- Email: [email protected]
Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?
Clinical Nutrition xxx (2015) 1e7
Contents lists available at ScienceDirect
Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu
Original article
Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia? lia Dreyfus a, Ce line Julie Fischer Fumeaux b, a, Laurent Remontet c, Le Murielle Christine Essomo Megnier Mbo Owono a, Sophie Laborie a, Delphine Maucort-Boulch c, d, e, Olivier Claris a, e, f, * ^pital Femme M Hospices Civils de Lyon, Service de R eanimation N eonatale et N eonatologie, Ho ere Enfant, Bron, France Centre Hospitalier Universitaire Vaudois et Universit e de Lausanne, Service de N eonatologie, D epartement M edico-Chirurgical de P ediatrie, Switzerland Hospices Civils de Lyon, Service de Biostatistique, Lyon, France d CNRS UMR5558, Laboratoire de Biom etrie et Biologie Evolutive, Equipe Biostatistique Sant e, Pierre-B enite, France e Universit e Claude-Bernard, Lyon, France f Equipe d'Accueil Mixte EAM 4128, Lyon, France a
b c
a r t i c l e i n f o
s u m m a r y
Article history: Received 23 November 2014 Accepted 27 July 2015
Background & aims: Hyperglycemia occurs in more than half of the extremely low birth weight (ELBW) neonates during the first weeks of life, and is correlated with an increased risk of morbi-mortality. Hypophosphatemia is another frequent metabolic disorder in this population. Data from animal, adult studies and clinical observation suggest that hypophosphatemia could induce glucose intolerance. Our aim was to determine whether a low phosphatemia is associated with hyperglycemia in ELBW neonates. Methods: This observational study included ELBW infants admitted in a tertiary neonatal care center (2010e2011). According to the center's policy, they received parenteral nutrition from birth and human milk from day 1. Phosphatemia and glycemia were measured routinely during parenteral nutrition. Hyperglycemia was defined by two consecutives values >8.3 mmol/L (150 mg/dL). Statistical analysis used a joint model combining a mixed-effects and a survival submodels to measure the association between phosphate and hyperglycemia. Results: The study included 148 patients. Mean gestational (Standard Deviation) age was 27.3 (1.6) weeks; mean birth weight was 803 (124) grams; 57% presented hyperglycemia. The multivariate joint model showed that the hazard of hyperglycemia at a given time was multiplied by 3 for each 0.41 mmol/L decrease of phosphate level at this time (p ¼ 0.002) and by 3.85 for the same decreased of phosphate the day before (p ¼ 0.0015). Conclusion: To our knowledge, this is the first study suggesting that low phosphatemia can be associated with hyperglycemia in ELBW neonates. Further studies will have to demonstrate whether better control of phosphatemia could help in preventing hyperglycemia. © 2015 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Keywords: Extremely low birth weight infant Preterm neonate Metabolic disorder Hypophosphatemia Hyperglycemia Insulin
1. Introduction
^ pital Femme Me re Enfants, * Corresponding author. Hospices Civils de Lyon, Ho onatologie et Re animation Ne onatale, 59 Boulevard Pinel, 69500 Bron, Service de Ne France. Tel.: þ33 4 27 85 52 83; fax: þ33 4 27 86 92 27. E-mail addresses: [email protected] (L. Dreyfus), celine-julie.fischer@ chuv.ch (C.J. Fischer Fumeaux), [email protected] (L. Remontet), [email protected] (S. Laborie), [email protected] (D. Maucort-Boulch), [email protected] (O. Claris).
Hyperglycemia is a frequent metabolic complication in critically ill neonates and was found to occur in more than a half of extremely low birth weight (ELBW) infants [1]. The frequency of this complication is inversely related to gestational age (GA) and birth weight (BW) [2]. Beside to induce an osmotic diuresis, dehydration, hyponatremia, hypokalemia and acidosis, studies suggest that hyperglycemia increases mortality [1e4] and several serious morbidities as sepsis, intraventricular hemorrhage [5] or retinopathy of prematurity [6,7]. Both insulin resistance and failure in pancreatic
http://dx.doi.org/10.1016/j.clnu.2015.07.019 0261-5614/© 2015 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019
2
L. Dreyfus et al. / Clinical Nutrition xxx (2015) 1e7
List of abbreviations VLBW ELBW ATP NICU GA BW SGA
Very Low Birth Weight Extremely Low Birth Weight Adenosine Tri-Phosphate Neonatal Intensive Care Unit Gestational Age Birth Weight Small for Gestational Age
islet b-cell processing of pro-insulin could contribute to hyperglycemia in extremely preterm infants [8], but the pathogenesis remains to be elucidated. The identification of risk factors and mechanisms associated to glucose intolerance is thus especially important in the context of neonatal care [1e3,9]. There is so far no consensus on safe and efficient management of glucose intolerance, as both insulin treatment and reduction in glucose intakes can be deleterious [10e13]. Hypophosphatemia is another frequent metabolic disorder in premature newborns, especially in very low birth-weight infants (VLBW) and in small for gestational age (SGA) neonates, although normal values at this age remain uncertain [14e17]. In the critically ill children, it has been associated with a longer duration of mechanical ventilation and with a longer duration of stay in the pediatric intensive care unit [18]. In the preterm newborn, hypophosphatemia is frequently associated to postnatal growth deficiency and osteopenia of the preterm [19,20] and has been recently incriminated in susceptibility for septicaemia in VLBW [21]. Various mechanisms have been raised, like limited storage, insufficient intakes in the first days or a re-feeding like syndrome in VLBW undergoing enhanced nutrition [22]. Due to our common clinical observation of hypophosphatemia preceding severe and persistent hyperglycemia, we drafted the hypothesis that hypophosphatemia could play a role in glucose intolerance in the preterm neonate. This hypothesis was sustained by some experimental studies and adult data, where hypophosphatemia could be involved in insulin resistance found in metabolic syndrome [23,24]. Two mechanisms have been involved; first, hypophosphatemia leads to a low adenosine triphosphate (ATP) intracellular rates; in the beta cells of the pancreas, the consequent dysfunction in ATP-ase activity may lead to a reduction in the insulin production [8,25]. Secondly, phosphorus is involved in phosphorylation of the insulin cell-receptor, which could thus be less efficient in case of low phosphatemia [25]. Whether hypophosphatemia is a risk factor for hyperglycemia in preterm neonates remains to be elucidated [26]. The aim of our study was to further investigate the association of low phosphate and the risk of glucose intolerance, defined as persistent hyperglycemia >8.3 mmol/L (>150 mg/dL) during the two first weeks of life in an ELBW cohort.
We included neonates without major congenital malformations, weighing <1000 g at birth and admitted at their first day of life in our NICU. Patients who did not have phosphate measurement between day 1 and day 14 were excluded. 2.2. Data collection Clinical data, nutritional intakes and laboratory measurements were extracted from computerized medical charts of the patients: ICCA® Philips (IntelliSpace Critical Care and Anesthesia). All available phosphate, calcium and glucose blood measurements from day 1 to 14 were recorded. Hyperglycemia was defined by at least two consecutives values >8.3 mmol/L (>150 mg/dL). Perinatal characteristics: Gestational age (GA), Birth weight (BW), multiple birth, delivery mode, antenatal steroids, small for gestational age (SGA), gender, Apgar score and main complications in the first two weeks of life as respiratory distress syndrome, intraventricular hemorrhage (grade 2), sepsis or death, were also collected. Exact daily parenteral and enteral intakes in glucose, amino acids, lipids, energy, phosphate and calcium were registered from day 1 to 14. 2.3. Nutritional regimens of the NICU
2. Methods
According to the nutritional policy of the centre, all the ELBW received individualized preparations of parenteral nutrition. Based on the ESPGHAN recommendations [28], our nutritional protocol recommended providing 1e1.5 mmol/kg/d of phosphate from day 2. Inorganic phosphate solutions were used (glucose-1-phosphate disodium (Phocytan®, Aguettant, France) or monobasic potassium phosphate 6.59%. Phosphorus intakes were thereafter adjusted to reach a serum phosphatemia between 1.5 and 2 mmol/L (4.7 and 6.2 mg/dL). Between 1 and 1.5 mmol/kg/d of calcium (solution of glucoheptonate calcium 10%) was provided from day 1 and was than adjusted from day 2 to target calcium serum level between 2.2 and 2.5 mmol/L. A daily dose of 55e110 UI per day of parenteral vitamin D was introduced from day 2 to 3 in a mixed vitamin solution (Cernevit®, Baxter, Maurepas, France). Parenteral glucose and amino acids were provided from birth. Lipids were introduced from day 2. Glucose was initiated at 6e8 g/kg/d and increased until 14e16 g/kg/d, with a progression rate of 1e2 g/kg/d; amino acids were initiated at 2 g/ kg/d and increased until 3.5 g/kg/d with a progression rate of 0.5 g/ kg/d; lipids were initiated at 0.5e1 g/kg/d and increased until 3e3.5 g/kg/d with a progression rate of 0.5e1 g/kg/d. Enteral nutrition was initiated in the first hours of following birth with 5e10 ml/kg/d of human milk (Mother's own milk or donor's milk) that was increased by 5e10 ml/kg/d depending on the enteral tolerance until the full ration of 160 ml/kg/d. Fortifiers (Eoprotine® and Liquigen®, Nutricia, Rueil Malmaison, France) were introduced when at least 100 ml/kg/d of human milk were tolerated. Enteral supplementation of 1000 units per day of vitamin D rol ADEC®) was started when parenteral nutrition was (Uveste discontinued.
