiring assisted ventilation if the responsible physician was uncertain about whether an infant might benefit from iNO.” iNO treatment was suggested to be initiated at 5 ppm and could be doubled up to a maximum of 40 ppm. There were 2 primary outcome criteria listed in the main publication, but the sample size was calculated on the basis of a reduction in the frequency of the combined outcome of death or severe disability at corrected postnatal age of 1 year. The study planned to enroll 200 infants at <34 weeks’ gestation and <28 days of age; 55 iNO-treated and 53 control infants were enrolled, and the study was stopped at the end of calendar year 2001, which was apparently preplanned. Follow-up data were complete for all except 1 infant and were not formally masked.
Dani et al19 performed a single-center study with infants who were undergoing ventilation and were experiencing severe respiratory distress despite surfactant treatment. The mean age of intervention initiation was 43 hours for the iNO group. This single-center study planned to enroll 52 infants but was terminated after 40 subjects had been enrolled, after a previously unplanned interim analysis that confirmed the investigators’ impression that BPD was less frequent in the iNO group. This early termination provided insufficient protection from type 1 errors.
Trials of iNO Treatment for Preterm Infants Eligible at >3 Days of Age Because of Elevated Risk of BPD
Subhedar et al23 performed an open randomized trial of iNO with 42 premature infants (<32 weeks’ gestation) who were enrolled at 96 hours of age if still intubated. The entry criteria included having a high risk for developing BPD, on the basis of a previously published risk score.26 This was a 2 × 2 factorial study that also investigated dexamethasone treatment; almost all data were presented as iNO versus no iNO, regardless of the use of dexamethasone. The initially planned sample size was 88 subjects, but the study was terminated at a sample size of 42 subjects because the incidence, at a predesignated 12-month review, of the primary outcome of death or BPD was much higher than planned, “which would have enabled the planned outcome to be detected with a much smaller group.” Oxygenation rates at baseline were not well matched between the groups, despite randomization. The median OI for control infants was 3.9 (range: 1.2–11.5), and that for iNO-treated infants was 7.9 (range: 1.6–46.7). There was also a greater proportion of male patients in the iNO group (12 of 20 infants vs 5 of 22 infants). Other baseline characteristics were similar.
Ballard et al24 performed a multicenter trial with masked allocation. The study gas (iNO or nitrogen) was masked to all except the study respiratory therapist. The study was overseen by a data safety-monitoring committee, with preplanned interim analyses, and the study enrolled the initially planned sample size. A total of 12% of the infants in the iNO group and 13% in the control group had respiratory severity scores of >10, indicating greater illness severity.
Trials of Routine Use of iNO for Intubated Preterm Infants
Schreiber et al21 performed a single-center study that enrolled intubated preterm infants (gestational age of <34 weeks and birth weight of <2 kg) at <72 hours of age, without additional criteria regarding severity of illness. This was a 2 × 2 factorial design examining 7 days of iNO treatment, compared with oxygen placebo treatment; the second comparison was the use of high-frequency oscillatory ventilation versus conventional ventilation. The planned sample size was achieved, and follow-up data were complete. Assessment of the primary outcome was performed in a masked manner. It was not stated whether the long-term neurodevelopmental follow-up monitoring was also masked.
Kinsella et al22 performed the largest of the multicenter trials completed to date. Subjects were <34 weeks’ gestational age, were undergoing ventilation because of respiratory failure in the first 48 hours, and were expected to remain intubated for >48 hours. There were no additional requirements regarding severity of illness. The planned sample size of 792 infants was achieved. The study was overseen by a data safety-monitoring committee, with interim analyses according to preplanned rules. Groups were well balanced. The follow-up data were complete with respect to assessment of the primary outcome, which was assessed in a masked manner.
The usefulness of overall analyses was considered to be limited, because of the differing entry criteria for the studies; the criteria regarding severity of illness and age at entry varied so greatly that pooling the results was not considered appropriate. Control group mortality rates also varied substantially (6%–64%), emphasizing the differences in the eligible patients. Therefore, we performed analyses only according to the aforementioned posthoc groupings.
