Hypoproteinemia on the First Day of Life and Adverse Outcome in Very

Home , Hypoalbuminemia, Hypoproteinemia, Serum total protein

Journal of Perinatology (2012) 32, 520–524 r 2012 Nature America, Inc. All rights reserved. 0743-8346/12 www.nature.com/jp ORIGINAL ARTICLE Hypoproteinemia on the ﬁrst day of life and adverse outcome in very preterm infants admitted to the neonatal intensive care unit

S Iacobelli1,7, F Bonsante1, C Lacoutie`re1, C Ferdynus2, J Cottenet3, C Binquet4,5,6, C Quantin3,7 and JB Gouyon1,2 1Neonatal Intensive Care Unit, Department of Pediatrics, Dijon University Hospital, Dijon, France; 2Centre d’Epide´miologie des Populations (EA4184), Universite´ de Dijon, Dijon, France; 3CHRU, Service de Biostatistique et d’Informatique Me´dicale, Dijon, France; 4INSERM CIE1, Dijon, France; 5CHRU Dijon, Centre d’Investigation Clinique - Epide´miologie Clinique/Essais Cliniques, Dijon, France; 6Universite´ de Bourgogne, Dijon, France and 7INSERM U866, Dijon, France

Introduction Objective: We aimed to investigate the relationship between day-1 Total protein concentration is the major determinant of colloid hypoproteinemia and severe adverse outcome (SAO) in very preterm infants oncotic pressure (COP) and it has been regarded as predictor of admitted to the neonatal intensive care unit (NICU). morbidity and mortality in acute illness. In critically ill adults Study Design: Retrospective study of all patients born from 24 to 31 admitted to the intensive care unit, hypoproteinemia is actually weeks gestation and cared for in our NICU over an 8-year period. Infants recognized as a prognostic factor of poor outcome.1 Similar results were excluded if the serum protein value on the first day of life was not are reported for hypoalbuminemia.2 available. Even if it is recognized that albumin is the most abundant Result: A total of 913 patients were included. In all, 14.6% presented protein in blood plasma, constituting up to two-thirds of total 3,4 with SAO (death or severe neurological injury on cranial ultrasound). protein in adults as in newborn babies, there is some evidence Hypoproteinemia (total protein level <40 g lÀ1) on day 1 of life occurred that in critically ill patients total protein concentration may reflect 4 in 19.5 % of all patients. The rate of SAO was 33.7% in patients with COP better than serum albumin concentration, thus representing hypoproteinemia and 9.9% in those with normoproteinemia (P<0.0001). one more interesting predictor for poor outcome. Logistic and multiple regression analysis confirmed that the association In newborn babies, the relationship between low serum protein hypoproteinemia–SAO remained significant after adjustment for the and unfavorable outcome has never been explored, while other major predictors of outcome present at baseline (odds ratio 3.4; 95% hypoalbuminemia has been related to increased morbidity and confidence interval 2.1–5.4; P<0.0001). mortality in some specific illness, such as gastrointestinal diseases.5,6 One retrospective report over a small sample size of very Conclusion: Hypoproteinemia was highly associated with SAO in this low birth weight (BW) infants has shown a significant association cohort of critically ill preterm infants. We are unable to explain the link between low albumin levels and mortality. In this study, the between hypoproteinemia and adverse outcome in our population. This recorded albumin measurements were not realized at standardized investigation serves as a hypothesis-generating report of a large preterm times, with some being performed more than a week after birth.7 infants sample, and suggests the need to assess the predictive accuracy for These observations have led us to perform a study to determine adverse outcome of hypoproteinemia in future prospective studies. the relationship between hypoproteinemia on the first day of life Journal of Perinatology (2012) 32, 520–524; doi:10.1038/jp.2011.137; and severe adverse outcome (SAO) in a large cohort of critically ill published online 29 September 2011 infants born below 32 weeks gestation. Keywords: very low birth weight infant; morbidity; mortality; total protein; hypoalbuminemia; neurological injury Patients and methods Design and study population Correspondence: Dr S Iacobelli, Neonatal Intensive Care Unit, Department of Pediatrics, The study design was an observational retrospective cohort analysis Dijon University Hospital, 10 Bd marećhal de Lattre de Tassigny, 21079 Dijon, France. of all the infants born from 24 to 31 weeks of gestational age (GA) E-mail: [email protected] Received 27 May 2011; revised 3 August 2011; accepted 29 August 2011; published online and admitted to the neonatal intensive care unit (NICU) of Dijon 29 September 2011 University Hospital during an 8-year period (1 January 2001 to 31 Hypoproteinemia in preterm infants S Iacobelli et al 521