2.1. Patients
2.4. Management of glucose intolerance
This study was conducted in a tertiary level Neonatal Intensive Care Unit (NICU) between January 2010 and December 2011 at the re Enfant”, Lyon, France. The study University Hospital “Femme Me population was a subset of a retrospective cohort study that was implemented in the NICU in order to monitor and assess several nutritional and growth issues in extremely low birth-weight infants [27]. It was approved by the Institutional Review Boards.
In case of glycemia 11 mmol/L (>200 mg/dL) confirmed on at least two blood checks in the same day, the nutritional protocol of the NICU was first to introduce insulin infusion. In case of refractory hyperglycemia despite progressive increase of insulin infusion rates from 0.01 U/kg/h to 0.03 UI/kg/h, the glucose infusion was reduced by 1e2 g/kg/d. Insulin was stopped when glycemia values were <4 mmol/L (<72 mg/dL).
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019
L. Dreyfus et al. / Clinical Nutrition xxx (2015) 1e7
2.5. Electrolyte and glucose monitoring During the study period, serum electrolytes and phosphate were routinely measured between 12 and 24 h of life, at day 3 and thereafter at least on a weekly basis, according to clinical evolution. During parenteral nutrition, capillary glycemia was checked (AccuChek®, Roche Diagnostics, Meylan) every 8 h at the first day, and then at least on a daily basis until parenteral nutrition was weaned. In case of glycemia 8 mmol/L, the measurement was repeated and urines were checked for glycosuria. The laboratory measurements were executed by the biochemistry department of the hospital. Blood samples were taken in a lithium heparinate tube and then centrifuged with 2500 g during 10 min at 18 C. Glycemia was measured by spectrophotometric enzymatic method (hexokinase/ G6PD). Phosphate was measured by spectrophotometric phosphomolybdate method. All measures were realized on spectrum analyzers: Architect C8000 & C16000, Abbott®. 2.6. Statistical analysis The analysis aimed to measure the association between Phosphate (as a longitudinal marker) and the risk of hyperglycemia, using a joint model for longitudinal and time-to-event data [29]. This approach consisted of two submodels: the longitudinal submodel and the survival submodel. In the longitudinal submodel, we used a suitable mixed-effects model to describe the subject-specific change over time of phosphate measurements. In the survival submodel, the event of interest was the occurrence of a hyperglycemia. The association between phosphate and hyperglycemia was quantified by the strength of the association between mi(t), the true unobserved value of phosphate at time t, and the hazard for an event at time t. The survival submodel can be extended by introducing lagged-effects [29]. We postulated that the risk at time t was associated with the true value of phosphate measurement at time t1 and thus tested a time-lag of one day. Covariates we introduced in the survival submodel for adjustment are the covariates that have significant effects (p-value <0.10) in the univariate analysis. The covariates (defined at baseline) considered were: gestational age, birth weight, small for gestational age, sex. The timedependent covariates considered were: nutritional intakes in lipids, proteins, carbohydrates, phosphate, and calcium. Discussion and further details on construction, validation and sensitivity analysis of this original joint model are provided in the electronic supplement. The analysis was carried out with the statistical software R 2.15 with package JM 1.1 written by C Rizopoulos [29]. 3. Results 3.1. Population characteristics In all, 156 ELBW patients were admitted on their first day of life during the study period; 8 of them were excluded (no available phosphatemia in 7 cases, and 1 major congenital malformation). The main characteristics of the 148 included patients are described in Table 1. 3.2. The longitudinal submodel In all, 469 phosphate measurements were available from the 148 patients; that is, 3.2 measurements per patient on average. Figure 1 shows these measurements over time as well as the mean curve based only on the fixed-effect estimates. On average, blood phosphate levels tended to decrease until day 6 and then to increase until day 13.
3
Total serum calcium levels and calcium intakes are described in the Fig. I in the supplementary material.
3.3. The survival submodel Among the 148 patients, 85 (57%) presented an episode of hyperglycemia (defined as at least two consecutives values >8.3 mmol/L) during the first 14 days of life. Figure 2a shows the KaplaneMeier curves of the survival without hyperglycemia, while Fig. 2b shows the hazard of hyperglycemia according to time. We noted high hazards at day 1 (20 patients experience the event, hazard ¼ 20/148 ¼ 0.135) and day 2 (13 patients experience the event, hazard ¼ 13/124 ¼ 0.105, 124 being the number of patients still at risk of a first hyperglycemia after excluding the former 20 and 4 censored patients). This means that the probability for a patient to present hyperglycemia on the two first days of life was 13.5% and 10%, respectively. We also noted high hazards at days 9 and 10 (10 and 9 patients with the event, respectively). After day 10, the hazard of hyperglycemia was low and the survival curve was flat with an estimate of 40% of patients without hyperglycemia. In the univariate Cox model, the baseline covariates that had significant effects and thus were part of the set of X1 variables of the survival submodel were: gestational age (p ¼ 0.03), birth weight (p < 0.0001), and small for gestational age (p ¼ 0.04). Calcium intake was the only time-dependent covariate that had a significant effect and represented X2(t) of the survival submodel. One should note that phosphate intake had a p-value close to our threshold (p ¼ 0.11). Table 2 shows the result relative to the multivariate survival submodel (from joint analysis) where all the covariates (except the binary variable “small for gestational age”) have been reduced (i.e., divided by their standard-deviations). Gestational age kept a significant effect, a low gestational age being a risk factor for hyperglycemia; every 1.6 week younger (1.6 being the standard deviation of the gestational ages) multiplied the hazard of hyperglycemia by 1.5 (1.5 ¼ 1/0.66). Low calcium intake seemed to be also a risk factor: every 0.17 mmol/kg/d less multiplied the hazard of hyperglycemia by 1.6 (1.6 ¼ 1/0.64). Our main interest lied in the effect of phosphate: the parameter of interest of the survival submodel was estimated at 1.101 (p ¼ 0.002); this means that a low phosphate level was a risk factor for hyperglycemia and that the hazard of hyperglycemia at a given time t would be multiplied by 3 (3 ¼ 1/0.33) for each 0.41 mmol/L decrease of phosphate level at time t. In other words, a patient A with a phosphate level PA at time t would have a 3 times higher hazard of hyperglycemia at time t than a patient B with a phosphate level PB ¼ PA þ 0.41. Moreover, we tested the joint model with a one-day lag-time: the pejorative effect of low phosphate levels was even higher, as the hazard of hyperglycemia at time t was multiplied by 3.85 for each 0.41 mmol/L decrease of phosphate level at time t1 day (p ¼ 0.0015). This means that a low phosphate level on a given day would be a strong predictor of hyperglycemia on the next day. On the basis of the joint model presented in Table 2, we calculated the probability of hyperglycemia in three infants: i) one example of infant who had high phosphate levels during first seven days (mean z 2 mmol/L); ii) one example of infant who had intermediate phosphate levels during first seven days (mean z 1 mmol/L); and, iii) example of infant who had low phosphate levels during first seven days (mean z 0.5 mmol/L). We supposed the three infants had the mean birth weight (800 g) and the mean gestational age (27 weeks) of the cohort. Calcium intake was not considered in the prediction process mainly because the values of this variable are time-dependent.
