Safety of Withdrawing Inhaled Nitric Oxide Therapy in Persistent Pulmonary Hypertension of the Newborn
Objective. Because of case reports describing hypoxemia on withdrawal of inhaled nitric oxide (I-NO), we prospectively examined this safety issue in newborns with persistent pulmonary hypertension who were classified as treatment successes or failures during a course of I-NO therapy.
Methods. Randomized, placebo-controlled, double-masked, dose-response clinical trial at 25 tertiary centers from April 1994 to June 1996. Change in oxygenation and outcome (death and/or extracorporeal membrane oxygenation) during or immediately after withdrawing I-NO were the principal endpoints. Patients (n = 155) were term infants, <3 days old at study entry with echocardiographic evidence of persistent pulmonary hypertension of the newborn. Exclusion criteria included previous surfactant treatment, high-frequency ventilation, or lung hypoplasia. Withdrawal from treatment gas (0, 5, 20, or 80 ppm) started once treatment success or failure criteria were met. Withdrawal of treatment gas occurred at 20% decrements at <4 hours between steps.
Results. The patient profile was similar for placebo and I-NO groups. Treatment started at an oxygenation index (OI) of 25 ± 10 (mean ± SD) at 26 ± 18 hours after birth. For infants classified as treatment successes (mean duration of therapy = 88 hours, OI <10), decreases in the arterial partial pressure of oxygen (Pao 2) were observed only at the final step of withdrawal. On cessation from 1, 4, and 16 ppm, patients receiving I-NO demonstrated a dose-related reduction in Pao 2 (−11 ± 23, −28 ± 24, and −50 ± 48 mm Hg, respectively). For infants classified as treatment failures (mean duration of therapy = 10 hours), no change in OI occurred for the placebo group (−13 ± 36%, OI of 31 ± 11 after the withdrawal process); however a 42 ± 101% increase in OI to 46 ± 21 occurred for the pooled nitric oxide doses. One death was possibly related to withdrawal of I-NO.
Conclusion. For infants classified as treatment successes, a dose response between the I-NO dose and decrease in Pao 2 after discontinuing I-NO was found. A reduction in I-NO to 1 ppm before discontinuation of the drug seems to minimize the decrease in Pao 2 seen. For infants failing treatment, discontinuation of I-NO could pose a life-threatening reduction in oxygenation should extracorporeal membrane oxygenation not be readily available or I-NO cannot be continued on transport.
- inhaled nitric oxide
- persistent pulmonary hypertension of the newborn
- extracorporeal membrane oxygenation
- neonatal transport
- PPHN =
- persistent pulmonary hypertension of the newborn •
- ECMO =
- extracorporeal membrane oxygenation •
- I-NO =
- inhaled nitric oxide •
- Fio2 =
- fractional inspired oxygenation concentration •
- Pao2 =
- arterial partial pressure of oxygen •
- SD =
- standard deviation •
- OI =
- oxygenation index
There are ∼10 000 term or near-term newborns in the United States per year who develop persistent pulmonary hypertension (PPHN) and/or hypoxemic respiratory failure.1 ,2 Therapy for these patients has undergone a major change within the last 5 years because of the increasing use of inhaled nitric oxide (I-NO), high-frequency ventilation, and surfactant to avoid rescue with extracorporeal membrane oxygenation (ECMO). Although still considered investigational, I-NO, acting as a selective pulmonary vasodilator, has been shown to produce a sustained improvement in oxygenation3 and reduce the need for ECMO.4 ,5Presently, use of I-NO therapy as an adjunct to conventional neonatal respiratory therapy is widespread.6 NO is used at many centers without ECMO facilities, because most newborns with PPHN or hypoxemic respiratory failure are born at non-ECMO centers and develop signs of these disorders shortly after birth.3
In 1990, a hidden mortality associated with neonatal transports to ECMO centers was described; labile oxygenation characteristic of PPHN and late transfers to ECMO centers were implicated.7 Now, with the increasing use of I-NO in all age groups, there have been case reports in older pediatric patients and adults describing serious and life-threatening pulmonary hypertension and hypoxemia on the withdrawal of this therapy.8–10 We hypothesized that with evolving protocols in the treatment of PPHN, withdrawing NO could lead to serious reductions in oxygenation. This hypothesis was studied in the second part of a randomized, double-masked, placebo-controlled, dose-response, multicenter trial which evaluated the efficacy and safety of I-NO for PPHN.3
Enrollment occurred between April 1994 and June 1996. Fifteen of the 25 neonatal intensive care units were ECMO centers. The protocol and consent forms were approved by the local institutional review board before patient enrollment. Written informed consent was obtained for each patient before enrollment. Equipment, treatment gases, and funding based on patient recruitment at each site was provided by Ohmeda, PPD (Liberty Corner, NJ).
