User:Chelalo89

Effect of Furosemide on Ductal Closure and Renal Function in Indomethacin-Treated Preterm Infants during the Early Neonatal Period Byong Sop Lee Shin Yun Byun Mi Lim Chung Ji Young Chang Hee Young Kim Ellen Ai-Rhan Kim Ki-Soo Kim Soo-Young Pi Division of Neonatology, Department of Pediatrics, Asan Medical Center, University of Ulsan College of Medicine, Seoul , Korea control group (3/29) (p ! 0.001). Compared with the control group, serum creatinine and cystatin C levels and fractional excretion of sodium were significantly increased in the furosemide group for 24–36 h after indomethacin therapy (p ! 0.01). There were no between-group differences in mortality and other neonatal morbidity rates. Conclusions: Use of furosemide in combination with indomethacin increased the incidence of ARF but did not affect the PDA closure rate in preterm infants. Copyright © 2010 S. Karger AG, Basel Introduction Indomethacin is a prostaglandin synthesis inhibitor widely used for pharmacologic closure of the patent ductus arteriosus (PDA) in preterm infants. Among the side effects associated with indomethacin, renal abnormalities are most common, with incidence rates of up to about 40% in low-birth-weight infants [1] . The major mechanism of this side effect is inhibition of prostaglandin synthesis by indomethacin and subsequent reduction in renal blood flow [2] . Furosemide increases renal prostaglandin synthesis, which would be expected to prevent indomethacin-associated renal toxicity [3, 4] . A few clinical trials, including two studies from the pre-surfactant era have investigated the effects of furosemide use in in- Key Words Indomethacin � Furosemide � Preterm infants � Acute renal failure � Creatinine � Cystatin C Abstract Background: Furosemide is known to increase renal prostaglandin synthesis. However, its influence on ductal closure and renal toxicities of indomethacin in preterm infants has not been conclusive, especially during the early neonatal period. Objectives: To identify the effects of furosemide after indomethacin administration on the rate of patent ductus arteriosus (PDA) closure and renal function in preterm infants. Methods: 68 infants (gestational age ! 34 weeks and birth weight ! 2,000 g) receiving indomethacin therapy (one course: 0.2–0.1–0.1 mg/kg q 12 h, mostly started ! 48 h after birth) were randomly assigned to the furosemide (n = 35) or control (n = 33) group. Each indomethacin dose was followed by furosemide (1.0 mg/kg) or placebo. The primary (PDA closure) and secondary (acute renal failure (ARF) and others) outcomes were assessed. Renal parameters before and 0–12 and 24–36 h after the first course of indomethacin were also investigated. Results: In an intention-to-treat analysis, there were no differences in the PDA closure rate between the furosemide (29/34) and the control (27/29) group (p = 0.437). The incidence of ARF (serum creatinine 1 1.6 mg/ dl) was greater in the furosemide group (20/34) than in the Received: March 30, 2009 Accepted after revision: September 4, 2009 Published online: March 16, 2010 formerly Biology of the Neonate Assist. Prof. Byong Sop Lee, MD Division of Neonatology, Department of Pediatrics Asan Medical Center, University of Ulsan College of Medicine 388-1, Poongnap-dong, Songpa-gu, Seoul 138-736 (South Korea) Tel. +82 23 010 3929, Fax +82 23 010 6978, E-Mail mdleebs @ amc.seoul.kr © 2010 S. Karger AG, Basel 1661–7800/10/0982–0191$26.00/0 Accessible online at: www.karger.com/neo Lee/Byun/Chung/Chang/Kim/Kim/Kim/ Pi Neonatology 192 2010;98:191–199 domethacin-treated preterm infants; however, conclusions regarding the role of furosemide in renal protection have been inconsistent [5–7] . Another clinical concern regarding furosemide administration is the increased incidence of PDA due to increased synthesis of renal prostaglandin [3, 4] . However, administration of furosemide did not affect the incidence of PDA closure in the abovementioned trials of small sample sizes [5–7] . Indomethacin was used as a ‘rescue’ therapy and administered only in infants with symptomatic PDA, and the time of administration was delayed until after the age of 3 days. Although the long-term benefit of early ( ! 2–3 days) or ‘prophylactic’ ( ! 24 h) indomethacin therapy is controversial, it was found to be more effective than ‘rescue’ therapy in inducing ductal closure and reducing the rate of PDA-associated morbidities such as severe intracranial hemorrhage in extremely-low-birth weight infants [8–10] . As preterm infants experience marked changes in renal function during the early days of life, the time of initiation of indomethacin therapy can be critical to conclusively determine the benefit or harm of furosemide infusion in combination with current indomethacin therapy for PDA closure. The aim of our study was to determine whether administration of furosemide immediately after indomethacin can influence the pharmacologic efficacy, in terms of ductal closure, or the renal toxicity of indomethacin in preterm infants during the early neonatal period. Methods Patients and Study Design The study was a single-center, prospective, randomized controlled trial. The Institutional Review Board of the Asan Medical Center approved the study protocol and written informed consent was obtained from the parents of each infant. The study period was between March 2007 and September 2008. Eligible criteria for this study were newborn infants with birth weight ! 