Melatonin reduces brain injury following inflammation‐amplified hypoxia–ischemia in a translational newborn piglet study of neonatal encephalopathy

There is a need to develop therapies for neonatal encephalopathy (NE) in low‐ and middle‐income countries (LMICs) where the burden of disease is greatest and therapeutic hypothermia (HT) is not effective. We aimed to assess the efficacy of melatonin following inflammation‐amplified hypoxia–ischaemia (IA‐HI) in the newborn piglet. The IA‐HI model accounts for the contribution of infection/inflammation in this setting and HT is not cytoprotective. We hypothesised that intravenous melatonin (5% ethanol, at 20 mg/kg over 2 h at 1 h after HI + 10 mg/kg/12 h between 24 and 60 h) is safe and associated with: (i) reduction in magnetic resonance spectroscopy lactate/N‐acetylaspartate (MRS Lac/sNAA); (ii) preservation of phosphorus MRS phosphocreatine/phosphate exchange pool (PCr/Epp); (iii) improved aEEG/EEG recovery and (iv) cytoprotection on immunohistochemistry. Male and female piglets underwent IA‐HI by carotid artery occlusion and reduction in FiO2 to 6% at 4 h into Escherichia coli lipopolysaccharide sensitisation (2 μg/kg bolus + 1 μg/kg/h over 12 h). At 1 h after IA‐HI, piglets were randomised to HI‐saline (n = 12) or melatonin (n = 11). There were no differences in insult severity between groups. Target melatonin levels (15–30 mg/L) were achieved within 3 h and blood ethanol levels were <0.25 g/L. At 60 h, compared to HI‐saline, melatonin was associated with a reduction of 0.197 log10 units (95% CrI [−0.366, −0.028], Pr(sup) 98.8%) in basal‐ganglia and thalamic Lac/NAA, and 0.257 (95% CrI [−0.676, 0.164], Pr(sup) 89.3%) in white matter Lac/NAA. PCr/Epp was higher in melatonin versus HI‐saline (Pr(sup) 97.6%). Melatonin was associated with earlier aEEG/EEG recovery from 19 to 24 h (Pr(sup) 95.4%). Compared to HI‐saline, melatonin was associated with increased NeuN+ cell density (Pr(sup) 99.3%) across five of eight regions and reduction in TUNEL‐positive cell death (Pr(sup) 89.7%). This study supports the translation of melatonin to early‐phase clinical trials. Melatonin is protective following IA‐HI where HT is not effective. These data guide the design of future dose‐escalation studies in the next phase of the translational pipeline.

density (Pr (sup) 99.3%) across five of eight regions and reduction in TUNELpositive cell death (Pr (sup) 89.7%).This study supports the translation of melatonin to early-phase clinical trials.Melatonin is protective following IA-HI where HT is not effective.These data guide the design of future doseescalation studies in the next phase of the translational pipeline.

