Aerosolized PGE₁: A Selective Pulmonary Vasodilator in Neonatal Hypoxemic Respiratory Failure Results of a Phase I/II Open Label Clinical Trial

Abstract

Objective

Phase I/II feasibility, safety and dose escalation study of inhaled PGE₁ (IPGE₁) in neonatal hypoxemic respiratory failure.

Design/Methods

Twenty term/near term neonates with hypoxemic respiratory failure and oxygenation index ≥ 20 were enrolled. IPGE₁ was delivered by a jet nebulizer in incremental doses (25, 50, 150, 300 ng/kg/min for 30 min each) for 2 hours followed by weaning over an hour (15 min for each dose). IPGE₁ was given to 13 patients before receiving INO (Group I) and to 7 patients who failed to respond to INO (Group II). Response was defined as an increase in P_aO₂ of ≥ 25 (full) or 10–25 (partial) torr. Exit criteria included an acute deterioration in oxygenation status in either group, a persistent OI above 35 in Group I and the availability of ECMO in Group II.

Results

The mean (SD) increase in P_aO₂ at the end of IPGE₁ administration was 63 (62.3) in Group I (p=0.024) and 40 (62.1) in Group II (p>0.05). In Group I, 8 of 13 neonates had a full response, but 4 deteriorated following discontinuation of IPGE₁. Two responded to INO and 2 were placed on ECMO. Five patients deteriorated before or during IPGE₁ and none responded to INO. In Group II, 3 of 7 patients had a full response to IPGE₁. The patient with a partial response and the patients exiting before or during IPGE₁ administration were placed on ECMO.

Conclusions

IPGE₁ improved oxygenation in neonatal hypoxemic respiratory failure when given for 3 hours with no short term adverse effects. Thus, IPGE₁, at a dose of 25 to 300 ng/kg/min appears to be a safe selective pulmonary vasodilator. Efficacy needs to be studied in a large multicenter randomized controlled trial.

INTRODUCTION

Hypoxemic respiratory failure in the newborn is usually associated with potentially reversible pulmonary hypertension that causes right-to-left shunting and profound hypoxemia. The goal of therapy is to selectively lower the pulmonary vascular resistance (PVR). Intravenously administered vasodilators lack pulmonary selectivity leading to systemic side effects. Inhaled nitric oxide (INO), a selective pulmonary vasodilator, has revolutionized the treatment of hypoxemic respiratory failure. However, there is lack of sustained improvement in 30–46% of infants (1–4); moreover, INO is a highly toxic molecule requiring expensive monitoring and scavenging systems for administration, making the treatment expensive and limiting availability. Aerosolized prostaglandins I₂ and E₁ have been reported to be effective selective pulmonary vasodilators in animals and human adults (5–19). In addition, inhaled PGI₂ (IPGI₂) has also been reported to be effective in preterm and term newborns and children with pulmonary hypertension (20–26). Although intravenous PGE₁ is widely used in neonates, the use of the inhaled form has not been reported in newborns with pulmonary hypertension. The high pulmonary clearance of PGE₁ (70 to 90%) contributes to its selectivity as a pulmonary vasodilator when administered as an aerosol (17). Compared to PGI₂, PGE₁ has a shorter half-life, lower pKa (6.3 versus 10.5), bronchodilator action, anti-proliferative and anti-inflammatory effects on the alveolar, interstitial and vascular spaces of the lung (17, 27–31). Prostaglandin nebulization can be used without the sophisticated technical equipment needed for controlled NO inhalation and hence is less expensive. It has no known toxic metabolites or toxic effects. Prostaglandins and nitric oxide relax the vascular smooth muscles through two different second-messenger systems; therefore, in combination, INO and IPGE₁ may have synergistic effect (32).

The existing literature suggests that inhaled PGE₁ is a potential effective vasodilator in the treatment of pulmonary hypertension of the newborn. We report the results of a phase I–II open-label pilot study of escalating doses of aerosolized PGE₁ in term/near-term neonates with hypoxemic respiratory failure. Our objectives were to establish the feasibility and safety of PGE₁ administered as an aerosol and to determine the effective dose.

