Early Parenteral Amino Acid Supplementation Reduces the Risk of Neonatal Hyperkalemia in Preterm Infants: Single-Center Retrospective Study.

The incidence of neonatal hyperkalemia in preterm infants was high in the early 2000s. However, the prevalence has declined recently, and the reason for the decline remains unclear. The development of neonatal hyperkalemia can be inuenced by suppression of insulin secretion due to hypercatabolism; thus, early parenteral amino acid supplementation, which is now widespread for nutritional management for preterm infants, may have contributed to the decline. We conducted a single-center, retrospective cohort study with 225 preterm infants inborn between 24 0/7 and 31 6/7 weeks of gestation during 2009– 2018. We compared the incidence of neonatal hyperkalemia between those with or without early parenteral amino acid supplementation (initiated within 24 h after birth). We included 150 infants with early AA supplementation from the rst day of life and 75 controls. The incidence of neonatal hyperkalemia was signicantly lower in the AA group than the control group (2.7% vs 12.4%, p<0.05). In multivariate analysis, the risk of hyperkalemia increased with delivery at <28 weeks and decreased with AA supplementation (adjusted odds ratio 0.11; 95% condence interval 0.02–0.56, p<0.01). In conclusion, early AA supplement from the rst day of life might have a prophylactic effect on neonatal hyperkalemia in preterm infants.


Introduction
Neonatal hyperkalemia is a severe, acute problem that can be life-threatening due to its effect on cardiac rhythm in premature infants. 1 The ECG manifestations of hyperkalemia include peaked T waves, widened QRS duration, bradycardia, ventricular tachycardia, and ventricular brillation. A rapid rise in serum potassium may be seen during the rst 24 h of life, especially in premature infants. 2 There are a few known mechanisms behind the development of hyperkalemia, including relative hypoaldosteronism, immaturity of the renal distal tubules, and low tissue K + intake due to low Na + -K + -ATPase activity. 1,3 Although the incidence of neonatal hyperkalemia remained high (50%) until the early 2000s, its prevalence has declined recently (25%) and the reasons of this decline remain unclear. 2,4,5 In the early 2000s, recommendations for early amino acid (AA) supplementation began, 6,7 and this practice gradually came into widespread use. Insulin plays a role in protein synthesis, as well as in the transport of glucose and K + into cells; thus, the secretion of insulin decreases under conditions of increased catabolism. 8 Conversely, infusion with parenteral AA supplementation increases insulin levels in preterm infants, which may promote K + shift 9 . However, the prophylactic effect of early AA supplementation on neonatal hyperkalemia has not been elucidated yet.
Tocolytic agents improve neonatal outcome when used for extending pregnancy up to 48 hours : meanwhile, pregnant women can be transferred to facilities with a high level of perinatal care and receive antenatal steroid to accelerate fetal lung maturation 10 . However, in Japan, once threatened preterm labor is diagnosed, gynecologists often continue to use tocolytic agents for a longer period just before delivery.
A longer treatment with both ritodrine and MgSO 4 has recently been reported to be a risk factor for hyperkalemia in preterm infants born at 32-36 weeks of gestation. 11 Antenatal steroid has been regarded as a potential prophylactic strategy against hyperkalemia as well as respiratory distress syndrome. 2,12,13 Omar et al reported that extremely low birth weight infants (ELBWI) with antenatal steroid showed signi cantly lower serum potassium levels than infants without one, probably because of Na + -K + ATPase maturation. 12 The aim of the present study was to assess the prophylactic effect of early parenteral AA supplementation on neonatal hyperkalemia in preterm infants, taking into account known risk factors and other prophylactic measures in use.

Results
During 2009-2018, a total of 261 preterm infants born at 24 0/7 -31 6/7 weeks of gestational age were admitted. We excluded 12 patients with congenital major malformations, 5 patients with chromosomal abnormalities, 4 with early deaths, 9 outborn patients, and 6 with missing K + values (Fig. 1). We included 150 preterm infants with early parenteral AA supplementation administered within the rst day of life and 75 controls without early AA supplementation. Of 225 infants, 13 (5.8%) were diagnosed with neonatal hyperkalemia. Of the 225, 106 were ELBWIs, of whom 10 (9.4%) showed neonatal hyperkalemia.
The effects of various risk factors for and prophylactic management on hyperkalemia occurrence We performed logistic regression with hyperkalemia occurrence as the dependent variable and risk factors and prophylactic management as the explanatory variables (Table 3). Both univariate analysis (crude odds ratio (OR) and multivariate analysis (adjusted odds ratio (aOR)) are presented. In multivariate analysis, independent risk factors and prophylactic management for hyperkalemia that were shown to be statistically signi cant were delivery at < 28 weeks of gestation (aOR: 10.93 (95% CI:1.78-66.79)) and AA supplementation from the rst day of life (aOR: 0.11 (95%CI 0.02-0.56, p<0.01)), adjusted for maternal aged>35, primipara, multiple pregnancy, antenatal steroid, tocolytic agents, cesarean section, infant sex, small for gestational age (SGA), non-reassuring fetal status (NRFS), Apgar score<3 at 1min, and calendar year. AA supplementation on the rst day of life was associated with an 89% risk reduction and may be protective against neonatal hyperkalemia.

