Neonatal Magnesemia in Preterm Neonates Unexposed to Maternal MgSO4 Administration and in Neonates Exposed for Fetal Neuroprotection or Maternal Eclampsia Prevention.

Objective - To compare neonatal magnesemia in the rst fteen days of neonatal life between three groups: a control group not exposed to MgSO 4 , a neuroprotection group, and an eclampsia prevention group, and to explore its’ associations with child outcomes. Design - Retrospective single-centre cohort study. Setting - Tertiary care setting. Population - Infants admitted at the neonatal intensive care unit born between 24 and 32 weeks’ gestation, regardless of etiology of preterm birth. Methods - Linear mixed regression of neonatal magnesemia on exposure group and day of life. Generalised estimating equations models of child outcomes on neonatal magnesemia according to exposure group and day of life. Main outcome measures - Neonatal magnesemia (mmol/l). Results - Neonatal magnesemia is signicantly higher in the preeclampsia group compared to the control and neuroprotection group. On the day of birth, this is irrespective of maternal magnesemia (preeclampsia vs control group), and the maternal total dose or duration of MgSO 4 administration (preeclampsia vs neuroprotection group). No differences were found in short-term composite outcome between the three groups. Conclusions - We found mean differences in neonatal magnesemia between children not exposed to MgSO 4 antenatally, children exposed for fetal neuroprotection, and children exposed for maternal eclampsia prevention. A 4g loading and 1g/h maintenance dose, for fetal neuroprotection and eclampsia prevention, appears to be safe on the short term for the neonate. the short term. on neonates at a gestational between and 31 gestation from were extracted from the hospital preterm birth register. The preterm birth register was created in 2016 and provides maternal and neonatal data on demographics, procedures/interventions, diagnoses, short- and long-term morbidities and mortality for preterm births between 24 +0 and 33 +6 weeks’ gestation. Obstetrical data were retrospectively entered in the database by senior clinicians. Neonatal data were extracted from an already operational neonatal database and imported into the register. The database was built and managed using the REDCap® electronic data capture tool. The preterm birth register was registered at clinicaltrials.gov (NCT03405116). Patient involvement was limited to consenting in registration and use of the data for scientic purposes. The funding body played no role in the creation of the manuscript. birth dened or start and preterm birth is no longer If an imminent risk of preterm delivery re-emerges, MgSO 4 administration is repeated. In the setting of (pre)eclampsia, MgSO 4 infusion is stopped 48 hours after delivery. MgSO 4 administration for fetal neuroprotection was introduced in 2014, all neonates not exposed to MgSO 4 make up the control group. MgSO 4 was not given to the neonates. 1.3, 4.9] versus the control group, and + 2.7 kg/m 2 [95%CI 0.5, 4.8] versus the neuroprotection group) and neonatal birth weight is, on average, lower (-283.0 g [95%CI -390.0, -176.0] versus the control group and -210.7 g [95%CI -348.0, -73.2] versus the neuroprotection group). There is a difference in number of multiple pregnancies, with less multiple pregnancies in the preeclampsia group (-22.7% [95%CI -36.5, -0.9] versus the control group, and -17.3% [95%CI -35.2, 0.5] versus the neuroprotection group). There are no other signicant differences between groups found in our sample. total maternal MgSO 4 dose and neonatal magnesemia during the rst four days and between duration of administration and neonatal magnesemia during the rst three days. Corrected for maternal magnesemia, total dose or duration of administration, the neonatal magnesemia was higher in the preeclampsia group compared to the neuroprotection group on the day of birth. No group differences were found in short-term outcome.