METHODS

Subjects

The study was conducted at five centers in the metropolitan Detroit area. The IRB at all 5 hospitals approved the study. All of the patients in this report were recruited at two of the five centers (Children’s Hospital of Michigan and Hutzel Women’s Hospital). INO was available only during transport to and at Children’s Hospital of Michigan. Newborn infants born at ≥ 34 week’s gestation requiring assisted ventilation for hypoxemic respiratory failure in the first two weeks of life were eligible to participate after an oxygenation index (OI) of ≥ 20 on two arterial blood gases at least 15 minutes apart in the preceding 12 hours. Additional requirements were an indwelling arterial catheter and informed parental consent. Although an attempt was made to obtain a head ultrasound and cardiac echocardiogram prior to enrollment, this was not a requirement for study participation. Neonates with congenital diaphragmatic hernia, congenital heart disease other than ductal or septal shunts, thrombocytopenia, or those in whom a decision to not provide full treatment (including ECMO) were considered ineligible for the study.

Two groups of patients were defined based on disease severity and prior treatment with INO at the time of enrollment. Patients in Group I (Pre-INO Group) were enrolled before they received INO. Group II patients (Post-INO Group) were enrolled after they were found to be refractory to INO. A patient was labeled as being ‘refractory to INO’ if there was no response to INO one hour after initiation or if there was a failure to sustain a response in P_aO₂ ≥ 25 mmHg above baseline at any time without weaning of INO or ventilator. Many of the patients in Group II were eligible and were waiting for the availability of ECMO.

Management of eligible infants was optimized prior to treatment with aerosolized PGE₁ by the clinical team. This included management decisions about the use of conventional or high frequency oscillatory ventilation, induction of alkalosis, administration of volume, surfactant, pressors, sedation and paralysis. Surfactant therapy was initiated prior to the initiation of IPGE₁. The mode of ventilation remained unchanged for the duration of the study.

Drug Dosing and Administration of aerosolized PGE₁

PGE₁ solution for aerosolization was prepared from 500 μg synthetic PGE₁ dissolved in 1 ml ethanol (Gensia Sicor Pharmaceuticals, Irvine, California) by dilution in 0.9% saline. Once diluted, the solution was used within 24 hrs. Four different doses of aerosolized PGE₁ (25, 50, 150, and 300 ng/kg/min) were prepared. These trial doses were selected based on the recommended intravenous dose of 50 to 100 ng/kg/min (0.05 to 0.1 μg/kg/min) for ductal patency in neonates with congenital cardiac lesions. The dose was varied by a factor of three from the recommended intravenous dose of 100 ng/kg/min. The PGE₁ concentration in the solution was varied to keep the nebulized volume constant at 2.2 to 2.6 cc/hour. Inhalation was begun at the lowest dose and each dose was administered for 30 min in all infants. Once the maximal dose was achieved (300 ng/kg/min), IPGE₁ was weaned in 15 minute steps (weaning phase). The entire study lasted for 3 hours unless the infant met exit criteria prior to completion.

The PGE₁ dosage refers to the total amount of nebulized drug placed in the nebulizer. It has previously been shown that the fraction deposited in the alveolar space during mechanical ventilation is less than 10 to 20% of the dose administered (10). The alveolar dose is even smaller in newborn infants as there is proportionally larger dead space (22).

Aerosols of PGE₁ with a mean particle size of 2 to 3 μm were generated with a jet nebulizer (miniHeart, Westmed Inc., Lakewood, Colorado). The nebulizer was connected to the inspiratory limb of the ventilator (Fig 1). During the inhalation period, the ventilator flow was adapted according to the additional flow of the nebulizer to maintain alveolar ventilation and inspired oxygen concentration. Baseline conditions were assessed by aerosolization of normal saline (2 cc/hour) for 15 minutes prior to the first dose of aerosolized PGE₁ (25 ng/kg/min).

