The New England
`Journal
` Medicine
`of
`
`© Copyright, 1997, by the Massachusetts Medical Societ y
`
`VOLUME 336
`
`F
`
`E B R U A R Y
`
` 27, 1997
`
`NUMBER 9
`
`INHALED NITRIC OXIDE IN FULL-TERM AND NEARLY FULL-TERM INFANTS
`WITH HYPOXIC RESPIRATORY FAILURE
`
`T
`HE
`
` N
`
`EONATAL
`
` I
`NHALED
`
` N
`ITRIC
`
` O
`
`XIDE
`
` S
`
`TUDY
`
` G
`ROUP
`
`*
`
`A
`BSTRACT
`Background
`Neonates with pulmonary hyperten-
`sion have been treated with inhaled nitric oxide be-
`cause of studies suggesting that it is a selective pul-
`monary vasodilator. We conducted a randomized,
`multicenter, controlled trial to determine whether in-
`haled nitric oxide would reduce mortality or the ini-
`tiation of extracorporeal membrane oxygenation in
`infants with hypoxic respiratory failure.
`Methods
`⭓
`34
`Infants born after a gestation of
`weeks who were 14 days old or less, had no struc-
`tural heart disease, and required assisted ventilation
`and whose oxygenation index was 25 or higher on
`two measurements were eligible for the study. The
`infants were randomly assigned to receive nitric ox-
`ide at a concentration of 20 ppm or 100 percent ox-
`ygen (as a control). Infants whose partial pressure of
`arterial oxygen (PaO
`) increased by 20 mm Hg or
`2
`less after 30 minutes were studied for a response to
`80-ppm nitric oxide or control gas.
`Results
`The 121 infants in the control group and
`the 114 in the nitric oxide group had similar base-line
`clinical characteristics. Sixty-four percent of the con-
`trol group and 46 percent of the nitric oxide group
`died within 120 days or were treated with extracorpo-
`⫽
`real membrane oxygenation (P
`0.006). Seventeen
`percent of the control group and 14 percent of the ni-
`tric oxide group died (P not significant), but signifi-
`cantly fewer in the nitric oxide group received extra-
`corporeal membrane oxygenation (39 percent vs. 54
`⫽
`0.014). The nitric oxide group had sig-
`percent, P
`⫾
`nificantly greater improvement in PaO
` (mean [
`SD]
`2
`⫾
`⫾
`85.2 mm Hg, vs. 9.7
`51.7 mm Hg in
`increase, 58.2
`⬍
`the controls; P
`0.001) and in the oxygenation index (a
`⫾
`⫾
`decrease of 14.1
`21.1, vs. an increase of 0.8
`21.1 in
`⬍
`0.001). The study gas was not discon-
`the controls; P
`tinued in any infant because of toxicity.
`Conclusions
`Nitric oxide therapy reduced the use
`of extracorporeal membrane oxygenation, but had
`no apparent effect on mortality, in critically ill infants
`with hypoxic respiratory failure. (N Engl J Med 1997;
`336:597-604.)
`©1997, Massachusetts Medical Society.
`
`H
`
`YPOXIC respiratory failure in neonates
`34 weeks’
`born at or near term (at
`⭓
`gestation) may be caused by conditions
`such as primary persistent pulmonary
`hypertension, respiratory distress syndrome, aspira-
`tion of meconium, pneumonia or sepsis, and con-
`genital diaphragmatic hernia.
` Conventional thera-
`1,2
`py, short of extracorporeal membrane oxygenation,
`involves support with oxygen, mechanical ventila-
`tion, and the induction of alkalosis, neuromuscular
` None of these therapies
`blockade, and sedation.
`3-6
`have been found to reduce mortality or the need for
`extracorporeal membrane oxygenation. To date, se-
`lective pulmonary vasodilators free of systemic side
`effects have not been studied in large trials of neo-
`nates.
`7
`Nitric oxide, or endothelium-derived relaxing fac-
`tor, is important in regulating vascular muscle tone.
`8-13
`In newborn lambs with pulmonary hypertension in-
`duced by hypoxia, the inhalation of 40 to 80 parts
`per million (ppm) of nitric oxide reversed pulmo-
`nary vasoconstriction without affecting the systemic
` Two recent studies of neonates with
`circulation.
`14-16
`severe persistent pulmonary hypertension have shown
`that inhaled nitric oxide rapidly improved preductal
`oxygen saturation, without detectable toxic effects.
`17,18
`A prospective study of multiple randomized doses of
`inhaled nitric oxide in infants referred for extra-
`corporeal membrane oxygenation did not find a
`correlation between the dose of nitric oxide and the
` We con-
`degree of improvement in oxygenation.
