Main

A PDA resulting in hemodynamically significant left to right shunting of blood is a common problem in sick preterm infants, increasing the risk for intraventricular hemorrhage, necrotizing enterocolitis, bronchopulmonary dysplasia, and death (1). The initial management conventionally includes treatment with a course of intravenous indomethacin and leads to PDA closure in 79% of infants (2). Indomethacin is also administered to preterm infants shortly after delivery as prophylaxis against the development of a hemodynamically significant PDA (3, 4) and for the prevention of intraventricular hemorrhage (5, 6).

However, there are concerns regarding the safety of indomethacin, particularly on the CNS. The drug causes a marked decline in CBF, COD, CBV, and CBVR and may also reduce cerebral intracellular oxygenation (7–9). Antenatal exposure to indomethacin for the treatment of preterm labor may be associated with cystic brain lesions (10) and increased incidence of intraventricular hemorrhage in the infant (11). Infants treated with indomethacin have a higher incidence of retinopathy of prematurity (12). These results in infants, together with data from animal studies showing an impairment of the normal compensatory effects that protect cerebral perfusion during hypoxia and hypotension (13, 14), suggest that indomethacin may lead to ischemic brain injury.

Significant improvements in long-term morbidity and mortality that could be expected from successful closure of PDA have also not been consistently demonstrated after treatment with indomethacin (15); this may be due to the counteracting effects of the drug's toxicity, which outside the CNS include necrotizing enterocolitis, gastrointestinal hemorrhage, and transient or permanent alterations in renal function (2, 16, 17). Reductions in superior mesenteric (18, 19) and renal arterial (20) blood flow velocities after indomethacin administration have been shown in preterm infants and may contribute to its toxic effects by producing ischemia.

Indomethacin and other NSAID inhibit the two isoforms of the enzyme cyclooxygenase to varying degrees, leading to reduced synthesis of prostaglandins (21). The mechanism by which indomethacin exerts its cerebral hemodynamic effects is not known, but, unlike its effects on the PDA, this may not be related to inhibition of prostaglandin synthesis. In the newborn piglet, the effects of indomethacin are not identical to those of other NSAID: CBF decreased after indomethacin but was unchanged after ibuprofen or naproxen despite marked reductions in cerebrovascular prostaglandins after all three agents (22). Further circumstantial evidence consistent with this has been obtained in newborn infants: indomethacin decreases CBF velocity within 2 min of administration (23) before any decrease in arterial prostaglandin occurs (24), although any effect on cerebral prostaglandin metabolism is not known.

Other NSAID may, therefore, close the PDA in preterm infants without cerebrovascular effects. In animal studies, ibuprofen, in contrast to indomethacin, does not affect basal CBF (22, 25, 26), cerebral metabolic rate (22, 25), or intestinal or renal hemodynamics (27). Ibuprofen may also enhance autoregulation of CBF (26) and has been shown to protect neurologic function after oxidative stresses in animal experiments (28). Ibuprofen constricts the lamb ductus arteriosus (29), and the administration of the drug within 3 h of birth reduces the incidence of subsequent PDA in preterm infants (30).

Our pilot data (31) and those of another group (32) indicate that unlike indomethacin, ibuprofen induces closure of hemodynamically significant PDA in infants without any immediate effects on CBV. Two small studies in preterm infants indicate that ibuprofen has similar efficacy to indomethacin at mediating closure of PDA (33, 34). No notable early adverse effects secondary to ibuprofen have been consistently reported (30–34).

We therefore performed a double-blind randomized controlled trial in which preterm infants with a hemodynamically significant PDA were randomized to receive either three doses of ibuprofen or indomethacin. The noninvasive optical technique of NIRS was used to monitor the effects of the drugs on CBF, COD, CBV, and CBVR. Echocardiography was performed by independent observers before and after drug treatment to assess the ductus arteriosus. The null hypothesis was that there were no significant differences between the effects of indomethacin and ibuprofen on CBF and CBV after the first dose of treatment.