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019
4
L. Dreyfus et al. / Clinical Nutrition xxx (2015) 1e7
Table 1 Clinical characteristics of the study population (N ¼ 148). Percent/rate Gestational age (weeks) Birth weight (grams) Female Small for gestational age Antenatal steroïds Apgar at 5 min Respiratory distress syndrome Intraventricular hemorragia >grade 2 Death Age at enteral feeding (hours) Insulin treatment Age at insulin treatment (days) Hyperglycemiaa Age at hyperglycemiaa (days)
Mean
Median
SD
min
max
27.3 803
27.3 800
1.6 124.4
23.7 514
32.6 999
7.8
8.5
2.3
0
10
58 44 81 78 15 21 15.6
14
9.2
2
44
10.6
14
4.3
0.75
14
7.8
9
4.9
1
14
36 57
SD: standard deviation. a At least 2 consecutives values >8.3 mmol/L.
Figure 3 shows the probabilities of occurrence of hyperglycemia in these three infants according to the number of first daily phosphate values we take into account (three or seven) and according to the smallness for gestational age. This figure shows great differences in the risk of occurrence of hyperglycemia according to the first phosphate levels. For example, in Fig. 3a, the probability of absence of hyperglycemia at Day 14 in the patient who had high phosphate values is close to 0.71 (95% confidence interval, CI: [0.35; 0.84]) whereas this probability is much lower in the patient who had low phosphate values: 0.22 (95% CI: [0.01; 0.59])
4. Discussion Our study reveals that a low phosphate during the first two weeks of life is a risk factor of hyperglycemia in ELBW infants. According to our results, the risk of persistent hyperglycemia >8.3 mmol/L (>150 mg/dL) is multiplied by 3 for a decrease of 0.41 mmol/L of phosphate and by 3.85 for the same decreased of phosphate the day before. These results were statistically significant and controlled for several nutritional and clinical features, as phosphate intakes, GA, BW and SGA. To our knowledge, this is the
Fig. 1. Observed measures of phosphate and marginal fitted values over time. Description: Figure shows all 469 phosphate measurements from the 148 patients during the first 14 days; the mean curve is based only on the fixed-effect estimates.
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019
L. Dreyfus et al. / Clinical Nutrition xxx (2015) 1e7
5
Fig. 2. : a. Survival without hyperglycemia. Description: Figure shows the Kaplan Meier curve describing probability of survival without hyperglycemia in the first 14 days of life. b. Rate of hyperglycemia according to time. Description: Figure shows the hazard of hyperglycemia according time during the first 14 days of life.
Table 2 Hazard ratios for hyperglycemia according to adjusting covariates and phosphate; results from the survival multivariate submodel. Covariates
Parameter
Standard-error
Hazard-ratioa
P-value
Gestational age Birth weight Small for gestational age Calcium intake Phosphate level
0.408 0.126 0.510 0.453 1.101
0.181 0.191 0.391 0.146 0.350
0.66 [0.47; 0.95] 0.89 [0.61; 1.28] 1.66 [0.77; 3.58] 0.64 [0.48; 0.85] 0 .33 [0.17; 0.66]
0.02 0.51 0.19 0.002 0.002
a
Exp (parameter) [95% CI].
first study to show such an association between low phosphate level and glucose intolerance in neonates. Other independent factors that were found to predict hyperglycemia in our model were: a lower gestational age, which is consistent with existing data [1,2], and a low calcium intake. The last one has not been reported before as far as we know. We postulate that it could be due to the fact that calcium intake is closely linked to phosphate intake. The link between hypophosphatemia and hyperglycemia is sustained by some physiological explanations and clinical observations. Experimental data in rats show that phosphate deficient diet causes liver glucose-6-phosphatase up-regulation which increased hepatic glucose production leading to glucose intolerance [24]. In human adults, one study suggests that a low phosphatemia impairs the phosphorylation of carbohydrate taking part in glycolysis and glycogenesis causing insulin resistance, while chronic hypophosphatemia may affect ATP production in the islet b-cells which causes a dysfunction in insulin production [23]. In ELBW, hyperglycemia has been shown yet to be linked to a defective islet b-cell production of pro-insulin and to a resistance to insulin action [9]. Finally, these findings are consistent with our preliminary observation in which hypophosphatemia <1.2 mmol/L in the first 72 h following birth was related to insulin requirement in ELBW. In the present study, the clinical and statistical strength of the association between low phosphatemia and hyperglycemia, as
well as the report of a one-day lag-time between these two metabolic disorders, reinforce the hypothesis of a causal link, although this remains speculative. To answer to our research question, we used a novel and original statistical joint model that enabled us to measure the association between a biological marker with repeated measurements and the hazard for the occurrence of an event. An advantage of this model is the use of the phosphatemia as a continuous longitudinal factor, allowing the results to be independent of any arbitrary value for hypophosphatemia as no consensual threshold emerges from the literature yet [16,17]. On the other hand, it does not allow defining a precise phosphatemia threshold that would directly predict hyperglycemia. However, the complementary analysis presented in Fig. 3 provides an estimation of this risk in different clinical situations: patients, with and without SGA, presenting low, normal and high phosphate values in the first three or seven days of life. Despite large confidence intervals, this figure helps in assessing the risk of subsequent hyperglycemia according to the early phosphatemia profile and may help clinicians in recognizing situations at high risk of hyperglycemia. This study has several limitations. First, its observational and retrospective design does not allow ruling out any confounding bias. Nevertheless, the analysis attempted to control for clinical
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019
6
L. Dreyfus et al. / Clinical Nutrition xxx (2015) 1e7
Fig. 3. a. Hyperglycemia according to the phosphate values of the first three days of life-AGA. b. Hyperglycemia according to the phosphate values of the first three days of life-SGA. c. Hyperglycemia according to the phosphate values of the first seven days of life-AGA. d. Hyperglycemia according to the phosphate values of the first seven days of life-SGA.
conditions like SGA or GA, as well as for nutritional intakes. Other biases as over-measurements in sick patients or patients with metabolic abnormalities are also possible. Properly designed prospective studies would then be necessary to confirm these first results. In the meanwhile, our study strongly suggests that hypophosphatemia contribute to glucose intolerance in ELBW patients. The results show moreover that the lowest phosphate values occurs around day 6, and the rate of hyperglycemia is the highest around day of life 8 to 9, suggesting that there is a window of opportunity to actively try to prevent hypophosphatemia by efficient supplementation. Valuable normograms and more aggressive nutritional strategies are then required. In conclusion, despite its limits, this study confirms our preliminary results and strongly suggests that low phosphatemia is associated with hyperglycemia in ELBW patients. The decrease in phosphate level seems to precede hyperglycemia around 24 h. These results are novel and relevant, as hyperglycemia remains a major risk factor in neonatal morbidity and mortality, that one can yet poorly predict, prevent or treat. Future research will have to define whether interventions aimed to sustain phosphatemia could improve glycemia control and reduce insulin requirement, contributing thus to a better survival and neurological outcome of these high risk neonates.
DMB and LR were responsible for the statistical analyses and were involved in the manuscript preparation and revisions. MCEMMO participated in the study development, the data collection and the manuscript revision. SL was involved in the study drafting and critically reviewed the manuscript. OC supervised the entire study, was involved in its design, conduct, result interpretation and critically reviewed the manuscript. All the authors approved the final manuscript as submitted. Conflict of interest On behalf of all authors, the corresponding author states that there is no conflict of interest. Acknowledgment The authors are grateful to Mr. Jean IWAZ (Hospices Civils de Lyon, Department of Biostatistics) for the editorial revision of the manuscript, and to Mrs. Anne BLESS (Scientific Communication, Corseaux, Switzerland) for her assistance in English writing. CF is grateful to the non-commercial foundations SICPA, SA, te Acade mique Vaudoise, Switzerland, for Switzerland and Socie their research grants obtained during the study period.