December 2008). According to previous published guidelines for Total protein values assessing fluid and electrolyte status in very low BW infants,8 all Laboratory values for serum total protein were available from babies born <32 weeks of GA and hospitalized in our NICU had a the infants’ medical records. blood sample at 12 h of life. This blood sample systematically Hypoproteinemia was defined as a total protein level of provided the serum total protein measurement. <40 glÀ1 on the first day of life.16 All the infants admitted to the NICU within 6 h after birth were eligible. Statistical analysis According to the NICU nutritional practices, all these infants The hypoproteinemic patients were compared with the had a total parenteral nutrition started into the first hours of life. normoproteinemic ones with respect to SAO. Patients were excluded if the serum protein value on the first Univariate analysis using one-factor analysis of variance was day of life was not available or if they had major congenital performed to investigate perinatal variables associated with malformations. hypoproteinemia. Variables significant at a P-level <0.20 at the univariate analysis were entered into a backward selection analysis Outcome measures of variance. SAO was defined as death and/or severe (grade 3 or 4) Univariate analysis was performed to estimate the association of intraventricular hemorrhage (IVH) and/or cystic periventricular SAO with all the other available perinatal variables and with leukomalacia (c-PVL) occurring before infant’s discharge from postnatal diseases occurring on the first day of life (RDS requiring hospital. IVH was graded at cerebral ultrasound according to Papile surfactant, EOS and birth anemia). et al.9 c-PVL was defined by cranial ultrasound as an area of To estimate the independent influence of hypoproteinemia on increased echogenicity of the periventricular white matter in acute SAO, we realized a logistic multiple regression model. The model phase, which subsequently evolved into cystic lesion.10 Cranial used eight predictor variablesFhypoproteinemia and seven ultrasounds were routinely performed during the infant hospital stay potentially confounding baseline variables. These were selected by experienced examiners (neonatologists or radiologists) according because they represented either factors identified as associated with to the following protocol: day 1, 3, 7, 10, 15 and then at least every SAO at a P-level <0.20 in the univariate analysis or as variables 2 weeks or more often as clinically indicated, until discharge. independently associated with hypoproteinemia, or both. We used Pearson w2 or Fischer exact test for qualitative Data collection variables, and one-way analysis of variance or Mann–Whitney test Clinical data. Clinical data on the population were obtained for comparing quantitative variables distributions between infants from the regional perinatal database of the Burgundy Perinatal groups. The multivariate regression model was realized using a Network.11,12 This perinatal database was set up with the approval backward elimination approach. The adjusted odds ratio and their of the National Committee of Informatics and Liberty, and confidence intervals were estimated. prospectively records clinical events for mothers and infants Statistical analyses were performed using SAS software 8.2 (SAS between birth and hospital discharge. Information was collected Institute, Cary, NY, USA). All hypotheses were tested at the 2-tailed about antenatal history: maternal age, hypertensive disease of 0.05 significance level. pregnancy, placental diseases (placental abruption, placenta praevia), preterm prelabor rupture of membranes, threatened preterm labor, diabetes, antenatal steroids administration, singleton Results or multiple pregnancy, clinical chorioamnionitis and mode of Study population delivery. The following data about birth and postnatal diseases were Among the 1067 infants admitted to the NICU during the study also collected: GA, BW, gender, appropriateness of BW for GA period, none had major congenital malformations, and total 13 (according to reference Hadlock FP et al. ), transfer after birth protein value on the first day of life was not available for 154 (outborn), 1 min Apgar score, anemia at birth (venous hemoglobin infants (14%). So, the study population included 913 infants. <13.3 g dlÀ1 at the admission in NICU),14 early onset sepsis (EOS), respiratory distress syndrome (RDS) requiring surfactant, Outcome and clinical data acute renal failure,15 hemodynamically significant patent ductus SAO was present in 14.6% of the study population, with 8.2% of arteriosus, late onset sepsis, necrotizing enterocolitis and death and 6.4% of survival with severe cerebral ultrasound findings bronchopulmonary dysplasia. The GA in completed weeks was (IVH grade 3–4 and/or c-PVL). Death occurred during the first assessed on the basis of the mother’s last menstrual period, and week of life in 3.2% of patients and later during the hospital stay confirmed or modified when necessary by routine early antenatal in 4.9%. ultrasound examination. GA was divided into three categories with Table 1 shows perinatal characteristics of the study population, cutoff points at 25, 27 and 31 weeks gestation. characteristics at birth and postnatal diseases.