METHODS

Subjects.

Between January and December 1996, 33 preterm infants on the neonatal units of Hammersmith, Queen Charlotte's, St George's, and St Mary's Hospitals, London, U.K., with a hemodynamically significant PDA diagnosed clinically and on echocardiographic criteria, were enrolled into the study. All infants had an initial diagnosis of respiratory distress syndrome. Written parental consent was obtained for all infants, and the study was approved by the Research Ethics Committee of each participating hospital.

Exclusion criteria and contraindications for administration of either drug were recent intraventricular hemorrhage (≤48 h), evidence of a bleeding diathesis, presence of major congenital malformations, clinical or radiographic evidence of necrotizing enterocolitis, platelet count <60 × 109 L−1, urine output ≤0.6 mL·kg−1·h−1 over the preceding 8 h, blood urea ≥14 mmol/L, and serum creatinine ≥140 μmol/L.

Diagnostic criteria for hemodynamically significant PDA.

As persisting patency of the ductus arteriosus is common in the postnatal period, only those infants with a hemodynamically significant PDA were studied. For the purposes of this trial, this was defined as left to right shunting of blood large enough to compromise cardiorespiratory status. The diagnosis of a hemodynamically significant PDA was considered on clinical grounds, taking into account the presence or absence of the following: a continuous or systolic murmur, bounding peripheral pulses, tachycardia (heart rate ≥170), tachypnea (respiratory rate ≥70), hepatomegaly (≥3 cm below costal margin), deterioration of respiratory status or failure to improve, unexplained persistent metabolic acidosis, and cardiomegaly with pulmonary plethora on chest radiograph.

A full echocardiographic assessment was then performed by a trained echocardiographer. The direction and severity of shunting through the PDA was evaluated using pulsed Doppler. Shunting was graded arbitrarily: mild if a small jet was detectable at the pulmonary end of the ductus arteriosus with no disturbed flow at the level of the pulmonary valves, moderate if a disturbed diastolic flow was easily detectable at all sites of the pulmonary trunk, and severe if diastolic back flow was additionally detectable in the aorta. Infants with moderate to severe shunting were considered to meet the criteria for a hemodynamically significant PDA.

Treatments and randomizations.

The attending clinical staff decided whether treatment was needed. The Pharmacy Department at Queen Charlotte's Hospital performed randomization in blocks of 12 (six, ibuprofen; six, indomethacin) for each hospital and also provided all trial medication. All other personnel were blinded to the identity of the drug administered.

The drug dosage schedule for all infants is shown in Table 1. Indomethacin (Indocid-PDA, Merck Sharp and Dohme), 2 mg, was dissolved in 2 mL of 0.9% saline. The initial dose of indomethacin was 0.2 mg/kg. Two further doses were administered after 12 and 24 h: infants aged 2 to 7 d at the time of the first dose received 0.2 mg/kg and those infants 8 d or older received 0.25 mg/kg.

Table 1 Drug dosage schedule
Full size table

Ibuprofen was provided as a clear solution of concentration 8 mg/mL (Merckle GmbH, Ulm, Germany). The initial dose was 10 mg/kg followed by 5 mg/kg at 24 and 48 h after the start of treatment.

To prevent identification of the drug administered from the timing schedule, all infants received a fourth dose containing 0.9% saline: in the indomethacin group, 48 h after the first dose and, in the ibuprofen group, 12 h after the first dose.

To adjust infusion volumes for both indomethacin and ibuprofen, 0.9% sodium chloride was used, enabling all infusions in an individual to be of identical volume. The clear and colorless drugs for each patient were provided in identical syringes. Peripheral intravenous infusion of all drugs was performed over 15 min by using an infusion pump.

NIRS.

Changes in the attenuation of near infrared light transmitted through the cranium were used to calculate changes in the concentration of HbO2 and Hb (35, 36).