Introduction
For several decades now, the administration of magnesium sulphate (MgSO4) is current practice in obstetrics. Historically MgSO 4 was used as a tocolytic agent. However, the utility of magnesium sulphate in this context is questionable [1]. Current guidelines recommend the use of MgSO4 for prevention or treatment of seizures in women with (pre)eclampsia or for fetal neuroprotection when extreme or early preterm birth is imminent. MgSO 4 administered for fetal neuroprotection reduces the risk of gross motor dysfunction and moderate to severe cerebral palsy [2][3][4][5][6][7][8].
While severe adverse effects (such as respiratory depression, cardiac arrest, coma and eventually death) due to an iatrogenic overdose are reported, most maternal adverse effects are minor and well tolerated [9]. Despite widespread use by obstetricians, controversy over unintended adverse neonatal outcomes following maternal magnesium therapy remains. Higher maternal serum magnesium concentrations have been signi cantly associated with neonatal complications, including low Apgar scores at 1 and 5 minutes, respiratory depression, hypotension, hypotonia, hypore exia, neonatal intensive care unit admission, intraventricular hemorrhage, and spontaneous intestinal perforation [10][11][12][13][14]. On the other hand, a recently published systematic review concluded that antenatal MgSO 4 administration was not associated with neonatal morbidities or perinatal death [15]. A limitation of most trials examining adverse neonatal effects of maternal MgSO4 administration is grouping all magnesium-exposed neonates together, regardless of its indication. The indication however may have an in uence on maternal and neonatal magnesemia.
The primary aim of this study was to compare neonatal magnesemia in the rst fteen days of neonatal life between three groups of infants born before 32 weeks' gestation: a control group that was not exposed to MgSO 4 antenatally, a neuroprotection group of neonates whose mothers received MgSO 4 for fetal neuroprotection, and a preeclampsia group of neonates whose mothers received MgSO 4 to prevent eclampsia.
Secondary aims were to explore correlations between neonatal magnesemia and maternal serum magnesium concentration, maternal total dose of magnesium, and duration of exposure to magnesium before delivery. We also explored if neonatal magnesemia was associated with adverse short-term outcomes.

Study design, data source and collection
We conducted a retrospective single-centre cohort study in Ghent University Hospital, a referral centre for high-risk pregnancies in Belgium.
Data on neonates born at a gestational age between 24 +0 and 31 +6 weeks' gestation from January 2012 to December 2015 were extracted from the hospital preterm birth register. The preterm birth register was created in 2016 and provides maternal and neonatal data on demographics, procedures/interventions, diagnoses, short-and long-term morbidities and mortality for preterm births between 24 +0 and 33 +6 weeks' gestation. Obstetrical data were retrospectively entered in the database by senior clinicians. Neonatal data were extracted from an already operational neonatal database and imported into the register. The database was built and managed using the REDCap® electronic data capture tool. The preterm birth register was registered at clinicaltrials.gov (NCT03405116). Patient involvement was limited to consenting in registration and use of the data for scienti c purposes. The funding body played no role in the creation of the manuscript.

Population
Neonates born at a gestational age between 24 +0 and 31 +6 weeks' gestation, regardless of the etiology of preterm birth, admitted at the neonatal intensive care unit were included in this study. Neonates with major congenital malformations are not included in the preterm birth register. A major congenital malformation is de ned as a malformation with higher neonatal morbidity or mortality.

Intervention and comparison
In Ghent University Hospital, the dosage of MgSO 4 for fetal neuroprotection is the same as the dosage for eclampsia prevention. A four gram loading dose is given over 15 to 20 minutes and is followed by a maintenance dose of one gram per hour (Zuspan's regimen) [15]. In the setting of fetal neuroprotection, the infusion is stopped when delivery does not occur within 24 hours after start and preterm birth is no longer threatening. If an imminent risk of preterm delivery re-emerges, MgSO 4 administration is repeated. In the setting of (pre)eclampsia, MgSO 4 infusion is stopped 48 hours after delivery. MgSO 4 administration for fetal neuroprotection was introduced in 2014, all neonates not exposed to MgSO 4 make up the control group. MgSO 4 was not given to the neonates.

Endpoints
The primary endpoint is neonatal magnesemia, measured repeatedly at un xed time points in the rst fteen days of neonatal life. If more than one serum magnesium value was available on one day, the rst value of that day was taken.
Short-term neonatal core outcome includes neonatal intensive care unit mortality, intraventricular hemorrhage (IVH) and/or periventricular leukomalacia (PVL). IVH and PVL were detected on routine brain ultrasound and respectively scored according to Papille classi cation criteria and the four-grade classi cation by de Vries et al [16,17].