| INTRODUCTION
Neonatal encephalopathy (NE) is a leading cause of mortality and morbidity across the world, affecting 1.4 million babies every year. 1 The burden of disease is greatest in low-resource settings, with 85% of cases in sub-Saharan Africa and Southeast Asia. 1 Therapeutic hypothermia (HT) is the only effective cerebroprotective therapy currently available, with an associated number needed to treat is 7-8 2 in the high-resource setting.Recently, in a large multicentre randomised control trial (RCT) of babies across India, Bangladesh and Sri Lanka, HT was associated with an increase in mortality and no reduction in disability. 3These findings realign the focus and urgent need to identify alternative therapies effective for babies with NE in low-and middle-income countries (LMICs).
Several factors may contribute to the lack of benefit of HT in LMICs, including perinatal infection and inflammation.Tann et al. 4 reported that neonatal bacteraemia and histological funisitis independently increase the risk of NE in Uganda.In our established piglet model of NE, 5 prior inflammation sensitisation with Escherichia coli lipopolysaccharide (LPS) followed by hypoxia-ischaemia (HI) was associated with increased mortality and TUNELpositive cell death, particularly in the white matter and cortical regions. 5Importantly, HT for 24 h did not improve neurological outcomes based on magnetic resonance spectroscopy (MRS) of lactate to N-acetyl aspartate peak area ratio (Lac/NAA), background amplitude-integrated encephalography (aEEG/EEG) activity and immunohistochemistry. 6 Similar findings have been reported by other groups [7][8][9] with evidence of a pathogen-specific response. 10elatonin (N-acetyl-5-methoxytryptamine) for NE has shown great promise in preclinical studies.Cytoprotective actions include free radical scavenging activity, antiapoptotic and anti-inflammatory properties.In a recent meta-analysis of preclinical studies of NE, melatonin was associated with significant reduction in brain infarct size, improved neurobehavioural outcomes and cell death reduction. 11The improvement in neurological outcomes occurred both with melatonin monotherapy and as an adjunct with HT.Preclinical evidence thus demonstrates robust protection following HI; however, the benefit with IA-HI is unknown.
The lipophilic nature of melatonin allows easy penetration across the blood-brain barrier, but limits its solubility in water.To date, no intravenous formulation of melatonin is available for clinical use.Excipients used in preclinical studies include ethanol, [12][13][14][15][16][17] cyclodextrin-derivates, 18,19 dimethylsulphoxide (DMSO) [20][21][22][23] and tween 24 to enhance solubility; however, their individual safety profiles need careful consideration.Improved outcomes in ethanolcontaining melatonin formulations are reported 11 ; partial cytoprotection was seen in piglets 12 and preterm lambs 25 receiving low-dose ethanol.The United Kingdom Medicines Health Regulatory Authority do not stipulate absolute blood alcohol concentration (BAC) limits in medicinal products; however, the American Academy of Paediatrics Committee on Drugs recommend levels should not exceed 0.25 g/L after a single dose. 26In this study, we have optimised melatonin administration to limit ethanol exposure while achieving target melatonin levels.
This translational study aimed to assess safety and efficacy of an optimised intravenous melatonin regimen as a monotherapy under normothermia in a piglet model of NE relevant to the low-resource setting.We hypothesised that intravenous melatonin (5% ethanol, at 20 mg/kg over 2 h at 1 h after HI + 10 mg/kg/12 h between 24 and 60 h) is cytoprotective and associated with: (i) reduction in MRS Lac/NAA; (ii) preservation of phosphorus ( 31 P) MRS phosphocreatine/exchangeable phosphate pool (PCr/Epp); (iii) improved aEEG/EEG recovery and (iv) cytoprotection on immunohistochemistry.The immunohistochemistry markers assessed included terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL), neuronal nuclear protein (NeuN), ionised calcium-binding adapter molecule-1 (Iba-1), oligodendrocyte transcription factor-2 (Olig2), and glial fibrillary acidic protein (GFAP).
The study was approved by the University College London ethics committee, conducted in accordance with the UK Home Office Regulations [Animals (Scientific Procedures) Act 1986] and reported under the ARRIVE 2.0 guidance.

| Sample size
In our previous neuroprotection piglet studies, we observed treatment differences between 0.5 and 0.77 19,27,28 log 10 units in the Lac/NAA peak ratio with a standard deviation of 0.3-0.4.A similar variance was observed in our IA-HI model (SD 0.34-0.42). 5,6Based on a pessimistic prediction of 0.5 Log 10 unit reduction in basal ganglia and thalamic (BGT) Lac/NAA and a standard deviation of 0.4, 12 piglets per group were required to achieve 80% power at a 5% type-I error rate.