Comparability of Studies Within Groups
Early Rescue Treatment Studies
The early rescue treatment studies all randomly assigned the infants to low-dose iNO or control treatment. The INNOVO study16 did not have a reproducible criterion for entry; despite this difference, the mortality and BPD rates were higher but not very dissimilar, compared with the other studies in the early rescue treatment group. The methods used for calculation of the oxygenation defect in the remaining studies were different, with some studies reporting OI and others reporting Pao2/alveolar oxygen pressure or Pao2/Fio2, which cannot be compared directly. In the early rescue treatment studies, the majority of patients were enrolled before 3 days of age, although some studies allowed enrollment up to 7 days of age. Of note, both Hascoet et al15 and the Franco-Belgium group18 allowed backup treatment of control subjects with iNO if their condition worsened to a prespecified degree. This might lead to underestimates of both benefits and risks. For this reason, sensitivity analyses were performed by excluding those 2 studies.
The early rescue treatment studies had comparable mortality rates for the control groups, with (except for the smallest study) mortality rates of >30%. The highest control group mortality rate was in the INNOVO study (64%),16 with the rest being between 30% and 44%.
Studies With Later Entry Based on BPD Risk
Two studies evaluated infants >3 days of age on the basis of elevated risk of BPD; the studies were quite different from each other. Subhedar et al23 investigated both iNO and dexamethasone therapy, by using a factorial design, for infants with an extremely high risk of BPD (almost 100% among survivors). Ballard et al24 enrolled infants with increased BPD risk solely on the basis of the continued need for respiratory support. Because of these differences in trial design, we performed sensitivity analyses with and without the trial by Subhedar et al,23 which showed no major effect because that trial was very small. The mortality rate for the control group in the study by Ballard et al24 was only 6%, which reflects the older age at entry than in the early rescue and early routine treatment studies and the lesser severity of illness than in the early rescue treatment studies.
Early Routine Treatment Studies
Two studies enrolled infants without specific criteria for disease severity. Schreiber et al21 randomly assigned preterm infants who were ventilator-dependent after receiving surfactant, and Kinsella et al22 enrolled infants with gestational ages of <34 weeks who were expected to undergo ventilation for >48 hours. The infants in this group of studies had much lower OIs than did the infants in the early rescue treatment studies. The control group mortality rates were quite similar between these 2 studies (23% and 25%, respectively).
Death Before Hospital Discharge
All trials assessed survival to discharge, and none of the individual trials showed a significant effect. The early rescue treatment studies had a typical RR of 1.05 (95% CLs: 0.91, 1.22; RD: 0.02; 95% CLs: −0.04, 0.09). The studies with entry after 3 days on the basis of BPD risk had a RR of 1.06 (95% CLs: 0.64, 1.74: RD: 0.00; 95% CLs: −0.04, 0.05). For the studies of early routine treatment, the typical estimate of the RR for death before hospital discharge showed a significant decrease, with an upper 95% CL that approached 1 (typical RR: 0.77; 95% CLs: 0.60, 0.98; RD: −0.06; 95% CLs: −0.11, 0.01) (Fig 1).
Death Before Postmenstrual Age of 36 Weeks
Six studies reported this outcome, 5 with early rescue treatment. There was no significant effect of iNO on this outcome (typical RR: 0.89; 95% CLs: 0.72, 1.11; RD: −0.05; 95% CLs: −0.13, 0.14). The study by Subhedar et al,23 with entry after 3 days on the basis of BPD risk, also reported this result and did not show a significant effect.
BPD (Oxygen Dependence Among Survivors at Corrected Age of 36 Weeks)
All of the published studies except those by Hascoet et al15 and Srisuparp et al20 reported BPD rates at 36 weeks, and none of the individual trials found a significant effect. There was substantial heterogeneity for each group of studies, none of which showed a statistically significant effect.