Statistical analysis
Maternal and neonatal characteristics were compared across the three exposure groups. Continuous variables were compared between groups with one-way ANOVA and pairwise signi cant differences were identi ed using the Tukey honest signi cant differences test. Categorical data on the level of the mother were compared using a Chi-square or Fisher's exact test. Correlations between two continuous variables were assessed using the Pearson correlation coe cient.
To account for clustering due to multiple pregnancies and repeated outcome assessments over days, linear mixed models (LMMs) with two random intercepts were tted for log-transformed neonatal magnesemia. For the main research question, neonatal day of life (day of birth (= day 0) until day 14), exposure group, and their two-way interaction were included in the xed effects part of the model. For the other research questions, the models included a three-way interaction between neonatal day of life, exposure group, and maternal magnesemia / total maternal dose of MgSO 4 / duration of MgSO 4 administration, and all underlying effects. The estimated marginal means with their 95% con dence interval (CI) are plotted by neonatal day of life according to exposure group. No indication of multi-collinearity was found based on the variance in ation factor (< 2.5).
Generalised estimated equation (GEE) models with an independence correlation structure, Gaussian distribution and identity link function were used to assess the association between neonatal magnesemia and a composite short-term outcome.
Neonatal and maternal magnesemia, total maternal dose of MgSO 4 , and duration of MgSO 4 administration were log-transformed for all analyses. Regression coe cients from analyses with log-transformed dependent outcome were exponentiated to infer associations regarding the geometric mean. With log-transformed independent variables, the estimated % change in geometric mean neonatal magnesemia for each 10% increase in the independent variable was reported.
Subgroup analyses were performed according to exposure group. No analyses regarding maternal magnesemia were performed in the neuroprotection subgroup, due to the limited number of measured maternal magnesemia values (n = 6). Evidently, no analyses regarding total maternal dose of MgSO 4 and duration of MgSO 4 administration were performed in the control subgroup.
All hypothesis testing was performed at the two-sided 5% signi cance level. No adjustment for multiple testing was done. All analyses were performed using R version 4.0.5. The package "lme4" was used to construct the LMMs and the "geepack" package to t the GEE models.

Ethical considerations
The preterm birth register was approved by the Medical Ethics Committee of Ghent University Hospital on May 5th 2017 with registration number BE670201732322. This study was approved on February 26th 2018 with registration number BE670201835532. Data were gathered after informed consent was obtained and were handled with professional con dentiality. Withdrawal from the study was possible at any time.

