B-Type Natriuretic Peptide Predicts Responses to Indomethacin inPremature Neonates with Patent Ductus Arteriosus

Jong-Hau Hsu, MD, San-Nan Yang, MD, PhD, Hsiu-Lin Chen, MD, Hsing-I. Tseng, MD, Zen-Kong Dai, MD, PhD,

and Jiunn-Ren Wu, MD

cts indomethacin responsiveness in pre-

ardiograhically large (diameter >1.5 mm)nderwent paired echocardiography andhacin. After treatment, neonates who re-

sponded (with closed or insignificant PDA) and neonates who did not respond (with persistent significant PDA re-PDA Patent ductus arteriosusquiring surgical ligation) were compared.Results Thirty-one premature neonates (mean gestational age, 30 weeks) underwent 119 paired echocardiogra-phy and BNP determinations. Mean BNP levels (1286 986 pg/mL) associated with significant PDA (n = 96) werehigher than those associated with closed or insignificant PDA (n = 23; 118 124 pg/mL; P < .001). Twenty-threeneonates responded and 8 neonates did not respond to indomethacin. Mean baseline BNP levels were higher inneonates who were non-responders (2234 991 pg/mL) than neonates who were responders (983 814 pg/mL; P = .001). A baseline BNP level >1805 pg/mL had a sensitivity rate of 88% and a specificity rate of 87% forpredicting indomethacin non-responsiveness (P = .003).Conclusions High baseline BNP levels predict poor responses to indomethacin and the need for surgery inpremature neonates with PDA. (J Pediatr 2010;157:79-84).

Patent ductus arteriosus (PDA) is associated with morbidity and mortality in premature infants. In addition to pulmo-nary congestion and heart failure caused by the left-to-right shunt of PDA, clinical sequelae of the ductal steal of bloodfrom the PDA include higher risks of adverse outcomes such as chronic lung disease,1 pulmonary hemorrhage,2 renal

hypoperfusion,3 necrotizing enterocolitis, and death.4 Therefore, early and proper management of the PDA has importantprognostic implications.Pharmacological closure of PDA with indomethacin is widely used in premature neonates. However, indomethacin has mul-

tiorgan adverse effects, including renal dysfunction,5 necrotizing enterocolitis,6 and decreased cerebral blood flow.7 In addition,10% to 40% of patients receiving indomethacin are non-responders and may require surgical ligation.8-11 Biomarkers predict-ing the response to indomethacin could be helpful in the management, but biomarkers, at present, are lacking.B-type natriuretic peptide (BNP) is a 32 amino acid polypeptide cardiac hormone with natriuretic, diuretic, and vasodilative

properties that is produced in response to increased myocyte stretch.12 BNP is a useful bedside screening tool for the presenceof PDA in premature infants.13,14 Recent studies further demonstrate the potential of BNP to guide an indomethacin therapeuticstrategy for these patients.15,16 However, whether there is a role for BNP determinations in evaluating the responsiveness to indo-methacin inprematureneonateswithPDAisunknown.Therefore,we report a prospective study to investigatewhetherBNPcanbea biomarker to differentiate responders from non-responders to indomethacin therapy in premature infants with PDA.

Methods

This study was conducted prospectively from April 2008 to May 2009, in a 15-bed neonatal intensive care unit (NICU) of a tertiary medical center with 323premature infants admitted to the NICU. The institutional review board of

From the Department of Pediatrics, Kaohsiung MedicalUniversity Hospital, Kaohsiung Medical University,Kaohsiung, Taiwan (J-H.H., S-N.Y., H-L.C., H-I.T., Z-K.D., J-R.W.); Department of Pediatrics, Faculty ofMedicine, College of Medicine, Kaohsiung MedicalUniversity, Kaohsiung, Taiwan (J-H.H., Z-K.D., J-R.W.);Graduate Institute of Medicine, Kaohsiung MedicalUniversity, Kaohsiung, Taiwan (J-H.H., S-N.Y); andFaculty of Respiratory Therapy, College of Medicine,Kaohsiung Medical University, Kaohsiung, Taiwan (H-L.C.)