The principal hypothesis was that withdrawal of I-NO, from newborns with PPHN who met treatment failure or success criteria, would lead to decreased oxygenation related to step-wise reductions of treatment gas dose, mandated by a weaning protocol. As a corollary hypothesis, the postulated decreases in oxygenation had clinical consequences, with regard to the need for ECMO or death.
Patient Entry Criteria
Patients were term infants (n = 155) with PPHN documented by echocardiography and were treated with an Infant Star conventional ventilator (Infrasonics, Inc, San Diego, CA). Inclusion criteria were fractional inspired oxygen concentration (Fio 2) of 1.0, mean airway pressure ≥10 cmH2O, and a postductal arterial partial pressure of oxygen (Pao 2) of 40 to 100 mm Hg. The major exclusion criteria were previous therapy with surfactant, concomitant high-frequency ventilation, and lung hypoplasia.
Protocol for Weaning Treatment Gas Levels
All patients received treatment gas using a delivery system (Ohmeda, PPD, Madison, WI) designed expressly to deliver either NO mixed with nitrogen or nitrogen alone (placebo), into the inspiratory limb of the ventilator system using a mass flow controller. Electrochemical detectors provided a continuous measurement of NO and NO2 (model EC90 NO monitor and model EC40 NO2monitor, Bedfort Scientific Ltd, Kent, England).3
Treatment gas consisted of either 0, 5, 20, or 80 ppm of I-NO, maintained until treatment success or failure criteria were met. The clinical investigators who managed the neonatal care were masked to treatment gas assignment through the 1-year follow-up. The unmasked laboratory investigator monitored NO, NO2, and methemoglobin levels. The masked clinical investigator was allowed to withdraw the patient from the study if it was judged to be in the best interests of a patient who was deteriorating but had not met specific failure criteria; this accounted for 20% of patients enrolled in the study.3 The data from this group was not included in the withdrawal analyses.
Patients were classified as a treatment failure when the Pao 2 was <40 mm Hg for 30 minutes and the Fio 2 was 0.95 on the conventional ventilator. In addition, patients were classified as treatment failures if they had refractory hypotension defined as a mean systemic arterial pressure of <35 mm Hg, independent of oxygenation. Patients were classified as a treatment success if they had a Pao 2 ≥60 mm Hg, Fio 2 <0.6, and mean airway pressure <10 cm H2O.
Immediately after reaching treatment failure or success criteria based on a blood gas result and ventilator settings, the masked, clinical investigator ordered a 20% reduction in treatment gas. An arterial blood gas with a record of hemodynamic and ventilatory status, was obtained 15 to 30 minutes after the change. No change in mechanical ventilation was permitted. Further 20% reductions could then be made immediately, or within 4 hours. This weaning process was continued until treatment gas was turned off. At each step the same data were collected, with no ventilatory changes immediately before, and 15 to 30 minutes after, the change. Arterial blood gases were obtained pre- and postchange during this 15- to 30-minute interval. Postchange data could be used as prechange data if a rapid weaning process was desired. If the protocol was followed, there would be 10 time points for data collection per patient during the weaning process. The treatment gas could be increased back up by 20% with appropriate increases in Fio 2 if the Pao 2 became <40 mm Hg during a weaning step. The final doses of I-NO were 20% of the starting dose (0, 5, 20, or 80 ppm), that is, 0, 1, 4, and 16 ppm.Figure 1 depicts the different dose reductions of treatment gas which were analyzed for oxygenation during the withdrawal process.