Fig 1. The nebulizer circuit.

Aerosols of PGE₁ were generated with a jet nebulizer (miniHeart, Westmed Inc) connected to the inspiratory limb of the ventilator. Different doses of IPGE₁ for nebulization were delivered by syringe pumps through a stopcock to the nebulizer chamber.

Monitoring During Administration of Aerosolized PGE₁

Hemodynamic parameters and gas exchange (arterial blood gas) were assessed prior to PGE₁ administration, after each dose change and 60 min after withdrawal of the final dose. Blood counts, serum electrolytes and liver enzymes were obtained before and 24 hrs after concluding the study. The infants were monitored clinically for hyperthermia, hypotension, dysrhythmias, signs of bleeding, and seizures.

Outcome Variables

The primary outcome variable was the change in P_aO₂ from baseline after 30 minutes exposure to aerosolized PGE₁ and at end of IPGE₁ administration. To determine this, an arterial blood gas analysis was performed after 30 min of exposure to each dose of IPGE₁. A response was defined as an increase in P_aO₂ ≥ 25 (full), 10–25 (partial) or <10 (none) mmHg. Secondary outcome variables included change in oxygenation index, need for nitric oxide and/or ECMO, mortality, and duration of mechanical ventilation, oxygen therapy, and hospitalization.

Exit Criteria

Infants in both groups exited the study if there was an acute drop in pulse oximeter saturation of > 10% for which no mechanical cause could be identified.. Additionally, infants in Group I exited the study if the OI exceeded 35 for more than 30 min during the study in the absence of mechanical cause for the worsening oxygenation. Infants in Group II awaiting ECMO, exited when ECMO was available.

Statistical Analyses

We estimated a sample size of 17 to determine minimum difference in P_aO₂ before and after IPGE₁ of 25 mmHg, a SD of 30 mmHg, assuming a power of 90, and alpha of 0.05 (2-tailed). The sample was increased to 20 to account for infants exiting the study.

Statistical analysis was done using an intention-to-treat analysis. We evaluated categorical variables using chi-square tests. Continuous variables were compared with paired or independent samples t tests and ANOVA. Longitudinal data was analyzed using the mixed procedure in SAS. Significance level was set at 0.05.

RESULTS

Baseline Characteristics

Twenty infants were enrolled in the trial; 13 in Group I and 7 in Group II. The baseline maternal, perinatal and postnatal characteristics are summarized (Table 1). Air leaks were present in 3 (15%), pulmonary hemorrhage in 2 (10%), and seizures in 2 infants (10%) before enrollment.

Table 1.

Baseline Characteristics:

	Group I (n=13)	Group II (n=7)
Maternal:
Age (years)	26.2±7.6	27.6±6.5
Positive GBS* status- no. (%)	5 (38.5)	2 (28.6)
Fetal distress present- no. (%)	5 (41.7)	1 (20)
Meconium stained amniotic fluid- no. (%)	4 (30.8)	4 (57.1)
Cesarean Delivery- no. (%)	6 (46.2)	3 (42.9)
Baby:
Birth weight (grams)	3005±588	3280±582
Gestational age (weeks)	37.2±2.2	38.3±2.8
Male sex - no. (%)	7 (53.8)	5 (71.4)
Race - no. (%)
Black	8 (61.5)	6 (85.7)
White	3 (23.1)	1 (14.3)
Out born - no. (%)	8 (61.5)	7 (100)
1-Minute Apgar score <3 - no. (%)	3 (23.1)	1 (16.7)
5-Minute Apgar score <3 - no. (%)	0 (0)	0 (0)
Delivery Room Resuscitation- no. (%)	10 (76.9)	4 (57.1)

Plus-minus values are mean ± SD

^*GBS Group B streptococcus

^†Data for Fetal Distress are based on 17 infants, 12 in Group I, 5 in Group II

^‡Data for 1-Minute Apgar is based on 19 infants, 13 in group I and 6 in group II

Cardio-respiratory variables at enrollment are summarized (Table 2). All infants in Group I had at least one OI ≥ 25, 10 had at least 2 such OI’s. Although all infants had an echocardiogram at some point in the acute course of their illness, eighty five percent (n=17) had an echocardiogram prior to enrollment; of these, 82.4% had evidence of pulmonary hypertension (right-to-left or bidirectional shunting, systemic or supra-systemic right ventricular pressure or a diagnosis by the cardiologist of pulmonary hypertension).