| Study design, eligibility criteria and randomisation
The experimental protocol is shown in Figure 1.Newborn male and female Large White piglets were assessed using the following eligibility criteria: (i) good baseline health, (ii) normal baseline aEEG/EEG following surgery, and (iii) isoelectric aEEG/EEG within the first hour of IA-HI.Piglets were randomised using a blinded computer-generated random allocation process at 1 h to receive intravenous (i) 0.9% sodium chloride (HI-saline) or (ii) melatonin.

| Animal care, surgical preparation and neurocritical care management
Piglets underwent surgical preparation as previously described and transferred to a purpose-built MRcompatible incubator for neonatal neurocritical care. 19ormothermia was maintained at a rectal temperature of F I G U R E 1 Experimental Protocol.In brief, newborn (male and female) large white piglets were examined for good health and surgically prepared as previously described. 6,19Inflammation-amplified hypoxia-ischaemia (IA-HI) was achieved by Escherichia coli liposaccharide (LPS) sensitisation (bolus of 2μg/kg bolus + 1μg/kg/h for 12 h) and hypoxia-ischaemia 4 h into LPS infusion (reduction in FiO2% to 6% and inflation of carotid artery occluders).At 1 h following IA-HI, piglets meeting the eligibility to enter the study were randomised to receive (i) 0.9% sodium chloride (HI-saline) (n = 12) or (ii) melatonin (loading dose 20 mg/kg, maintenance dose 10 mg/kg every 12 h after 24 h) (n = 11).All piglets received full neonatal neurocritical care management for the duration of the experiments, including continuous aEEG/EEG monitoring and 3 T MRI/MRS at 62-65 h.After 65 h, piglets were euthanised and brain dissected for immunohistochemistry.
PANG ET AL.
38°C using a servo-controlled system (Criticool, Belmont Medica).Complications of neonatal HI were treated following neonatal intensive care guidelines as previously described. 19

| Inflammation-amplified hypoxia-ischaemia
Piglets underwent inflammation sensitisation with an intravenous bolus of 2μg/kg Escherichia coli LPS O55:B5 (Sigma-Aldrich) followed by a continuous infusion of 1μg/kg/h for 12 h.At 4 h into the LPS infusion, piglets underwent an IA-HI insult. 19The carotid artery occluders were inflated and FiO 2 was titrated to 6% in a stepwise manner over 3 min.The duration of the HI insult was determined in real time by the blood lactate (target 10-12 mmol/L), duration of isoelectric aEEG and duration of hypotension (mean arterial blood pressure [MABP] 27-30 mmHg).In severe hypotension (MABP < 27 mmHg), the FiO2 was titrated up by 1% every minute until MABP > 27 mmHg was achieved.Following HI, piglets were resuscitated with ventilation breaths in air.

| Cytoprotective interventions
Melatonin (100 mg) (Sigma-Aldrich) was dissolved in 100% ethanol (2 mL) in the dark and added to 5% glucose to a final concentration of 2.5 mg/mL (5% v/v ethanol) for infusion.A melatonin loading dose of 20 mg/kg (ethanol 320 mg/kg) was administered intravenously over 2 h at 1 h after IA-HI via a central venous line sited in the antecubital fossa.Maintenance doses of 10 mg/kg over 2 h were administered every 12 h from 24 to 60 h.The comparison (HI-saline) group received the equivalent intravenous 0.9% sodium chloride infusion at the same volume and rate.