To measure CBF, a small sudden change in arterial concentration of oxyhemoglobin was induced by increasing the inspired oxygen concentration for a few breaths (37). CBF was then calculated from the changes in SaO2, HbO2, and Hb concentrations:MATH 1 where K1 is a constant reflecting the molecular mass of hemoglobin (64 500) and cerebral tissue density (1.05) (38) and [tHb] is the total arterial hemoglobin concentration.

COD was calculated as the product of physically dissolved and hemoglobin-bound oxygen and CBF:MATH 2 where PaO2 is the arterial oxygen tension.

Changes in CBV were calculated by directly observing the changes in total cerebral hemoglobin (39):MATH 3 where K2 is a constant reflecting the molecular mass of hemoglobin, cerebral tissue density, and cerebral to large vessel hematocrit ratio (0.69) (40).

Experimental procedure.

The first, 24-, and 48-h doses were studied using NIRS. Measurements were commenced 1 h before the administration of the drug and continued for 1 h after the end of the infusion.

Optodes placed biparietally on the head at least 3.5 cm apart allowed changes in HbO2 and Hb to be measured by NIRS (NIRO 500, Hammamatsu, Japan). The interoptode distance was measured using calipers and the optical path length taken as 4.39 times the interoptode distance (41). NIRS data were collected every 0.5 s for CBF measurements and every 10 s for CBV-related data.

CBF measurements were performed by the method described and calculated using an integration time of 8 s. One to four (median, 3) CBF measurements were performed before administration of each dose and one to four times (median, 2) afterward. The average SaO2 measured while the infant was quiet and well oxygenated was used to calculate COD. Changes in CBV were calculated from changes in total cerebral hemoglobin concentration between the 5 min before each dose and the 5 min after completion of each dose. CBVR was measured once before each dose and once afterward by observation of the change in CBV in response to a spontaneous or induced (by changing ventilator rate) change in PaCO2 of approximately 1 kPa within or toward the normal range over several minutes. CBF and CBVR measurements were not performed during the period commencing 5 min before each dose and ending 5 min after completion of each dose. CBF data were rejected if they did not meet accepted predetermined criteria (42).

Heart rate, transcutaneous oxygen, and carbon dioxide tensions were recorded continuously. SaO2 was monitored by a pulse oximeter sited on the right hand with measurements performed in the beat to beat mode for CBF measurements. Mean arterial blood pressure was monitored before and after drug administration, either via an indwelling arterial line or by an oscillometric method. Arterial blood sampling was performed before and after drug administration for determination of pH, PaO2, PaCO2, and blood glucose.

Serum urea, electrolytes, creatinine, and total blood hemoglobin and platelet count were measured before each dose. Daily fluid input was recorded, and all infants were weighed daily if possible.

Cranial ultrasound scans were performed in all infants before the first dose, at 48 h after the last dose, and subsequently at least once weekly. All scans were classified on a standardized basis (43).

Reassessment for ductal patency was performed within 24–48 h of the dose at 48 h. The study was double blinded: NIRS data and echocardiographic data were obtained by independent observers and withheld from each other. If criteria for a hemodynamically significant PDA were no longer met, the ductus was considered closed and no further treatment was provided. If clinical signs of a hemodynamically significant PDA were still present or developed while the infant was <28 d of postnatal age, then further treatment was with indomethacin. The dosage of indomethacin was calculated on the same basis as for initial treatment. These further doses were not studied by NIRS.

Statistical analysis.

The mean CBF and COD were calculated before and after every dose for each individual. CBV before each dose was defined as the average volume in the 5 min immediately before commencing the infusion, and the CBV afterward as the average volume in the 5 min starting immediately on completion of the drug infusion. CBVR before and after each dose was defined as the difference between the average volume in the 5 min after stabilization of the alteration in PaCO2 and the average volume in the 5 min before the alteration in PaCO2 divided by the change in PaCO2.