Table 2.

Cardio-respiratory Variables at Enrollment:

	Group I (n=13)	Group II (n=7)
Primary diagnosis - no. (%)
Aspiration Syndrome	5 (38.5)	4 (57.1)
RDS	4 (30.8)	1 (14.3)
Idiopathic PPHN	2 (15.4)	1 (14.3)
Suspect pulmonary hypoplasia	2 (15.4)	0 (0)
Pneumonia/Sepsis	0 (0)	1 (14.3)
Systolic blood pressure (mmHg)	66.5±12.9	80.4±13.6
First qualifying arterial-blood gas value
Oxygenation index (OI)	29.8±4.9	44.1±5.1§
Mean airway pressure (cm of water)	15.4±4.4	19.3±3.2*
FiO2 (mmHg)	1.0±0.0	1.0±0.0
PaO2 (mmHg)	51.9±10.4	44.1±9.7
Alveolar-arterial oxygen gradient (mmHg)	618.1±19.8	624.6±21.2
Age at admission - hrs (Median)	9.7	9.8
Age at enrollment - hrs (Median)	24.2	12.7
Median time, enrollment to IPGE1 initiation (min)‡	19	25

Plus-minus values are mean±SD.

FiO₂ denotes fraction of inspired oxygen, and P_aO₂ partial pressure of arterial oxygen

^*p<0.1

^**p<0.05

^§p<0.001

Conventional therapy prior to enrollment included volume resuscitation (100%), sedation/analgesia (100%), vasopressors (95%), steroids for hypotension (40%), surfactant (60%), neuromuscular blockade (45%), and intravenous PGE₁ (5%). Twenty-five percent of the infants had respiratory alkalosis (defined as pH>7.45 at enrollment). Ninety percent of the infants had a trial of high frequency oscillatory ventilation prior to enrollment; however only 55% were on an oscillatory ventilator at the time of enrollment.

Primary Outcome: Change in P_aO₂ from Baseline (Fig 2, Table 3)

Fig 2.

Flow Diagram of Enrolled Patients.

Table 3.

Description of Patients;