| Amplitude-integrated electroencephalography (aEEG/EEG)
Continuous multichannel 6-lead aEEG/EEG (Nicolet, Natus) was acquired over the full duration.At the end of the study, the hourly background aEEG/EEG activity was classified by two lab members blinded to the study group using the voltage criteria described by Hellstrom-Westas. 29An isoelectric trace was scored 0 and a normal voltage trace scored 4. Seizures were detected by a rise in baseline on aEEG followed by a review of the raw EEG trace demonstrating evolving, repetitive spike and/or wave pattern lasting at least 10 s.Treatment of seizures followed standard neonatal seizure guidelines. 307 | Magnetic resonance spectroscopy 1 H and 31 P MRS were acquired by medical physicists blinded to the treatment group at 60 h using a clinical 3-Tesla MRI scanner (Philips Achieva) as previously described.19 1 H MRS was acquired using chemical shift imaging with an 8 × 8 matrix in two 8 × 8 × 10 mm 3 voxels: the BGT voxel over the left thalamic region and the white matter (WM) voxel at the level of the centrum semiovale. spctral acquisition was acquired with a relaxation time of 2000 ms and a long echo time (TE) of 288 ms as standard.Final post-imaging processing was carried out at the end of the study using TARQUIN and the Lac/NAA peak area ratio was calculated (lactate+threonine/N-acetylaspartate+N-acetylaspartylglutamate).The 31 P MRS metabolites were acquired using a 6-cm diameter circular transmit-receive coil (PulseTeq) placed above the piglet's head with a single-pulse acquisition at an average repetition time of 10 and 32 s.

| Immunohistochemistry
Piglets were euthanised at 65 h with intravenous pentobarbital followed immediately by intracardiac cold phosphate-buffered saline perfusion and 5% paraformaldehyde (PFA).The brain was subsequently dissected, immersed in 4% PFA for 1 month and then embedded in paraffin.For immunohistochemistry, 8 μm sections were cut from two coronal slices at the level of the optic chiasm and hippocampus from the right hemisphere.Slides were stained for TUNEL+ cell density for cell death, NeuN+ cell density for neuronal survival, Iba-1 cell count for neuro-inflammation, OLIG2+ cell density for oligodendrocyte survival and GFAP for astrocytosis (see Supporting Information for detailed methodology).

| Statistical analysis
Statistical analysis was performed using JMPv14 (SAS), R-software (v4.0.5) and GraphPad Prism (v9).Physiological data, biochemical and haematological results were assessed using an ANOVA model with fixed factor effects of treatment, time and treatment × time interaction plus a random effect subject to account for repeated measures where necessary as previously described. 19Treatment groups were compared using 95% confidence interval (CI) for the difference in the least square (LS) means and p values deduced.A Bayesian statistical approach was used to assess neurological outcomes (MRS, aEEG and histology) to determine the probability of treatment benefit using non-informative priors.Unlike frequentist statistical methods, [31][32][33] the Bayesian approach avoids the dichotomisation of results, enables direct hypothesis testing (such as probability of treatment superiority) and therefore produces inferences that are more intuitive and meaningful for decision-making purposes.Detailed methodology is included in the Supporting Information.At the final analysis, a Pr (sup) of 94.8% was considered the threshold for treatment superiority.
We used the 'nlmxr2' package in R (v4.3.1) to fit a population-based nonlinear mixed-effects model to the melatonin pharmacokinetic data.Age (hours from the insult) and weight were co-variants tested and included in the final model.Graphical inspection of goodness of fit was evaluated and shown in the Supporting Information.

| Baseline characteristics
The baseline characteristics and insult parameters are shown in Table 1.Twenty-four animals were recruited to enter the study; however, one animal in the melatonin arm was excluded a priori due to severe hyperglycaemia before the IA-HI insult, contributing to brain injury.
Twenty-three animals were studied and randomised to either (i) HI-saline (n = 12) or (ii) melatonin (n = 11).Physiological and biochemical parameters for the IA-HI insult were similar across the groups (p > .05)(Table 1).