Data groups were inspected and parametric or nonparametric tests used as appropriate. The primary end points of the trial were the effects of the first dose on CBF and CBV. Secondary analyses were made to assess the data further. The effects of treatment on cerebral circulation variables were tested in three ways. First, individual mean values for CBF, COD, and CBVR before and after each dose were tested using the paired t test or the Wilcoxon signed-rank test. Second, comparisons were made by group between infants receiving indomethacin and ibuprofen. The group values for CBF, COD, and CBVR before and after each dose of ibuprofen and change in CBV with each dose of ibuprofen were compared with those of the correspondingly timed dose in the indomethacin group using the unpaired t test or the Mann-Whitney rank sum test. Third, comparisons were performed using grouped data to test whether there were differences between the effects of the three doses within each treatment arm of the study. In the indomethacin group, the 48-h dose was saline, and, therefore, the first and 24-h doses were compared with this control group by using 1-way ANOVA with Dunn's test for multiple comparisons. In infants receiving ibuprofen, this was performed by 1-way ANOVA for all cerebral circulation variables. The ibuprofen data were also analyzed using ANOVA with Dunn's test for multiple comparisons and gave similar results.

Fisher's exact test was used to compare the efficacy of the drugs at inducing PDA closure. A data monitor was employed by the research team to review the data at predetermined points to decide when the primary end points of the study were reached. Statistical calculations were performed using Sigmastat (version 1.01, Jandel Corporation, USA). A p value <0.05 was considered significant.

RESULTS

The clinical details of the study infants before the first dose are shown in Table 2. The clinical data for the dose at 24 and 48 h were similar but are not shown. There were no significant differences between the two groups.

Table 2 Clinical details of study infants before the first dose of treatment * One other receiving nasal continuous positive airway pressure. ** Two others receiving additional oxygen in a headbox.
Full size table

There were no significant changes in heart rate, mean arterial blood pressure, blood glucose, blood pH, PaO2, or PaCO2 after the administration of indomethacin, ibuprofen, or saline. Blood glucose ranged from 2.8 to 11.0 mmol/L during the study period.

NIRS measurements were attempted in all infants. During treatment with the first dose, CBF measurements were achieved in 21 infants and CBV measurements in all 33 infants. Successful measurements for at least one cerebral circulation variable were achieved in 22 infants (11, indomethacin; 11, ibuprofen) during the dose at 24 h and 25 infants (12, saline; 13, ibuprofen) during the dose at 48 h. There were no significant clinical differences within each treatment group between those infants in whom NIRS measurements could and could not be analyzed.

The individual values of CBF, change in CBV, COD, and CBVR after the first dose of indomethacin and ibuprofen and the dose of saline at 48 h in the indomethacin group are shown in Figure 1.

Figure 1
figure 1

Individual mean values for cerebral circulation variables before and after the first doses of indomethacin and ibuprofen and the saline dose in the indomethacin group. Change in CBV with treatment has been related to an arbitrary zero.

Full size image

Individual and group CBF values fell significantly after the administration of the first dose of indomethacin (p < 0.001), but there were no significant changes after the first dose of ibuprofen. The respective before and after group mean (SD) values were 13.6 (4.1) and 8.3 (3.1) mL·100 g−1·min−1 in the indomethacin group and 13.3 (3.2) and 14.9 (4.7) mL·100 g−1·min−1 in the ibuprofen group (p < 0.001).

After the first dose, CBV fell significantly in infants treated with indomethacin but not in those treated with ibuprofen. The median (interquartile range) values for change in CBV were −0.4 (−0.3 to −0.6) mL/100 g in the indomethacin group and 0.0 (0.1 to −0.1) mL/100 g in the ibuprofen group (p < 0.001).

Individual and group CBF values fell significantly after the administration of the 24-h dose of indomethacin, but there were no significant changes after the administration of the 24-h dose of ibuprofen. The respective before and after group mean (SD) values were 16.7 (4.4) and 11.8 (5.1) mL·100 g−1·min−1 in the indomethacin group and 15.0 (3.9) and 15.5 (3.9) mL·100 g−1·min−1 in the ibuprofen group (p < 0.001). There were no significant changes in CBF after the 48-h doses of saline or ibuprofen.