2,000 g and who were delivered in Asan Medical Center at ! 34 0 / 7 weeks’ gestation and received indomethacin therapy before the age of 96 h. Exclusion criteria were: fetal hydrops, major congenital anomalies including structural heart or kidney diseases as shown by prenatal ultrasound, twin-to-twin transfusion syndrome and suspected congenital infection. None of mothers of the patients received tocolytic therapy with indomethacin or other prostaglandin inhibitors. The patients were randomly assigned to two groups by a computer-generated randomization scheme (in random blocks of two) at a 1: 1 ratio. Investigators were blind to the designated group until data analysis. In patients in the furosemide group, furosemide (1.0 mg/kg) was intravenously infused slowly following each dose of indomethacin. In patients in the control group, placebo (the same volume of 0.9% saline) was administered instead of furosemide. Each drug (furosemide or placebo) was handled and delivered by a hospital study pharmacist who had a randomization scheme indicating the sequence of treatment group assignment. The intervention was limited to the days of the first course of indomethacin therapy (‘intervention days’, see below). Furosemide use during subsequent courses of indomethacin therapy was at the attending neonatologist’s discretion. The primary outcome of the present study was ductal closure after the first course of indomethacin therapy (‘primary PDA closure’, see below). Secondary outcome measures included parameters associated with acute renal failure (ARF, see below) after the first course of indomethacin therapy and other parameters associated with neonatal mortality and morbidities. Sample Size Based on previous reports [4, 7] , we hypothesized that the difference in ductal closure rate between the furosemide and control group would be 20% and that the primary PDA closure rate after one course of indomethacin therapy would be about 85% [11] . Thus, to reduce this rate to 65%, 144 infants would be required in each group with a two-tailed type I error rate of 0.05 and a power of 80%. However, recruitment was halted after enrollment of 68 infants, because interim data analysis revealed a higher rate of ARF in the furosemide group. Indomethacin Therapy and Evaluation of PDA Contraindications for indomethacin therapy were: urine output ! 1.0 ml/kg/h for the preceding 8 h; a serum creatinine concentration 1 1.6 mg/dl; platelet count ! 60,000/mm 3 ; bleeding tendency (gross hematuria, fresh blood in gastric aspirate, tracheal aspirate or stool), and signs of necrotizing enterocolitis (NEC) or bowel perforation. Indomethacin treatment for PDA in preterm infants comprised ‘prophylactic’ or ‘rescue’ therapy. ‘Prophylactic’ therapy was indicated for infants at high risk of persistent PDA (see below) and was started within 24 h of birth. At the start of the therapy, echocardiography (ECHO) was not routinely performed in the patients receiving ‘prophylactic therapy’ because it was assumed that more than 80% of this age group would have a PDA opened during this period [12, 13] . Indications for the ‘prophylactic’ indomethacin therapy were: (1) birth weight ! 1,000 g and (2) weighing between 1,000 and 1,499 g with respiratory distress syndrome and/or assisted ventilator therapy. ‘Rescue’ therapy was indicated in infants weighing at least 1,000 g and diagnosed with ‘significant’ PDA (see below) 24 h after being born. The indomethacin therapy protocol did not differ between the ‘prophylactic’ and the ‘rescue’ therapy. One course of indomethacin therapy comprised three doses of indomethacin administered with an interval of 12 h. Doses were determined according to the patient’s age at the time of the first dose: patients aged ! 