Results
Demographics and characteristics of the study cohort Between 2012 and 2015, 345 neonates were born alive before 32 weeks' gestation and admitted to the neonatal intensive care unit. One hundred and three neonates (29.9%) were part of a twin and ve of a triplet (1.4%). Two hundred and ninety six mothers were included.
There are 218 neonates in the control group (63.2%), 68 in the neuroprotection group (19.7%), and 59 in the preeclampsia group (17.1%). Demographics and characteristics are summarised in Table 1  There is a moderately positive correlation between the total maternal dose of MgSO 4 and the maternal magnesemia in the whole population (pearson = 0.64) and in the preeclampsia group (pearson = 0.60). The neonatal magnesemia is, for all days and in all groups, correlated with the neonatal magnesemia of the subsequent day. Table 2 provides a summary of the observed maternal and neonatal magnesium values. Primary aim (Online Resource S1) The distribution of neonatal magnesemia according to neonatal day of life (day 0-14) and indication for MgSO 4 administration is visualised in  Table S1.
In all three groups, there is no indication of an association of maternal BMI and neonatal birth weight with neonatal magnesemia.
Maternal magnesemia (Online Resource S1) In our cohort, there is a signi cant association between maternal magnesemia and neonatal magnesemia during the rst four days of life (day 0-3) in both the control and the preeclampsia group. The association remains signi cant until the eight day of life (day 0-7) in the preeclampsia group (Online Resource Table S3).
The estimated geometric mean neonatal magnesemia per neonatal day of life per exposure group for a maternal magnesemia corresponding to the overall geometric mean of 1.30 mmol/l is depicted in Figure 2b and numerically summarised in Online Resource Table S2. Only on the rst day of neonatal life (day 0), independent of the maternal magnesemia, neonatal magnesemia is signi cantly higher in the preeclampsia group compared to the control group.
Total maternal dose of MgSO 4 (Online Resource S1) Most women in the neuroprotection group received MgSO 4 for less than 24 hours; only 12 out of 68 women in the neuroprotective group received the maximum dose of 28 grams.
There is a signi cant association between total maternal dose and neonatal magnesemia during the rst four days of life (day 0-3) in both the neuroprotection and the preeclampsia group. The association is signi cant until the seventh day of life (day 0-6) in the preeclampsia group (Online Resource Table S3).
The estimated geometric mean neonatal magnesemia per neonatal day of life per exposure group for a total maternal dose of MgSO 4 corresponding the overall geometric mean in these two exposure groups of 23.82 grams, is depicted in Figure 2c and numerically summarised in Online Resource Table S2. In our cohort, on the two rst days of neonatal life (day 0-1) neonatal magnesemia is signi cantly higher in the preeclampsia group compared to the neuroprotection group, independent of the total maternal MgSO 4 dose.
Duration of MgSO 4 exposure (Online Resource S1) There is a signi cant association between the duration of MgSO 4 administration and the neonatal magnesemia during the rst three days of life (day 0-2) in both the neuroprotection and the preeclampsia group. The association remains signi cant until the seventh day of life (day 0-6) in the preeclampsia group (Online Resource Table S3).  Figure S1c).
The estimated geometric mean neonatal magnesemia per neonatal day of life per exposure group for a duration of MgSO 4 administration corresponding to the overall geometric mean of 850 minutes in these two exposure groups is depicted in Figure 2d and numerically summarised in Online Resource Table S2. In our cohort, on the two rst days of neonatal life (day 0-1) neonatal magnesemia is signi cantly higher in the preeclampsia group compared to the neuroprotection group, independent of the duration of MgSO 4 administration.
Short-term outcome (Online Resource S1) Ninety three neonates (27.0%) experienced the composite short-term neonatal outcome (mortality, IVH and/or PVL). No signi cant differences in short-term neonatal outcome between the three exposure groups were identi ed in our cohort.

Main ndings
We used LMMs to explore the association of neonatal magnesemia with antenatal MgSO 4 exposure. In our cohort, neonatal magnesemia is signi cantly higher in the preeclampsia group compared to the neuroprotection and control group, and in the neuroprotection group compared to the control group, during the rst four days of life. There was an association between maternal and neonatal magnesemia during the rst four days in the control and preeclampsia group (too few maternal magnesemia values in neuroprotection group). Within the preeclampsia and neuroprotection groups, there was an association between total maternal MgSO 4 dose and neonatal magnesemia during the rst four days and between duration of administration and neonatal magnesemia during the rst three days. Corrected for maternal magnesemia, total dose or duration of administration, the neonatal magnesemia was higher in the preeclampsia group compared to the neuroprotection group on the day of birth. No group differences were found in short-term outcome.

Strengths and limitations
This is the rst study exploring neonatal magnesemia in three distinct groups during the rst fteen days of life. Most studies only take into account the day of birth and/or the day after. Furthermore, few studies compare neonates not exposed to MgSO 4 to neonates primarily exposed for neuroprotection or preeclampsia.
We recognise the limitations of a single-centre retrospective study design with a relatively small sample size. Maternal pre-delivery magnesemia levels were missing in the majority of neuroprotection group patients. We could not explore long-term outcomes due to considerable loss to follow-up (Online Resource S2).
Since few women in the neuroprotection group received the maximum dose (28 grams), caution is warranted in making safety statements about 'neuroprotective dosage'. However, results from the preeclampsia group, with higher doses in more pathological circumstances, were reassuring. done. On the other hand, a lack of power might be present due to small sample sizes, resulting in less associations found than truly present. Our analyses should be considered exploratory.