| Pharmacokinetics
Serum melatonin, ethanol and acetaldehyde levels are shown in Figure 3.Following an intravenous loading dose of 20 mg/kg melatonin over 2 h, the peak serum melatonin (mean ± SD) level was 33.5 ± 7.6 mg/L at 3 h after IA-HI with a trough of 14.0 ± 8.1 mg/L at 24 h.Peak serum levels were ~10 4 -fold higher compared to HIsaline animals (~0.4 μg/L).Subsequent maintenance doses of 10 mg/kg every 12 h achieved peak serum levels of 26.4 ± 12.6 to 27.7 ± 12.0 mg/L.The median melatonin level in the brain tissue (n = 6) was ~10 3 -fold higher compared to HI-saline.
A population-based pharmacokinetic model was derived using a one-compartment model and the PK parameter estimates are shown in Table 2.The inclusion of age and weight covariates with the combined additive and proportional error provided the best fit as demonstrated by a reduction in the objective function value (OFV).This was used as the final model with the goodness of fit plots shown in the Supporting Information.As shown in Table 2, the parameter estimates for clearance (CL), volume distribution (Vd) and half-life were 0.086 L/h, 1.23 L and 10 h, respectively.

| aEEG/EEG
aEEG/EEG data were available for all 23 animals and shown in Figure 5. Melatonin was associated with a probability of 86.1% to improve background aEEG/ EEG activity.Melatonin was associated with earlier aEEG/EEG recovery from 19 to 24 h (score difference   5A).Five animals in HI-saline (41.7%) versus one animal in the melatonin group (9%) did not recover to at least an aEEG/EEG score of 3. Seizures were detected in five animals (21.7%): four in the HI-saline arm (from 11 to 25 h after IA-HI) and one in the melatonin arm (from 13 h after insult).Four animals required treatment with phenobarbitone (HI-saline  B, E, H).The posterior distribution is shown in panels (C, F, I) from Bayesian analysis using a non-informative prior.The probability of treatment superiority was 89.8, 98.8% and 97.6% for WM Lac/NAA, BGT/NAA and PCr/Epp, respectively.Probability of treatment superiority was defined by posterior probability of Lac/NAA peak area ratio <0 and PCr/Epp ratio >0 in melatonin versus HI-saline.n = 3, melatonin n = 1) and one terminated spontaneously without treatment (HI-saline).
We observed no significant difference in the overall or regional density of % GFAP positivity or OLIG2+ve cell counts (see Figure 6G-I).

| Peripheral markers of inflammation
We observed no significant difference in the total white cell, platelet, neutrophil and lymphocyte counts between melatonin and HI-saline-treated animals (see Figure 7A-D).SIRI and SII (see Figure 7E,F) reduced over time (p < .05),however, did not differ significantly between the groups at all time points.Plasma pro-(IL-6) and anti-inflammatory cytokines (IL-1ra, IL-4, IL-10, IL-12) reduced over the time from HI (p < .001)(see Figure 7G-L).We observed no significant differences in cytokines between the groups except lower IL-10 levels in melatonin-treated animals at 48 h (LS mean difference 0.87, 95% CI 0.14-1.59,p = .019)(see Figure 6J).

| DISCUSSION
In this newborn piglet study relevant to the low-resource setting, intravenous melatonin monotherapy, commenced at 1 h after IA-HI, demonstrated treatment efficacy across neurometabolic (MRS), neurophysiological (aEEG/EEG) and neuropathological (immunohistochemistry) outcome measures.Findings from this study add to the body of evidence 11 F I G U R E 5 Amplitude-integrated encephalography (EEG/aEEG).aEEG scores were classified in accordance with the voltage criteria 29 (see panel), averaged over 6 h intervals and compared (mean ± SEM).Melatonin-treated animals showed early EEG/aEEG recovery at 19-24 h (*probability of superiority of 95.4%).Probability of treatment superiority was defined by posterior probability of aEEG slope >0 in melatonin versus HI-saline.Box showing example micrographs.