Journal of Perinatology Hypoproteinemia in preterm infants S Iacobelli et al 522

Table 1 Characteristics of the study population (913 infants born at 24–31 weeks gestation)

Characteristics %

Prenatal characteristics Multiple pregnancy 31.3 Hypertensive disease of pregnancy 22.1 Maternal diabetes 5.4 PPROM 31.5 Placental disease 7.2 Clinical chorioamnionitis 18.8 Threatened preterm labour 51.3 Antenatal steroids 85.7 Figure 1 Protein levels on the first day of life according to the outcome in the Characteristics at birth study population. Data are expressed as median, 101,251,751 and 901 percentile. Birth weight (g) 1184±324a General P<0.001 (analysis of variance for repeated measures). *P<0.05 versus survival without severe abnormal ultrasound (US). 1P<0.05 versus survival with GA (weeks) 28.8±1.9a severe abnormal US and late death. Shaded circles and squares are minimal, GA (24–25 weeks) 5.6 maximal and median values. GA (26–27 weeks) 17.3 GA (28–31 weeks) 76.9 Small for GA 38.5 Table 2 Variables associated with hypoproteinemia at multivariate analysis in Male gender 55.5 the study population (913 infants born at 24–31 weeks gestation) Apgar score <3 at 1 min of life 11.1 Caesarean section/vaginal delivery 69.7/30.3 Variables OR 95% CI P Outborn 12.1 Multiple pregnancy 2.3 1.5–3.6 0.001 Antenatal steroids 0.5 0.3–0.8 0.007 Postnatal diseases GA 24–25 versus 28–31 weeks 6.05 2.8–12.9 <0.0001 RDS requiring surfactant 83.5 GA 26–27 versus 28–31 weeks 4.7 3.0–7.3 <0.0001 Early/late onset sepsis 4.1/17 Male gender 1.9 1.2–2.8 0.0015 Anemia at birth 12.7 Apgar score <3 at 1 min of life 1.8 1.0–3.0 0.04 Acute renal failure 12 Birth anemia 2.1 1.3–3.6 0.005 HsPDA requiring treatment 31.7 RDS requiring surfactant 2.1 1.0–4.2 0.02 Necrotizing enterocolitis 2.3 BPD at 36 weeks 20.2 Abbreviations: CI, confidence intervals; GA, gestational age; OR, odds ratio; RDS, respiratory distress syndrome. Abbreviations: BDP, bronchopulmonary dysplasia; GA, gestational age; HsPDA, hemodynamically significant patent ductus arteriosus; PPROM, preterm prelabor rupture of membranes; RDS, respiratory distress syndrome. aValues are mean±s.d. Serum total protein values (mean±s.d.) on day 1 were significantly different in infants who died during the first week of life (34.8±6.6 g lÀ1), compared with infants who died later Patients included in the analysis were not different with regard during hospital stay (41.8±5.4 g lÀ1), infants surviving with to GA, BW and frequency of SAO when compared with those severe abnormal ultrasound (42.3±5.4 g lÀ1) and infants excluded owing to missing data for serum protein values surviving without severe abnormal ultrasound (45.7±6.3 g lÀ1) (data not shown). (Data shown on Figure 1). Variables associated with hypoproteinemia at the univariate Hypoproteinemia, total protein values and variables analysis were: multiple pregnancy, hypertensive disease of associated with SAO pregnancy, threatened preterm labor, lack of antenatal steroids, Hypoproteinemia on day 1 of life occurred in 19.5% of all patients. low GA, small for GA BW, male gender, Apgar score <3 at 1 min of Protein values (mean±s.d.) were 36.06±3.33 g lÀ1 in life, cesarean section, outborn, EOS and RDS requiring surfactant hypoproteinemic and 47.15±5.26 g lÀ1 in normoproteinemic (data not shown). patients. The blood sample was realized at 12.3±5.9 Variables that remained associated with hypoproteinemia at (mean±s.d.) h after birth. the multivariate analysis are detailed in Table 2. The rate of SAO was 33.7% in patients with hypoproteinemia Clinical factors significantly associated with SAO at the and 9.9% in those with normoproteinemia (P<0.0001). univariate analysis were: lack of antenatal steroids, birth without