Interpretation
Higher levels of neonatal magnesemia in the neuroprotection group compared to unexposed neonates have been reported [11,18,19].  [20]. Mean neonatal and maternal magnesemia were signi cantly associated. It is not clear when neonatal magnesemia was measured. We found an association between neonatal and maternal magnesemia from day 0 to 3 in both control and preeclampsia group.
Choi et al. explored the association between maternal BMI and maternal and umbilical cord magnesemia in children born at less than 32 weeks. Maternal and adjusted umbilical cord magnesemia were not signi cantly different between BMI categories [21]. In our cohort, there was no association between BMI and neonatal magnesemia.
Two studies found a correlation between the total maternal dose and neonatal magnesemia on day 0 when MgSO 4 is given for neuroprotection, as we did [18,22]. Borja-Del-Rosario et al. excluded patients with preeclampsia [18]. They did not detect a correlation between maternal and neonatal magnesemia, nor between total MgSO 4 dose and maternal magnesemia [18,22]. We have too few values in the neuroprotection group to con rm or refute these ndings.
Nassar et al. reported a higher neonatal magnesemia in a group exposed more than 48 hours to MgSO 4 as tocolysis compared to shorter exposure [23].
We cannot explain why neonatal magnesemia was higher in the preeclampsia on the day of birth. It might be due to placental dysfunction, or could be multifactorial.
In general, antenatal MgSO 4 is considered to be safe for the neonate [15,24,25]. However, there are reports of an association between high neonatal magnesemia and neonatal morbidity and mortality [11,12,26,27]. Basu et al. found that neonatal mortality, in children born between 24 and 32 weeks, increased with increasing neonatal magnesemia. Mortality was highest when neonatal magnesemia exceeded 2.25 mmol/L during the rst day of life [11]. In our cohort, none of the neonates in the neuroprotection group had a value exceeding 2.25 mmol/L. In the preeclampsia group, ve neonates had a higher magnesemia. There were no cases of neonatal mortality. They found no association with survival without IVH and/or PVL. In the meta-analysis of Shepherd et al., no differences in perinatal death were identi ed between exposed and unexposed neonates [15]. Only in one (of 11 non-randomised trials) cohort study, with moderate to high bias risk, an increased risk of perinatal death was observed when the dose was more than 48 grams. Possible harms were mostly seen in studies not correcting for confounders, studies with small sample sizes, or in subgroup analyses [15]. Mittendorf et al. concluded that exposure to 50 grams or more of tocolytic MgSO 4 is indirectly associated with IVH [28]. We didn't nd any between group differences in the proportions of our short-term outcome, which included IVH. In the preeclampsia and neuroprotection group, respectively 36 and two women received a dose of 50g or more. Garcia-Alonso et al. studied outcomes in exposed and unexposed children born before 29 weeks. Eighteen per cent of exposed neonates had preeclamptic mothers. They found a signi cant correlation between MgSO 4 dose and neonatal magnesemia on day 0 in the whole group, as we did in the preeclampsia and neuroprotection group. They reported a lower mortality in the exposed group and no differences in neonatal morbidity [29].

Conclusion
We found mean differences in neonatal magnesemia between children not exposed to MgSO 4 , children exposed for fetal neuroprotection, and children exposed for maternal eclampsia prevention. After correction for maternal magnesemia, a higher neonatal magnesemia was still present in the preeclampsia group compared to the control group on the day of birth (no comparison with neuroprotection group). After correction for total dose or duration of administration, a higher neonatal magnesemia was still present in the preeclampsia compared to the neuroprotection group till day two of life. When considering neonatal magnesemia per day of neonatal life by exposure group, there was an association with maternal magnesemia, total dose, and duration of administration the rst days. A 4g loading and 1g/h maintenance dose, for fetal neuroprotection and eclampsia prevention, appears to be safe on the short term for the neonate.