F I G U R E 6 (See caption on next page).
supporting translation of melatonin to early-phase clinical trials in babies with NE.
This study provided further evidence of the strong safety profile of high-dose, intravenous melatonin administration.We observed no significant differences in the physiological and biochemical parameters between melatonin and HI-saline animals.The inotropic support requirements were marginally higher in melatonintreated animals in the first 3-12 h following HI (increase in dopamine of ∼3 µg/kg/min).The implications of this marginal increase in inotropic requirements is unclear; however, it highlights the need to study melatonin in a phase I dose escalation study in high-income settings where babies have continuous invasive blood pressure measurement, starting at a quarter (5 mg/kg) dose, increasing incrementally while carefully assessing the effect on blood pressure and inotrope requirement.A phase I safety study will need to be repeated in LMICs.In this optimised melatonin administration regimen, the loading dose of 20 mg/kg achieved therapeutic target levels rapidly within 3 h of IA-HI.Subsequent 12-hourly maintenance doses of 10 mg/kg ensured levels of 15-30 mg/L were maintained.This therapeutic target was established in previous studies in piglets and is summarised in Figure 8. 34 In this current study of IA-HI, where cooling is not protective, 6,9,35 melatonin monotherapy targeting this therapeutic range is cytoprotective, highlighting its diverse benefits.Aridas and colleagues 13 observed improvement in Lac/NAA, histology and neurobehavioural outcomes with melatonin doses of ∼15 mg/kg/day.While serum melatonin levels were 10fold lower than our target range, melatonin administration was earlier (at 30 min) and given as multiple 2hourly 5 mg boluses.Lower melatonin levels may suffice if melatonin is administered earlier following HI; however, our 2-h infusion every 12-24 h is more feasible in the clinical setting.The anticipated delay between birth and recruitment to clinical trials necessitates rapid attainment of therapeutic levels with a loading dose.
The population-PK parameters following intravenous melatonin administration are not well characterised in the target population of term infants with NE.The similarities between pigs and humans in anatomy and physiology, including drug metabolising enzymes, suggest pigs are well suited to predicting the PK profile in humans. 36Melatonin is metabolised by the hepatic CYP450 system 37 and renally cleared, 38 processes which may be affected by the systemic HI injury.The volume of drug distribution in this term-equivalent piglet study was similar to that in a study of intravenous melatonin administration to preterm infants by Merchant et al., 39 although the half-life in piglets (10.2 h) was shorter than preterm babies (16.9 h).In term infants, PK data are limited to a study by Balduini et al., 40 who reported a prolonged half-life of 26 h in cooled infants with HIE following oral melatonin. 39The melatonin half-life in babies and our piglets was significantly prolonged compared to the adult half-life of 30-60 min. 41n important consideration for clinical translation is limiting the risk associated with excipient use.Our preclinical meta-analysis suggested ethanol-containing formulations may provide improved protection. 11This is supported by the partial benefit of low-dose ethanol observed in newborn piglets 12 and foetal sheep 25 following HI.The concerns with chronic alcohol exposure during foetal development and developmental impairment likely relate to the chronicity of exposure and BAC, with a U-shaped dose-response relationship suggested.In preclinical adult stroke models, ethanol doses of 1.5 g/kg resulting in BAC of 0.89 g/L [42][43][44] were protective, but a higher dose of 3 g/kg (BAC 2 g/L) was toxic. 45In preclinical studies of the immature brain, toxicity was observed at a lower BAC threshold of 1 g/L, [46][47][48] highlighting increased vulnerability of the developing brain.Ethanol-containing medicines are already commonplace in drugs prescribed in neonatal units including furosemide, iron and phenobarbital. 