`19
`ducted a prospective, multicenter, randomized, con-
`trolled, double-blind trial to evaluate whether inhaled
`
`Address reprint requests to Dr. Richard A. Ehrenkranz at the Depart-
`ment of Pediatrics, Yale University School of Medicine, P.O. Box 208064,
`333 Cedar St., New Haven, CT 06520-8064.
`Dr. Ehrenkranz, as co-principal investigator of the study, assumes re-
`sponsibility for the overall content and integrity of the article.
`*The members of the Neonatal Inhaled Nitric Oxide Study Group are
`listed in the Appendix.
`
`Volume 336 Number 9
`
`ⴢ
`
`597
`
`The New England Journal of Medicine
`
`Downloaded from nejm.org on September 17, 2014. For personal use only. No other uses without permission.
`
` Copyright © 1997 Massachusetts Medical Society. All rights reserved.
`
`597
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`The New England Journal of Medicine
`
`nitric oxide would reduce mortality or the need for
`extracorporeal membrane oxygenation in infants
`born at or near term who had hypoxic respiratory
`failure that was unresponsive to aggressive conven-
`tional therapy.
`
`METHODS
`
`Study Hypotheses
`The primary hypothesis in the study was that administering in-
`haled nitric oxide to infants born at 34 or more weeks of gesta-
`tion who had hypoxic respiratory failure and an oxygenation in-
`dex of 25 or higher would reduce the risk of death by day 120
`or the initiation of extracorporeal membrane oxygenation from
`50 percent in control infants to 30 percent in infants given nitric
`oxide, a relative reduction of 40 percent. The oxygenation index
`was calculated as the mean airway pressure times the fraction of
`) divided by the partial pressure of arterial
`inspired oxygen (FiO
`2
`oxygen (PaO
`) times 100.
`2
`The secondary hypothesis was that 30 minutes after the start
`of treatment, inhaled nitric oxide would increase PaO
` and de-
`2
`crease the oxygenation index and the alveolar–arterial oxygen gra-
`dient. We hypothesized that among the surviving infants, treat-
`ment with inhaled nitric oxide would shorten hospitalization
`without increasing the duration of assisted ventilation or the in-
`cidence of air leakage, bronchopulmonary dysplasia, or neurode-
`velopmental disability at 18 to 24 months.
`
`Study Patients
`Infants born at 34 or more weeks of gestation who required
`assisted ventilation for hypoxic respiratory failure and had an oxy-
`genation index of at least 25 on two measurements made at least
`15 minutes apart were eligible for the trial. Hypoxic respiratory
`failure was caused by persistent pulmonary hypertension, meco-
`nium aspiration, pneumonia or sepsis, respiratory distress syn-
`drome, or suspected pulmonary hypoplasia associated with oligo-
`hydramnios and premature rupture of the membranes. All the
`infants were required to have an indwelling catheter and to un-
`dergo echocardiography before randomization. Echocardiographic
`evidence of pulmonary hypertension was not required, because
`studies have shown that inhaled nitric oxide improves the match-
`ing of ventilation with perfusion and may reduce intrapulmonary
`shunting in the absence of a direct intracardiac shunt.
`20,21
`Infants were considered ineligible for the study if they were
`more than 14 days old, had a congenital diaphragmatic hernia, or
`were known to have congenital heart disease, or if it had been de-
`cided not to provide full treatment. The study centers attempted
`to obtain a cranial ultrasonogram before enrolling an infant in the
`study. Consent was obtained from the parents or guardians before
`the infants underwent randomization, and each study center ob-
`tained approval from the institutional review board before enroll-
`ment began. Copies of the study protocol are available from the
`authors on request.
`
`Guidelines for Management
`The approach to care before enrollment was not specified by
`the study protocol. Each participating center developed general
`management guidelines to be used throughout the study and
`agreed to use the most aggressive forms of conventional therapy
`before randomization. These guidelines included the maintenance
`of a mean arterial blood pressure above 45 mm Hg, the induction
`of alkalosis (range of target pH, 7.45 to 7.6), and treatment with
`bovine surfactant (BLES, BLES Biochemicals, London, Ont.,
`Canada; or Survanta, Abbott Laboratories, Columbus, Ohio) be-
`fore the start of treatment with the study gas. The protocol spec-
`ified that the mode of ventilation (conventional or high frequen-
`cy) could not be changed after randomization, except as part of
`weaning from assisted ventilation.
`
`598
`
`ⴢ
`
`Februar y 27, 1997
`
`Randomization
`The infants were stratified according to study center and ran-
`domly assigned by telephone to receive either 100 percent oxygen
`(the control treatment) or nitric oxide according to a permuted-
`block design developed and implemented by the coordinating
`center.