48 h received 0.2, 0.1, and 0.1 mg/kg, whereas patients aged 1 48 h received 0.2, 0.2, and 0.2 mg/kg [14] . Indomethacin was slowly infused intravenously over 30 min by syringe pump. ‘Primary PDA closure’ was diagnosed when PDA closure was diagnosed on the first follow-up ECHO 12–24 h after completion of the third dose of the first course of indomethacin therapy. In cases where the ductus was still open and ‘significant’ on ECHO, a second indomethacin course was indicated. Meanwhile, in patients with open but not ‘significant’ ductus, a second follow-up ECHO was performed 12– 24 h later without additional indomethacin therapy. If the PDA was still not closed in this ECHO, the second indomethacin course was indicated. PDA ligation was performed in patients Furosemide and Renal Function in Indomethacin-Treated Preterm Infants Neonatology 2010;98:191–199 193 who still had ‘significant’ PDA after two or three courses of indomethacin therapy, or when indomethacin was contraindicated. Color-Doppler ECHO with M-mode was performed and double checked by two attending physicians (B.S.L. and S.Y.B.) blinded to the assigned study group. Using a standard method, diagnosis of ‘significant’ PDA was made according to the presence of a continuous left to right ductal flow with a ductal lumen size 1 1.5 mm and a ratio of the left atrium and aortic root (LA/Ao ratio) 1 1.4 on the parasternal long-axis M-mode view at the level of the aortic valve [15] . All patients enrolled in our study underwent ECHO at the time of hospital discharge for confirmation of PDA closure. Other Management During the intervention days, infants were kept in incubators with servocontrol of temperature. For all infants with birth weight ! 1,500 g, humidification was initiated with an ambient humidity of about 95% from birth until the fourth day of life and, thereafter, was reduced by 5–10% each day until a humidity of 50% was reached. During the first day of birth, fluid intake was restricted to 70–80 ml/kg/day with 5 or 10% dextrose solution and amino acid solution of 1.5 g/kg/day and adjusted to maintain serum sodium concentration within 135–145 mEq/l, by monitoring body weight, serum sodium and urine volume. In general, fluid intake was increased by 10 ml/kg/day after confirmation of PDA closure, reaching 130–140 ml/kg/day at 10–14 postnatal days. In patients with decreased urine output or weight gain during the first few hospital days, the daily increase in fluid intake was withheld or reduced. If symptomatic PDA was diagnosed, fluid was restricted to 70–80 ml/kg/day for the first 4 days after birth and thereafter to 90–100 ml/kg/day. All infants in our study received prophylactic antibiotics (ampicillin and gentamicin) until the initial blood culture was reported to be negative, usually for about 2 or 3 days after birth. For infants in whom sepsis was diagnosed or clinically suspected, antibiotics were changed to vancomycin and amikacin, and in a few cases of suspected or proven fungal septicemia, amphotericin-B was administered. Data Collection Urine collection was performed in all patients during the first course of indomethacin therapy using urine bags for determining urine volume and urine sodium and creatinine. If urine spillage occurs, urine volume was estimated by weighing the diaper. Suprapubic compression was performed at the beginning and end of each collection. Urine samples were collected three times at the following hours: (1) 8 h (for prophylactic therapy) or 12 h (for rescue therapy) before, and (2) 0–12 h and (3) 24–36 h after completion of the third indomethacin dose of the first course of therapy. Serum data including electrolytes, blood urea nitrogen (BUN), creatinine and cystatin C levels, and urine electrolytes and creatinine were obtained at the same period for urine collection and thereafter, serum electrolytes and creatinine levels were measured every day for the first 2 weeks after birth and on alternate or every 3 days during the third week of life and thereafter. Glomerular filtration rate (GFR) was calculated and corrected for body surface area according to the Haycock formula [16] . ARF was defined as increased serum creatinine 1 1.6 mg/dl following the first indomethacin course during the first 2 weeks after birth [17, 18] . Serum cystatin C was measured using a particle-enhanced immunonephelometric immunoassay (Dade Behring Diagnostics, Marburg, Germany). Oliguria was defined as urine volume ! 1.0 ml/kg/h and hyperkalemia was defined as serum potassium level 1 6.0 mEq/l. The following clinical parameters associated with ARF were obtained: percentage of maximum weight loss and days to regain birth weight after a week. Other outcomes of neonatal mortality and morbidities were also obtained: bronchopulmonary dysplasia (BPD) was defined by the National Institutes of Health consensus definition [19] and NEC was defined according to Bell’s classification II or greater, including so-called ‘spontaneous intestinal perforation’ [20] . Patients with retinopathy of prematurity (ROP) who underwent laser photocoagulation were recorded. Intracranial hemorrhage was classified using the fourlevel grading system of Papile et al. [21] . Hemorrhages of grade 3 or 4 were considered severe. Statistical Analysis Continuous, normally distributed data, non-parametric continuous data and categorical data were compared using the t test, Mann-Whitney test and � 2 test (or Fisher’s exact test), respectively. The correlation between serum cystatin C and serum creatinine concentrations was assessed using Pearson’s correlation coefficients (r p ). Repeated measures ANOVA was used to compare the renal parameters before and after the first indomethacin therapy, with each parameter (e.g. serum creatinine) as within-subject factor and the study groups as between-subject factor. The effects of furosemide exposure and ARF on fractional excretion of sodium (FE Na ) were assessed by two-way ANOVA. Multiple logistic regression analyses were performed to identify the risk factor for ARF after indomethacin therapy using the backward elimination method. We selected continuous or categorical variables with p value ! 