Authors' contributions Appendix A. Supplementary data LD and CJFF designed the study and the data collection methods, participated in data collection and analysis, and prepared the manuscript.
Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.clnu.2015.07.019.
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019
L. Dreyfus et al. / Clinical Nutrition xxx (2015) 1e7
References [1] Hays SP, Smith EO, Sunehag AL. Hyperglycemia is a risk factor for early death and morbidity in extremely low birth-weight infants. Pediatrics 2006;1118(5):1811e8. http://dx.doi.org/10.1542/peds.2006-0628. [2] Van der Lugt NM, Smits-Wintjens VEHJ, van Zwieten PHT, Walther FJ. Short and long term outcome of neonatal hyperglycemia in very preterm infants: a retrospective follow-up study. BMC Pediatr 2010;10:52. http://dx.doi.org/ 10.1186/1471-2431-10-52. [3] Hall NJ, Peters M, Eaton S, Pierro A. Hyperglycemia is associated with increased morbidity and mortality rates in neonates with necrotizing enterocolitis. J Pediatr Surg 2004;39(6):898e901. [4] Blanco CL, McGill-Vargas LL, McCurnin D, Quinn AR. Hyperglycemia increases the risk of death in extremely preterm baboons. Pediatr Res 2012. http:// dx.doi.org/10.1038/pr.2012.184. [5] Linder N, Haskin O, Levit O, Klinger G, Prince T, Naor N, et al. Risk factors for intraventricular hemorrhage in very low birth weight premature infants: a retrospective case-control study. Pediatrics 2003;111(5 Pt 1):e590e595. [6] Kermorvant-Duchemin E, Pinel AC, Lavalette S, Lenne D, Raoul W, Calippe B, et al. Neonatal hyperglycemia inhibits angiogenesis and induces inflammation and neuronal degeneration in the retina. PLoS One 2013;8(11):e79545. http:// dx.doi.org/10.1371/0079545. [7] Kaempf JW, Kaempf AJ, Wu Y, Stawarz M, Niemeyer J, Grunkemeier G. Hyperglycemia, insulin and slower growth velocity may increase the risk of retinopathy of prematurity. J Perinatol 2011;31(4):251e7. http://dx.doi.org/ 10.1038/jp.2010.152. [8] Mitanchez-Mokhtari D, Lahlou N, Kieffer F, Magny J-F, Roger M, Voyer M. Both relative insulin resistance and defective islet beta-cell processing of proinsulin are responsible for transient hyperglycemia in extremely preterm infants. Pediatrics 2004;113(3 Pt 1):537e41. €ld B, Karle n J, Tessma MK, Norman M, Ade n U, et al. Early [9] Alexandrou G, Skio hyperglycemia is a risk factor for death and white matter reduction in preterm infants. Pediatrics 2010;125(3):e584e591. http://dx.doi.org/10.1542/ peds.2009-0449. [10] Sinclair JC, Bottino M, Cowett RM. Interventions for prevention of neonatal hyperglycemia in very low birth weight infants. Cochrane Database Syst Rev 2011;(10). http://dx.doi.org/10.1002/14651858. CD007615. [11] Ogilvy-Stuart AL, Beardsall K. Management of hyperglycaemia in the preterm infant. Arch Dis Child Fetal Neonatal 2010;95(2):F126e31. http://dx.doi.org/ 10.1136/adc.2008.154716. [12] Le Compte AJ, Lynn AM, Lin J, Pretty CG, Shaw GM, Chase JG. Pilot study of a model-based approach to blood glucose control in very-low-birthweight neonates. BMC Pediatr 2012;12:117. http://dx.doi.org/10.1186/1471-2431-12117. [13] Beardsall K, Ogilvy-Stuart AL, Frystyk J, Chen J-W, Thompson M, Ahluwalia J, et al. Early elective insulin therapy can reduce hyperglycemia and increase insulin-like growth factor-I levels in very low birth weight infants. J Pediatr 2007;151(6):611e7. http://dx.doi.org/10.1016/j.jpeds.2007.04.068. 617.e1. €m O, Larsson L. Plasma ionized calcium, phosphate and [14] Nelson N, Finnstro magnesium in preterm and small for gestational age infants. Acta Paediatr Scand 1989;78(3):351e7.
7
[15] Ichikawa G, Watabe Y, Suzumura H, Sairenchi T, Muto T, Arisaka O. Hypophosphatemia in small for gestational age extremely low birth weight infants receiving parenteral nutrition in the first week after birth. J Pediatr Endocrinol Metab 2012;25(3e4):317e21. [16] El Hassan N, Ryan R, Nirupama L. Premature infants: analysis of serum phosphate during the first 4 weeks of life. Am J Perinatol 2007;24(5):327e30. http://dx.doi.org/10.1055/s-2007-981431. [17] Fenton TR, Lyon AW, Rose MS. Cord blood calcium, phosphate, magnesium, and alkaline phosphatase gestational age-specific reference intervals for preterm infants. BMC Pediatr 2011;11:76. http://dx.doi.org/10.1186/14712431-11-76. [18] Kilic O, Demirkol D, Ucsel R, Citak A, Karabocuoglu M. Hypophosphatemia and its clinical implications in critically ill children: a retrospective study. J Crit Care 2012;27(5):474e9. http://dx.doi.org/10.1016/j.jcrc.2012.03.005. [19] Rigo J, Pieltain C, Salle B, Senterre J. Enteral calcium, phosphate and vitamin D requirements and bone mineralization in preterm infants. Acta Paediatr 2007;96(7):969e74. http://dx.doi.org/10.1111/j.1651-2227.2007.00336. [20] Rigo J, De Curtis M, Pieltain C, Picaud JC, Salle BL, Senterre J. Bone mineral metabolism in the micropremie. Clin Perinatol 2000;27(1):147e70. [21] Moltu SJ, Strømmen K, Blakstad EW, Almaas AN, Westerberg AC, Brække K, et al. Enhanced feeding in very-low-birth-weight infants may cause electrolyte disturbances and septicemia e a randomized, controlled trial. Clin Nutr 2013;32(2):207e12. http://dx.doi.org/10.1016/j.clnu.2012.09.004. [22] Ross JR, Finch C, Ebeling M, Taylor SN. Refeeding syndrome in very-low-birthweight intrauterine growth-restricted neonates. J Perinatol 2013;33(9): 717e20. http://dx.doi.org/10.1038/jp.2013.28. [23] Håglin L. Hypophosphataemia: cause of the disturbed metabolism in the metabolic syndrome. Med Hypotheses 2001;56(6):657e63. http://dx.doi.org/ 10.1054/mehy.2000.1272. chin MC, Van De Werve G. Up-regulation of liver glucose-6[24] Xie W, Li Y, Me phosphatase in rats fed with a P(i)-deficient diet. Biochem 1999;343(2): 393e6. [25] DeFronzo RA, Lang R. Hypophosphatemia and glucose intolerance: evidence for tissue insensitivity to insulin. N Engl J Med 1980;303(22):1259e63. http:// dx.doi.org/10.1056/NEJM198011273032203. [26] Dreyfus L, Fischer C, Maucort-Boulch D, Essomo Megnier Mbo Owono M, Laborie S, Claris O. Hypophosphatemia a risk factor for insulin requirement in ELBW infants?. ESPNIC/ESPR: Nutrition in the Critically Ill. 2012. [27] Fischer CJ, Maucort-Boulch D, Essomo Megnier-Mbo CM, Remontet L, Claris O. Early parenteral lipids and growth velocity in extremely-low-birth-weight infants. Clin Nutr 2014;33(3):502e8. http://dx.doi.org/10.1016/ j.clnu.2013.07.007. [28] Koletzko B, Goulet O, Hunt J, Krohn K, Shamir R. Guidelines on paediatric parenteral nutrition of the European society of paediatric gastroenterology, hepatology and nutrition (ESPGHAN) and the European society for clinical nutrition and metabolism (ESPEN), supported by the European society of paediatric research (ESPR). J Pediatr Gastroenterol Nutr 2005;41(Suppl. 2). S1e87. [29] Rizopoulos D. Joint models for longitudinal and time-to-event data (with applications in R). CRC biostatistics series. Chapman & Hall; 2012.