Discussion
This retrospective cohort study showed that parenteral AA supplementation on the rst day of life was associate with reduced risk of hyperkalemia in preterm infants.
Blanco et al reported that a small randomized controlled study did not show a signi cant association between serum insulin levels and the incidence of hyperkalemia among preterm infants when a higher AA supplementation group (2 g/kg/day) was compared with a lower AA supplementation group (0.5 g/kg/day) 5 . However, they described the incidence of hyperkalemia as lower than expected, which may have reduced the statical power needed to show intervention effects. We also suggested that any AA supplementation (i.e., both lower and higher levels) could help prevent catabolism as compared to no supplementation at all. Bonsante et al. conversely reported that early AA supplementation strongly reduced neonatal hyperkalemia among preterm infants in a prospective observational study. 14 Tocolytic agents did not affect the development of neonatal hyperkalemia in this study. We think that in preterm infants before 32 weeks of gestation, the effects of prematurity on hyperkalemia appeared to be greater than those of tocolytic agents.
In multivariate analysis, antenatal steroid did not show a signi cant effect in preventing hyperkalemia. One of the reasons was that the use of antenatal steroid in the present study was quite lower than that of previous reports (42 % vs. 80-90 %) 15 . This is because long-term tocolysis and bed rest masked the threatened preterm labor, and the obstetricians in our hospital did not administer antenatal steroid for pregnant women with hypertension disorder and infection. Also, even with maternal steroids, the continuous tocolytics often result in a signi cant delay in labor and might reduce the effect of maternal steroid.
A key limitation in this study is long-term enrolment. Early AA supplement increased in 2014-2018 compared to 2009-2013. Despite multivariate analysis using set with calendar year, epoch effects could not be zero. Changes in management, i.e., respiratory management, postnatal steroid use, and antivascular endothelial growth factor therapy for ROP might have affected the results for mortality or morbidity. 16,17 Potential confounders that were not included as variables might have affected the incidence of neonatal hyperkalemia. Therefore, multicenter observational studies are required to con rm the present ndings.
In conclusion, our results indicate that early AA supplementation from the rst day of life may have a prophylactic effect on neonatal hyperkalemia in preterm infants, in addition to the known bene t as nutritional support.

Ethics
All aspects of the study were approved by the ethics committee of Tokushima University (approval number 3755) and were carried out in accordance with the approved guidelines. Because this is a retrospective study, it was di cult to obtain appropriate informed consents from each subject. Therefore, we gained consents using opt-out, which is a way for investigators to give subjects an opportunity for all participants or their guardians to refuse to participate in this study by announcing the detail of this study on the Tokushima University website (http://www.tokushima-hosp.jp/about/disclosure_document.html). The review board of ethics committee of Tokushima University Hospital has approved the way of opt-out consent in this retrospective study.

Enrollment
Inclusion criteria for this study were preterm infants who were inborn at 24 0/7 -31 6/7 weeks of gestation and were admitted between 2009 and 2018 to the neonatal intensive care unit (NICU) at Tokushima University Hospital. Patients were excluded if any of following was applicable: chromosomal abnormalities, congenital major malformations, outborn, or death prior to 48 h of life. Basic clinical information was extracted from the database in the NICU, and individual medical records were reviewed by multiple authors (MT, ST, OT, and SM).

Fluid therapy starting from the rst day of life
Fluid therapy was principally determined based on the judgment of attending physician. Early parenteral AA supplementation was de ned as having started within 24 h after birth. Controls were infants supplied with parenteral AA after 24 h of life. The other regimen of uid therapy on the rst of life was also obtained including glucose and gluconate calcium.

Neonatal hyperkalemia
The primary outcome was the occurrence of neonatal hyperkalemia. Peak K + levels during the rst 72 h were recorded, and neonatal hyperkalemia was de ned as K + > 6.5 mEq/l with urine output ≥1 ml/kg/h . 2,18 Other data for blood examinations were also obtained at the time of the highest serum K + concentration.

Neurological impairments and morbidites
Neurological impairments were cerebral palsy, mental retardation, attention de cit hyperactivity disorder, autism spectrum disorder, and learning disorder. These neurological impairments were adjudicated by senior pediatricians who were not otherwiseinvolved in this study. Neonatal morbidities were also recorded, including BPD, IVH, PVL, NEC, ligation of PDA, and ROP requiring treatment. BPD was de ned as oxygen or respiratory support at a corrected age of 36 weeks. IVH above grade 2 according to the Papile classi cation was extracted in this study. 19

Statistical analysis
Continuous variables are shown as medians and interquartile ranges, and categorical variables are shown as counts and percentages. The chi-squared test and Fisher's exact test were used to compare categorical variables appropriately, and the Wilcoxon rank-sumtest was used to compare continuous variables, because all continuous variables were determined to be nonparametric by the Shapiro-Wilk normality test. p<0.05 was considered the threshold for statistical signi cance. Multivariate analysis were performed with the following explanatory variables: AA supplementation that started within 24 h after birth, risk factors and possible prophylactic management for the occurrence of hyperkalemia determined by clinical relevance and the settings used in previous reports 2, 11 . These included maternal age, primipara, multiple pregnancy, antenatal steroid, tocolytic agents, cesarean section, delivery before 28 weeks of gestation, non-reassuring fetal status, infant sex, small for gestational age (de ned as an infant with weight below the 10th percentile for gestational age and height below the 10th percentile for gestational age), 20 Apgar score at 1 min after birth <3, and calendar year (in 5 year intervals). JMP version 15.2 (SAS Institute, Cary, NC) was used for all statistical analyses.

Data availability
Data and materials used in this study are available upon reasonable request to the corresponding author and under a collaboration agreement.