Journal of Perinatology Hypoproteinemia in preterm infants S Iacobelli et al 523

Table 3 Variables associated with severe adverse outcome at multivariate are recognized risk factors for cerebral hemorrhage and ischemia analysis in the study population (913 infants born at 24–31 weeks gestation) in very low BW infants.20,21 The finding that serum protein levels Variables OR 95% CI P were lower in infants who died early during hospital stay supports this hypothesis. GA 24–25 versus 28–31 weeks 6.1 3.0–12.5 <0.0001 Regrettably, the lack of information about time of cord GA 26–27 versus 28–31 weeks 2.0 1.2–3.4 <0.0001 clamping and hypotension requiring treatment by volume Hypoproteinemia on day 1 of life 3.4 2.1–5.4 <0.0001 expansion during the first hours of life does not allow any Cesarean section 0.5 0.3–0.7 0.001 conclusion with respect of this point, and this is a limitation of our Apgar score < 3 at 1 min of life 2.3 1.4–4.0 0.002 study. Abbreviations: CI, confidence intervals; GA, gestational age; OR, odds ratio. Our study cannot draw conclusions about the cause underlying hypoproteinemia. As reported by previous papers,16 low protein levels were cesarean section, low GA, 1 min Apgar score <3, anemia at birth, associated with lower GA in our population. RDS requiring surfactant and EOS (data not shown). It is also interesting to note that the lack of antenatal At the multivariate analysis, hypoproteinemia on day 1 of life corticosteroids was an independent variable associated with remained a strong independent factor positively associated with hypoproteinemia in our cohort, and this is consistent with the SAO. Higher GA, cesarean section and Apgar score >3 at 1 min of results of Bunt et al., showing that albumin synthesis in preterm life showed a significant protective effect (results are shown in infants on the first day of life tended to increase after Table 3). administration of antenatal corticosteroids.22 The same authors have found that albumin synthesis rates were significantly lower in preterm babies with intrauterine growth retardation, and this is in Discussion contrast with our results that show no difference with regard to the This is to our knowledge the first paper showing an independent occurrence of hypoproteinemia in small for GA babies when significant association between low protein levels on the first day of compared with babies appropriate for GA. life and adverse outcome in very preterm infants. Of course, the retrospective design of this study restricts the Of course, the existence of this correlation does not establish interpretation of our results; but in spite of the above limitations, causality, but demonstrating such an association is the first step in the strong association between hypoproteinemia and severe determining whether or not a causal relationship exists. impaired outcome depicted in this large population remains valid. There is evidence that low serum protein levels accompany Moreover, our data raise the question whether there is an acute RDS in critically ill adults: in this population, the higher interest in increasing total protein levels in critically ill preterm mortality related to hypoproteinemia is driven by increased RDS infants admitted to the NICU. In our series, total protein severity.1 In our series, hypoproteinemia was associated to determinations were available in the absence of serum albumin increased rate of RDS requiring surfactant, but this latter did not levels and this represents a further limitation of the present study. represent a risk factor for SAO, so a different mechanism by which However, we know that in newborn infants born preterm, albumin low serum protein could affect outcome may be involved. accounts for >60% of the total blood body protein,4,23 so that we One of the main physiological properties of serum protein can suppose that in our population protein levels paralleled resides on the maintenance of volemia by influencing the COP, albumin levels. and the relationship of COP with total protein is stronger in Nowadays, the use of albumin as a fluid for volume critically ill patients than in stable ones.4 In newborn babies, a replacement for hypotension or as a treatment for low serum significant linear correlation between COP and total protein in cord albumin levels in premature infants is an ongoing debate and it blood has been described by Kero et al.,17 and even in sick has been the subject of evidence-based reviews.24,25 Whether neonates, COP is correlated with total protein during the first days albumin infusion would be beneficial in some circumstances in of life, as reported by Bhat et al.18 In one previous investigation in preterm babies is still controversial25–27 and our findings further newborn babies with RDS, we have described a significant positive raise this issue. correlation between COP, total protein and hypotension.19 Of course, additional investigations are warranted in order to Based on all above, we can suppose that in our population answer to this and to other questions that arise as a consequence of protein levels reflected plasma volemia, and the decreased oncotic our results. In an attempt to better explore the associated risk factor pressure could have impaired maintenance of intravascular volume that we have identified in the current work, we suggest several and adequate blood flow to vital organs. This could also explain proposals for future prospective studies: (1) identify the predictive the association between hypoproteinemia and severe neurological accuracy for adverse outcome of hypoproteinemia compared with injury as both subclinical hypoperfusion and neonatal hypotension other valid risk-adjusted scores of illness severity in very preterm

