537523
research-article2014
PENXXX10.1177/0148607114537523Journal of Parenteral and Enteral NutritionLaborie et al
Brief Communication
Shielding Parenteral Nutrition Solutions From Light: A Randomized Controlled Trial Sophie Laborie, MD1,2; Angélique Denis, MS3; Gilles Dassieu, MD4; Antoine Bedu, MD5; Pierre Tourneux, MD6; Didier Pinquier, MD7; Elsa Kermorvant, PhD, MD8,9; Véronique Millet, MD10; Serge Klosowski, MD11; Hugues Patural, MD, PhD12; Catherine Clamadieu, MD1; Anne Brunhes, MD13; Marie Walther, MD14; Isabelle Jaisson-Hot, MD3; Bruno Mandy15; and Olivier Claris1,2,16
Journal of Parenteral and Enteral Nutrition Volume 39 Number 6 August 2015 729–737 © 2014 American Society for Parenteral and Enteral Nutrition DOI: 10.1177/0148607114537523 jpen.sagepub.com hosted at online.sagepub.com
Abstract Introduction: Oxidant stress is implicated in the pathogenesis of bronchopulmonary dysplasia (BPD). Light induces peroxide generation in parenteral nutrition (PN) solutions, creating an oxidant stress. Shielding PN from light decreases its peroxide content, which has nutrition and biochemical benefits in animals and humans. This study aims at determining whether full light protection of PN decreases the rate of bronchopulmonary dysplasia and/or death in very low-birth-weight infants. Methods: Multicenter randomized controlled trial of photoprotection, using amber bags and tubing initiated during compounding of PN and maintained throughout infusion in the light-protected (LP) group. The control group (light exposed [LE]) received PN exposed to ambient light. Depending on centers, lipids were infused either separately or as all-in-one PN. Results: In total, 590 infants born 30% supplemental oxygen or the need for positive-pressure support or, in the case of infants requiring less than 30% oxygen, the need for any supplemental oxygen after an attempt at withdrawing oxygen.24 Other end points included death, intraventricular hemorrhage, retinopathy, periventricular leukomalacia, necrotizing enterocolitis, and infection. Sequential transfontanelle cranial ultrasounds were examined and interpreted by an independent reviewer.
Intraventricular hemorrhages were graded according to the Papile scale.25 Periventricular leukomalacia was classified as described by de Vries et al.26 All stages of retinopathy were recorded according to the international classification.27 Necrotizing enterocolitis was classified according to Bell’s stages.28
Sample Size Assuming an incidence of BPD/death at 28 days of life of 55%,29,30 a total sample size of 460 infants would be required to detect a 25% relative reduction in the risk of the primary composite end point with 80% power. The 25% relative reduction was chosen based on the 30% decrease measured in a previous post hoc analysis.20 To allow for potential withdrawal from the study, the sample size was increased by 25%.
Statistical Analysis Analyses were conducted according to the intention-to-treat (ITT) principle using the Statistical Analysis System (SAS) software version 9.1 (SAS Institute, Cary, NC). They were followed by per-protocol (PP) analyses. Baseline characteristics and outcomes variables were compared in the 2 study groups using χ2 and 2-tailed t tests or the Wilcoxon test. Outcome variables were analyzed using a 3-step multivariate analysis as follows. First, to select independent variables associated with the risk of BPD or death, a linear logistic regression model was applied to all clinically pertinent preselected covariables of interest (photoprotection, gestational age, birth weight, sex, antenatal steroids, Clinical Risk Index for Babies [CRIB], chorioamnionitis, caffeine exposure before inclusion, invasive ventilation before inclusion, maximal bilirubin, erythropoietin use, surgery for patent ductus arteriosus, duration of exposure to PN with lipids and/or vitamins, and type of PN: all-in-one or lipids infused separately). Second, a 1-way analysis of variance was conducted to examine the distribution of model residuals between the study centers. Third, a mixed-effects logistic regression model using study center as a random effect was developed to predict the risk of BPD or death after controlling for the photoprotection of PN solution adjusted for the other selected covariables. The mixed-effects logistic regression model was fit using the SAS PROC NLMIXED application.31 A Wald statistic was used to test the significance of each coefficient in the model. Comparisons of MDA and cysteine concentrations between PN solutions exposed to or protected from ambient light were performed using the Wilcoxon signed rank test. A P value 2, and those transfused before randomization. The MDA and cysteine data suggest that photoprotection is technically feasible and that we had overall good compliance.