For early rescue treatment studies, the typical RR was 0.89 (95% CLs: 0.76, 1.05; I2 = 47.8%; RD: −0.05; 95% CLs: −0.12, 0.02). For studies with entry after 3 days on the basis of BPD risk, the typical RR was 0.89 (95% CLs: 0.78, 1.02; I2 = 85.5%; RD: −0.07; 95% CLs: −0.15, 0.01). For studies of routine use for intubated preterm infants, the typical RR was 0.96 (95% CLs: 0.85, 1.08; I2 = 64.4%; RD: −0.02; 95% CLs: −0.09, 0.04).
Death or BPD
Data on the combined outcome of death or BPD (or its converse, survival without BPD) were available for all of the studies (Fig 2). None of the individual early rescue treatment trials found a significant effect, and this group of studies showed no effect (typical RR: 0.95; 95% CLs: 0.88, 1.02: RD: −0.04; 95% CLs: −0.09, 0.02). Similarly, the studies with entry after 3 days on the basis of BPD risk did not show a significant effect individually, and the group results were not significant (typical RR: 0.90; 95% CLs: 0.80, 1.02: RD: −0.06; 95% CLs: −0.14, 0.01). The studies of routine use for intubated preterm neonates showed a significant reduction; of note, the upper 95% CL was 0.99 (RR: 0.91; 95% CLs: 0.84, 0.99). The RD was −0.06 (95% CLs: −0.12, −0.01), and the number needed to treat was 17 (95% CLs: 8, 100).
Any Intracranial Hemorrhage
Three studies reported this outcome, all of which were early rescue treatment studies. There was no evidence of an effect of iNO on overall intracranial hemorrhage frequency (typical RR: 1.0; 95% CLs: 0.73, 1.37).
Severe Intracranial Hemorrhage
Six of the early rescue treatment studies reported this outcome, and they showed a trend toward increased incidence of severe intracranial hemorrhage (RR: 1.27; 95% CLs: 0.99, 1.62; RD: 0.06; 95% CLs: 0.00, 0.13) (Fig 3). Because most intracranial hemorrhage occurs in the first 3 days of life, the studies with later entry would not be expected to demonstrate an effect on intracranial hemorrhage. Evolution of preexisting abnormalities, development of hydrocephalus, and occurrence of periventricular leukomalacia were reported as a single variable by Ballard et al,24 and results were not different between the groups. Of the studies of routine use for intubated preterm infants, only Kinsella et al22 reported severe intracranial hemorrhage as a separate outcome, which was not affected (RR: 0.77; 95% CLs: 0.55, 1.09; RD: −0.04; 95% CLs: −0.08, 0.01).
Severe Intracranial Hemorrhage or Periventricular Leukomalacia
The early rescue treatment studies showed no significant effect, but there was a trend toward an increase in the frequency of this adverse outcome (RR: 1.16; 95% CLs: 0.93, 1.44; RD: 0.04; 95% CLs: −0.02, 0.10) (Fig 4). The studies of routine use for intubated preterm infants showed a reduction in the frequency of this outcome (RR: 0.70; 95% CLs: 0.53, 0.91; RD: −0.07; 95% CLs: −0.12, −0.02; number needed to treat: 14; 95% CLs: 8, 50).
To date, the only studies to report on neurodevelopmental outcomes were the studies by Mestan et al27 and Bennett et al28 and the INNOVO trial.16 From the original study by Subhedar et al,23 22 children were still alive at 30 months of age and 21 (7 iNO-treated infants and 14 control infants) were examined formally.28 There were no significant differences in outcomes. The definition of “severe neurodisability” in that report was very similar to our definition of neurodevelopmental disability; the 5 infants with severe neurodisability (Mental Developmental Index or Psychomotor Development Index of <71, cerebral palsy, or sensorineural impairment) were all control infants.
The study by Mestan et al27 showed a significant reduction, at corrected age of 2 years, in the frequency of a composite outcome of neurodevelopmental disability (cerebral palsy, bilateral blindness, bilateral hearing loss, or a score on the Bayley Scales of Infant Development >2 SDs below the mean). This improvement was largely the result of a decrease in the incidence of Bayley Scales of Infant Development scores >2 SDs below the mean. Cerebral palsy rates were not different between the groups.