`
`Administration and Monitoring of Study Gas
`If treatment with the study gas could be started within 15 min-
`utes after the second qualifying oxygenation-index score was ob-
`tained, the arterial-blood gas values from that measurement
`served as the base-line values in assessing the response to the
`study treatment. If the treatment could not be started within the
`15-minute period, a third measurement of arterial-blood gas,
`obtained before the administration of the study gas, was used to
`determine the base-line value. Primary-grade nitric oxide was sup-
`plied in a concentration of 800 ppm in balanced nitrogen
`(Canadian Liquid Air, Montreal; and Ohmeda, Liberty Corner,
`N.J.); the gas was certified to be within
`1 percent of the stated
`⫾
`nitric oxide content and to contain less than 5 ppm of nitrogen
`dioxide. The gas mixture was sampled after it entered the injec-
`tion site of the inspiratory circuit and before it reached the in-
`fant’s endotracheal tube and was analyzed continuously for ni-
`tric oxide and nitrogen dioxide with chemiluminescence (model
`42H, Thermo Environmental Instruments, Franklin, Mass.; and
`model CLD 700AL, ECO Physics, Durten, Switzerland) or with
`electrochemical analyzers (Pulmonox II, Pulmonox, Tolfield,
`Alta., Canada; and Dräger Prac II, Dräger, Chantilly, Va.). Qual-
`ity-control procedures ensured accurate calibration and prevented
`the supply tank of nitric oxide gas from being contaminated.
`Except when the treatment was initiated and when the concen-
`tration of the study gas was changed, the infants were cared for
`by clinical teams unaware of each infant’s treatment assignment;
`the randomization was performed, the gas administered, and safe-
`ty monitored by designated persons who were not involved in the
`clinical care. Levels of inspired oxygen, nitric oxide, and nitrogen
`dioxide were recorded every two hours and after the settings of
`the ventilator were changed. We kept the clinical teams unaware
`of the treatment assignments by making mock adjustments in the
`case of the control infants, covering the analyzer readings and the
`gas tanks, and sampling the supply of oxygen before the injection
`site of the study gas.
`A response to treatment was defined according to the change
` 30 minutes after the initial exposure
`from base line in the PaO
`2
`to the study gas (a complete response was defined as an increase
`of more than 20 mm Hg; a partial response, as an increase of
`10 to 20 mm Hg; and no response, as an increase of less than 10
`mm Hg) when the two measurements were made at comparable
`sampling sites. When an infant had a complete response, treat-
`ment with the study gas (either nitric oxide at a concentration of
`20 ppm or 100 percent oxygen) was continued. When an infant
`had less than a complete response, the treatment was stopped for
`15 minutes if the stoppage was tolerated, the arterial-blood gases
`were measured again, and then the study gas was administered at
`a maximal concentration of 80 ppm. Arterial-blood gases were
`measured again 30 minutes later. Infants who had complete re-
`sponses to the maximal concentration continued to be treated at
`that concentration; in infants with partial responses, treatment
`was continued at the lowest concentration of gas that produced
`at least a partial response. If an infant had no response with either
`the 20-ppm or the 80-ppm concentration of gas, treatment was
`discontinued. Gas was also discontinued in any infant whose con-
`dition deteriorated (absolute decrease in oxygen saturation,
`10
`⬎
`percent) before the end of the initial phase of administration at
`either the high or the low concentration, and such infants were
`classified as having no response. When an infant did not respond
`to the initial administration of the study gas, the treatment could
`be attempted again as many as three times at six-hour intervals.
`No crossover between study groups was allowed.
`If an infant continued to receive the study gas after the initial
`
`The New England Journal of Medicine
`
`Downloaded from nejm.org on September 17, 2014. For personal use only. No other uses without permission.
`
` Copyright © 1997 Massachusetts Medical Society. All rights reserved.
`
`598
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`INHALED NITRIC OXIDE IN FULL-TERM AND NEARLY FULL-TERM INFANTS WITH HYPOXIC RESPIRATORY FAILURE
`
`dosing algorithm, the gas was monitored in an unmasked fashion
`by designated persons who were not involved with the infant’s
`clinical care. The protocol suggested algorithms for weaning in-
`fants from the study gas, escalating the dose of gas after the oc-
`currence of clinical deterioration, and starting treatment again af-
`ter successful weaning. The study protocol permitted treatment
`with the study gas for a cumulative maximum of 336 hours (14
`days). Decisions about initiating extracorporeal membrane oxy-
`genation were made by the blinded clinical team on the basis of
`center-specific criteria.