An independent data and monitoring board was composed of pediatric subspecialists and statisticians. A safety analysis, in which the board was masked from treatment gas assignment, was performed using data from the first 100 patients.
Sample Size and Statistics
Changes in oxygenation on withdrawal of treatment gas to 0 ppm were studied separately for treatment successes and failures using the Wilcoxin rank sum test with Bonferonni correction for multiple doses of I-NO. The overall α level was 0.05. Data were not used from a pre- or postchange value in oxygenation if one of the data points was missing. The Cochran-Mantel-Haenszel χ2 test was used for categorical data such as death and ECMO.
A detailed patient profile for this trial has been previously described.3 The birth weight was 3.4 ± 0.5 kg (mean ± SD), 93% of the patients were outborn, and 63% had the diagnosis of meconium aspiration syndrome. Treatment gas was started at 26 ± 18 hours, when the oxygenation index (OI) (Fio 2 × Pa̅w̅ × 100/postductal Pao 2) was 25 ± 10.
Onset and Duration of Withdrawal
Table 1 demonstrates that the OI was similar for the control or the pooled NO groups, both for patients classified as treatment success or treatment failure. Because there were no differences in OI or duration of therapy between NO doses, pooled NO data are shown.
The duration of treatment gas before reaching either treatment success or failure criteria is shown in Table 2. The average time to reach treatment failure criteria was 10 ± 13 hours for the I-NO group and this was not different from the placebo group. Patients classified as a treatment success took 81 ± 59 hours to reach weaning criteria in the I-NO group; and although it seemed to take longer to reach weaning criteria in the placebo group, this was not statistically significant (P = .13).
Acute Changes in Oxygenation
Assessments of oxygenation were obtained on an average of 9.8 of the weaning steps (5 steps with assessments before, and 15 to 30 minutes after a reduction) in patients classified as treatment successes. No changes in oxygenation were observed for reductions in treatment gas made in a step-wise manner down to 20% of the starting treatment level. However, on stopping the treatment gas, a dose-related reduction in Pao 2 was observed (Fig 2). The baseline Pao 2 (at 20% treatment gas) was not different between groups: 105 ± 39, 92 ± 26, 115 ± 41, and 129 ± 75 mm Hg (mean ± SE) for the 0-, 1-, 4-, and 16-ppm groups, respectively. On cessation of the treatment gas no change in Pao 2 for the control gas was observed. When inhaled NO at 1 ppm was stopped, there was a trend toward decreased Pao 2 (−11 ± 23 mm Hg, uncorrectedP = .04, statistically not significant). Statistically significant decreases in oxygenation, −28 ± 24 mm Hg for the 4-ppm dose and −50 ± 48 mm Hg for the 16-ppm dose, occurred. The decrease in Pao 2 was statistically greater for the 16-ppm vs 1-ppm group.
Because patients classified as treatment failures were acutely deteriorating, many infants had rapid weaning of the NO concentration and only 3.1 oxygenation assessments were recorded for these infants. Therefore, data were available from the start of the weaning process to cessation of treatment gas as shown in Fig 3. Because ventilator changes were made from start to finish of the weaning process, the OI was used to examine the effect of NO withdrawal. There was no significant change from baseline OI index for the placebo group after the weaning process, in patients classified as treatment failures. For the pooled NO group there was a 42 ± 101% (Fig 3) increase in OI to 46 ± 21, and this was significantly different from the placebo group (P = .03).
The incidence of death or need for ECMO after withdrawal from NO was examined for treatment failures. There were no statistically significant differences in the placebo and pooled NO group for death or ECMO (Table 3). Of the 4 deaths in the I-NO group, 3 patients expired 7 to 21 days after the cessation of NO and investigators believed there was no relation to the treatment gas.3 However, 1 treatment failure patient in the 20-ppm NO group was started on high-frequency oscillation rescue therapy after I-NO was discontinued. This patient expired because, in part, of a pneumothorax 1½ hours after NO was stopped. The masked investigators determined that the death was possibly related to the treatment gas.