The INNOVO trial investigators reported the rates of major disabilities at 1 year of age, which were not different between the groups.16 Severe disability was defined as no/minimal head control or inability to sit unsupported or no/minimal responses to visual stimuli (equivalent to a developmental quotient of <50). There was no difference between the groups (7 of 55 patients vs 3 of 53 patients).
Severe Retinopathy of Prematurity
There was no evidence of an effect on severe retinopathy of prematurity (reported only by Mestan et al27).
Retinopathy Requiring Surgery
None of the groups of studies showed an effect on this outcome. For the studies with early rescue treatment, the typical RR was 0.86 (95% CLs: 0.58, 1.29; RD: −0.02; 95% CLs: −0.07, 0.03). For the studies with entry after 3 days on the basis of BPD risk, the typical RR was 1.04 (95% CLs: 0.78, 1.38; RD: 0.01; 95% CLs: −0.06, 0.08). For the studies of routine use for intubated preterm infants, the typical RR was 1.09 (95% CLs: 0.79, 1.50; RD: 0.01; 95% CLs: −0.04, 0.06).
Repeating the analyses for the early rescue treatment group after exclusion of the 2 studies that allowed backup treatment of control subjects did not affect any of the analyses substantially. There remained no significant effect on the incidence of survival without BPD (RR: 0.94; 95% CLs: 0.88, 1.02), with a similar trend toward increased incidence of grade 3 or 4 intraventricular hemorrhage or periventricular leukomalacia (RR: 1.15; 95% CLs: 0.89, 1.48).
This review suggests that there may be identifiable groups of preterm infants who receive substantial benefits from iNO therapy, with reduction in the incidence of brain injuries visible on ultrasound scans and potential reduction in mortality rates; however, the precision of the estimates of these effects is low and the number needed to treat may be large, if these effects are confirmed. There are other groups of infants with evidence of adverse effects (specifically, increased rates of severe intracranial hemorrhage) without evidence of benefits.
None of the individual trials showed a reduction in mortality rates. Only the meta-analysis of the 2 trials that evaluated routine use for intubated infants demonstrated a potential difference. In that case, the effect was marginally significant. The typical RR was 0.77 (95% CLs: 0.60, 0.98). The RD was −0.06 (95% CLs: −0.11, −0.01); although this is potentially an important magnitude of effect, the estimate lacks precision. The number needed to treat to save 1 infant may be as few as 9 infants or as many as 100 infants. The groups of studies of early rescue treatment and later treatment for infants at risk of BPD showed no effect on mortality rates.
Survival Without BPD
There was no apparent benefit of iNO in early rescue treatment studies. A preplanned subgroup analysis of the largest of the rescue studies, that reported by Van Meurs et al,17 suggested that there might be a reduction in the incidence of the combined outcome of death or BPD among infants with birth weights of >1 kg but not either outcome separately. The early routine treatment studies showed a modest and barely significant reduction in the incidence of the combined outcome of death or BPD. There was heterogeneity in this outcome (I2 = 64.6%), with the larger study by Kinsella et al22 showing no significant benefit in their overall analysis. A preplanned subgroup analysis from that study suggested that the infants at lower risk (birth weights of >1000 g) experienced benefit in this outcome. In the study by Schreiber et al,21 although the overall analysis indicated significance, the subgroup analysis suggested that it was only the less-sick infants (with OIs less than the median) who benefited. Therefore, it may be that, to reduce the risk of BPD, therapy must be instituted before there is major lung injury.
The study by Ballard et al24 reported a significant benefit of iNO in improving rates of survival without BPD. However, the figures given in the article (165 of 294 vs 182 of 288 patients) were not significant when RevMan software was used to calculate the RR and its CLs, as shown above, or when SPSS software (SPSS, Chicago, IL) was used to perform a simple χ2 test (without correction for continuity; χ12 = 2.841; P = .092) or to perform unadjusted, univariate, logistic regression analysis (odds ratio: 0.75; 95% CLs: 0.54, 1.048; P = .092). The reason for this discrepancy is not clear.
