Antenatal magnesium sulfate treatment and risk of necrotizing enterocolitis in preterm infants born at less than 32 weeks of gestation

Antenatal magnesium sulfate (MgSO4) treatment is widely used for fetal neuroprotection in women at risk of preterm delivery. However, some studies have recently suggested that in utero MgSO4 exposure is associated with an increased risk of necrotizing enterocolitis (NEC). This study aimed to investigate the association between antenatal MgSO4 treatment and risk of NEC. This retrospective cohort study included 756 infants born at 24‒31 weeks’ gestation. Subjects were classified into three groups: period 1, when MgSO4 treatment protocol for fetal neuroprotection was not adopted (n = 267); period 2, when the protocol was adopted (n = 261); and period 3, when the protocol was withdrawn because of concern of risk of NEC (n = 228). Rates of NEC (≥ stage 2b) were analyzed according to time period and exposure to antenatal MgSO4. Significant difference in the rate of NEC was not found across the three time periods (2.6% vs. 6.5% vs. 4.8% in periods 1, 2 and 3, respectively, p = 0.103). The rate of NEC was comparable between the infants exposed and unexposed to antenatal MgSO4 (5.1% vs. 3.6%, p = 0.369). These results showed that antenatal MgSO4 treatment was not associated with risk of NEC in our study population.

www.nature.com/scientificreports/ gastrointestinal complications have been controversial, the role of antenatal MgSO 4 in association with gastrointestinal complications is not well established yet 8 . In our institute, the protocol of antenatal MgSO 4 treatment for fetal neuroprotection was adopted in 2014. However, concern about the risk of neonatal gastrointestinal complications including NEC and SIP has led to treatment withdrawal after two years. Therefore, the objective of this study was to investigate whether these protocol changes might have any effect on the risk of NEC by comparing outcomes according to time periods classified by the adoption and withdrawal of antenatal MgSO 4 treatment protocol for fetal neuroprotection. In addition, we investigated whether antenatal MgSO 4 treatment per se or for fetal neuroprotection was associated with the risk of NEC in preterm neonates at 24-31 weeks of gestation.

Results
A total of 756 neonates were enrolled in this study, including 267 (35.3%) during period 1, 261 (34.5%) during period 2, and 228 (30.1%) during period 3. Maternal characteristics including maternal age, parity, and body mass index were similar across the three time periods. However, distributions of singleton, twin, and triplet pregnancies were significantly different among the three time periods (Table 1). Rates of antenatal MgSO 4 treatment were significantly different across the three time periods (20.9%, 62.1% and 14.9% in periods 1, 2, and 3, respectively, p < 0.001). The most common indication of MgSO 4 treatment during period 2 was fetal neuroprotection whereas that during periods 1 and 3 was severe preeclampsia. A total dose of MgSO 4 used during period 2 was significantly lower than that during period 1 or period 3 because the period of antenatal MgSO 4 treatment for fetal neuroprotection was shorter than those used for severe preeclampsia or tocolysis.
Neonatal outcomes including gestational age at delivery, birth weight, sex, and rate of 1-min Apgar score < 4 were similar across the three time periods. However, the rate of 5-min Apgar score < 7 was significantly lower during period 3 than that during period 1 ( Table 2). NEC (≥ stage 2b) occurred in 41 among 756 neonates (5.4%). There was no significant difference in the rate of NEC (≥ stage 2b) across the three time periods (2.6% vs. 6.5% vs. 4.8% in periods 1, 2, and 3, respectively, p = 0.103). The median postnatal age at the onset of NEC was 18.5 days Table 1. Maternal baseline characteristics and pregnancy outcomes according to time period. Data are presented as number (percentage) or mean ± standard deviation. MgSO 4 magnesium sulfate, PTL preterm labor, PPROM preterm premature rupture of membrane. 1 Significantly different between period 2 and period 3 (Bonferroni test). 2 Significantly different between period 1 and period 2 (Bonferroni test). 3 Significantly different between period 1 and period 3 (Bonferroni test). www.nature.com/scientificreports/ (range, 3-63 days). Early-onset NEC occurred in 14 (40.0%), and late-onset NEC occurred in 21 (60.0%), but there was no significant difference in the age of onset of NEC across the three time periods. Laboratory findings of NEC including metabolic acidosis (pH < 7.2) and thrombocytopenia (platelet < 150,000/μL), and radiologic and sonographic findings of NEC including fixed loop or severe ileus, portal venous gas, pneumoperitoneum/ pneumatosis intestinalis, and ascites were similar across the three time periods. Among 35 neonates with NEC (≥ stage 2b), 23 had NEC perforation which was confirmed by pathologic examination after surgery. The most common site of NEC perforation was jejunum, but the locations of NEC perforation were not significant different across the three time periods. The rate of SIP during period 3 was higher than that during period 1.