PULMONARY DIAGNOSIS	RESPONSE To IPGE1	IPGE1 – No. of doses*	MAXIMAL SUPPORT§	COMMENTS	CRANIAL IMAGING BEFORE DISCHARGE /DEATH	AGE AT DISCHARGE /DEATH (DAYS)	SURVIVAL STATUS**
GROUP I
IDIOPATHIC	FULL	COMPLETE	HFOV	---		16.2	S
RDS	FULL	COMPLETE	CMV	---		16.4	S
RDS	FULL	COMPLETE	HFOV	---		10	S
RDS	FULL	COMPLETE	CMV	Intractable hypoglycemia, seizures	NORMAL	37.2	S
IDIOPATHIC	FULL	COMPLETE	NO	Down's syndrome		17.6	S
ASPIRATION	FULL	COMPLETE	NO	---		13.8	S
ASPIRATION	FULL	COMPLETE	ECMO	---	VM, EAF	34.6	S
ASPIRATION	FULL	COMPLETE	ECMO	---	VM, EAF	31.5	S
HYPOPLASIA	EXIT	1 DOSE	ECMO	Hydrops	PVL, EAF	31.9	D (Sepsis, pulmonary hypoplasia)
ASPIRATION	EXIT	1 DOSE	ECMO	---	EAF	22.2	S
ASPIRATION	EXIT	2 DOSES	ECMO	Deterioration from mechanical cause	EAF	26	S
HYPOPLASIA	EXIT	NO PGE1	NO	Deteriorated during placement of urinary catheter, IPCKD, not candidate for ECMO		2.7	D (IPCKD, Pulmonary hypoplasia)
RDS	EXIT	NO PGE1	ECMO	Rapidly progressive course, died during cannulation for ECMO 10 hours later		5.1	D (Pseudomonas sepsis, HMD)
GROUP II
IDIOPATHIC	FULL	COMPLETE	NO	Skeletal dysplasia, reservoir placed for HC	HC, EAF	115.2	D (Pulmonary hypoplasia)
RDS	FULL	COMPLETE	NO	PaO2 improved after enrollment but before IPGE1		19.1	S
ASPIRATION	FULL	COMPLETE	NO	---	VM, EAF	89.6	S
PNEUMONIA	PARTIAL	COMPLETE	ECMO	---	VM, EAF	54.2	S
ASPIRATION	EXIT	4 DOSES	ECMO	On intravenous PGE1 prior to study Study stopped as ECMO team ready	INFARCT, PVL, VM, EAF	29.3	S
ASPIRATION	EXIT	3 DOSES	ECMO	Study stopped as ECMO team ready	VM	30.6	S
ASPIRATION	EXIT	NO PGE1	ECMO	PaO2 deteriorated after enrollment but before starting IPGE1	VM, EAF	22.7	S

GA=gestational age, VM=stable ventriculomegaly, EAF=extra-axial fluid. PVL=periventriclar leukomalacia, HC=hydrocephalus, IPCKD=Infantile polycystic kidney disease

^*The study design included 8 doses of IPGE₁ – 4 in the escalation phase and 4 in the weaning phase. COMPLETE refers to a patient completing both the escalation and weaning phase of the study. NO indicates a patient who exited the study prior to receiving any IPGE₁. For babies exiting during the study, the number of doses administered is indicated.

^**S=survived, D=died; significant autopsy findings given in parentheses for infants who died

^§Babies with hypoxemicc respiratory failure receive CMV initially, followed by trial of HFOV, INO and ECMO, generally in that order.

Fig 2 is the flow diagram of enrolled subjects whose clinical descriptions are shown in Table 3.

Group I

Two infants in Group I met exit criteria prior to receiving IPGE₁. Three infants met exit criteria during the study. Eight infants completed the study and had a full response to IPGE₁. At the end of the study, there was a significant increase in P_aO₂ (63.0±62.3, p=0.024), decrease in OI (13.0±8.8, p=0.004) and decrease in alveolar-arterial oxygen gradient (58.3±65.2, p=0.039).

Group II

All infants in this group had demonstrated a lack of response to INO. Five received INO during transport from birth hospital to the referral center. Duration of INO therapy prior to enrollment was 6.1 hours (median) (range 1.5 to 91 hours). One infant exited the study without receiving IPGE₁ and two infants exited during the study as ECMO was available. Four infants in Group II completed the IPGE₁ study. One infant improved prior to receiving IPGE₁. This infant demonstrated an additional full response to IPGE₁. The mean±SD increase in P_aO₂ (39.8±62.1), decrease in OI (7.6±13.4) and alveolar-arterial oxygen gradient (39.3±66.6), was not statistically significant compared to baseline (p>0.05 for all) in these 4 infants.

Dose Response Effect (Fig 3)

Fig 3.