A posthoc subgroup analysis of the study by Ballard et al24 showed a significant reduction in the incidence of the combined outcome of death or BPD for infants who were 7 to 14 days of age at randomization; infants with less-severe disease also might be more likely to benefit (although the interaction term for that analysis was not significant). These findings suggest that additional studies, focusing on infants who seem, from these subgroup analyses, to receive the greatest benefit, may be worthwhile.
The early rescue treatment studies showed no significant effect on brain injury, as documented on ultrasound scans, but there was a trend toward increased rates of serious intracranial hemorrhage. Some of the studies, including the largest study in the group,17 did not require head ultrasound scans before enrollment; however, there were no major baseline group differences in that randomized trial,17 and the only likely explanation for the finding is that it was indeed a treatment effect. Although the result was not significant according to the conventional strict threshold of P < .05, potential evidence of harm should always be taken seriously, especially when there is no evidence of benefit. The studies with later entry on the basis of BPD risk would not be expected to demonstrate effects on intracranial hemorrhage incidence.
The studies of early routine use for intubated preterm infants showed a reduction in the incidence of serious, ultrasonographically diagnosed, brain injury (either severe intracranial hemorrhage or the combined outcome of severe hemorrhage or periventricular leukomalacia). The number needed to treat for this important outcome was a modest 14, but the confidence interval was wide (95% CLs: 8, 50). Although the criteria for study entry reported by Kinsella et al22 included gestational age of ≤34 weeks, the mean gestational age was actually 25.6 weeks for each group and the mean birth weight was <800 g, demonstrating that higher-risk groups were enrolled. Similarly, in the study by Schreiber et al,21 the actual birth weight was ∼1 kg for the 2 groups and the gestational age was <28 weeks; also, that hospital serves a somewhat deprived, inner-city neighborhood with a low rate of prenatal steroid use (<60%).
Clearly, if a population at very low risk for serious brain injury were treated, then the absolute benefit of iNO would be substantially less; for example, it would be very difficult to show a reduction in the incidence of severe brain injury in mildly ill infants born at 30 to 32 weeks’ gestation. Additional analysis of the patient characteristics that predict a beneficial response is therefore important. It has been suggested21 that there may be ethnic differences in responses to nitric oxide, accounting for some of the different results of the studies; however, the study of Kinsella et al22 did not show an effect of ethnic group. Infants who are less severely sick and of higher birth weight but still at risk of adverse outcomes seem to have the greatest benefit, on the basis of the currently available data.
Neurologic and Developmental Outcomes
Neurodevelopmental outcomes were not improved in the only early rescue treatment study to report outcomes.16 Outcomes have not been reported by most of the studies performed to date. For the early routine use studies, Schreiber et al21 demonstrated a reduction in the incidence of abnormal neurodevelopmental outcomes at 2 years of age, largely attributable to an improvement in Bayley Scales of Infant Development scores. Kinsella et al,22 who also showed an improvement in the ultrasound appearance of the brain, have not yet reported longer-term outcomes.
Our systematic review of the literature revealed that, in the group of studies of rescue therapy of very sick preterm infants who met criteria for poor oxygenation, iNO did not improve rates of survival, survival without BPD, or brain injury. In fact there was some evidence of an increase in the frequency of severe intracranial hemorrhage and of the combined outcome of severe intracranial hemorrhage or periventricular leukomalacia. In view of these findings, iNO should not be used routinely as rescue therapy in cases of severe respiratory failure among preterm infants. Although there are variations in eligibility criteria, doses of iNO, and duration of therapy, there is no clear indication from the early rescue treatment studies that this approach to treatment is promising.
In view of the lack of statistically significant benefit and the lack of long-term follow-up data from the group of studies of the later use of iNO for infants at risk of BPD, iNO use in this clinical situation cannot be recommended presently. Additional studies restricted to infants determined to be most likely to benefit, on the basis of the subgroup analyses described above, are warranted.