`
`Monitoring of Safety
`Blood methemoglobin concentrations were measured 1, 3, 6,
`and 12 hours after the start of treatment with the study gas and
`every 12 hours thereafter until 24 hours after the treatment end-
`ed. Methemoglobin levels of 5 to 10 percent were managed by
`reducing the concentration of study gas by half until the level fell
`below 5 percent. The study gas was discontinued if the methe-
`moglobin level exceeded 10 percent. If the concentration of ni-
`trogen dioxide exceeded 7 ppm, the study gas was discontinued;
`the gas was decreased by half if the concentration was 5 to 7 ppm.
`The infants were monitored for signs of bleeding. Cranial ul-
`trasonography was performed before randomization and 24
`hours after the final discontinuation of the study gas. All the read-
`ings were done by local ultrasonographers.
`22
`
`Statistical Analysis
`According to the data from the participating centers, we esti-
`mated that mortality or the use of extracorporeal membrane oxy-
`genation in infants with an oxygenation-index score between 25
`and 40 would be 50 percent. To demonstrate a 40 percent reduc-
`tion in the primary outcome with a power of 0.90 and a two-
`tailed alpha of 0.05, 125 patients were required in each group.
`The primary analysis was an intention-to-treat analysis.
`Continuous variables were compared by t-tests or Wilcoxon
`tests, and discrete variables were compared by chi-square tests.
`The Gart test was used to evaluate the homogeneity of relative
`risks.
`23
`The trial was monitored by an independent Data Safety and
`Monitoring Committee, which planned evaluations after approxi-
`mately one third and two thirds of the study patients were enrolled.
`To reduce the overall probability of a type I error as much as pos-
`sible, significance was tested at each interim analysis by the group-
`sequential method of Lan and DeMets with the O’Brien–Fleming
`spending function.
` Results are presented as means
`SD.
`⫾
`24
`
`RESULTS
`The trial was terminated at the recommendation
`of the Data Safety and Monitoring Committee after
`the second planned review of data, which showed
`that the z value had crossed the predetermined
`boundary of statistical significance. After the recom-
`mendation was reviewed and accepted by the Na-
`tional Institute of Child Health and Human Devel-
`opment and the investigators, recruitment ceased on
`May 2, 1996.
`
`Base-Line Characteristics
`Two hundred thirty-five infants were enrolled in
`the trial. There were no significant differences be-
`tween the study groups in the characteristics of the
`patients (Table 1), treatment methods, or status at
`the time of randomization (Table 2). Seventy-two
`percent of the controls and 71 percent of the treated
`infants received surfactant before randomization, and
`
`T
` 1.
`ABLE
`
` C
`
`HARACTERISTICS
`
`
`OF
`
` P
`.*
`THE
`ATIENTS
`
`C
`HARACTERISTIC
`
`Birth weight — g
`Gestational age — wk
`Male sex — no. (%)
`Race — no. (%)†
`Black
`White
`Hispanic
`Other
`Not born in treating facility — no. (%)
`Age at admission‡
`12 hr
`⬍
`12–24 hr
`24 hr
`⬎
`3 — no. (%)§
`1-Minute Apgar score
`⬍
`Primary diagnosis — no. (%)
`Persistent pulmonary hypertension of
`the newborn
`Respiratory distress syndrome
`Meconium aspiration syndrome
`Pneumonia or sepsis
`Other
`
`C
`ONTROL
`G
`ROUP
`ⴝ
`(N
`121)
`
`597
`3359
`⫾
`38.9
`2.2
`⫾
`76 (62.8)
`
`19 (16.0)
`72 (60.5)
`17 (14.3)
`11 (9.2)
`93 (76.9)
`
`47 (50.5)
`20 (21.5)
`26 (28.0)
`25 (20.8)
`
`N
` O
`XIDE
`ITRIC
`G
`ROUP
`ⴝ
`(N
`114)
`
`3460
`578
`⫾
`39.3
`1.8
`⫾
`63 (55.3)
`
`19 (17.1)
`70 (63.1)
`13 (11.7)
`9 (8.1)
`92 (80.7)
`
`41 (45.1)
`20 (22.0)
`30 (33.0)
`28 (24.8)
`
`22 (18.2)
`
`19 (16.7)
`
`15 (12.4)
`58 (47.9)
`24 (19.8)
`2 (1.7)
`
`10 (8.8)
`58 (50.9)
`26 (22.8)
`1 (0.9)
`
`SD.
`*Plus–minus values are means
`⫾
`†Data on race are based on 119 infants in the control group and 111
`infants in the nitric oxide group.
`‡Data for this variable are based on 93 infants in the control group and
`91 infants in the nitric oxide group.
`§Data for this variable are based on 120 infants in the control group and
`113 infants in the nitric oxide group.