0.2 for inclusion in the regression model. All statistical analyses were performed with SPSS 13.0 for Windows. Results Study Population and Outcomes of PDA A total of 68 patients were enrolled and allocated to the control (n = 33) or furosemide (n = 35) group. There were no between-group differences in baseline characteristics or in therapeutic interventions within 2 weeks of birth that might have affected PDA closure or renal function ( table 1 ). The cumulative furosemide doses, other than those indicated by the study protocol, also did not differ between groups. This was more evident in the 52 patients subjected to ‘renal data analyses’ (see below). The first course of indomethacin therapy could not be completed in 5 patients (4 in the control and 1 in the furosemide group); all 5 were extremely-low-birth-weight infants and died during the first 5 days of life (median 2 days, range 1–5) due to grade 4 intracranial hemorrhage (n = 4) and massive pulmonary hemorrhage (n = 1). Thus, the primary outcome variable and several secondary outcome variables (such as ARF, BPD and ROP with laser therapy) could not be determined in these patients. Another 5 patients in the control group were considered to Lee/Byun/Chung/Chang/Kim/Kim/Kim/ Pi Neonatology 194 2010;98:191–199 have violated treatment protocol by receiving one (n = 4) or two (n = 1) doses of furosemide during the intervention days (‘furosemide contamination’). None of the patients in the furosemide group received additional furosemide during the intervention days other than those indicated by the study protocol. As a result, 58 patients were available for the as-treated analysis. Another 6 patients who had ‘significant’ PDA on the first follow-up ECHO were indicated for two or more courses of repeated indomethacin therapy (n = 5) or surgical ligation (n = 1, infants with NEC). These 6 patients were excluded from renal data analyses because decreased renal perfusion caused by ductal shunting, NEC and/or additional doses of indomethacin may have contributed to renal dysfunction [22] . Meanwhile, 7 patients who experienced ‘primary PDA closure’ but later PDA reopening were included in the renal data analyses because PDA reopening occurred at least 10 days after birth (median 15, range 10–22); furthermore, after reopening, the serum creatinine level was not greater than the maximum creatinine level observed before the diagnosis of PDA reopening. Consequently, ‘renal data analyses’ were performed for the 52 patients in whom ‘primary PDA closure’ was identified on the first follow-up ECHO ( fig. 1 ). Study Outcomes In the intention-to-treat analysis of all enrolled patients, the rate of primary PDA closure after the first course of indomethacin therapy did not differ between the furosemide group (29/34, 85%) and the control group (27/29, 93%) (p = 0.437). Of the 56 patients who experienced primary PDA closure, 7 (13%) showed PDA reopening. The incidence of PDA reopening was 6/29 (21%) in the furosemide group and 1/27 (4%) in the control group (p = 0.103). Post-hoc analysis of enrolled patients revealed that the percentage receiving two or more courses of indomethacin therapy or PDA ligations was higher in the furosemide group (11/34, 32%) than in the control group (3/29, 10%) (OR 4.15, 95% CI 1.028–17.72, p = 0.036). In the subgroup analysis of infants with gestational age ! 29 weeks, the trend of increased indomethacin courses and/or PDA ligation in the furosemide group (10/19, 53%) versus the control group (3/17, 18%) was also evident (OR 5.19, 95% CI 1.11–24.14, p = 0.041). The rate of PDA ligation did not differ between the control (3/33, 9%) and furosemide (5/35, 14%) groups (p = 0.710). The rate of ARF was significantly greater in the furosemide group than in the control group ( table 2 ). This was also observed in the subgroup analyses of more premature ( ! 29 weeks’ gestation) or smaller ( ! 