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019
Contents lists available at ScienceDirect
Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu
Original article
Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia? lia Dreyfus a, Ce line Julie Fischer Fumeaux b, a, Laurent Remontet c, Le Murielle Christine Essomo Megnier Mbo Owono a, Sophie Laborie a, Delphine Maucort-Boulch c, d, e, Olivier Claris a, e, f, * ^pital Femme M Hospices Civils de Lyon, Service de R eanimation N eonatale et N eonatologie, Ho ere Enfant, Bron, France Centre Hospitalier Universitaire Vaudois et Universit e de Lausanne, Service de N eonatologie, D epartement M edico-Chirurgical de P ediatrie, Switzerland Hospices Civils de Lyon, Service de Biostatistique, Lyon, France d CNRS UMR5558, Laboratoire de Biom etrie et Biologie Evolutive, Equipe Biostatistique Sant e, Pierre-B enite, France e Universit e Claude-Bernard, Lyon, France f Equipe d'Accueil Mixte EAM 4128, Lyon, France a
b c
a r t i c l e i n f o
s u m m a r y
Article history: Received 23 November 2014 Accepted 27 July 2015
Background & aims: Hyperglycemia occurs in more than half of the extremely low birth weight (ELBW) neonates during the first weeks of life, and is correlated with an increased risk of morbi-mortality. Hypophosphatemia is another frequent metabolic disorder in this population. Data from animal, adult studies and clinical observation suggest that hypophosphatemia could induce glucose intolerance. Our aim was to determine whether a low phosphatemia is associated with hyperglycemia in ELBW neonates. Methods: This observational study included ELBW infants admitted in a tertiary neonatal care center (2010e2011). According to the center's policy, they received parenteral nutrition from birth and human milk from day 1. Phosphatemia and glycemia were measured routinely during parenteral nutrition. Hyperglycemia was defined by two consecutives values >8.3 mmol/L (150 mg/dL). Statistical analysis used a joint model combining a mixed-effects and a survival submodels to measure the association between phosphate and hyperglycemia. Results: The study included 148 patients. Mean gestational (Standard Deviation) age was 27.3 (1.6) weeks; mean birth weight was 803 (124) grams; 57% presented hyperglycemia. The multivariate joint model showed that the hazard of hyperglycemia at a given time was multiplied by 3 for each 0.41 mmol/L decrease of phosphate level at this time (p ¼ 0.002) and by 3.85 for the same decreased of phosphate the day before (p ¼ 0.0015). Conclusion: To our knowledge, this is the first study suggesting that low phosphatemia can be associated with hyperglycemia in ELBW neonates. Further studies will have to demonstrate whether better control of phosphatemia could help in preventing hyperglycemia. © 2015 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Keywords: Extremely low birth weight infant Preterm neonate Metabolic disorder Hypophosphatemia Hyperglycemia Insulin
1. Introduction
^ pital Femme Me re Enfants, * Corresponding author. Hospices Civils de Lyon, Ho onatologie et Re animation Ne onatale, 59 Boulevard Pinel, 69500 Bron, Service de Ne France. Tel.: þ33 4 27 85 52 83; fax: þ33 4 27 86 92 27. E-mail addresses: [email protected] (L. Dreyfus), celine-julie.fischer@ chuv.ch (C.J. Fischer Fumeaux), [email protected] (L. Remontet), [email protected] (S. Laborie), [email protected] (D. Maucort-Boulch), [email protected] (O. Claris).
Hyperglycemia is a frequent metabolic complication in critically ill neonates and was found to occur in more than a half of extremely low birth weight (ELBW) infants [1]. The frequency of this complication is inversely related to gestational age (GA) and birth weight (BW) [2]. Beside to induce an osmotic diuresis, dehydration, hyponatremia, hypokalemia and acidosis, studies suggest that hyperglycemia increases mortality [1e4] and several serious morbidities as sepsis, intraventricular hemorrhage [5] or retinopathy of prematurity [6,7]. Both insulin resistance and failure in pancreatic
http://dx.doi.org/10.1016/j.clnu.2015.07.019 0261-5614/© 2015 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019
2
L. Dreyfus et al. / Clinical Nutrition xxx (2015) 1e7
List of abbreviations VLBW ELBW ATP NICU GA BW SGA
Very Low Birth Weight Extremely Low Birth Weight Adenosine Tri-Phosphate Neonatal Intensive Care Unit Gestational Age Birth Weight Small for Gestational Age
islet b-cell processing of pro-insulin could contribute to hyperglycemia in extremely preterm infants [8], but the pathogenesis remains to be elucidated. The identification of risk factors and mechanisms associated to glucose intolerance is thus especially important in the context of neonatal care [1e3,9]. There is so far no consensus on safe and efficient management of glucose intolerance, as both insulin treatment and reduction in glucose intakes can be deleterious [10e13]. Hypophosphatemia is another frequent metabolic disorder in premature newborns, especially in very low birth-weight infants (VLBW) and in small for gestational age (SGA) neonates, although normal values at this age remain uncertain [14e17]. In the critically ill children, it has been associated with a longer duration of mechanical ventilation and with a longer duration of stay in the pediatric intensive care unit [18]. In the preterm newborn, hypophosphatemia is frequently associated to postnatal growth deficiency and osteopenia of the preterm [19,20] and has been recently incriminated in susceptibility for septicaemia in VLBW [21]. Various mechanisms have been raised, like limited storage, insufficient intakes in the first days or a re-feeding like syndrome in VLBW undergoing enhanced nutrition [22]. Due to our common clinical observation of hypophosphatemia preceding severe and persistent hyperglycemia, we drafted the hypothesis that hypophosphatemia could play a role in glucose intolerance in the preterm neonate. This hypothesis was sustained by some experimental studies and adult data, where hypophosphatemia could be involved in insulin resistance found in metabolic syndrome [23,24]. Two mechanisms have been involved; first, hypophosphatemia leads to a low adenosine triphosphate (ATP) intracellular rates; in the beta cells of the pancreas, the consequent dysfunction in ATP-ase activity may lead to a reduction in the insulin production [8,25]. Secondly, phosphorus is involved in phosphorylation of the insulin cell-receptor, which could thus be less efficient in case of low phosphatemia [25]. Whether hypophosphatemia is a risk factor for hyperglycemia in preterm neonates remains to be elucidated [26]. The aim of our study was to further investigate the association of low phosphate and the risk of glucose intolerance, defined as persistent hyperglycemia >8.3 mmol/L (>150 mg/dL) during the two first weeks of life in an ELBW cohort.
We included neonates without major congenital malformations, weighing <1000 g at birth and admitted at their first day of life in our NICU. Patients who did not have phosphate measurement between day 1 and day 14 were excluded. 2.2. Data collection Clinical data, nutritional intakes and laboratory measurements were extracted from computerized medical charts of the patients: ICCA® Philips (IntelliSpace Critical Care and Anesthesia). All available phosphate, calcium and glucose blood measurements from day 1 to 14 were recorded. Hyperglycemia was defined by at least two consecutives values >8.3 mmol/L (>150 mg/dL). Perinatal characteristics: Gestational age (GA), Birth weight (BW), multiple birth, delivery mode, antenatal steroids, small for gestational age (SGA), gender, Apgar score and main complications in the first two weeks of life as respiratory distress syndrome, intraventricular hemorrhage (grade 2), sepsis or death, were also collected. Exact daily parenteral and enteral intakes in glucose, amino acids, lipids, energy, phosphate and calcium were registered from day 1 to 14. 2.3. Nutritional regimens of the NICU
2. Methods
According to the nutritional policy of the centre, all the ELBW received individualized preparations of parenteral nutrition. Based on the ESPGHAN recommendations [28], our nutritional protocol recommended providing 1e1.5 mmol/kg/d of phosphate from day 2. Inorganic phosphate solutions were used (glucose-1-phosphate disodium (Phocytan®, Aguettant, France) or monobasic potassium phosphate 6.59%. Phosphorus intakes were thereafter adjusted to reach a serum phosphatemia between 1.5 and 2 mmol/L (4.7 and 6.2 mg/dL). Between 1 and 1.5 mmol/kg/d of calcium (solution of glucoheptonate calcium 10%) was provided from day 1 and was than adjusted from day 2 to target calcium serum level between 2.2 and 2.5 mmol/L. A daily dose of 55e110 UI per day of parenteral vitamin D was introduced from day 2 to 3 in a mixed vitamin solution (Cernevit®, Baxter, Maurepas, France). Parenteral glucose and amino acids were provided from birth. Lipids were introduced from day 2. Glucose was initiated at 6e8 g/kg/d and increased until 14e16 g/kg/d, with a progression rate of 1e2 g/kg/d; amino acids were initiated at 2 g/ kg/d and increased until 3.5 g/kg/d with a progression rate of 0.5 g/ kg/d; lipids were initiated at 0.5e1 g/kg/d and increased until 3e3.5 g/kg/d with a progression rate of 0.5e1 g/kg/d. Enteral nutrition was initiated in the first hours of following birth with 5e10 ml/kg/d of human milk (Mother's own milk or donor's milk) that was increased by 5e10 ml/kg/d depending on the enteral tolerance until the full ration of 160 ml/kg/d. Fortifiers (Eoprotine® and Liquigen®, Nutricia, Rueil Malmaison, France) were introduced when at least 100 ml/kg/d of human milk were tolerated. Enteral supplementation of 1000 units per day of vitamin D rol ADEC®) was started when parenteral nutrition was (Uveste discontinued.