After the 24-h dose, CBV fell significantly in infants treated with indomethacin but not in those treated with ibuprofen, the median (interquartile range) values for change in CBV being −0.2 (−0.3 to −0.1) mL/100 g in the indomethacin group and −0.1 (−0.1 to 0.1) mL/100 g in the ibuprofen group (p < 0.01). There were no significant changes in CBV after the 48-h doses of saline or ibuprofen.

Individual and group COD values fell significantly after the first and 24-h doses of indomethacin, but there were no significant changes after the 48-h dose of saline or with any of the three doses of ibuprofen. The respective before and after group mean (SD) values for the first doses were 2.6 (0.8) and 1.5 (0.5) mL·100 g−1·min−1 in the indomethacin group and 2.2 (0.5) and 2.5 (0.6) mL·100 g−1·min−1 in the ibuprofen group (p < 0.001). The before and after values for the 24-h doses were 2.8 (0.6) and 1.9 (0.5) mL·100 g−1·min−1 in the indomethacin group and 2.3 (1.0) and 2.5 (0.8) mL·100 g−1·min−1 in the ibuprofen group (p < 0.05).

In those infants who received indomethacin, comparison of changes in CBF, COD, and CBV by group after the first and 24-h doses of indomethacin with those occurring after saline showed a significant difference. There were no significant differences among the effects of the three doses of ibuprofen. The CBF, COD, and CBV response to the first dose of treatment was not significantly different between infants of birth weight ≤700 g or above or of postnatal age ≤7 d or above in both the indomethacin and ibuprofen groups.

A limited number of CBVR measurements were also performed. There was a significant decrease (p < 0.01) in CBVR after the first dose of indomethacin, but there was no significant change after the first dose of ibuprofen or with the saline dose at 48 h in the indomethacin group. The respective before and after group mean (SD) values were 0.24 (0.08) and 0.06 (0.08) mL·100 g−1·kPa−1 in the indomethacin group, 0.31 (0.05) and 0.28 (0.04) mL·100 g−1·kPa−1 in the ibuprofen group, and 0.26 (0.08) and 0.28 (0.03) mL·100 g−1·kPa−1 for the saline dose in the indomethacin group.

There appeared to be no significant difference in the efficacy of indomethacin and ibuprofen at mediating PDA closure, the rates being 93 and 78%, respectively, although it is not possible to state this with confidence. There was no apparent difference in the cerebral hemodynamic effects of the drugs between those infants in whom the PDA closed after a course of treatment and in those in whom it did not, but this cannot be stated with confidence.

Five infants continued to meet the criteria for a hemodynamically significant PDA after the first course of treatment (one in the indomethacin group and four in the ibuprofen group). The PDA closed in three infants after retreatment with indomethacin (one in the indomethacin group and two in the ibuprofen group).

In three infants, a hemodynamically significant PDA recurred 4–10 d after completion of treatment and was associated with suspected sepsis (two in the indomethacin group and one in the ibuprofen group). In one infant previously treated with indomethacin, PDA closure followed retreatment with indomethacin. In another infant previously treated with indomethacin who developed necrotizing enterocolitis, no retreatment was given, and the PDA closed spontaneously later. In the one infant from the ibuprofen group, retreatment with indomethacin did not lead to PDA closure.

Ductal ligation was performed in three infants in the ibuprofen group. In two infants in whom oliguria and serum creatinine >140 μmol/L followed the first dose of indomethacin, the attending clinician only was allowed access to the code. In these infants, the subsequent doses of indomethacin were 0.1 mg/kg and delayed by 18–24 h and were not studied with NIRS. No adverse effects were noted after the administration of ibuprofen.