Journal of Perinatology Hypoproteinemia in preterm infants S Iacobelli et al 524 infants; (2) consider the potential mechanism by which low total 12 Quantin C, Allaert FA, Gouyon B, Cohen O. Proposal for the creation of a European protein may affect outcome in preterm babies and (3) investigate healthcare identifier. Stud Health Technol Inform 2005; 116: 949–954. whether albumin infusion may have potential benefits in preterm 13 Hadlock FP, Harrist RB, Martinez-Poyer J. In utero analysis of fetal growth: a infants presenting with early hypoproteinemia. sonographic weight standard. Radiology 1991; 181(1): 129–133. 14 Doyle JJ, Schmidt B, Blanchette V, Zipursky A. Hematology In: Avery GB, Fletcher MA, Mac Donald MG (eds). Neonatology, Pathophysiology and Management of the Newborn. Lippincott Williams & Wilkins: Philadelphia, 1999, pp 1045–1092. Conflict of interest 15 Choker G, Gouyon JB. Diagnosis of acute renal failure in very preterm infants. Biol The authors declare no conflict of interest. Neonate 2004; 86(3): 212–216. 16 Reading RF, Ellis R, Fleetwwod A. Plasma albumin and total protein in preterm babies from birth to eight weeks. Early Hum Dev 1990; 22(2): 81–87. References 17 Kero P, Korvenranta H, Alamaakala P, Sela¨nne P, Kiilholma P, Va¨lima¨ki I. Colloid osmotic pressure of cord blood in relation to neonatal outcome and mode of delivery. 1 Mangialardi MJ, Martin GS, Bernard GS, Wheeler AP, Christman BW, Dupont WD et al. Acta Paediatr Scand Suppl 1983; 305: 88–91. Hypoproteinemia predicts acute respiratory distress syndrome development, weight 18 Bhat R, Javed S, Malalis L, Vidyasagar D. Critical care problems in neonates. gain, and death in patients with sepsis. Crit Care Med 2000; 28(9): 3137–3145. Colloid osmotic pressure in healthy and sick neonates. Crit Care Med 1981; 9(8): 2 Vincent Jl, Dubois MJ, Navickis RJ, Wilkes MM. Hypoalbuminemia in acute illness: is 563–567. there a rationale for intervention? A meta-analysis of cohort studies and controlled 19 Zimmermann B, Françoise M, Germain GF, Lallemant C, Gouyon JB. Colloid trials. Ann Surg 2003; 237: 319–334. oncoticpressurein neonatal respiratory distress syndrome. Arch Pediatr 1997; 4: 3 Horowitz IN, Ta K. Hypoalbuminemia in critically ill children. Arch Pediatr Adolesc 952–958. Med 2007; 161(11): 1048–1052. 20 Verhagen EA, Ter Horst HJ, Keating P, Martijn A, Van Braeckel KN, Bos AF. Cerebral 4 Uhing MR. The albumin controversy. Clin Perinatol 2004; 31: 475–488. oxygenation in preterm infants with germinal matrix-intraventricular hemorrhages. 