Rates of NEC (≥ stage 2b), NEC perforation, and SIP were not significantly different between the infants exposed and unexposed to antenatal MgSO 4 ( Table 3). And the results were similar in the subgroup analyses  www.nature.com/scientificreports/ stratified by gestational age at delivery (preterm birth less than 26 weeks of gestation and preterm birth at 26-31 weeks of gestation). Rates of NEC (≥ stage 2b), NEC perforation, and SIP were also not significantly different between the infants exposed and unexposed to antenatal MgSO 4 for fetal neuroprotection, and the results were similar in the subgroup analyses stratified by gestational age at delivery (Table 4).

Discussion
In this study, we investigated the effect of antenatal MgSO 4 treatment on the risk of NEC by comparing neonatal outcomes before and after two major changes in the antenatal MgSO 4 treatment protocol (adoption and withdrawal of antenatal MgSO 4 treatment for fetal neuroprotection). We found that protocol changes had no effect on the rate of NEC in this population. We also found that neither antenatal MgSO 4 treatment per se or for fetal neuroprotection was associated with the risk of NEC in preterm neonates born at 24-31 weeks of gestation in this population. These findings are consistent with previous randomized control trials and other recent studies reporting that there is no significant difference in the incidence of NEC between placebo and antenatal MgSO 4 treatment groups 9,13-20 . If magnesium causes intestinal injury, the pathogenesis for that injury is unclear. However, the biologic effect of magnesium on intestinal muscle contractility may be a possible mechanism to cause gastrointestinal complications. Magnesium ions can cross the placenta rapidly and its fetal levels increase in proportion with its maternal levels 21,22 . Magnesium ions can replace calcium ions, disrupt actin and myosin interactions, and reduce contractility, thus generating atony of intestine and fecal impaction 23 . They can also increase mesenteric arterial resistance and reduce mesenteric blood flow 24 . The hypomotility of intestine is also due to cholinergic effect of magnesium 24 . Hypomotility may lead to increased water absorption, the formation of stool plugs, and overgrowth of bacteria. An intestinal plug can increase proximal intraluminal pressure which can be particularly detrimental in setting of immature intestine along with decreased blood flow to the intestine of preterm neonates 8 . However, the effect of antenatal MgSO 4 exposure on development of NEC is not fully understandable, especially in infants who developed late-onset NEC.
Feeding issue, including enteral or parenteral feeding, breast milk or formula feeding, and feeding intolerance, is another important factor for development of NEC in premature infants 25 . In our study, early-onset NEC (within 14 days after birth) occurred in 14/35 (40.0%), and this incidence was similar to that of a large population-based cohort of 16,669 infants with gestational age < 33 weeks of gestation 26 . The development of these early-onset NEC might be associated with failure of or delayed breast milk feeding 27 . Cesarean section, especially emergency cesarean section, is associated with reduced breast milk feeding success or delayed breast milk feeding initiation 28 . In our study, cesarean section rate was over 70% and most of them are performed as emergency surgery. However, although there were no changes in the feeding protocol during the entire study period, we were not able to collect the individual feeding data because this study was a retrospective medical record review. This is one of the main limitations of this study.