Supported by the grant from Kaohsiung Medical Uni-versity Hospital (KMUH96-6G09), and the grant fromKaohsiung Medical University (KMU-Q098016). Theauthors declare no conflicts of interest

AO Aorta

BNP B-type natriuretic peptide

E/A Ratio of early (E) to late (A) transmitral inflow velocities

LA Left atrium

LVEDD Left ventricular end-diastolic dimension

NICU Neonatal intensive care unitObjectives To determine whether B-type natriuretic peptide (BNP) predimature neonates with patent ductus arteriosus (PDA).Study design Premature neonates receiving indomethacin for an echocand clinically significant PDA were prospectively studied. All neonates uBNP measurements at baseline and 24 hours after each dose of indomet0022-3476/$ - see front matter. Copyright 2010 Mosby Inc.All rights reserved. 10.1016/j.jpeds.2009.12.045

RPA Right pulmonary artery

79

THE JOURNAL OF PEDIATRICS www.jpeds.com Vol. 157, No. 1our institution approved the study, and the families of allsubjects gave informed consent before enrollment.Inclusion criteria for the study were: (1) born at a gesta-

tional age 1.5 mm; (2) the need for positive pressureventilation; and (3) a worsening cardiopulmonary status,which included a least 1 of these conditions: increased cardio-thoracic ratio or pulmonary vascular markings on radio-graphs, decreased urine output, increased need of inotropes,or escalating settings on ventilators.17 A PDA not fulfillingthe indications for indomethacin administration was definedas an insignificant PDA. Exclusion criteria for this study werethe presence of any other congenital structural heart defects,severe infection, history of maternal indomethacin adminis-tration, or contraindication for indomethacin.The treatment strategy for all these patients followed stan-

dard institutional practices. Prophylactic indomethacin is notused in premature infants in this institution; therefore, onlyPDA verified with echocardiography and associated with clin-ical significance is treated. In addition, after PDA was initiallydiagnosed, all patients were treated with fluid restriction anddiuresis for 24 to 48 hours until repeated echocardiographyand clinical reassessment weremade.When PDA remained he-modynamically significant, the initial course of indomethacinwas administered. In this study, only patients admitted within24 hours after birthwere included. Themedical teams involvedin the treatment of the patients were blinded to the BNP values.

Indomethacin AdministrationIndomethacin (Indocid,Merck,West Point, Pennsylvania)wasgiven as a 60-minute intravenous infusion at 24-hour intervalsfor each course (3 doses). For the initial indomethacin dose, 0.2mg/kg was administered to all patients regardless of their age.17

For the second or third dose, 0.1mg/kgwas administeredwhenthe patient was 7 days old. Echocardiographyand clinical assessments were performed at baseline (within30 minutes before the first dose) and 24 hours after eachdose (before the next dose of indomethacin) to determinewhether the subsequent dose was still needed.After the initial 3-dose course of indomethacin, when the di-

agnostic criteria for hemodynamically significant PDA werefulfilled again, an additional second course of indomethacinwas administered. When the criteria were no longer fulfilled,no additional indomethacin was administered. Surgical liga-tion was performed when the hemodynamically significantPDA remained despite 2 courses of indomethacin or whenthe patient had contraindications for indomethacin treatment.

EchocardiographyEchocardiographic studies were performed with a PhillipsSONOS 7500 echocardiographic scanner (Phillips, Andover,80Massachusetts). All echocardiographic studies were per-formed by the same investigator (J-H. H.) to avoid any inter-observer variability. A screening echocardiography wasperformed to exclude patients with congenital heart defectsor PDA with bidirectional or right-to-left shunt. After enroll-ment, studies were performed at baseline and at 24 hours af-ter each dose of indomethacin.In each study, these echocardiographic variables were re-

corded and analyzed : (1) PDA diameter; (2) PDA to rightpulmonary artery (PDA/RPA) ratio; (3) left atrial to aorticroot (LA/AO) ratio; (4) fractional shortening of the left ven-tricle (LV); (5) left ventricular end-diastolic dimension(LVEDD); and (6) E/A ratio: the ratio of early (E) to late(A) transmitral inflow velocities. The diameter of the PDAwas determined in the 2-dimensional parasternal short axisview, measured at the junction of the PDA and main pulmo-nary artery. The left- to-right shunt of ductal flow was con-firmed with color Doppler ultrasound scanning. Similarly,the RPA diameter was determined in the parasternal shortaxis view and measured at the junction of the RPA andmain pulmonary artery. The LA/AO ratio was determinedin the parasternal long axis with M-mode.