1,500 or Table 1. Baseline characteristics of the study groupsa Control (n = 33) Furosemide (n = 35) p valuesb Male 22 (67) 19 (54) 0.297 Cesarean section 21 (64) 22 (63) 1.000 Gestational age, weeks 28.082.8 27.782.8 0.592 Ranges, weeks 23.4–33.7 23.7–33.0 Birth weight, g 1,0938419 1,0318387 0.532 <1,000 g 18 (55) 22 (63) 0.653 <1,500 g 26 (79) 30 (86) 0.454 Small for gestational age 2 (6) 8 (23) (0.085) Apgar score at 5 min Median, interquartile range 7 [6–8] 7 [5–8] 0.970 5-min Apgar <7 16 (49) 16 (49) 1.000 Base deficit, mEq/l 6.084.5 6.984.4 0.424 Base deficit >10 mEq/l 4 (17) 6 (18) 1.000 RDS 17 (52) 19 (54) 0.819 Intubation at birth 21 (64) 19 (54) 0.434 PROM >18 h 8 (24) 12 (34) 0.364 Chorioamnionitis 9 (27) 13 (37) 0.385 Betamethasone 30 (91) 32 (91) (1.000) Fluid, ml/kg birth weight Day 1 76.288.5 (n = 32) 77.6810.3 (n = 35) 0.536 Day 2 76.989.9 (n = 31) 80.5814.4 (n = 34) 0.249 Day 3 81.2810.1 (n = 31) 85.7810.0 (n = 34) 0.077 Day 4 83.8810.6 (n = 30) 87.3812.8 (n = 34) 0.247 UAC insertion 25 (76) 32 (91) (0.105) Indwelling days 383 583 0.084 Indwelling days ≥7 7 (21) 13 (37) 0.150 Antibioticsc <2 weeks 10 (30) 16 (46) 0.191 Indomethacin Start hours 20.8818.0 18.2820.4 0.587 Start <24 h of birth 25 (76) 28 (80) 0.673 Cumulative doses of furosemided Until 1 week 1.482.1 [1.681.8] 2.482.5 [2.082.0] 0.092 [0.426] Until 2 weeks 3.082.9 [3.282.9] 4.884.6 [4.384.1] 0.051 [0.299] Severe ICH 5 (15) 6 (17) 0.824 R DS = Respiratory distress syndrome; PROM = prolonged rupture of membrane; UAC = umbilical artery; ICH = intracranial hemorrhage. a Data are presented as number (%) and mean 8 SD unless otherwise specified. b p values were calculated by the �2 test (or Fisher’s exact test), the t test, or the Mann-Whitney test for categorical data, continuous data, or non-parametric data, respectively. c Vancomycin + amikacin. d Cumulative furosemide doses, other than indicated by the study protocol. Values in brackets represent the data and p value in 52 cases eligible for renal data analyses (see text). Furosemide and Renal Function in Indomethacin-Treated Preterm Infants Neonatology 2010;98:191–199 195 ! 1,000 g) infants, except for the as-treated analysis for a subgroup of infants weighing ! 1,000 g ( table 2 ). Subgroup analysis of infants weighing 6 1,000 g showed that the rate of ARF was significantly greater in the furosemide group (7/13, 54%) than in the control group (0/15, 0%) (p = 0.005). Neonatal mortality did not differ between the furosemide group (3/35, 9%) and the control group (4/33, 12%) (p = 0.705). There were no betweengroup differences in the rate of other neonatal morbidities including BPD, ROP requiring laser photocoagulation, NEC, sepsis, days of parenteral nutrition, days of mechanical ventilator and hospital days (data not shown). Renal Data Analyses Table 3 shows serial changes in renal parameters over the 36 h following indomethacin therapy in the 52 patients eligible for renal data analyses. Repeated measures ANOVA revealed that there were significant differences in each parameter between the three times of measurement before and after the first indomethacin course (p ! 0.001). There was a significant effect of furosemide on the change in renal parameters associated with ARF except for urine output. A two-way ANOVA model demonstrated that furosemide and ARF had different roles in increasing FE Na after the first course of indomethacin; at Eligible (n = 97) Randomized (n = 68) Not randomized (n = 29) Not approached (n = 18) No consent (n = 11) Control (n = 33) 4 died before primary response was evaluated Control (n = 29) Protocol violators (n = 5) Furosemide (n = 35) 1 died before primary response was evaluated Furosemide (n = 34) 1 course – closed (n = 27)* Protocol violators (n = 4) 1 course – not closed (n = 2) Protocol violator (n = 1) 1 course – not closed (n = 5) No longer open (n = 26) 2 more INDO NC SL 2 more INDO NC SL t t t t Reopen (n = 1) NCtSL (ARF) No longer open (n = 23) Reopen (n = 6) 1 more INDO closed 2 more INDO closed 2 more INDO NC comfort care (Gr 4 ICH) 2 more INDO closed 2 more INDO NC SL SL (NEC) t t t t t t t 2 more INDO NC SL 1 more INDO closed 2 more INDO NC SL 1 more INDO closed SL (NEC) t t t t t t 1 course – closed (n = 29)* Fig. 1. Flow of the study population and the outcome of patent ductus arteriosus. NC = Not closed; SL = surgical ligation; INDO = indomethacin course; ARF = acute renal failure; ICH = intracranial hemorrhage; NEC = necrotizing enterocolitis. * A total of 52 patients without protocol violation were included in the renal data analysis. Lee/Byun/Chung/Chang/Kim/Kim/Kim/ Pi Neonatology 196 2010;98:191–199 0–12 h, increased FE Na was associated with furosemide infusion (control vs. furosemide: 6.5 8 4.3 vs. 12.4 8 8.1, p = 0.029) rather than with ARF (no ARF vs. ARF: 7.7 8 4.6 vs. 13.5 8 8.8, p = 0.165). At 24–36 h after, however, increased FE Na was associated with ARF (no ARF vs. ARF: 7.8 8 5.3 vs. 15.7 8 10.3, p = 0.032) rather than with furosemide infusion (control vs. furosemide: 7.3 8 5.3 vs. 13.4 8 11.2, p = 0.176). Before indomethacin treatment, the mean FE Na level was greater in infants with gestational age ! 