Platelet transfusions were administered to two infants in both the indomethacin and ibuprofen groups before the first dose to correct a low platelet count. Adverse hematologic effects were not seen with either drug during the study.

Mean (SD) daily fluid intake during the 72-h period after the first dose of treatment in the indomethacin group was 145 (18) mL·kg−1·d−1 and in the ibuprofen group was 154 (16) mL·kg−1·d−1. These values were not significantly different. The results of the cranial ultrasound scans before the first dose are included in Table 2. No obvious changes on cranial ultrasonography followed treatment with either drug.

DISCUSSION

This study in preterm infants being treated for a hemodynamically significant PDA showed that indomethacin caused a substantial reduction in CBF, COD, CBV, and CBVR but that these variables remained unaffected by ibuprofen. The effects of multiple doses of each drug on CBF, COD, and CBV were also studied, and the changes were noted to be similar on repeated administration.

The values for measured cerebral hemodynamic variables and the magnitude of their changes after indomethacin were similar to those reported previously (7, 8). This study confirmed our previously published preliminary data regarding the absence of any effects of ibuprofen on CBV and CBVR (31). Another study showed that ibuprofen had no effect on CBV immediately after its administration, but, in contrast to our study, an increase in CBV was seen at 1 h, the reasons for which are not clear (32).

The PDA closed in most infants with treatment, and there was no apparent difference in the efficacy of the two drugs at mediating PDA closure. Although the results of this study and those reported previously are encouraging concerning the efficacy of ibuprofen at mediating PDA closure, none of these studies have involved more than 103 infants (31–34). To demonstrate a difference in efficacy between the two drugs of 15% with a significance level of 5 and 80% power, assuming that the efficacy of indomethacin at mediating PDA closure is 79% (2), a larger study with 220 infants would be needed.

The dosage and timing of indomethacin were based on the results of a large study that demonstrated efficacy of the drug at PDA closure (2) and also from pharmacokinetic data in preterm infants (2, 44). Pharmacokinetic data for ibuprofen in preterm infants are only available for the first 72 h of life (45), but theoretical calculations indicate that plasma levels needed to inhibit prostaglandin synthesis inhibition would have been attained in our study (J.V. Aranda, personal communication). Ibuprofen administered to preterm infants in early postnatal life at the dosage and timing used in our study reduced the subsequent incidence of PDA (30). The same dosage schedule of ibuprofen also appears to be effective at inducing PDA closure in infants commencing treatment between 48–72 h of age (33).

The mechanism by which indomethacin exerts its effect on cerebral hemodynamics is unknown, but our findings support previous studies that this may not be related to inhibition of prostaglandin synthesis (22–24). Topical application of arachidonic acid to the brains of newborn piglets leads to prostanoid production and dilatation of pial arteries that can be blocked by indomethacin (46), but equipotent doses of other inhibitors of prostaglandin synthesis have failed to reproduce the effects of indomethacin (47), and a specific action of indomethacin on the vascular endothelium is possible. Indomethacin increases the levels of circulating endothelins, suggesting an indirect vasoconstrictive effect of the drug (48). The reduction in CBF after indomethacin in adult humans has been found to be independent of decreased cerebral metabolic rate for oxygen (49).

Abnormalities of cerebral perfusion play an important role in the development of cerebral injury in newborn infants. Ibuprofen enhances CBF autoregulation in newborn piglets (26), and, thus, by stabilizing CBF, a reduction in the incidence of intraventricular hemorrhage in preterm infants may be expected. In one small study, early administration of ibuprofen to preterm infants was associated with a trend toward reduction in the incidence of intraventricular hemorrhage, although this was not statistically significant (30).

This study shows that ibuprofen does not cause the changes in CBF or CBV that are characteristic of indomethacin treatment. Further studies to define the potential role of ibuprofen as a treatment for newborn infants are warranted. As with any new treatment, long-term follow up of treated infants will be needed to detect possible adverse effects, e.g. newborn piglets treated with ibuprofen show changes in the electroretinogram.