Measurement of Plasma B-type NatriureticPeptide LevelsBNPwasmeasured in blood samples from an indwelling cath-eter and collected at the same time as performing echocardi-ography.18 In brief, the samples were placed immediately onice in chilled ethylenediamine tetraacetic acid-treated tubesand centrifuged at 3000 rpm for 15 minutes at 4C. Separatedplasma was stored at 20C. Within 4 days, the plasma wasthawed to room temperature, and BNP levels were measuredwith a commercially available fluorescence immunoassay(Triage Meter Plus, Biosite Diagnostic, San Diego, Califor-nia). The measurable range of BNP is between 5 and 5000pg/mL. The estimated coefficient of variation for the assayis 9.2% to 11.4%.

Data CollectionThe clinical and biochemical data were prospectively col-lected by an observer blinded to the BNP results. The clinicaldata included patient demographics, Apgar score, antenataluse of steroids and indomethacin, presence of pre-eclampsia,vaginal delivery, use of surfactant, adverse effects associatedwith indomethacin, and mortality. The hemodynamic datacollected included systemic blood pressure and heart rate ateach sampling point. The biochemical data collected in-cluded complete blood cell counts, arterial blood gas levels,and serum creatinine values.

Indomethacin ResponsivenessThe patients were classified as responders and non-responders. Responders were defined as patients whose PDAbecame closed or insignificant after indomethacin therapy.Non-responders were defined as patients with persistent sig-nificant PDA after indomethacin therapy, requiring surgicalligation during the same hospital stay.Hsu et al

Table I. Demographic data of responders and non-responders

Responders Non-responders P value

Patient number (%) 23 (74%) 8 (26%)Neonatal characteristics

Gestational age, wk 30.1 3.1 (25-35) 29.6 2.6 (26-35) .73Male, n (%) 11 (48%) 5 (63%) .69Birth weight, g 1412 94 (670-2220) 1350 154 (940-2250) .73Apgar score at 1 minute 6 2 (3-8) 6 1 (5-8) .90Apgar score at 5 minutes 8 1 (5-10) 7 1 (6-9) .54Small for gestational age 2 (9%) 1 (13%) 1.0Use of surfactant 6 (26%) 3 (37%) .66Age at diagnosis of PDA, days 2.3 0.9 (1-5) 2.5 0.7 (1-3) .62Age at the first dose, days 4.2 2.1 (3-11) 4.7 1.4 (3-7) .52

Maternal historyAntenatal corticosteroid 14 (61%) 6 (75%) .68Pre-eclampsia 3 (13%) 0 (0%) .55Vaginal delivery 9 (39%) 4 (50%) .69Gestational DM 1 (4%) 1 (13%) .46

Data are presented as mean SD (range) or number (percent).DM, Diabetes mellitus.

July 2010 ORIGINAL ARTICLESStatistical AnalysisDifferences in the continuous variables between groups weretested with the student t test or Mann-WhitneyU test. Differ-ences in the categorical variables between groups were testedwith the Fisher exact test. Correlations between variableswere performed with the Spearman rank correlation method.Changes in BNP levels with time were compared with paired ttests. The usefulness of BNP as a predictor of indomethacinresponsiveness was evaluated by using receiver operatingcharacteristic curves. Statistical analyses were performedwith Prism software version 4.0 (GraphPad Software, SanDiego, California) and SPSS software for Windows version12 (SPSS Inc., Chicago, Illinois).

Results

Thirty-one consecutive infants were enrolled (mean gesta-tional age, 30 weeks; range, 25-35 weeks; mean body weight,Table II. Baseline hemodynamic and echocardiographic chartreatment with indomethacin

Responders

Patient number (%) 23 (74%)Hemodynamic data

Heart rate, minutes1 150 10 (124-164)MAP, mm Hg 43 7 (33-58)Arterial blood pH 7.32 0.07 (7.22-7.56)PaCO2, mm Hg 45.3 9.0 (22-56)Hemoglobin, g/dL 12.9 2.3 (9.8-17.3)Serum creatinine 1.0 0.3 (0.4-1.5)

Echocardiographic dataPDA diameter, mm 3.9 0.7 (2.2-5.6)LA/AO ratio 1.7 0.1 (1.1-2.5)PDA/RPA ratio 0.90 0.15 (0.46-1.10)LVEDD, mm 15.5 2.0 (10.4-18.6)FS, % 34.8 7.3 (16.1-48.7)E/A ratio 1.06 0.13 (1.01-1.23)

Data are presented as mean SD (range) or number (percent).MAP, Mean arterial pressure; FS, fraction of shortening.