Downloaded from pen.sagepub.com at UNIVERSITE DE MONTREAL on July 30, 2015
734
Journal of Parenteral and Enteral Nutrition 39(6)
Table 2. Outcome of the Infants Depending on the Group of Randomization. Characteristic
Light Protected (n = 293)
BPD or death At 28 d At 36 wk GA BPD At 28 d At 36 wk GA Death At 28 d At 36 wk GA Intraventricular hemorrhage None Grades 1–2 Grades 3–4 Periventricular leukomalacia None Grades 1–2 Grades 3–4 Necrotizing enterocolitis Retinopathy of prematurity Retinopathy of prematurity ≥III Sepsis episodes 0 1 2 3 and more Volume of blood transfused, mean ± SD, mL Duration of exposure to PN with lipids or vitamins, mean ± SD, d Cholestasis If yes, maximal bilirubin, umol/l, median [Q1-Q3]
Light Exposed (n = 294)
227 (77.5) 89 (30.4)
213 (72.4) 80 (27.2)
217/283 (76.7) 69/274 (25.2)
193/274 (70.4) 55/269 (20.4)
10 (3.4) 16 (5.5)
20 (6.8) 25 (8.5)
255 (90.1) 24 (8.5) 4 (1.4)
247 (88.2) 29 (10.4) 4 (1.4)
263 (96.0) 9 (3.3) 2 (0.7) 19 (6.5) 10 (3.9) 2 (0.8)
264 (97.8) 5 (1.9) 1 (0.4) 24 (8.1) 10 (3.9) 2 (0.8)
136 (46.4) 120 (40.9) 27 (9.2) 10 (3.4) 46.0 ± 29.4 28.1 ± 13.5
145 (49.3) 100 (34.0) 37 (12.6) 10 (3.4) 43.3 ± 31.1 27.7 ± 13.6
14 (5) 147 [57–173]
P Value
12 (4) 109 [82–137]
.16 .35 .09 .09 .06 .15 .76 .56 .44 .65 .30 .42 .56 .695 .591
Values are presented as number (%) unless otherwise indicated. BPD, bronchopulmonary dysplasia; GA, gestational age; PN, parenteral nutrition.
But it does not inform on how effective the technique was for individual study participants or on the magnitude of the oxidant stress for each infant. An individual assessment of oxidant stress in a nested cohort of “healthy” preterm infants was planned. Due to insufficient inclusion, analysis was not feasible. The heterogeneity of PN protocols, especially of vitamins protocols, accounts for the fact that infants of different centers might have been exposed to varying levels of oxidant load from PN. However, the stratified design limits the effect of this heterogeneity on the results. The difference in MDA levels measured between LE and LP solutions is lower than that previously published by Picaud et al.13 This could be explained partly by differences in composition of the tested PN solutions. It could also be due to differences between centers in the use of multivitamins. Indeed, Lavoie demonstrated lower levels of peroxides with vitalipide and soluvit than with cernevit in photoexposed and photoprotected PN (unpublished data). No center used intravenous iron saccharate, a further source of peroxidation that accounts for an important increase in MDA.36
In healthy preterm neonates, the transition to extrauterine life is associated with an imbalance between the oxidative load and immature antioxidant defenses,37 as demonstrated by an increase in urinary markers of oxidative stress38 compared with term infants. In extremely low-birth-weight infants who underwent resuscitation at birth, exposure to 100% oxygen was associated with higher rates of BPD and urinary markers of oxidant stress compared with 30%.7 Cord blood glutathione, the intracellular central antioxidant implicated in peroxide detoxication, is increased in preterm neonates compared with term infants. However, the concentration falls rapidly after birth, with a rise after amino acid infusion.39 The consequences of infusing peroxides may be more deleterious during this critical period of low glutathione level than later. The negative results observed in this study may therefore also be due to the design, in which, during the first days of life before parental consent, all infants received photoprotected PN or PN without vitamins or lipids. If oxidant-related determinants of BPD are effective very early in life, the difference in outcome between the study
Downloaded from pen.sagepub.com at UNIVERSITE DE MONTREAL on July 30, 2015
Laborie et al
735
Table 3. Analysis of the Rate of Death or BPD at 28 Days Using a Logistic Regression Model (n = 554).
Table 5. Analysis of the Rate of Death or BPD at 28 Days Using a Mixed-Effects Logistic Regression Model (n = 587).
Characteristic
Characteristic
Photoprotection Gestational age, wk Male sex Birth weight, g Chorioamnionitis Antenatal steroids CRIB Invasive ventilation before inclusion Maximal bilirubin, µmol/L Erythropoietin Duration of exposure to PN with lipids or vitamins, d PN (all-in-one vs lipids separately)
Adjusted OR (95% CI)
P Value
1.21 (0.74–1.98) 0.36 (0.26–0.49) 2.13 (1.28–3.53) 0.82 (0.67–0.91)b 0.97 (0.34–2.79) 0.90 (0.42–1.93) 1.03 (0.89–1.21)c 1.72 (0.92–3.19)
.44