The development of NEC is associated with many other contributing factors 25,29 . Among them, the most important risk factor is prematurity. The earlier the gestational age at delivery, the higher the risk of NEC 25 . In our study, the incidences of NEC and SIP in the infants born at less than 26 weeks of gestation were higher than those in the infants born at 26 weeks of gestation or more. Previous studies reporting the association of antenatal MgSO 4 with NEC and/or SIP have included extremely preterm babies born at less than 25-26 weeks of gestation 11,12 . Other studies reporting a lack of association between antenatal MgSO 4 treatment and NEC or SIP have included preterm infants born at ≤ 28-34 weeks of gestation [13][14][15][16] . We included infants born at 24-31 weeks of gestation, because the indication of antenatal MgSO 4 treatment protocol for fetal neuroprotection in our institute was those who were at risk of preterm birth at 24-31 weeks of gestation, and the objective of this study was to investigate whether these protocol changes might have any effect on the risk of NEC in this population. Our results showed that antenatal MgSO4 treatment per se or for fetal neuroprotection was not associated with increased risk of NEC or SIP in babies born at less than 26 weeks of gestation, as well as in babies born at www.nature.com/scientificreports/ 26-31 weeks of gestation. However, our sample size was limited to analyze the association between antenatal MgSO 4 treatment and NEC or SIP in extreme preterm neonates born at less than 26 weeks of gestation. Therefore, a further study is needed to assess the effect of antenatal MgSO 4 treatment on gastrointestinal complications such as NEC, SIP, and meconium ileus in extreme preterm infants. The optimal antenatal MgSO 4 regimen for fetal neuroprotection in terms of dose, maintenance dose, duration, timing, or repeat treatment has not been established yet 30 . Different protocols of antenatal MgSO 4 treatment might be another plausible cause of conflicting results about the effect of antenatal MgSO 4 treatment on the risk of NEC. The antenatal MgSO 4 treatment protocol for fetal neuroprotection in previous studies consisted of loading dose varying from 4 to 6 g intravenous over 20-30 min and maintenance dose varying from none to 1 g or 2 g per hour for 12 h or 24 h 30 . In studies showing adverse neonatal gastrointestinal outcomes in the setting of antenatal MgSO 4 exposure 11,12 , higher loading dose (6 g) and higher maintenance dose (2 g per hour) were used than those in our study. Total dose of antenatal MgSO 4 treatment for fetal neuroprotection was higher (50.9 ± 45.7 g or 33.0 ± 19.8 g) 11,12 than that of our study (6.3 ± 10.3 g). However, studies that used lower dose of MgSO 4 (4 g of loading dose only or 4 g of loading dose and 1 g per hour of maintenance dose for 24 h), including the Australasian collaborative trial (ACTOMgSO4), European Trial (PREMAG), and our current study, found no relationship between antenatal MgSO 4 treatment and NEC or SIP 16,18,31 . Therefore, it is necessary to establish optimum regimen for fetal neuroprotection to achieve maximal effectiveness with minimal adverse effects for both fetus and mother, especially for those at risk of preterm birth at less than 26 weeks of gestation.
The strength of our study is that it is a relatively large cohort study examining the association of antenatal MgSO 4 treatment with risk of NEC in preterm neonates delivered from 24 + 0 weeks to 31 + 6 weeks. Our study investigated neonates with a large sample size of 756 in spite of a single center study like previous two studies 13,14 . Whereas previous studies investigated neonates with a relatively small sample size to demonstrate the association of antenatal MgSO 4 with NEC and/or SIP 11,12 . However, our study may be underpowered because the sample size was not enough to show a difference in neonatal outcome, especially for the subgroup analyses stratified by gestational age at delivery. Besides, by comparing neonatal outcome before and after two major changes in the antenatal MgSO 4 treatment protocol, we were able to analyze the effect of antenatal MgSO 4 treatment on the risk of NEC. However, subjects in the three time period groups were not comparable regarding baseline characteristics and other predictor variables. Therefore, we performed multivariable analyses to adjust for potential confounding factors. In addition, there might be other unidentifiable confounding factors such as feeding issues and practice changes in obstetrics and neonatology over the study period that should have been controlled for the multivariable analysis.
This study was further limited by its inherent disadvantage of a retrospective study design including information bias. However, we tried to minimize information bias by blinding chart reviewers to study question and hypothesis. This study was also limited by inability to determine the total dose, timing, and retreatment of MgSO 4 administration, especially in women who were transferred from other hospitals.
In conclusion, antenatal MgSO 4 treatment for fetal neuroprotection was not associated with NEC, mortality, or other short-term neonatal outcomes in this population. A further study is needed to uncover the long-term neurodevelopmental outcome of these babies.

Methods
This was a retrospective cohort study of neonates delivered from 24 + 0 weeks to 31 + 6 weeks of gestation in Samsung Medical Center, Seoul, Korea from January 2012 to June 2018. Fetal death and major congenital anomalies were excluded. Medical records of mothers and babies were examined independently by obstetricians and neonatologists. Researchers who were blinded to maternal data including antenatal MgSO 4 treatment reviewed the neonatal data for outcome. This study was approved by the Institutional Review Board for Clinical Research at Samsung Medical Center (IRB No 2019-04-051) and exemption for informed consent was granted because this was a retrospective chart review study. All methods were carried out in accordance with relevant guidelines and regulations.
Subjects were classified into three groups: period 1, from January 2012 to December 2013 when antenatal MgSO 4 treatment for fetal neuroprotection was not adopted; period 2, from January 2014 to March 2016 when the treatment was routinely used; and period 3, from April 2016 to June 2018 when the treatment was withdrawn because of concern of increased risk of NEC.