B-Type Natriuretic Peptide Predicts Responses to Indomethacin1393 g; range, 670-2250 g). After treatment with indometha-cin, there were 23 infants who were responders and 8infants who were non-responders with no mortality. Com-parisons of demographic data between responders and non-responders are shown in Table I. Comparisons of baselinehemodynamic and echocardiographic characteristics beforeindomethacin administration in these 2 groups are shownin Table II.All responders completed their therapy with a single treat-

ment course (3 doses), without the need for a second courseof indomethacin. Among these 23 responders, 11 (48%) re-ceived 1 dose, 6 (26%) received 2 doses, and 6 (26%) received3 doses of indomethacin. The PDA reopened in 2 responderswithPDA, but the PDA was insignificant, and no further in-domethacin treatment or surgical ligation were needed. All 8of the non-responders had surgical ligations because of per-sistent significant PDA. Seven of the non-responders received2 complete courses (6 doses), and 1 received 5 doses.acteristics of responders and non-responders before

Non-responders P value

8 (26%)

145 10 (130-160) .2341 5 (32-49) .38

7.29 0.08 (7.14-7.36) .3749.3 7.7 (40-66) .2712.6 2.8 (8.3-17.1) .761.1 0.2 (0.8-1.5) .46

4.1 0.8 (2.9-5.1) .341.6 0.2 (1.1-2.2) .61

0.99 0.25 (0.60-1.33) .3914.6 2.3 (10.2-17.1) .3638.3 4.5 (30.3-43.6) .24

1.16 0.12 (0.84-1.22) .17

in Premature Neonates with Patent Ductus Arteriosus 81

= 23) and non-responders (n = 8) measured at the first 3 time. C, n = 12 in responders because 11 responders received only 1

THE JOURNAL OF PEDIATRICS www.jpeds.com Vol. 157, No. 1BNP Levels and Echocardiographic FindingsThere were a total of 119 paired echocardiography and BNPdeterminations. In the 96 echocardiograms with PDA, BNPlevels ranged from 16 to 3380 pg/mL (1286 986 pg/mL).The BNP levels for the 23 ultrasound scanning studies withclosed (n = 21) or insignificant PDAs (n = 2) were 118 124 pg/mL (P < .001). In the 96 measurements with signifi-cant PDA, BNP levels correlated positively with ductal size(r = 0.83, P < .001), LA/AO ratio (r = 0.47, P = .001), LVEDD(r = 0.31, P = .02), and PDA/RPA ratio (r = 0.45, P = .001),BNP levels did not correlate with serum creatinine level, ageat treatment, fraction shortening of LV, or mitral E/A ratio.

BNP Levels and Indomethacin ResponsivenessAfter treatment with indomethacin, BNP levels of the re-sponders decreased significantly from baseline to 24 hours af-ter the final dose (983 914 pg/mL versus 121 133 pg/mL,P< .001), whereas BNP levels of non-responders did not changesignificantly (2234 991 pg/mL versus 2037 737 pg/mL,

Figure 1. Comparisons of BNP levels between responders (npoints. Vertical lines represent the median. *P < .01; **P < .001dose of indomethacin.P = .66). BNP levels at baseline and 24 hours after the firstand the second doses were all higher in non-responders thanresponders (Figure 1). In responders, those who received 3doses of indomethacin had higher baseline BNP levels thanpatients who received 1 or 2 doses (Figure 2). In addition,we found that a baseline BNP level >1805 pg/mL predictedindomethacin non-responsiveness with a sensitivity rate of88%, a specificity rate of 87%, a positive predictive rate of70%, and a negative predictive rate of 95% (P =.003, areaunder the curve = 0.85). Similarly, a BNP level >665 pg/mL24 hours after the first dose predicted indomethacin non-responsiveness with a sensitivity rate of 75%, a specificityrate of 95%, a positive predictive rate of 86%, and a negativepredictive rate of 92% (P = .01, area under the curve = 0.83).