This prospective, randomized, placebo-controlled, double-masked, dose-response trial, is unique among studies of I-NO therapy. The study results are specific to withdrawal of I-NO from patients with PPHN on intermittent mandatory ventilation and not confounded by surfactant and high-frequency oscillatory ventilation.11 The patients in the present study were full-term newborns started on treatment gas at a mean age of 26 hours with a mean OI of 25.3 Their treatment was started early while they were seriously ill but not meeting ECMO oxygenation criteria. Data related to withdrawing treatment gas were analyzed for two prospectively-defined groups, infants experiencing treatment success and those with treatment failure. Accordingly, the results of this study should be useful in the development of practice guidelines for treating PPHN with I-NO as an adjunct to conventional respiratory therapy.
Patients classified as treatment successes were weaned off treatment gas generally after 3 to 4 days and investigators took 12 hours on average, to make the five step-wise, 20% reductions in treatment gas. As designed, the weaning process began when respiratory status appreciably improved (OI <10). There was excellent compliance to weaning guidelines for treatment successes. No significant changes in oxygenation occurred at 30 minutes after any of the 20% reductions from the initial treatment doses of 0, 5, 20, or 80 ppm, except when the final step of discontinuing the gas was reached. The masked, final doses of treatment gas were 0, 1, 4, and 16 ppm. On cessation of the treatment gas, oxygenation decreased for the three groups receiving NO in a dose-related manner. The decrease in oxygenation was statistically and clinically appreciable in the 4- and 16-ppm group but not the 1-ppm group. The decreases in oxygenation on withdrawal of I-NO for treatment successes did not result in any adverse outcomes such as death or need for ECMO.
The data from the present study suggest that I-NO can be safely withdrawn in infants who have improved their oxygenation status and could be classified as treatment successes. Specifically, once the OI is <10, I-NO can be weaned to 1 ppm without severe reductions in oxygenation in most patients. In a retrospective review of PPHN treatment successes,12 inhaled N0 was discontinued at 5 ppm. The investigators observed that by increasing the Fio 2 by 0.4, hypoxemia could be avoided. However, the present study indicates that by gradually reducing inhaled NO from the usual treatment dose of 5 to 20 ppm down to 1 ppm before cessation of therapy, substantial compensatory changes in Fio 2 or conventional ventilation may be unnecessary. There was 11 ± 23 mm Hg (mean ± SD) reduction in Pao 2 within one-half hour of stopping the treatment from 1 ppm of NO. In other words, most patients had a reduction in Pao 2 of <34 mm Hg (mean −1 SD) or a drop in Pao 2 from a mean of 105 mm Hg (prechange) to 71 (postchange), when the Fio 2was of 0.5 to 0.6. This reduction in oxygenation could be explained by an ∼10% increase in right-to-left shunt13 ,14 which could be theoretically prevented by a 20% increase in Fio 2. Cessation of I-NO from higher levels of inhaled NO could lead to a greater hypoxemia that may not be easily correctable by an increase in Fio 2 alone.
Treatment failures occurred in 34% of placebo and 25% of the pooled I-NO group at 10 ± 11 and 10 ± 13 hours, respectively (mean ± SD). At the time of treatment failure, patients were near ECMO criteria, with oxygenation indices of ∼40. Then, investigators took usually <12 hours to wean off the treatment gas, with a trend toward faster weaning for placebo. In contrast to patient successes, there was only fair compliance to completing the withdrawal protocol for patients declared as treatment failures. At this critical stage of PPHN, in which protocol considerations were outweighed by patient interests and preparing rescue management, less than half of the requested data were captured. Accordingly, only a single comparison was made between patient data for placebo and the pooled NO groups.