2.1. Patients
2.4. Management of glucose intolerance
This study was conducted in a tertiary level Neonatal Intensive Care Unit (NICU) between January 2010 and December 2011 at the re Enfant”, Lyon, France. The study University Hospital “Femme Me population was a subset of a retrospective cohort study that was implemented in the NICU in order to monitor and assess several nutritional and growth issues in extremely low birth-weight infants [27]. It was approved by the Institutional Review Boards.
In case of glycemia 11 mmol/L (>200 mg/dL) confirmed on at least two blood checks in the same day, the nutritional protocol of the NICU was first to introduce insulin infusion. In case of refractory hyperglycemia despite progressive increase of insulin infusion rates from 0.01 U/kg/h to 0.03 UI/kg/h, the glucose infusion was reduced by 1e2 g/kg/d. Insulin was stopped when glycemia values were <4 mmol/L (<72 mg/dL).
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019
L. Dreyfus et al. / Clinical Nutrition xxx (2015) 1e7
2.5. Electrolyte and glucose monitoring During the study period, serum electrolytes and phosphate were routinely measured between 12 and 24 h of life, at day 3 and thereafter at least on a weekly basis, according to clinical evolution. During parenteral nutrition, capillary glycemia was checked (AccuChek®, Roche Diagnostics, Meylan) every 8 h at the first day, and then at least on a daily basis until parenteral nutrition was weaned. In case of glycemia 8 mmol/L, the measurement was repeated and urines were checked for glycosuria. The laboratory measurements were executed by the biochemistry department of the hospital. Blood samples were taken in a lithium heparinate tube and then centrifuged with 2500 g during 10 min at 18 C. Glycemia was measured by spectrophotometric enzymatic method (hexokinase/ G6PD). Phosphate was measured by spectrophotometric phosphomolybdate method. All measures were realized on spectrum analyzers: Architect C8000 & C16000, Abbott®. 2.6. Statistical analysis The analysis aimed to measure the association between Phosphate (as a longitudinal marker) and the risk of hyperglycemia, using a joint model for longitudinal and time-to-event data [29]. This approach consisted of two submodels: the longitudinal submodel and the survival submodel. In the longitudinal submodel, we used a suitable mixed-effects model to describe the subject-specific change over time of phosphate measurements. In the survival submodel, the event of interest was the occurrence of a hyperglycemia. The association between phosphate and hyperglycemia was quantified by the strength of the association between mi(t), the true unobserved value of phosphate at time t, and the hazard for an event at time t. The survival submodel can be extended by introducing lagged-effects [29]. We postulated that the risk at time t was associated with the true value of phosphate measurement at time t1 and thus tested a time-lag of one day. Covariates we introduced in the survival submodel for adjustment are the covariates that have significant effects (p-value <0.10) in the univariate analysis. The covariates (defined at baseline) considered were: gestational age, birth weight, small for gestational age, sex. The timedependent covariates considered were: nutritional intakes in lipids, proteins, carbohydrates, phosphate, and calcium. Discussion and further details on construction, validation and sensitivity analysis of this original joint model are provided in the electronic supplement. The analysis was carried out with the statistical software R 2.15 with package JM 1.1 written by C Rizopoulos [29]. 3. Results 3.1. Population characteristics In all, 156 ELBW patients were admitted on their first day of life during the study period; 8 of them were excluded (no available phosphatemia in 7 cases, and 1 major congenital malformation). The main characteristics of the 148 included patients are described in Table 1. 3.2. The longitudinal submodel In all, 469 phosphate measurements were available from the 148 patients; that is, 3.2 measurements per patient on average. Figure 1 shows these measurements over time as well as the mean curve based only on the fixed-effect estimates. On average, blood phosphate levels tended to decrease until day 6 and then to increase until day 13.
3
Total serum calcium levels and calcium intakes are described in the Fig. I in the supplementary material.
3.3. The survival submodel Among the 148 patients, 85 (57%) presented an episode of hyperglycemia (defined as at least two consecutives values >8.3 mmol/L) during the first 14 days of life. Figure 2a shows the KaplaneMeier curves of the survival without hyperglycemia, while Fig. 2b shows the hazard of hyperglycemia according to time. We noted high hazards at day 1 (20 patients experience the event, hazard ¼ 20/148 ¼ 0.135) and day 2 (13 patients experience the event, hazard ¼ 13/124 ¼ 0.105, 124 being the number of patients still at risk of a first hyperglycemia after excluding the former 20 and 4 censored patients). This means that the probability for a patient to present hyperglycemia on the two first days of life was 13.5% and 10%, respectively. We also noted high hazards at days 9 and 10 (10 and 9 patients with the event, respectively). After day 10, the hazard of hyperglycemia was low and the survival curve was flat with an estimate of 40% of patients without hyperglycemia. In the univariate Cox model, the baseline covariates that had significant effects and thus were part of the set of X1 variables of the survival submodel were: gestational age (p ¼ 0.03), birth weight (p < 0.0001), and small for gestational age (p ¼ 0.04). Calcium intake was the only time-dependent covariate that had a significant effect and represented X2(t) of the survival submodel. One should note that phosphate intake had a p-value close to our threshold (p ¼ 0.11). Table 2 shows the result relative to the multivariate survival submodel (from joint analysis) where all the covariates (except the binary variable “small for gestational age”) have been reduced (i.e., divided by their standard-deviations). Gestational age kept a significant effect, a low gestational age being a risk factor for hyperglycemia; every 1.6 week younger (1.6 being the standard deviation of the gestational ages) multiplied the hazard of hyperglycemia by 1.5 (1.5 ¼ 1/0.66). Low calcium intake seemed to be also a risk factor: every 0.17 mmol/kg/d less multiplied the hazard of hyperglycemia by 1.6 (1.6 ¼ 1/0.64). Our main interest lied in the effect of phosphate: the parameter of interest of the survival submodel was estimated at 1.101 (p ¼ 0.002); this means that a low phosphate level was a risk factor for hyperglycemia and that the hazard of hyperglycemia at a given time t would be multiplied by 3 (3 ¼ 1/0.33) for each 0.41 mmol/L decrease of phosphate level at time t. In other words, a patient A with a phosphate level PA at time t would have a 3 times higher hazard of hyperglycemia at time t than a patient B with a phosphate level PB ¼ PA þ 0.41. Moreover, we tested the joint model with a one-day lag-time: the pejorative effect of low phosphate levels was even higher, as the hazard of hyperglycemia at time t was multiplied by 3.85 for each 0.41 mmol/L decrease of phosphate level at time t1 day (p ¼ 0.0015). This means that a low phosphate level on a given day would be a strong predictor of hyperglycemia on the next day. On the basis of the joint model presented in Table 2, we calculated the probability of hyperglycemia in three infants: i) one example of infant who had high phosphate levels during first seven days (mean z 2 mmol/L); ii) one example of infant who had intermediate phosphate levels during first seven days (mean z 1 mmol/L); and, iii) example of infant who had low phosphate levels during first seven days (mean z 0.5 mmol/L). We supposed the three infants had the mean birth weight (800 g) and the mean gestational age (27 weeks) of the cohort. Calcium intake was not considered in the prediction process mainly because the values of this variable are time-dependent.