5 Kenny SE, Pierro A, Isherwood D, Donnell SC, Van Saene HK, Lloyd DA et al. Stroke 2010; 41(12): 2901–2907. Hypoalbuminaemia in surgical neonates receiving parenteral nutrition. J Pediatr Surg 21 Watkins AM, West CR, Cooke RW. Blood pressure and cerebral haemorrhage 1995; 30(3): 454–457. and ischaemia in very low birthweight infants. Early Hum Dev 1989; 19(2): 6 Atkinson SD, Tuggle DW, Tunell WP. Hypoalbuminemia may predispose infants to 103–110. necrotizing enterocolitis. J Pediatr Surg 1989; 24(7): 674–676. 22 Bunt JE, Rietveld T, Schierbeek H, Wattimena JL, Zimmermann LJ, van Goudoever JB. 7 Morris I, McCallion N, El-Khuffash A, Molloy EJ. Serum albumin and mortality in very Albumin synthesis in preterm infants on the first day of life studied with [1-13C]. Am J low birth weight infants. Arch Dis Child Fetal Neonatal Ed 2008; 93(4): F310–F312. Physiol Gastroinest Liver Physiol 2007; 292(4): G1157–G1161. 8 Gouyon JB, Bouziane M, Semama DS. Objectifs des apports hydrosode´s dans la 23 Scott PH, Wharton BA. Biochemical values in the newborn In: Roberton RNC (ed). premie`re semaine de vie du pre´mature´: comment les atteindre? In: Gouyon JB, Textbook of Neonatology. Roberton: New York, 1992, pp 1213–1227. Guignard JP, Simeoni U (eds). Equilibre hydroe´lectrolytique du grand pre´mature´. 24 Jardine LA, Jenkins-Manning S, Davies MW. Albumin infusion for low serum albumin Arnette: Paris, 1997, pp 101–118. in preterm newborn infants. Cochrane Database Syst Rev 2004; (3): CD004208. 9 Papile LA, Burstein J, Burstein R, Koffler H. Incidence and evolution of subependymal 25 Lynch SK, Mullet MD, Graeber JE, Polak MJ. A comparison of albumin-bolus therapy and intraventricular hemorrhage: a study of infants with birth weights less than versus normal saline-bolus therapy for hypotension in neonates. J Perinatol 2008; 1500 gm. J Pediatr 1978; 92(4): 529–534. 28(1): 29–33. 10 deVries LS, Eken P, Dubowitz LMS. The spectrum of leukomalacia using cranial 26 Ghirardello S, Mosca F. Albumin versus saline: an open question. J Perinatol 2008; ultrasound. Behav Brain Res 1992; 49:1–6. 28(8): 586. 11 Cornet B, Gouyon JB, Binquet C, Sagot P, Ferdynus C, Metral P et al. Using discharge 27 Bignall S, Bailey PC, Bass CA, Cramb R, Rivers RP, Wadsworth J. The cardiovascular abstracts as a tool to assess a regional perinatal network. Rev Epidemiol Sante and oncotic effects of albumin infusion in premature infants. Early Hum Dev 1989; Publique 2001; 49(6): 583–593. 20(3–4): 191–201.

Journal of Perinatology