29 weeks (5.9 8 3.3) than in infants aged 6 29 weeks (4.1 8 1.4) (p = 0.017). After indomethacin therapy, however, the mean FE Na levels were similar in infants aged ! 29 weeks and in those aged 6 29 weeks at 0–12 h (10.0 8 7.3 vs. 9.7 8 6.8, p = 0.883) and at 24–36 h (11.7 8 9.4 vs. 9.7 8 7.2, p = 0.383). Figure 2 shows the serial change in mean serum creatinine level for each group at representative days. Peak serum creatinine level was achieved at approximately 6 days after the first indomethacin dose in each group (furosemide vs. control: 6 8 4 vs. 6 8 5 days, p = 0.523). It was greater in the furosemide group (1.76 8 0.55 mg/dl) than in the control group (1.29 8 0.37 mg/dl) (p ! 0.001). Thereafter, however, the mean serum creatinine level decreased and there was no difference between the two groups from 10 days after the first indomethacin dose. Although the serum creatinine level in 2 patients in the Table 2. Subgroup analyses for acute renal failurea Subgroups Acute renal failure intention-to-treat analysis (n = 63) a s-treated analysis (n = 58b) control furosemide odds ratio p valuesb control furosemide odds ratio p valuesc All infants 3/29 (10) 20/34 (59) 12.38 (3.13–49.04) <0.001 3/24 (13) 20/34 (59) 10.00 (2.49–40.12) <0.001 <29 weeks’ gestation 3/21 (14) 16/24 (67) 12.00 (2.71–53.14) (0.001) 2/13 (15) 13/19 (68) 11.92 (1.99–71.41) (0.005) <1,500 g 3/22 (14) 18/29 (62) 10.36 (2.48–43.3) (0.001) 3/18 (17) 18/29 (62) 8.18 (1.92–13.84) (0.003) <1,000 g 3/14 (21) 13/21 (62) 5.96 (1.26–28.10) (0.036) 3/11 (27) 13/21 (62) 4.33 (0.88–21.31) (0.135) PD A = Patent ductus arteriosus; ARF = acute renal failure. a Data from the 5 infants of early death (4 in control and 1 in furosemide group) could not be determined for each outcome variable. Comparative data are provided for infants in subgroups. Data are presented as number (%) unless otherwise specified. b Five patients with protocol violation (furosemide contamination) were excluded from the data analyses. c p values were calculated by the �2 test (or Fisher’s exact test). Table 3. Comparison of renal parameters before and after the first course of indomethacin therapya Pretreatment 0 –12 h after 24–36 h after p valuesb control furosemide control furosemide control furosemide Urine output, ml/kg/h 2.881.7 2.381.0 3.781.7c 4.981.9 c 3.081.4 3.281.3 0.298 Serum creatinine, mg/dl 0.8480.30 0.8580.30 1.0580.29d 1.2880.21d 1.0880.29e 1.4180.29e 0.003 Serum cystatin C, mg/l 1.3680.22 1.4480.17 1.4980.30e 1.8780.33e 1.5680.35e 1.9580.29e 0.001 Calculated GFR, ml/min/1.73 m2 16.484.6 15.684.1 13.184.3d 9.881.7d 12.784.2e 9.182.5e 0.002 BUN 14.2810.1 14.888.6 22.689.0c 30.6814.9c 24.689.2d 36.1819.1d 0.028 FENa, % 5.282.9 4.882.8 6.584.3d 12.488.1d 7.385.3c 13.4811.2c 0.009 GFR = Glomerular filtration rate; BUN = blood urea nitrogen; FENa = fractional excretion of sodium. p < 0.001 for each parameter in each group. a Data from the 52 patients (23 in the control group, 29 in the furosemide group) eligible for renal data analyses (see text). Data are presented as mean 8 SD. b Between the study group. p values were calculated by repeated measures ANOVA. c p < 0.05, d p < 0.01, e p < 0.001 between the three times of measurement in each parameter. p values were calculated by repeated measures ANOVA. Furosemide and Renal Function in Indomethacin-Treated Preterm Infants Neonatology 2010;98:191–199 197 furosemide group was 1 1.6 mg/dl at 21 days after intervention, all patients had serum creatinine ! 0.6 mg/dl at the time of hospital discharge. The incidence of oliguria did not differ between furosemide (4/29, 13.8%) and control (1/23, 4.3%) groups (p = 0.368). The incidence of hyperkalemia within 2 weeks was 6/29 (21%) in the furosemide group and 1/23 (4%) in the control group (p = 0.117). The maximum percent of weight loss (10.3 8 5.9 vs. 11.3 8 6.0%, p = 0.560) and days to regain birth weight (15 8 6 vs. 15 8 4 days, p = 0.538) did not differ between the furosemide and the control group. Although there was no correlation between serum cystatin C and creatinine levels before indomethacin therapy (r p = 0.105, p = 0.483), a strong correlation was observed at 0–12 h (r p = 0.629, p ! 0.001) and at 24–36 h (r p = 0.646, p ! 