`
`50 percent and 49 percent, respectively, received it
`within six hours before randomization. High-fre-
`quency ventilation, primarily oscillatory, was used in
`55 percent of both groups; 37 percent of the con-
`trols and 32 percent of the treated infants received
`such treatment at randomization. Over 90 percent
`of all the infants received volume support, vasopres-
`sor support, neuromuscular blockade, and sedation
`before randomization (Table 2).
`The causes of hypoxic respiratory failure are shown
`in Table 1. Forty-nine percent of all randomized in-
`fants had meconium aspiration syndrome; 17 percent
`had persistent pulmonary hypertension. Echocardi-
`ography was performed before randomization in 228
`infants (97 percent); of the 226 infants for whom
`complete data were available, 78 percent had evi-
`dence of pulmonary hypertension (right-to-left or bi-
`directional shunting, tricuspid-valve regurgitation, or
`both). There was no difference in the prevalence of
`pulmonary hypertension between the study groups.
`2.3 days after birth
`Randomization occurred 1.7
`⫾
`for the controls and 1.7
`1.8 days after birth for the
`⫾
`treated infants (Table 2). Data from the first quali-
`fying arterial-blood gas measurement are also shown
`in Table 2; on the second qualifying measurement,
`the oxygenation index was 46.3
`19.9 in the control
`⫾
`
`Volume 336 Number 9
`
`ⴢ
`
`599
`
`The New England Journal of Medicine
`
`Downloaded from nejm.org on September 17, 2014. For personal use only. No other uses without permission.
`
` Copyright © 1997 Massachusetts Medical Society. All rights reserved.
`
`599
`
`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`The New England Journal of Medicine
`
`T
`
` 2.
`ABLE
`
` S
` T
` V
`
`
`TATUS
`AND
`REATMENT
`ARIABLES
` R
`.*
`AT
`ANDOMIZATION
`
`
`OF
`
`THE
`
` P
`ATIENTS
`
`V
`ARIABLE
`
`Treatment — no. of patients (%)
`Volume support
`Vasopressor support
`Tolazoline
`Sedation or analgesia
`Neuromuscular blockade
`Alkalosis
`Surfactant
`High-frequency ventilation
`Air leaks — no. of patients (%)
`Pulmonary hemorrhage — no. of patients (%)
`First qualifying arterial-blood gas value
`Oxygenation index
`Mean airway pressure (cm of water)
` (mm Hg)
`FiO
`2
`PaO
` (mm Hg)
`2
`Alveolar–arterial oxygen gradient
`(mm Hg)
`Age at randomization (days)
`Median time, randomization to study-gas
`initiation (min)‡
`
`C
`ONTROL
`G
`ROUP
`ⴝ
`(N
`121)
`
`N
` O
`XIDE
`ITRIC
`G
`ROUP
`ⴝ
`(N
`114)
`
`116 (96.7)†
`121 (100.0)
`16 (13.3)†
`120 (99.2)
`115 (95.0)
`106 (87.6)
`87 (71.9)
`67 (55.4)
`25 (20.7)
`22 (18.2)
`
`108 (94.7)
`108 (94.7)
`23 (20.2)
`113 (99.1)
`107 (93.9)
`88 (77.2)
`81 (71.1)
`63 (55.3)
`20 (17.5)
`18 (15.8)
`
`45.1
`22.4
`⫾
`18.3
`4.4
`⫾
`1.0
`0.0
`⫾
`13.9
`45.5
`⫾
`613.7
`40.3
`⫾
`
`17.6
`43.0
`⫾
`18.3
`4.3
`⫾
`1.0
`0.0
`⫾
`46.8
`15.5
`⫾
`616.1
`33.5
`⫾
`
`1.7⫾2.3
`10.0
`
`1.7⫾1.8
`15.0
`
` denotes fraction of inspired
`SD. FiO
`*Plus–minus values are means
`⫾
`2
`oxygen, and PaO
` partial pressure of arterial oxygen.
`2
`†A total of 120 patients were studied for this variable.
`‡Data for this variable are based on 117 infants in the control group and
`113 in the nitric oxide group.
`
`31.3 in the nitric oxide group. Six-
`group and 47.3
`⫾
`ty-two percent of the control group and 64 percent
`of the nitric oxide group had a third arterial-blood
`gas measurement before treatment with the study
`gas was begun. The median time from randomiza-
`tion to the administration of the study gas was 10
`minutes in the control group and 15 minutes in the
`nitric oxide group (Table 2). Five randomized in-
`fants (four in the control group and one in the nitric
`oxide group) did not receive study gas.