Maternal demographic characteristics included maternal age at delivery, body mass index, parity, and plurality. Pregnancy outcomes reported included indications of antenatal MgSO 4 treatment, type and cycle of antenatal corticosteroids therapy, type of tocolysis treatment, indications for delivery, mode of delivery, and histologic chorioamnionitis. Indications for delivery were classified into preterm labor, preterm premature rupture of membranes, and maternal-fetal indication such as preeclampsia, placenta previa, placenta abruption, and intrauterine growth restriction. Indications of antenatal MgSO 4 treatment use were fetal neuroprotection, severe preeclampsia, and tocolysis.
MgSO 4 treatment protocol for fetal neuroprotection consisted of intravenous loading dose of 4 g over 15 to 20 min followed by maintenance infusion of 1 g per hour for 24 h for women having imminent preterm birth within 24 h. Maintenance dose was discontinued if delivery did not occur within 24 h and delivery was no longer considered imminent. It was resumed if the risk of imminent delivery recurred within 6 h. Repeat loading dose and subsequent maintenance therapy were given if risk of imminent delivery recurred after 6 h. MgSO 4 for treatment of severe preeclampsia and eclampsia was given with the same loading dose followed by a maintenance infusion of 1 g per hour and discontinued at 24 h after delivery. MgSO 4 was not used for tocolysis in our institute. Women who were given MgSO 4 for tocolysis in our study population were transferred from other hospitals. Therefore, the exact protocol or dose of MgSO 4 used in these women was unknown. www.nature.com/scientificreports/ Neonatal outcome measures included gestational age at delivery, sex, birth weight, Apgar scores, and neonatal mortality and morbidities. The primary outcome of our analysis was occurrence of NEC (≥ stage 2b). NEC was diagnosed based on modified Bell's staging criteria 29 . Stage 2b was defined as mild metabolic acidosis (pH < 7.2), mild thrombocytopenia (platelet < 150,000/μL), absent bowel sounds, definite abdominal tenderness with or without abdominal cellulitis or right lower quadrant mass, and portal venous gas with or without ascites. Stage 3a was defined as systemic, intestinal, radiologic signs of stage 2b plus hypotension, bradycardia, disseminated intravascular coagulation, neutropenia, signs of generalized peritonitis, marked tenderness, and distention of abdomen, and definite ascites. Stage 3a was defined as systemic, intestinal, radiologic signs of stage 3a plus pneumoperitoneum. Early-onset NEC was defined as NEC diagnosed within 14 days after birth, and late-onset NEC was defined as NEC diagnosed more than 14 days after birth 26 . Spontaneous intestinal perforation (SIP) was defined as radiological evidence of perforation, a lack of clinical features of NEC, an absence of radiological features of intestinal ischemia (fixed dilated bowel loops, pneumatosis intestinalis), and/or an intra-operative surgical report and/or histopathology assessment indicating a perforation located in the ileum and on the antimesenteric border 32 . As a unit policy, minimal enteral feedings (20 mL/kg/day) were initiated within the first 3-4 h after birth and usually maintained for a week for infants with gestational age of less than 26 weeks. After that, feedings were gradually increased to reach full enteral feeding (150 mL/kg/day) according to the infant's tolerance. For infants with gestational age of 26 weeks or greater, feedings were started within the first 3-4 h and advanced by 20 mL/kg/day until full enteral feedings. Infants were fed human breast milk; preterm formula milk was given in the absence of mother's own milk. The management of feeding intolerance and the decision of continuing or discontinuing feedings were done by the attending neonatologists. There were no changes in the feeding protocol during the entire study period.
Obtained data were analyzed using the Statistical Package for Social Sciences version 24 (SPSS Statistics; IBM, Armonk, NY, USA). Maternal characteristics, pregnancy, and neonatal outcomes were compared across the three time periods. They were also compared between the infants exposed and unexposed to antenatal MgSO 4 and between the infants exposed and unexposed to antenatal MgSO 4 for fetal neuroprotection. Continuous variables were compared using independent-sample parametric (analysis of variance) or nonparametric (Kruskal-Wallis test) tests depending on data normality. Categorical variables were compared using Chi-square test or Fisher's exact test when one or more expected value was less than 5. Bonferroni test was used for post-hoc analysis to correct for multiple comparisons. Multiple logistic regression analysis was performed to evaluate effects of potential confounding variables such as plurality, antenatal corticosteroids treatment, nifedipine treatment, and gestational age at delivery. Results were considered statistically significant when p value was less than 0.05. For multiple comparisons, p value was adjusted to 0.017 (0.05/3) by Bonferroni correction.