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019
4
L. Dreyfus et al. / Clinical Nutrition xxx (2015) 1e7
Table 1 Clinical characteristics of the study population (N ¼ 148). Percent/rate Gestational age (weeks) Birth weight (grams) Female Small for gestational age Antenatal steroïds Apgar at 5 min Respiratory distress syndrome Intraventricular hemorragia >grade 2 Death Age at enteral feeding (hours) Insulin treatment Age at insulin treatment (days) Hyperglycemiaa Age at hyperglycemiaa (days)
Mean
Median
SD
min
max
27.3 803
27.3 800
1.6 124.4
23.7 514
32.6 999
7.8
8.5
2.3
0
10
58 44 81 78 15 21 15.6
14
9.2
2
44
10.6
14
4.3
0.75
14
7.8
9
4.9
1
14
36 57
SD: standard deviation. a At least 2 consecutives values >8.3 mmol/L.
Figure 3 shows the probabilities of occurrence of hyperglycemia in these three infants according to the number of first daily phosphate values we take into account (three or seven) and according to the smallness for gestational age. This figure shows great differences in the risk of occurrence of hyperglycemia according to the first phosphate levels. For example, in Fig. 3a, the probability of absence of hyperglycemia at Day 14 in the patient who had high phosphate values is close to 0.71 (95% confidence interval, CI: [0.35; 0.84]) whereas this probability is much lower in the patient who had low phosphate values: 0.22 (95% CI: [0.01; 0.59])
4. Discussion Our study reveals that a low phosphate during the first two weeks of life is a risk factor of hyperglycemia in ELBW infants. According to our results, the risk of persistent hyperglycemia >8.3 mmol/L (>150 mg/dL) is multiplied by 3 for a decrease of 0.41 mmol/L of phosphate and by 3.85 for the same decreased of phosphate the day before. These results were statistically significant and controlled for several nutritional and clinical features, as phosphate intakes, GA, BW and SGA. To our knowledge, this is the
Fig. 1. Observed measures of phosphate and marginal fitted values over time. Description: Figure shows all 469 phosphate measurements from the 148 patients during the first 14 days; the mean curve is based only on the fixed-effect estimates.
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019
L. Dreyfus et al. / Clinical Nutrition xxx (2015) 1e7
5
Fig. 2. : a. Survival without hyperglycemia. Description: Figure shows the Kaplan Meier curve describing probability of survival without hyperglycemia in the first 14 days of life. b. Rate of hyperglycemia according to time. Description: Figure shows the hazard of hyperglycemia according time during the first 14 days of life.
Table 2 Hazard ratios for hyperglycemia according to adjusting covariates and phosphate; results from the survival multivariate submodel. Covariates
Parameter
Standard-error
Hazard-ratioa
P-value
Gestational age Birth weight Small for gestational age Calcium intake Phosphate level
0.408 0.126 0.510 0.453 1.101
0.181 0.191 0.391 0.146 0.350
0.66 [0.47; 0.95] 0.89 [0.61; 1.28] 1.66 [0.77; 3.58] 0.64 [0.48; 0.85] 0 .33 [0.17; 0.66]
0.02 0.51 0.19 0.002 0.002
a
Exp (parameter) [95% CI].
first study to show such an association between low phosphate level and glucose intolerance in neonates. Other independent factors that were found to predict hyperglycemia in our model were: a lower gestational age, which is consistent with existing data [1,2], and a low calcium intake. The last one has not been reported before as far as we know. We postulate that it could be due to the fact that calcium intake is closely linked to phosphate intake. The link between hypophosphatemia and hyperglycemia is sustained by some physiological explanations and clinical observations. Experimental data in rats show that phosphate deficient diet causes liver glucose-6-phosphatase up-regulation which increased hepatic glucose production leading to glucose intolerance [24]. In human adults, one study suggests that a low phosphatemia impairs the phosphorylation of carbohydrate taking part in glycolysis and glycogenesis causing insulin resistance, while chronic hypophosphatemia may affect ATP production in the islet b-cells which causes a dysfunction in insulin production [23]. In ELBW, hyperglycemia has been shown yet to be linked to a defective islet b-cell production of pro-insulin and to a resistance to insulin action [9]. Finally, these findings are consistent with our preliminary observation in which hypophosphatemia <1.2 mmol/L in the first 72 h following birth was related to insulin requirement in ELBW. In the present study, the clinical and statistical strength of the association between low phosphatemia and hyperglycemia, as
well as the report of a one-day lag-time between these two metabolic disorders, reinforce the hypothesis of a causal link, although this remains speculative. To answer to our research question, we used a novel and original statistical joint model that enabled us to measure the association between a biological marker with repeated measurements and the hazard for the occurrence of an event. An advantage of this model is the use of the phosphatemia as a continuous longitudinal factor, allowing the results to be independent of any arbitrary value for hypophosphatemia as no consensual threshold emerges from the literature yet [16,17]. On the other hand, it does not allow defining a precise phosphatemia threshold that would directly predict hyperglycemia. However, the complementary analysis presented in Fig. 3 provides an estimation of this risk in different clinical situations: patients, with and without SGA, presenting low, normal and high phosphate values in the first three or seven days of life. Despite large confidence intervals, this figure helps in assessing the risk of subsequent hyperglycemia according to the early phosphatemia profile and may help clinicians in recognizing situations at high risk of hyperglycemia. This study has several limitations. First, its observational and retrospective design does not allow ruling out any confounding bias. Nevertheless, the analysis attempted to control for clinical
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019
6
L. Dreyfus et al. / Clinical Nutrition xxx (2015) 1e7
Fig. 3. a. Hyperglycemia according to the phosphate values of the first three days of life-AGA. b. Hyperglycemia according to the phosphate values of the first three days of life-SGA. c. Hyperglycemia according to the phosphate values of the first seven days of life-AGA. d. Hyperglycemia according to the phosphate values of the first seven days of life-SGA.
conditions like SGA or GA, as well as for nutritional intakes. Other biases as over-measurements in sick patients or patients with metabolic abnormalities are also possible. Properly designed prospective studies would then be necessary to confirm these first results. In the meanwhile, our study strongly suggests that hypophosphatemia contribute to glucose intolerance in ELBW patients. The results show moreover that the lowest phosphate values occurs around day 6, and the rate of hyperglycemia is the highest around day of life 8 to 9, suggesting that there is a window of opportunity to actively try to prevent hypophosphatemia by efficient supplementation. Valuable normograms and more aggressive nutritional strategies are then required. In conclusion, despite its limits, this study confirms our preliminary results and strongly suggests that low phosphatemia is associated with hyperglycemia in ELBW patients. The decrease in phosphate level seems to precede hyperglycemia around 24 h. These results are novel and relevant, as hyperglycemia remains a major risk factor in neonatal morbidity and mortality, that one can yet poorly predict, prevent or treat. Future research will have to define whether interventions aimed to sustain phosphatemia could improve glycemia control and reduce insulin requirement, contributing thus to a better survival and neurological outcome of these high risk neonates.
DMB and LR were responsible for the statistical analyses and were involved in the manuscript preparation and revisions. MCEMMO participated in the study development, the data collection and the manuscript revision. SL was involved in the study drafting and critically reviewed the manuscript. OC supervised the entire study, was involved in its design, conduct, result interpretation and critically reviewed the manuscript. All the authors approved the final manuscript as submitted. Conflict of interest On behalf of all authors, the corresponding author states that there is no conflict of interest. Acknowledgment The authors are grateful to Mr. Jean IWAZ (Hospices Civils de Lyon, Department of Biostatistics) for the editorial revision of the manuscript, and to Mrs. Anne BLESS (Scientific Communication, Corseaux, Switzerland) for her assistance in English writing. CF is grateful to the non-commercial foundations SICPA, SA, te Acade mique Vaudoise, Switzerland, for Switzerland and Socie their research grants obtained during the study period.
Authors' contributions Appendix A. Supplementary data LD and CJFF designed the study and the data collection methods, participated in data collection and analysis, and prepared the manuscript.
Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.clnu.2015.07.019.