`
`Primary Outcome
`The incidence of the primary outcome (death by
`120 days of age or the initiation of extracorporeal
`membrane oxygenation) was significantly lower in
`the nitric oxide group than in the control group (46
`percent vs. 64 percent; relative risk, 0.72; 95 percent
`0.006, a sig-
`confidence interval, 0.57 to 0.91; P
`⫽
`nificant difference given the Lan–DeMets cutoff of
`0.044) (Table 3). Thirty-six infants died, among
`whom 17 (9 in the control group and 8 in the ni-
`tric oxide group) received extracorporeal membrane
`oxygenation. Among the other 19 infants who died,
`10 (5 in each group) had contraindications to extra-
`corporeal membrane oxygenation; 5 (3 in the con-
`trol group and 2 in the nitric oxide group) had their
`life support withdrawn; and 4 (3 and 1 in the respec-
`tive groups) did not meet center-specific criteria for
`
`TABLE 3. OUTCOMES OF ADMINISTRATION OF THE STUDY GAS, ACCORDING TO GROUP.*
`
`OUTCOME
`
`Death by day 120 or ECMO — no. (%)
`Death — no. (%)
`ECMO — no. (%)
`Change in PaO2 — mm Hg
`Change in oxygenation index
`Change in alveolar–arterial oxygen gradient — mm Hg
`Outcomes in surviving infants
`Length of hospitalization — days
`Duration of assisted ventilation — days
`Air leak after randomization — no. (%)
`Bronchopulmonary dysplasia — no. (%)†
`
`CONTROL
`GROUP
`(Nⴝ121)
`
`NITRIC OXIDE
`GROUP
`(Nⴝ114)
`
`P VALUE
`
`0.006
`52 (45.6)
`77 (63.6)
`0.60
`16 (14.0)
`20 (16.5)
`0.014
`44 (38.6)
`66 (54.5)
`58.2⫾85.2 ⬍0.001
`9.7⫾51.7
`0.8⫾21.1 ⫺14.1⫾21.1 ⬍0.001
`⫺6.7⫾57.5 ⫺60.0⫾85.1 ⬍0.001
`
`29.5⫾22.6
`11.7⫾13.0
`5 (5.1)
`12 (11.9)
`
`36.4⫾44.8
`11.6⫾7.0
`5 (5.2)
`15 (15.3)
`
`0.17
`0.97
`0.96
`0.48
`
`*Plus–minus values are means ⫾SD. ECMO denotes extracorporeal membrane oxygenation, and
`PaO2 partial pressure of arterial oxygen.
`†This condition was considered to be present when there was dependence on oxygen at the age
`of 28 days accompanied by abnormal results on chest radiography.
`
`600
`
`ⴢ
`
`Februar y 27, 1997
`
`The New England Journal of Medicine
`
`Downloaded from nejm.org on September 17, 2014. For personal use only. No other uses without permission.
`
` Copyright © 1997 Massachusetts Medical Society. All rights reserved.
`
`600
`
`

`
`INHALED NITRIC OXIDE IN FULL-TERM AND NEARLY FULL-TERM INFANTS WITH HYPOXIC RESPIRATORY FAILURE
`
`extracorporeal membrane oxygenation. There were
`no differences between the groups in the causes of
`death. The infants in the nitric oxide group received
`extracorporeal membrane oxygenation less often (39
`percent) than the controls (55 percent, P⫽0.014)
`(Table 3). The median time from randomization to
`the initiation of extracorporeal membrane oxygena-
`tion was 4.4 hours in the control group and 6.7
`hours in the nitric oxide group (P⫽0.04).
`
`Secondary Outcomes
`Among the surviving infants, there were no differ-
`ences between the groups with respect to the length
`of hospitalization, the number of days of respiratory
`support (assisted ventilation, continuous positive air-
`way pressure, or oxygen), or the incidence of air
`leakage or bronchopulmonary dysplasia (Table 3).
`Thirty minutes after the administration of the
`study gas began, the infants in the nitric oxide group
`had a significantly greater mean increase in PaO2
`than the controls (58.2⫾85.2 vs. 9.7⫾51.7 mm Hg),
`a significantly greater change in the oxygenation
`index (a decrease of 14.1⫾21.1 as compared with
`an increase of 0.8⫾21.1), and a significantly greater
`decrease in the alveolar–arterial oxygen gradient
`(60.0⫾85.1 vs. 6.7⫾57.5 mm Hg; P⬍0.001 for all
`three comparisons) (Table 3).
`More infants in the nitric oxide group than in the
`
`control group had at least a partial response to the
`initial administration of the study gas (66 percent vs.