0.001, Pearson’s correlation) after the first course of indomethacin therapy. When serum cystatin C 1 2.0 mg/l at 24–48 h after intervention was used as an alternative marker of ARF [23] , the incidence of ARF was also significantly greater in the furosemide group (12/29, 41%) than in the control group (3/23, 13%) (OR 4.71, 95% CI 1.14–19.48, p = 0.033). Univariate analysis was performed among the variables listed in the table 1 to identify risk factors predictive of ARF after the first course of indomethacin. The incidence (or mean or median value) of each variable selected for multivariate analysis in the subgroup of renal data analyses with (n = 19) versus without (n = 33) ARF were as follows: female (12/19 vs. 11/33, p = 0.172), gestational age (27.7 8 2.7 vs. 29.3 8 2.6 weeks, p = 0.031), birth weight (1,002 8 367 vs. 1,189 8 421 g, p = 0.039), Apgar score at 5 min (median [interquartile range] 6 (5–7) vs. 6 (5–7), p = 0.041), intubation at birth (13/19 vs. 13/33, p = 0.044), indwelling of umbilical artery catheter for 7 days or more (9/19 vs. 8/33, p = 0.087) and furosemide use (16/19 vs. 13/33, p = 0.003). Multivariate analysis revealed that furosemide use (OR 12.28, 95% CI 2.38–63.39, p = 0.003) and intubation at birth (OR 4.41, 95% CI 1.00– 19.39, p = 0.049) were associated with significantly increased risks of ARF after the first course of indomethacin therapy. Discussion In the present study, routine furosemide administration after early indomethacin therapy was a significant risk factor for ARF during the initial weeks of life in preterm infants. The role of furosemide in preventing renal toxicity has been proposed by two previous studies with small sample sizes, but the results are less applicable to current practice because only a single dose of indomethacin was used and indomethacin therapy was delayed even to about 10 days after birth [5, 6] . Only the trial by Romagnoli et al. [7] in which three doses of indomethacin were administered at 12-hour intervals has shown the same result as ours, namely a reduced GFR in the furosemide group. The reason for these conflicting results is not clear. In a meta-analysis of the three controlled trials, patients were divided into normal or dehydration subgroups that were based on the serum BUN/creatinine ratio (BCR) before indomethacin therapy, categorized as BCR ! 20 or 1 20 mg/mg, respectively [24] . This study indicated that, although furosemide decreased GFR in the dehydration group, it improved GFR in the subgroup without dehydration. In the present study, however, furosemide decreased GFR despite the mean BCR of furosemide group of ! 20 mg/mg. Assessment of volume status by BCR only in preterm infants, particularly during the early days of life, may be inaccurate because serum BUN level is affected by protein intake of the infants and the serum creatinine level is also influenced by the maternal serum creatinine level. At the time of this writing, Andriessen et al. [25] reported an observational study assessing the interaction between furosemide and indomethacin in preterm infants in two different centers. The clinical setting and, in particular, the indomethacin dosing strategies were similar to those in our study. However, in this previous study, 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Serum creatinine (mg/dl) 0 0.941 0.85 0.85 2 5 8 11 14 <0.001 <0.001 0.005 0.150 0.562 1.08 1.10 0.96 0.76 0.71 1.41 1.62 1.28 0.90 0.78 Days after the first indomethacin dose Control Furosemide p values* Fig. 2. Changes in serum creatinine at representative days following the first indomethacin dose. * Data from the 52 patients (23 in the control group, 29 in the furosemide group) eligible for renal data analyses (see text). Data are presented as mean. Each rectangle and error bar represents mean and range of SD. Lee/Byun/Chung/Chang/Kim/Kim/Kim/ Pi Neonatology 198 2010;98:191–199 indomethacin was indicated only in patients with significant PDA diagnosed 48 h after birth, thus the mean starting day of indomethacin ( ; 3.5 day) was later than in the present study ( ; 1 day). The renal outcomes in the previous study were also similar to those reported here; a significant increase in serum creatinine concentration and a concomitant decrease in serum sodium concentration in the furosemide group. The latter result was thought to be caused by increased natriuresis, although FE Na was not reported. In our study, the FE Na increase in the furosemide group was almost double than in the control group, and this increase lasted until 24–36 h after the last furosemide infusion. This may have been caused, at least in part, by persistent furosemide-induced natriuresis, because the half-life of furosemide can be 6 24 h in particularly premature infants [26, 27] . However, two-way ANOVA showed that the delayed increase in FE Na (at 24– 36 h after indomethacin therapy) was associated with ARF rather than with furosemide infusion. Serum sodium concentrations did not differ between the control and the furosemide group (data not shown). We have demonstrated a role for increased cystatin C in the diagnosis of ARF in preterm infants and its strong correlation with the serum creatinine level after indomethacin therapy. Cystatin C is a low-molecular-weight protein and has recently been considered an alternative marker useful in the assessment of renal function. Cystatin C does not pass through the placenta, hence it is indicative of the GFR of neonates and is not affected by the mother’s GFR [28] . However, its advantage over serum creatinine levels in the estimation of true GFR of infants has not yet been established, particularly in neonates [23, 29] . In addition, the reference value for cystatin C in preterm infants has not yet been determined, although one previous study suggested upper limits of normal GFR of 2.10 and 1.95 mg/l at days 1 and 3 after birth [30] . Moreover, no study has yet examined whether increased serum cystatin C level can be a more useful marker in the diagnosis of intrinsic ARF than an increased serum creatinine level. A previous pediatric study demonstrated that serum cystatin C level was not influenced by prerenal factors in healthy children aged more than 1 year [31] . However, it is not known whether this result is applicable to preterm infants. Measurement of serum cystatin C levels beyond 36 h after indomethacin therapy was not included in the present study, so we do not know whether the peak serum cystatin C level coincides with that of serum creatinine. We found that furosemide infusion after indomethacin therapy did not increase the incidence of PDA, although the rate of primary PDA closure was approximately 10% higher in the control group than in the furosemide group. This finding differs from the results of a previous controlled trial comparing the effects of furosemide and chlorothiazide on PDA patency [4] , but it is in agreement with the results of the two studies using similar indomethacin and furosemide dosing strategies [5, 25] . The effects of furosemide-induced prostaglandin synthesis on patency of the ductus may have been offset by the antagonistic action of indomethacin [3] . This possibility is also supported by the results of a previous study on the dosing approach of indomethacin therapy followed by furosemide infusion in which the PDA closure rate was significantly greater in the group that received a higher dose of indomethacin [32] . Our finding that furosemide had no effect on the patency of PDA, however, may have been due simply to the small difference in PDA closure rate between the groups, despite the randomized study design. Even in the furosemide group, the primary PDA closure rate was about 85%. This relatively high rate might, at least in part, be attributable to the early initiation of indomethacin therapy [8, 10] , or to the extended indication of indomethacin for more mature preterm infants with respiratory distress. A recent review suggested more conservative guidelines for indomethacin therapy because prophylactic or early pre-symptomatic treatment may unnecessarily expose infants, particularly more mature and heavier infants with higher chance of spontaneous closure of PDA, to pharmacologic agents and their adverse effects [33] . Although about 18 and 41% of our patients weighed 1 1,500 and 1 1,000 g, respectively, all were indicated for indomethacin therapy due to significant PDA diagnosed by our study criteria, and there was no between-group difference in the distribution of these heavier infants. Our study had several limitations. First, the dropout rate was higher in the control group than in the furosemide group. This unexpected selection bias may have affected the characteristics of the control group by excluding infants with more serious conditions. Second, furosemide infusion after intervention days could not be completely controlled. After the first course of indomethacin therapy, patients who still had patent PDA may have increased chance to receive greater amounts of furosemide during subsequent courses of indomethacin and this may have affected the renal outcome. However, the cumulative doses of furosemide in each study group within 2 weeks after birth, other than indicated by the study protocol, was similar in the 52 patients eligible for renal data analyses. In addition, influence on renal outcomes by the subsequent doses of furosemide during the second or the third courses of indomethacin therapy, particularly in the Furosemide and Renal Function in Indomethacin-Treated Preterm Infants Neonatology 2010;98:191–199 199 patients of furosemide group, was negligible because most of PDA reopening occurred at least 10 days after birth. In conclusion, administration of furosemide after indomethacin increased the incidence of ARF in preterm infants without long-term renal morbidities. Furosemide infusion did not affect the therapeutic efficacy of indomethacin for PDA closure. These findings indicate that furosemide should not be administered routinely after each dose of indomethacin to preterm infants during the early neonatal period. Acknowledgements The authors specially thank all nurses of AMC NICU who helped to collect the blood and urine samples. The first author expresses his gratitude to Dr. Soo-Young Pi, the mentor for teaching him about neonatology, and Dr. Yong Choi for stimulating interest about neonatal nephrology.