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019
L. Dreyfus et al. / Clinical Nutrition xxx (2015) 1e7
References [1] Hays SP, Smith EO, Sunehag AL. Hyperglycemia is a risk factor for early death and morbidity in extremely low birth-weight infants. Pediatrics 2006;1118(5):1811e8. http://dx.doi.org/10.1542/peds.2006-0628. [2] Van der Lugt NM, Smits-Wintjens VEHJ, van Zwieten PHT, Walther FJ. Short and long term outcome of neonatal hyperglycemia in very preterm infants: a retrospective follow-up study. BMC Pediatr 2010;10:52. http://dx.doi.org/ 10.1186/1471-2431-10-52. [3] Hall NJ, Peters M, Eaton S, Pierro A. Hyperglycemia is associated with increased morbidity and mortality rates in neonates with necrotizing enterocolitis. J Pediatr Surg 2004;39(6):898e901. [4] Blanco CL, McGill-Vargas LL, McCurnin D, Quinn AR. Hyperglycemia increases the risk of death in extremely preterm baboons. Pediatr Res 2012. http:// dx.doi.org/10.1038/pr.2012.184. [5] Linder N, Haskin O, Levit O, Klinger G, Prince T, Naor N, et al. Risk factors for intraventricular hemorrhage in very low birth weight premature infants: a retrospective case-control study. Pediatrics 2003;111(5 Pt 1):e590e595. [6] Kermorvant-Duchemin E, Pinel AC, Lavalette S, Lenne D, Raoul W, Calippe B, et al. Neonatal hyperglycemia inhibits angiogenesis and induces inflammation and neuronal degeneration in the retina. PLoS One 2013;8(11):e79545. http:// dx.doi.org/10.1371/0079545. [7] Kaempf JW, Kaempf AJ, Wu Y, Stawarz M, Niemeyer J, Grunkemeier G. Hyperglycemia, insulin and slower growth velocity may increase the risk of retinopathy of prematurity. J Perinatol 2011;31(4):251e7. http://dx.doi.org/ 10.1038/jp.2010.152. [8] Mitanchez-Mokhtari D, Lahlou N, Kieffer F, Magny J-F, Roger M, Voyer M. Both relative insulin resistance and defective islet beta-cell processing of proinsulin are responsible for transient hyperglycemia in extremely preterm infants. Pediatrics 2004;113(3 Pt 1):537e41. €ld B, Karle n J, Tessma MK, Norman M, Ade n U, et al. Early [9] Alexandrou G, Skio hyperglycemia is a risk factor for death and white matter reduction in preterm infants. Pediatrics 2010;125(3):e584e591. http://dx.doi.org/10.1542/ peds.2009-0449. [10] Sinclair JC, Bottino M, Cowett RM. Interventions for prevention of neonatal hyperglycemia in very low birth weight infants. Cochrane Database Syst Rev 2011;(10). http://dx.doi.org/10.1002/14651858. CD007615. [11] Ogilvy-Stuart AL, Beardsall K. Management of hyperglycaemia in the preterm infant. Arch Dis Child Fetal Neonatal 2010;95(2):F126e31. http://dx.doi.org/ 10.1136/adc.2008.154716. [12] Le Compte AJ, Lynn AM, Lin J, Pretty CG, Shaw GM, Chase JG. Pilot study of a model-based approach to blood glucose control in very-low-birthweight neonates. BMC Pediatr 2012;12:117. http://dx.doi.org/10.1186/1471-2431-12117. [13] Beardsall K, Ogilvy-Stuart AL, Frystyk J, Chen J-W, Thompson M, Ahluwalia J, et al. Early elective insulin therapy can reduce hyperglycemia and increase insulin-like growth factor-I levels in very low birth weight infants. J Pediatr 2007;151(6):611e7. http://dx.doi.org/10.1016/j.jpeds.2007.04.068. 617.e1. €m O, Larsson L. Plasma ionized calcium, phosphate and [14] Nelson N, Finnstro magnesium in preterm and small for gestational age infants. Acta Paediatr Scand 1989;78(3):351e7.
7
[15] Ichikawa G, Watabe Y, Suzumura H, Sairenchi T, Muto T, Arisaka O. Hypophosphatemia in small for gestational age extremely low birth weight infants receiving parenteral nutrition in the first week after birth. J Pediatr Endocrinol Metab 2012;25(3e4):317e21. [16] El Hassan N, Ryan R, Nirupama L. Premature infants: analysis of serum phosphate during the first 4 weeks of life. Am J Perinatol 2007;24(5):327e30. http://dx.doi.org/10.1055/s-2007-981431. [17] Fenton TR, Lyon AW, Rose MS. Cord blood calcium, phosphate, magnesium, and alkaline phosphatase gestational age-specific reference intervals for preterm infants. BMC Pediatr 2011;11:76. http://dx.doi.org/10.1186/14712431-11-76. [18] Kilic O, Demirkol D, Ucsel R, Citak A, Karabocuoglu M. Hypophosphatemia and its clinical implications in critically ill children: a retrospective study. J Crit Care 2012;27(5):474e9. http://dx.doi.org/10.1016/j.jcrc.2012.03.005. [19] Rigo J, Pieltain C, Salle B, Senterre J. Enteral calcium, phosphate and vitamin D requirements and bone mineralization in preterm infants. Acta Paediatr 2007;96(7):969e74. http://dx.doi.org/10.1111/j.1651-2227.2007.00336. [20] Rigo J, De Curtis M, Pieltain C, Picaud JC, Salle BL, Senterre J. Bone mineral metabolism in the micropremie. Clin Perinatol 2000;27(1):147e70. [21] Moltu SJ, Strømmen K, Blakstad EW, Almaas AN, Westerberg AC, Brække K, et al. Enhanced feeding in very-low-birth-weight infants may cause electrolyte disturbances and septicemia e a randomized, controlled trial. Clin Nutr 2013;32(2):207e12. http://dx.doi.org/10.1016/j.clnu.2012.09.004. [22] Ross JR, Finch C, Ebeling M, Taylor SN. Refeeding syndrome in very-low-birthweight intrauterine growth-restricted neonates. J Perinatol 2013;33(9): 717e20. http://dx.doi.org/10.1038/jp.2013.28. [23] Håglin L. Hypophosphataemia: cause of the disturbed metabolism in the metabolic syndrome. Med Hypotheses 2001;56(6):657e63. http://dx.doi.org/ 10.1054/mehy.2000.1272. chin MC, Van De Werve G. Up-regulation of liver glucose-6[24] Xie W, Li Y, Me phosphatase in rats fed with a P(i)-deficient diet. Biochem 1999;343(2): 393e6. [25] DeFronzo RA, Lang R. Hypophosphatemia and glucose intolerance: evidence for tissue insensitivity to insulin. N Engl J Med 1980;303(22):1259e63. http:// dx.doi.org/10.1056/NEJM198011273032203. [26] Dreyfus L, Fischer C, Maucort-Boulch D, Essomo Megnier Mbo Owono M, Laborie S, Claris O. Hypophosphatemia a risk factor for insulin requirement in ELBW infants?. ESPNIC/ESPR: Nutrition in the Critically Ill. 2012. [27] Fischer CJ, Maucort-Boulch D, Essomo Megnier-Mbo CM, Remontet L, Claris O. Early parenteral lipids and growth velocity in extremely-low-birth-weight infants. Clin Nutr 2014;33(3):502e8. http://dx.doi.org/10.1016/ j.clnu.2013.07.007. [28] Koletzko B, Goulet O, Hunt J, Krohn K, Shamir R. Guidelines on paediatric parenteral nutrition of the European society of paediatric gastroenterology, hepatology and nutrition (ESPGHAN) and the European society for clinical nutrition and metabolism (ESPEN), supported by the European society of paediatric research (ESPR). J Pediatr Gastroenterol Nutr 2005;41(Suppl. 2). S1e87. [29] Rizopoulos D. Joint models for longitudinal and time-to-event data (with applications in R). CRC biostatistics series. Chapman & Hall; 2012.
Please cite this article in press as: Dreyfus L, et al., Low phosphatemia in extremely low birth weight neonates: A risk factor for hyperglycemia?, Clinical Nutrition (2015), http://dx.doi.org/10.1016/j.clnu.2015.07.019