`26 percent, P⬍0.001) (Table 4). Of the 125 infants
`who had no response to 20-ppm nitric oxide or con-
`trol gas, similar proportions of the nitric oxide group
`(18 percent [7 of 38]) and the control group (20
`percent [17 of 87]) had at least partial responses to
`80-ppm nitric oxide or control gas (P⫽0.30). Of the
`30 infants who had partial responses to the study gas
`at 20 ppm, 29 percent of the nitric oxide group (5 of
`17) and 8 percent of the control group (1 of 13)
`had at least a partial response at 80 ppm (P⫽0.34).
`Therefore, a majority of the infants who did not have
`complete responses at the 20-ppm concentration and
`who were evaluated at the 80-ppm concentration
`had no response to the study gas at the higher con-
`centration (nitric oxide group, 77 percent [41 of 53];
`control group, 81 percent [75 of 93]).
`According to the study protocol, three additional
`trials were permitted, but only 10 infants (6 in the
`control group and 4 in the nitric oxide group) under-
`went such trials. Twenty-eight infants assigned to the
`control group (23 percent) received the study gas for
`more than 24 hours, as compared with 64 infants as-
`signed to the nitric oxide group (56 percent) (median
`duration of gas administration, 2 hours vs. 40 hours;
`P⬍0.001). Among the infants who had responses to
`either the 20-ppm or the 80-ppm concentration of
`
`TABLE 4. RESPONSES TO THE INITIAL ADMINISTRATION OF 20-ppm NITRIC OXIDE
`OR OXYGEN, AND SUBSEQUENT RESPONSES TO 80-ppm CONCENTRATIONS
`OF STUDY GAS BY INFANTS WHOSE RESPONSES TO THE INITIAL TREATMENT
`WERE LESS THAN COMPLETE.*
`
`VARIABLE
`
`Response to treatment at 20 ppm
`No. of infants
`None
`Partial
`Complete
`Subsequent response to treatment at 80 ppm
`Infants with no response at 20 ppm
`None
`Partial
`Complete
`80 ppm not tried
`Infants with partial responses at 20 ppm
`None
`Partial
`Complete
`80 ppm not tried
`
`CONTROL
`GROUP
`
`NITRIC OXIDE
`GROUP
`
`P VALUE†
`
`no. of patients (%)
`
`117
`87 (74.4)
`13 (11.1)
`17 (14.5)
`
`112
`38 (33.9)
`17 (15.2)
`57 (50.9)
`
`⬍0.001
`
`64 (73.6)
`5 (5.7)
`12 (13.8)
`6 (6.9)
`
`11 (84.6)
`1 (7.7)
`0
`1 (7.7)
`
`29 (76.3)
`5 (13.2)
`2 (5.3)
`2 (5.3)
`
`12 (70.6)
`4 (23.5)
`1 (5.9)
`0
`
`0.30
`
`0.34
`
`*Data on 229 infants are shown because 4 infants in the control group and 1 in the nitric oxide group
`did not receive study gas and data on 1 infant treated with nitric oxide were unavailable because of me-
`chanical problems with gas delivery. Seven infants (six in the nitric oxide group and one in the control
`group) who received the wrong study gas are included in the table under their assigned treatments.
`†P values are for the comparison between groups with respect to the number of infants with either
`a partial or a complete response to the study gas.
`
`Volume 336 Number 9
`
`ⴢ 601
`
`The New England Journal of Medicine
`
`Downloaded from nejm.org on September 17, 2014. For personal use only. No other uses without permission.
`
` Copyright © 1997 Massachusetts Medical Society. All rights reserved.
`
`601
`
`

`
`The New England Journal of Medicine
`
`TABLE 5. RESULTS OF THE SUBGROUP ANALYSIS.