For treatment failure patients, the OI decreased 13% for the placebo (not significant), but increased 42% (P = .03) for the pooled NO group. There was no statistical difference in death or ECMO for treatment failures in the placebo versus the pooled NO group. Because a majority of patients were treated at ECMO centers it was unlikely that an increase in deaths would have occurred. Initially investigators reported no serious adverse events related to NO. For treatment failures there were 4 deaths in the pooled NO group and no deaths in the placebo group. This was not a statistically significant increase because patients were initially randomized to NO versus placebo in a 3-to-1 ratio. One patient had severe meconium aspiration syndrome and was declared a treatment failure because of hypoxemia after 56 hours on treatment gas. On cessation of the treatment gas, worsening hypoxemia led to the need for high-frequency oscillation and positive pressure ventilation by hand. A pneumothorax occurred and was treated, but the infant died 1.5 hours after stopping NO. This patient received I-NO at 20 ppm, decreased in 20% decrements throughout 13 minutes.
The present study did not address the causes for the decrease in oxygenation on withdrawal of I-NO. The dose-related reduction in Pao 2 on cessation could have been simply related to significant shunting or ventilation-perfusion mismatch that was still present when NO was withdrawn. For the treatment failure group, the reduced oxygenation on withdrawal could have been related to a gradual, unrecognizable, positive effect of I-NO on patients who eventually failed beyond several hours after the start of therapy.15 Other explanations for a decrease in oxygenation include a possible down-regulation of constitutive, endothelial, NO synthase activity.16 Unopposed vasoconstrictor action17 on the pulmonary circulation, at the time of withdrawal, is another possibility if the underlying disease is not completely resolved.
In summary, our study has shown withdrawal of I-NO is not problematic for treatment successes if the OI is <10 and the I-NO dose is gradually reduced to 1 ppm before cessation. If this method of weaning I-NO is followed, a 20% increase in Fio 2 at cessation of I-NO should prevent transient hypoxemia. As previously reported,3 25% of patients with PPHN will go on to become classified as treatment failures using inhaled NO as an adjunct to conventional ventilation. These patients may have rapid and life-threatening reductions in oxygenation (OI >40) when I-NO is discontinued. Therefore, we recommend that patients with continued deterioration after starting I-NO: 1) be transferred to ECMO centers on I-NO18 if the OI becomes sustained >25 on conventional ventilation; and 2) if further rescue therapy fails, I-NO should be continued until the patient is on circulatory bypass support. These guidelines should help avoid setbacks in oxygenation after successful treatment and the potential of peritransport mortality for patients in need of ECMO support in the era of NO as an adjunctive therapy to conventional or high-frequency ventilation.
This study was sponsored by Ohmeda PPD.
The scientific and technical support of Anne Berssenbrugge, Phd, and Greg Beltrand from Ohmeda Medical Systems Division and Michael Karol, RRT, from the Long Island Jewish Medical Center is gratefully acknowledged. The investigators are indebted to the neonatologists, nursing staffs, and respiratory therapists at their respective sites. The assistance of Irene Barling, Barbara Wilkens, RN, and Diane Davidson, RN, at the coordinating study center (Schneider Children's Hospital) is greatly appreciated.