The x-axis represents the dose of IPGE₁ in ng/kg/min. The data points represent oxygenation at the end of each dose. ‘A’ and ‘B’ represent oxygenation in two arterial blood gases immediately prior to enrollment. ‘C’ represents oxygenation at the end of nebulized normal saline. ‘D’ represents oxygenation 1 hour after discontinuing IPGE₁. The hatched vertical line demarcates the escalation and weaning phases. The individual lines represent the subject-specific P_aO₂ profiles over time. In Group I, there is significant increase in P_aO₂ with increasing IPGE₁ dose in the escalation phase; the response is sustained in the weaning phase in the group as a whole. In Group II, there is no association between the IPGE₁ dose and P_aO₂ response in either the escalation or weaning phase. The between subjects variability is more marked in Group II compared to Group I

Among the 8 subjects completing the study in Group I, a full response was observed in 50% after a dose of 50 ng/kg/min, and in 87.5% after a dose of 150 ng/kg/min. One patient (12.5%) had a partial response in the escalation phase, but showed a complete response during the weaning phase (Figure 3). At the end of the weaning phase, 5 infants sustained a full response.

In the four patients in Group II completing the IPGE₁ study, a full response was observed in 25% after a dose of 50 ng/kg/min, in 50% after a dose of 150 ng/kg/min, and in 75% after a dose of 300 ng/kg/min (Figure 3). At the end of the weaning phase, the 3 infants with a full response sustained it.

Analysis using Mixed Models (SAS) (Fig 3) revealed a significant association of IPGE₁ dose and P_aO₂ during the escalation phase in Group I (β =11.4, SE=3.5, p=0.009) but not during weaning. In Group II, no association was observed either in the escalation phase or during weaning. A more complex model adjusting for time on IPGE₁ and IPGE₁ dose revealed a significant effect of both dose (β =7.1, SE=2.4, p=0.02) and time (β =6.2, SE=1.1, p<0.0001) for Group I and only a significant effect of time (β =2.7, SE=1.3, p=0.0461) but not dose (β =6.5, SE=19.8, p=0.8) for Group II.

Need for INO or ECMO

Nine of the 13 infants in Group I (69.2%) received INO and 6 (46.2%) eventually were placed on ECMO. All infants in Group I who did not respond to IPGE₁ (n=5) also failed to respond to INO and all qualified for ECMO. Fifty percent (n=4) of the infants with full response to IPGE₁ in Group I continued to improve without need for INO or ECMO. The median time from the discontinuation of IPGE₁ to initiation of INO in the 4 patients who did not sustain a response was 8.2 hours (range 0.3 to 30 hours). Of these, 2 did not respond to INO and were placed on ECMO 13.5 and 75 hours after conclusion of the IPGE₁ study. Four of the 7 infants in Group II (57.1%) were placed on ECMO.

Mortality

Three infants in Group I (23.1%) and 1 infant in Group II (14.3%) died before hospital discharge or 120 days of life (Table 4). Causes of death included fulminant sepsis (n=1) and pulmonary hypoplasia (n=3).

Table 4.

Adverse Effects:

	Before IPGE1 Mean (SD)	After IPGE1 Mean (SD)	p
Alanine aminotransferase*	26.3 (10.2)	25.3 (14.1)	NS
Aspartate aminotransferase*	74.7 (65.6)	47.2 (29.4)	0.04
Alkaline phosphatase*	193.4 (272.5)	93.6 (54.9)	NS
Bilirubin**	4.6 (2.9)	5.3 (4.3)	NS
Blood urea nitrogen**	7 (3.3)	9.9 (5.5)	0.011
Creatinine**	0.7 (0.18)	0.7 (0.23)	NS
Hematocrit (%)	46.8 (6.4)	40.1 (6.6)	0.001
Platelets†	219.3 (116.7)	187 (77.5)	NS
Total leukocyte count†	16.2 (6.8)	12.6 (6.4)	0.039

^*U/L

^**(mg/dL)

^†10³/mm³

Results of Cranial Imaging Studies

A cranial ultrasound was obtained in all patients during the acute phase of the illness. It was obtained prior to enrollment in 9 of 13 (69.2%) infants in Group I and 6/7 (85.7%) infants in Group II. Abnormal findings included cerebral edema, ventriculomegaly, Grade I IVH and periventricular echogenicity. Abnormal findings in the initial sonogram obtained within 24 hours after the IPGE₁ study (n=5) included Grade I IVH. No new findings or progression was observed in patients who had a sonogram both prior to enrollment and after the IPGE₁ study in the acute phase. Cranial imaging findings prior to discharge or death included hydrocephalus (n=1), stable ventriculomegaly (n=7), extra-axial fluid (n=10), infarction (n=1) and periventricular leukomalacia (n=2).