`
`VARIABLE
`
`Primary diagnosis
`Persistent pulmonary hypertension
`Respiratory distress syndrome
`Meconium aspiration
`Pneumonia or sepsis
`Pulmonary hypertension found by
`echocardiography
`Yes
`No
`Surfactant
`Yes
`No
`High-frequency ventilation
`Yes
`No
`Surfactant and high-frequency
`ventilation
`Both
`Neither
`Oxygenation index
`25.0–29.9
`30.0–39.9
`40.0–59.9
`⭓60.0
`
`NO. OF
`PATIENTS*
`
`PERCENT WITH
`IMPROVED
`OXYGENATION†
`NITRIC
`OXIDE
`
`CONTROL
`
`PERCENT WITH
`PRIMARY
`OUTCOME‡
`NITRIC
`OXIDE
`
`CONTROL
`
`RELATIVE RISK
`(95% CI)§
`
`41
`25
`116
`50
`
`176
`50
`
`168
`67
`
`130
`105
`
`88
`25
`
`53
`72
`74
`35
`
`14
`8
`16
`17
`
`15
`14
`
`14
`15
`
`14
`16
`
`13
`14
`
`7
`27
`11
`11
`
`61
`60
`47
`52
`
`54
`44
`
`51
`50
`
`57
`43
`
`59
`40
`
`76
`42
`51
`27
`
`73
`47
`62
`67
`
`65
`50
`
`54
`88
`
`66
`61
`
`57
`93
`
`61
`42
`76
`84
`
`32
`50
`52
`39
`
`47
`39
`
`38
`64
`
`46
`45
`
`37
`64
`
`28
`47
`47
`69
`
`0.43 (0.23–0.81)
`1.07 (0.46–2.49)
`0.83 (0.61–1.15)
`0.58 (0.33–1.00)
`
`0.72 (0.55–0.94)
`0.79 (0.42–1.48)
`
`0.71 (0.51–0.99)
`0.72 (0.55–0.95)
`
`0.70 (0.51–0.96)
`0.74 (0.51–1.06)
`
`0.64 (0.40–1.00)
`0.69 (0.45–1.04)
`
`0.46 (0.24–0.88)
`1.13 (0.67–1.91)
`0.62 (0.43–0.89)
`0.82 (0.56–1.18)
`
`*Data on primary diagnosis are based on 232 patients; on pulmonary hypertension found by echocardiography, 226;
`and on oxygenation index, 234.
`†Improved oxygenation was defined as a complete response (an increase of more than 20 mm Hg in the partial pressure
`of arterial oxygen) to the administration of 20-ppm nitric oxide or control gas at 30 minutes.
`‡The primary study outcome was death by 120 days of age or the initiation of extracorporeal membrane oxygenation.
`§Relative risks shown are for the occurrence of the primary study outcome in the nitric oxide group as compared with
`the control group. CI denotes confidence interval.
`
`study gas when it was first administered, 62 percent
`of those in the nitric oxide group (50 of 81) were
`successfully weaned, as compared with 40 percent of
`those in the control group (19 of 47). Among the in-
`fants successfully weaned, three of those in the con-
`trol group and two of those in the nitric oxide group
`had the study gas administered again.
`Post hoc subgroup analyses were performed to
`evaluate the relations between each of several vari-
`ables — the primary diagnosis, the presence or ab-
`sence of echocardiographic evidence of pulmonary
`hypertension, the first qualifying oxygenation-index
`score, and treatment with surfactant before random-
`ization, the use of high-frequency ventilation at the
`time of randomization or earlier, or the use of both
`surfactant and ventilation — and the incidence of
`the primary outcome and a complete response to
`the study gas (Table 5). Tests of homogeneity did
`not show significant differences between the relative
`risks. Therefore, there was no conclusive evidence,
`when the nitric oxide group was compared with the
`control group, that the relative risk either of the pri-
`mary outcome or of a complete response to nitric
`
`oxide was related to any of the variables studied in
`the subgroup analysis.
`
`Safety and Toxicity
`The study gas was not discontinued in any infant
`because of toxic effects. In the nitric oxide group,
`the mean peak level of nitrogen dioxide was 0.8⫾1.2
`ppm, and the mean peak methemoglobin level was
`2.4⫾1.8 percent. The concentration of inhaled ni-
`tric oxide was reduced in 11 infants in the nitric ox-
`ide group because of elevated methemoglobin levels
`(5 to 10 percent).
`There were no significant differences between the
`groups after randomization in the overall incidence
`or severity of intracranial hemorrhage (total number,
`19 in the control group and 18 in the nitric oxide
`group; grade IV, 8 and 5, respectively). There were
`also no significant differences between the control
`group and the nitric oxide group in the occurrence
`of periventricular leukomalacia (6 vs. 3), brain in-
`farction (7 vs. 7), seizures requiring anticonvulsive
`therapy (16 vs. 24), and either pulmonary (4 vs. 6)
`or gastrointestinal (1 vs. 1) hemorrhage.
`
`602 ⴢ
`
`Februar y 27, 1997
`
`The New England Journal of Medicine
`
`Downloaded from nejm.org on September 17, 2014. For personal use only. No other uses without permission.
`
` Copyright © 1997 Massachusetts Medical Society. All rights reserved.
`
`602
`
`

`
`INHALED NITRIC OXIDE IN FULL-TERM AND NEARLY FULL-TERM INFANTS WITH HYPOXIC RESPIRATORY FAILURE
`
`There were 21 deviations from the protocol. Two
`infants who were ineligible for the study were ran-
`domized: one had a cystic adenomatoid malforma-
`tion, and the other had a qualifying oxygenation-
`index score of 24.4. One infant randomly assigned
`to nitric oxide received oxygen, and six co