In contrast, the group of studies of early routine use of iNO for preterm infants undergoing ventilation who were not severely ill but nevertheless were at risk for serious brain injury or BPD showed promise. There was heterogeneity in the effects on BPD, with the main benefit being evident from a single-center study.21 Other studies to confirm this effect and to demonstrate its generalizability are required. Furthermore, only that study reported longer-term neurodevelopmental outcomes, and caution is suggested before more-widespread implementation of this use of iNO. Clear criteria for which patients in this population should be considered for treatment do not exist currently. The apparent decreases in the frequency of the combined outcome of death or BPD and in ultrasonographically detected brain injury suggest that infants who are at significant risk for these 2 outcomes but are not seriously ill would be the most appropriate target subjects for additional studies with long-term follow-up monitoring. Confirmation of the efficacy of such an approach would raise questions regarding why such infants would benefit but more-seriously ill infants would not. It is possible that infants who are sick enough to fulfill the entry criteria of the rescue studies have already suffered brain and pulmonary injuries that are too severe to be improved with iNO, whereas routine “prophylactic” use may be able to reduce the incidence of such injuries. This possibility warrants additional research.
- Accepted May 24, 2007.
- Address correspondence to Keith J. Barrington, MB, ChB, Department of Pediatrics, McGill University, Royal Victoria Hospital, 687 Pine Ave W, Montreal, Quebec, Canada H3A 1A1. E-mail:
Financial Disclosure: Dr Barrington was chairperson of the Canadian Medical Advisory Committee for iNO Therapeutics for a meeting in 2004; Dr Finer has indicated he has no financial relationships relevant to this article to disclose.
- ↵Horbar JD, Badger GJ, Carpenter JH, et al. Trends in mortality and morbidity for very low birth weight infants, 1991–1999. Pediatrics.2002;110 :143– 151
- ↵Schmidt B, Asztalos EV, Roberts RS, Robertson CM, Sauve RS, Whitfield MF. Impact of bronchopulmonary dysplasia, brain injury, and severe retinopathy on the outcome of extremely low-birth-weight infants at 18 months: results from the trial of indomethacin prophylaxis in preterms. JAMA.2003;289 :1124– 1129
- ↵Walther FJ, Benders MJ. Persistent pulmonary hypertension in premature neonates with severe respiratory distress syndrome. Pediatrics.1992;90 :899– 904
- ↵Finer NN, Barrington KJ. Nitric oxide for respiratory failure in infants born at or near term. Cochrane Database Syst Rev.2006;(4) :CD000399
- ↵Vohr BR, Wright LL, Poole WK, McDonald SA, NICHD Neonatal Research Network Follow-up Study. Neurodevelopmental outcomes of extremely low birth weight infants <32 weeks’ gestation between 1993 and 1998. Pediatrics.2005;116 :635– 643
- ↵Wood NS, Costeloe K, Gibson AT, et al. The EPICure study: associations and antecedents of neurological and developmental disability at 30 months of age following extremely preterm birth. Arch Dis Child Fetal Neonatal Ed.2005;90 :F134– F140
- ↵Abman SH, Kinsella JP, Schaffer MS, Wilkening RB. Inhaled nitric oxide in the management of a premature newborn with severe respiratory distress and pulmonary hypertension. Pediatrics.1993;92 :606– 609
- ↵Field D, Elbourne D, Truesdale A, et al. Neonatal Ventilation With Inhaled Nitric Oxide Versus Ventilatory Support Without Inhaled Nitric Oxide for Preterm Infants With Severe Respiratory Failure: the INNOVO Multicentre Randomised Controlled Trial (ISRCTN 17821339). Pediatrics.2005;115 :926– 936
- ↵Srisuparp P, Heitschmidt M, Schreiber MD. Inhaled nitric oxide therapy in premature infants with mild to moderate respiratory distress syndrome. J Med Assoc Thai.2002;85(suppl 2) :S469– S478
- ↵Subhedar N, Ryan S, Shaw N. Open randomised controlled trial of inhaled nitric oxide and early dexamethasone in high risk preterm infants. Arch Dis Child Fetal Neonatal Ed.1997;77 :F185– F190
- Copyright © 2007 by the American Academy of Pediatrics