Discussion

This study broadens the application of BNP from the diagno-sis of PDA to prognostic stratification in these patients.

82Specifically, our results demonstrated that a cutoff baselineBNP level >1805 pg/mL was predictive of non-responsive-ness to indomethacin and the need for surgical ligation inpremature neonates. In addition, higher baseline BNP levelswere associated with the requirement of more doses in re-sponders. After administration of indomethacin, a highBNP level was also associated with a poor pharmacologic re-sponse. Finally, the BNP levels during the treatment courseswere correlated with degrees of left-to-right shunt as shownwith echocardiography. BNP can be a prognostic biomarkerin the management of premature infants with PDA treatedwith indomethacin.BNP has been widely used as a diagnostic biomarker of

cardiac dysfunction in adult patients. In recent years, the use-fulness of BNP in pediatric patients, especially in neonateswith PDA, has gained attention. For example, BNP levelsstrongly correlated with the ductal size in premature neo-nates,19 and BNP levels decreased significantly in respondersto indomethacin.20 Consistent with earlier studies, we foundFigure 2. Comparisons of baseline BNP levels in responderswho received 1 dose (n = 11), 2 doses (n = 6), and 3 doses(n = 6). Vertical lines represent the median. * P < .05; **P < .01.

Hsu et al

July 2010 ORIGINAL ARTICLESthat BNP levels correlated with ductal size and indexes of leftventricular volume loading such as LA/AO ratio and LVEDD,and these levels declined after treatment in responders. How-ever, the wide variations of BNP levels in patients with signif-icant PDA observed in this study and others suggest thatthere are other unidentified factors influencing BNP levelsin these patients.In addition to the diagnostic value, there is an emerging

role for BNP levels in guiding PDA therapy. Attridge et al re-cently reported that BNP levels could be used to guide thelength of indomethacin treatment byminimizing the numberof indomethacin doses given in a single course of therapy.15

In addition, Czernik et al found that in infants 550 pg/mL on the second day of life predictedthe need for ductus intervention (indomethacin alone or in-domethacin followed by surgery).16 However, in contrast toour results, Czernik et al did not find significant differencesof BNP levels between responders to indomethacin and pa-tients requiring surgery. This discrepancy may be explainedby several factors. First, there are still no standard criteriafor the indications and optimal timing for starting indometh-acin. The BNP levels measured before interventions may re-flect different hemodynamic status in different studies. One-third of their patients started indomethacin therapy within 3days (35-52 hours) of birth, and all our patients started indo-methacin therapy at $3 days of age. Second, patients withsepsis were excluded from our study, but consisted of 25%of their patients, which might affect BNP levels before inter-ventions. Finally, our patients were not limited to infants

and indomethacin: a consecutive experience. J Pediatr 1978;93:

639-46.

10. Siassi B, Blanco C, Cabal LA, Coran AG. Incidence and clinical features

of patent ductus arteriosus in low-birthweight infants: a prospective

analysis of 150 consecutively born infants. Pediatrics 1976;57:347-51.

11. Yeh TF, Luken JA, Thalji A, Raval D, Carr I, Pildes RS. Intravenous in-

domethacin therapy in premature infants with persistent ductus arterio-

susa double-blind controlled study. J Pediatr 1981;98:137-45.

12. Levin ER, Gardner DG, SamsonWK. Natriuretic peptides. N Engl J Med

1998;339:321-8.

13. Puddy VF, Amirmansour C, Williams AF, Singer DR. Plasma brain na-

triuretic peptide as a predictor of haemodynamically significant patent

ductus arteriosus in preterm infants. Clin Sci (Lond) 2002;103:75-7.

14. Sanjeev S, Pettersen M, Lua J, Thomas R, Shankaran S, LEcuyer T. Role

of plasma B-type natriuretic peptide in screening for hemodynamically

significant patent ductus arteriosus in preterm neonates. J Perinatol

2005;25:709-13.

15. Attridge JT, Kaufman DA, Lim DS. B-type natriuretic peptide concen-

trations to guide treatment of patent ductus arteriosus. Arch Dis Child

Fetal Neonatal Ed 2009;94:F178-82.

16. Czernik C, Lemmer J, Metze B, Koehne PS, Mueller C, Obladen M. B-

type natriuretic peptide to predict ductus intervention in infants