49In this study, the ethanol dose (320 mg/kg) was comparable to the partially protective dose observed in our previous piglet neuroprotection study (285 mg/kg).7][48] We limited the accumulation of ethanol by reducing the infusion rate of the maintenance doses to below the rate of ethanol metabolism (0.1 g/kg/h 50 ).The BAC levels were low and similar to the exposure associated with furosemide administration in preterm infants (0-73 mg/L), 51 and F I G U R E 6 Immunohistochemistry. Eight regions were assessed for TUNEL-positive cell density (A, B), NeuN+ve cell density (C, D), Iba-1 cell density (7 regions) (E, F), GFAP % positive area (G, H) and OLIG2+ve cell density (I, J).Data shown as overall and regional means.The probability of treatment superiority is shown where PCr > 80% with Pr > 94.8% (superiority threshold) in green.cCTX, cingulate gyrus; CAUD, caudate nucleus; HIP, hippocampus; IC, internal capsule; PvWM, periventricular white matter; PTMN, putamen; sCTX, sensorimotor cortex; THAL, thalamus.Probability of treatment superiority was defined by posterior probability of count <0 in TUNEL, Iba-1 and GFAP and count >0 in NeuN and OLIG2 with melatonin versus HI-saline.Example micrograph shown from putamen (see box).
other medicines in the NICU. 52While the by-product of ethanol metabolism, acetaldehyde, may contribute to ethanol-induced toxicity, 53 levels were not significantly higher than those in preterms prescribed iron and furosemide (range 0-9 mg/L). 51Acetaldehyde did not accumulate following the maintenance doses and was comparable to levels in preterm infants without ethanol exposure (0-0.14 mg/L). 51elatonin was associated with improvement in clinically important surrogate biomarkers of neonatal Peripheral Blood Markers of Inflammation.Plasma samples were measured for blood counts including total white cells (A), platelets (B), neutrophils (C) and lymphocytes (D).The systemic inflammation response index (SIRI) and systemic immune-inflammation index (SII) were deduced.Plasma cytokines including IL-1ra (G), IL-4 (H), IL-6 (I), IL-10 (J), IL-12 (K) and GFAP (L) were measured at 0 (end of HI), 24, 48 and 60 h.Data presented as LS means ± SD where *p < .05 in melatonin versus HIsaline (ANOVA analysis).
outcomes.Lac/NAA peak area ratio predicts 2-year neurodevelopmental outcomes in babies with NE with high accuracy 54 and has been validated as an early quantitative read-out for use in neonatal neuroprotection trials. 55In piglets, Lac/NAA correlates closely with immunohistochemistry markers of cell death and neuro-inflammation. 56Extrapolating from the logistic regression model comparing Lac/NAA to outcomes in babies with moderate to severe NE published in Mitra et al., 54 the observed BGT Lac/ NAA reduction of 0.23 log 10 units in melatonin versus HI-saline in this study translates to a 87%, 50% and 34% improvement in motor, cognitive and language outcomes, respectively.These are important outcomes as ~40% of infants develop adverse cognitive outcomes. 57We observed a reduction in WM Lac/NAA despite the reduced statistical power secondary to the large variance from CSF partial volume effect.Improved neurophysiological recovery on aEEG/ EEG at 18-24 h was consistent with previous piglet studies of melatonin with cooling. 12,199][60] Over half of the HI-saline animals did not recover to at least a discontinuous normal voltage (normal trace) background pattern compared to only one piglet in the melatonin group.In a clinical study by Thoresen, infants who never regained a normal background trace within 72 h of birth had poor outcomes compared all infants who recovered by 24 h having good outcomes. 58The potential for melatonin to improve EEG activity within this critical time interval provides promise that this may translate to improvement in clinically important outcomes.The strong benefit of melatonin on Lac/NAA and aEEG/EEG is complemented by the improvement in NeuN+ cell density suggesting improved neuronal survival.Interestingly, we observed a more modest reduction in TUNEL-positive cell death.TUNEL is a nonspecific marker of several cell death pathways, including apoptosis, necrosis and autolysis. 61TUNEL-negative cell death pathways have been reported 62 and cells F I G U R E 8 Summary of melatonin PK profile in newborn piglet studies. 12,18,19,28Attainment of serum melatonin levels of 15-30 mg/L within the latent phase of injury (<6 h after HI) is necessary for reduction in Lac/NAA.