This clinical trial was a multicenter collaboration sponsored by Ohmeda, PPD. The Steering Committee was composed of the co-authors of this article. In addition, the following institutions and investigators participated in this trial: the University of Virginia School of Medicine, Charlottesville, VA—M. Pamela Griffin, MD; and Martina Compton, RRT; the Children's Hospital of Orange County, CA—Robert Langdon Hillyard, MD; Elizabeth Phyllis Drake, RN, PNP; and Albert Rocky Stone, RRT; the Children's Hospital and Health Center, San Diego, CA—Marva L. Evans, MD; and Gail R. Knight, MD; the Schneider Children's Hospital, New Hyde Park, NY—Andrew M. Steele, MD; Robert Koppel, MD; and Angela Romano, MD; the University of Kentucky Children's Hospital, Lexington, KY—Thomas H. Pauly, MD; John R. Walker, DO; and Vicki Whitehead, RN; the University of Alabama at Birmingham—Virginia A. Karle, MD; and Monica V. Collins, RN; the St. Joseph's Hospital and Medical Center, Phoenix, AZ—Mark L. Shwer, MD; and Donna C. Hamburg, MD; the Boston Perinatal Center, Floating Hospital for Children at New England Medical Center, Boston, MA—Barbara A. Shephard, MD; John M. Fiascone, MD; and Ivan D. Frantz, MD; the Legacy Emanuel Children's Hospital, Portland, OR—John V. McDonald, MD; the Medical College of Georgia Hospital and Clinics, Augusta, GA—D. Spencer Brudno, MD; and William Grayson, RRT; the Children's Hospital of Oklahoma, University of Oklahoma, Oklahoma City, OK—K. C. Sekar, MD; Mary Anne McCaffree, MD; and Mike McCoy, ARNP; the Pennsylvania Hospital, Philadelphia, PA—Vinod K. Bhutani, MD; Emidio M. Sivieri, and Mary Kay Grous; the Columbia Wesley Medical Center, Wichita, KS—Barry T. Bloom, MD; Pamela S. Keyes, RN, ARNP; and Tom L. Rose, RRT; the Crouse Irving Memorial Hospital, Syracuse, NY—Ellen M. Bifano, MD; the Children's Hospital of the King's Daughters/Eastern Virginia Medical School, Norfolk, VA—Jamil H. Khan, MD; and Marilyn M. Reininger, RN; the Bowman Gray School of Medicine, Winston-Salem, NC—Steven M. Block, MD; the Children's Hospital of New Jersey, Newark, NJ—Morris Cohen, MD; and David Brown, MD; the University of California, Davis Medical Center, Davis, CA—T. Allen Merritt, MD; and Boyd Goetzman, MD; the Ochsner Foundation Hospital, New Orleans, LA—Erick M. Fajardo, MD; Janet E. Larsen, MD; and Marie C. McGettigan, MD; the East Carolina University School of Medicine, Pitt County Memorial Hospital, Greenville, NC—Steven C. Covington, RRT; John E. Wimmer, MD; and Arthur E. Kopelman, MD; the Columbia-Presbyterian/St.Luke's Medical Center, Denver, CO—Barbara J. Quissell, MD; and Delphine M. Eichorst, MD; the Hope Children's Hospital, Christ Hospital and Medical Center, Oak Lawn, IL—Arvind Shukla, MD; Manohar Rathi, MD; and Alison Miklos, RNC; the Duke University Medical Center, Durham, NC–Richard L. Auten, MD; and Kathy J. Auten; the University of South Dakota School of Medicine, Sioux Falls, SD—Dennis C. Stevens, MD; David P. Munson, MD; and Rachel D. Klinghagen, RN, CNP; the Santa Rosa Children's Hospital, San Antonio, TX—Anthony Corbet, MD; Sue Pape, RN; and William Rapp, RRT. The Data Safety and Monitoring Board included: Statistics Collaborative, Inc, Washington, DC—Janet Wittes, PhD (Chair); the Medical College of Virginia, Richmond, VA—Stephen Ayres, MD; the Boston Floating Hospital, Boston, MA—Jonathan Rhodes, MD; the St. Vincent's Hospital, New York, NY—Jayne Rivas, MD; the Yale School of Medicine, New Haven, CT—Stanley Rosenblum, MD; and the Children's National Medical Center, Washington, DC—Billie Short, MD.
- Received September 30, 1998.
- Accepted January 26, 1999.
Reprint requests to (D.D.) Division of Neonatal-Perinatal Medicine, Schneider Children's Hospital, Long Island Jewish Medical Center, New Hyde Park, NY 11040. E-mail:
Members of the I-NO/PPHN Study Group are listed in the “Acknowledgments” section.
- Linde-Zwirble WT,
- Angus DC,
- Clermont G,
- et al.
- Davidson D,
- Barefield ES,
- Kattwinkel J,
- et al.
- Stark AR,
- Davidson D
- ↵Goldsmith JP, Karotkin EH. Introduction to assisted ventilation. In: Goldsmith JP, Karotkin EH, eds. Assisted Ventilation of the Neonate. 3rd ed. Philadelphia, PA: WB Saunders Inc; 1996:14
- Boga GM,
- Griscavage JM,
- Rogers NE,
- Ignarro LJ
- Kinsella JP,
- Schmnidt JM,
- Griebel J,
- Abman SH
- Copyright © 1999 American Academy of Pediatrics