Toxicity/Adverse Effects

Inhaled PGE₁ was not discontinued in any infant because of adverse effects. No episodes of exacerbation of hypotension, hyperthermia, dysrhythmias, or bleeding tendency were observed. Paired t tests did not reveal significant differences in temperature, systolic blood pressure or heart rate at baseline and end of IPGE₁ administration in either group. Results of blood counts and serum chemistries obtained before and 24 hours after study conclusion are presented in Table 4. The hematocrit, total leukocyte count and aspartate aminotransferase were lower after IPGE₁ whereas blood urea nitrogen was higher; however all values were within normal limits and the differences were not clinically significant. Seizures and ventriculomegaly were documented in two infants prior to enrollment in study. No new onset seizures were documented in any of the patients within 1 week after enrollment. Mild ventriculomegaly was documented in babies who had undergone ECMO and is probably not attributable to the use of IPGE₁ in these infants.

DISCUSSION

The results of this phase I–II study suggest that IPGE₁ is a safe selective pulmonary vasodilator in hypoxemic respiratory failure at the concentrations and durations used in the trial. No evidence of tolerance, rebound pulmonary hypertension between doses and after cessation of IPGE₁ or systemic side effects was detected.

Although a few studies have reported on the use of aerosolized PGI₂ in newborns with hypoxemic respiratory failure, there are no reports on the use of IPGE₁ in this population. De Jaegere et al (20) reported improved oxygenation without adverse systemic effects in 4 preterm infants following endotracheal instillation of PGI₂. Similar benefits were reported by Soditt et al in a preterm newborn with pulmonary hypertension (21). Aerosolized PGI₂ was associated with improvement in oxygenation in 2 term infants with persistent pulmonary hypertension and one infant with congenital heart disease (22, 24). Sustained improvement in oxygenation was also reported by Kelly et al in 3 of 4 term infants who had failed to respond to INO following administration of iPGI₂ and milrinone (23). Milrinone has been shown to amplify the pulmonary vasodilatory response to inhaled PGI₂ (33).

A clinically significant improvement in oxygenation was observed in the patients in Group I when IPGE₁ was given for 3 hours. This improvement was sustained during the weaning phase in this group. The magnitude of the improvement in oxygenation in this group is comparable to that reported previously for INO (1, 2, 34). In Group II, at least one infant had already received intravenous PGE₁ at the referring hospital. This may have interfered with the selective pulmonary action of IPGE₁. Three infants in this group did show a full response to IPGE₁ and never qualified for ECMO. This provides credence to the speculation that INO and IPGE₁ may have additive effects because of activation of different cellular mechanisms (32). The improvement in P_aO₂ was observed earlier and at lower IPGE₁ dose in Group I patients compared to Group II patients. Previous studies have referred to the benefits of earlier institution of therapy for hypoxemic respiratory failure (1, 35). The improvement in P_aO₂ was predicted by both dose and time in study for Group I but only by time in study for Group II. It is possible that a greater number of patients might have responded and sustained the response had the IPGE₁ been administered for longer.

This study was designed to find a safe and effective dose of inhaled PGE₁ and found that a dose of 150 to 300 ng /kg/min may be used safely and effectively in neonates with hypoxemic respiratory failure. The sample size is relatively small and heterogeneous without a placebo control group, not atypical for clinical phase 1 and phase 2 drug trials and can only detect large differences. The drug was administered for a short duration (3 hours) and the efficacy and safety of a longer drug inhalation remains to be evaluated. This study describes for the first time the use of inhaled PGE₁ in neonates with hypoxemic respiratory failure and suggests further placebo controlled randomized studies to establish efficacy and safety of this drug in this neonatal population.