can escape TUNEL staining, underestimating the population of cell death, particularly where several pathways co-exist. 63urprisingly, we observed a limited biological effect of melatonin on neuro-inflammation in this study. 16,64,65elatonin was associated with a reduction in Iba-1 positive cells in the deep grey matter but we observed no difference in circulating plasma cytokine levels or differential full blood count.As Iba-1 is a pan-microglia marker, characterising the balance between the M1 pro-inflammatory versus M2 antiinflammatory phenotypes may define the immunomodulatory properties of melatonin more clearly, as in a recent preclinical adult stroke study. 66A cytokine storm may be triggered in this IA-HI model as plasma IL-6 and IL-10 levels were 100-fold higher compared to a previous HI piglet study. 67These excessively high cytokine levels, leading to overwhelming immune hyperactivation, may impair the ability of melatonin to modulate the immune response.Indeed, O'Dea et al. reported relative immune hyporesponsiveness to endotoxin challenge among infants with NE with elevated cytokine levels. 68Additional therapies targeting immunomodulation such as azithromycin 69,70 may complement the neuroprotective effect of melatonin.
There are limitations to our study.First, while both sexes were included, the study was not powered to assess the sexual dimorphic response.The study was designed to assess the effect of melatonin compared to HI alone and a control group of non-injured animals was not included.We are therefore unable to determine whether the observed protection is partial compared the non-injured animals.This was a pragmatic decision based on the need to limit animal use in preclinical research.Previous studies in foetal sheep demonstrated complete protection following HI on MRS and immunohistochemistry with melatonin administration to levels similar to non-injured animals. 13,16,17Interestingly, Malhotra et al. 71 observed an over-abundance of myelin basic protein and 2′,3′-cyclicnucleotide 3′-phosphodiesterase (CNPase) in growthrestricted foetal sheep following maternal melatonin administration compared to non-injured animals.This provided insight into the potential promotion of myelination and oligodendrocyte maturation associated with melatonin, also reported by other groups. 72Neurobehavioural outcomes were not assessed in this study; however, Lac/NAA and aEEG/EEG are clinical surrogate biomarkers that accurately predict 2-year outcomes in infants with NE.Neurobehavioural assessment of piglets may not translate to neurodevelopmental outcomes in humans and not without risks, including intractable seizures and aspiration pneumonia. 73NE is a heterogeneous disease and inflammation may not be the only contributing factor in LMICs.A partial prolonged HI model may better mimic the situation in low-resource settings, given the observed high frequency of seizures in the HELIX trial 3 ; however, such a model may necessitate a large sample size. 74he decision to translate therapies to early-phase trials requires confirmation of safety and benefit in multiple preclinical models.Indeed, over the last decade, melatonin has demonstrated protection on multiple outcome measures in large and small animal models of HI 11 .In this newborn piglet of IA-HI, under circumstances where HT is not protective, intravenous melatonin administration (achieving serum melatonin of 15-30 mg/L within 3 h) was associated with robust cytoprotection across neurometabolic, neurophysiological and histological outcomes.No significant physiological safety concerns were observed with this optimised melatonin (with ethanol excipient) and BAC was below safety limits.Overall, the preclinical evidence for melatonin as a monotherapy in babies with NE in LMICs is robust, supporting clinical translation.
1 H and31 P MRS results were available for 22 animals and are shown in Figure4.One animal in the HIsaline group was excluded due to poor spectral fit.

F I G U R E 3
Pharmacokinetics.Target serum melatonin levels (mean ± SD) of 15-30 mg/L were achieved within 3 h of HI and maintained throughout the study and were 10 4 higher than HI-saline (A).Brain tissue melatonin levels were 10 3 -fold higher in melatonin-treated animals versus saline (B).Serum ethanol levels were below 0.25 g/L through the experiments (C) and with the exception of the loading dose, peak acetaldehyde levels were similar compared to the HI-saline group (D).**p < .01.T A B L E 2 Population-based pharmacokinetic parameter estimates for melatonin.