Hypoglycaemia and neonatal brain injury

The level of blood glucose (BG) concentration that leads to cerebral injury in newborns and adverse neurodevelopmental outcome is unknown: low BG is observed during postnatal adaptation of healthy term infants without apparent adverse consequence, and the capacity to mobilise and use alternative cerebral fuels when BG is low varies between patient groups.1 Because of this uncertainty an ‘operational threshold’ approach to the management of neonates with hypoglycaemia has been proposed, which defines ‘the concentration of plasma or whole BG at which clinicians should intervene based on evidence currently available in the literature.’ In this model infants at risk of neurological sequelae from hypoglycaemia are identified (see algorithm), and interventions to raise the BG are recommended at specified thresholds, with the caveat that acute neurological dysfunction in association with low BG at any level should prompt urgent investigation and treatment.2

A review of the literature supports the previous consensus that BG levels below 1.0 mmol/l that are persistent beyond 1–2 h (or are recurrent) and are associated with acute neurological dysfunction present the greatest risk for cerebral injury, and that brief episodes of hypoglycaemia in the absence of acute neurological dysfunction or an associated disorder are less likely to lead to cerebral injury and poor outcome.3

The spectrum of cerebral injury associated with hypoglycaemia is wide and includes: white matter injury including parenchymal haemorrhage and ischaemic stroke, cortical neuronal injury, and sometimes signal change in the basal ganglia (mainly the globus pallidus) and thalami4–15 (figure 1). Vulnerability of the white matter and cortex of the posterior parietal and occipital lobes has been well reported in human imaging studies,6 ,13 ,16 but the site of injury is more widespread in pathological and experimental studies of neonatal hypoglycaemia.4 ,5 ,17 In the largest series …


BACKGROUND
The level of blood glucose (BG) concentration that leads to cerebral injury in newborns and adverse neurodevelopmental outcome is unknown: low BG is observed during postnatal adaptation of healthy term infants without apparent adverse consequence, and the capacity to mobilise and use alternative cerebral fuels when BG is low varies between patient groups. 1 Because of this uncertainty an 'operational threshold' approach to the management of neonates with hypoglycaemia has been proposed, which defines 'the concentration of plasma or whole BG at which clinicians should intervene based on evidence currently available in the literature.' In this model infants at risk of neurological sequelae from hypoglycaemia are identified (see algorithm), and interventions to raise the BG are recommended at specified thresholds, with the caveat that acute neurological dysfunction in association with low BG at any level should prompt urgent investigation and treatment. 2 A review of the literature supports the previous consensus that BG levels below 1.0 mmol/l that are persistent beyond 1-2 h (or are recurrent) and are associated with acute neurological dysfunction present the greatest risk for cerebral injury, and that brief episodes of hypoglycaemia in the absence of acute neurological dysfunction or an associated disorder are less likely to lead to cerebral injury and poor outcome. 3 The spectrum of cerebral injury associated with hypoglycaemia is wide and includes: white matter injury including parenchymal haemorrhage and ischaemic stroke, cortical neuronal injury, and sometimes signal change in the basal ganglia (mainly the globus pallidus) and thalami 4-15 (figure 1). Vulnerability of the white matter and cortex of the posterior parietal and occipital lobes has been well reported in human imaging studies, 6 13 16 but the site of injury is more widespread in pathological and experimental studies of neonatal hypoglycaemia. 4 5 17 In the largest series of infants with isolated neonatal hypoglycaemia and acute neurological dysfunction, there was an association with a predominantly posterior pattern of injury in one third of the cohort, and a more extensive distribution of lesions was common. 18 Sampling and measurement methods are important when assessing neonatal hypoglycaemia. It is desirable to sample whole blood because the glucose concentration of plasma is higher than that of whole blood, and it is essential that accurate methods of measurement are used to make the diagnosis of neonatal hypoglycaemia: measurements from glucose reagent strips are not reliable in this range (for discussion, refer to Beardsall 19 ).
The epidemiology of neonatal hypoglycaemia and associated brain injury is poorly understood due to the controversies surrounding definition, and a paucity of long-term follow-up studies of humans.

CLINICAL PRESENTATION
Signs may be non-specific but include altered consciousness, lethargy, irritability, high-pitched cry and tremor, sweating, poor feeding after initially feeding well, hypotonia, apnoea and hypothermia, and there can be rapid progression to seizures and coma.

INVESTIGATION
It is important to obtain a blood sample during the period of hypoglycaemia and the first available urine to test for the cause of hypoglycaemia. Investigations for severe or persistent hypoglycaemia have been proposed by Hawdon 21 and are listed in the algorithm (figure 2). Complex cases such as hyperinsulinaemic hypoglycaemia should be discussed with specialist teams at an early stage.

Cranial ultrasound
Early ultrasound is helpful for assessing associated structural malformations specifically associated with hypoglycaemia (eg, septo-optic dysplasia spectrum) or other problems that might be associated with poor feeding (eg, other developmental malformations or established injury), and is useful later in the neonatal period for detecting parenchymal or deep grey matter echodensities that could represent evolving haemorrhage or stroke. 22 Detecting the more posterior injury thought typical for hypoglycaemia is not easy and a normally appearing ultrasound scan would not exclude this-imaging white matter and cortex through the posterior fontanelle can be helpful.

Electroencephalography
The amplitude integrated EEG is useful for detection and management of encephalopathy and seizures, and specific EEG measures may be sensitive to changes in BG concentration. 23 A cerebral function monitor should be applied as soon as acute neurological dysfunction is diagnosed. Full array EEG should be considered if any aspect of the neurological examination remains abnormal 5-7 days after the hypoglycaemic event, or if seizures are ongoing. Focal seizures occurring in the occipital lobes may not be captured by single or two channel amplitude integrated EEG.

MRI
This is useful for characterising more precisely the extent and nature of cerebral injury, explaining neurological findings, identifying confounding factors such as injury more typical of acute hypoxia-ischaemia, identifying arterial territory stroke and involvement of the internal capsule, investigating associated structural abnormalities such as septo-optic dysplasia or pituitary abnormality, detecting MR features suggestive of specific metabolic disorders, detection of associated venous thrombosis, predicting outcome, counselling parents and informing medicolegal proceedings. There are widespread areas of cystic change (long T2) within the parietal and posterior temporal and occipital regions with involvement of both optic radiations and primary visual areas, and there is some high SI in the frontal white matter. The basal ganglia look normal but the thalami are slightly small though with normal SI; the SI from the posterior limb of the internal capsule (PLIC) was normal on T1-weighted imaging (not shown). (C-E) (C) Shows extensive white matter injury with loss of grey/white matter differentiation (transverse T2-weighted imaging at mid-ventricular level), and there is marked asymmetry with the left side being more severely affected than the right (arrow). There are some small darker areas of punctate haemorrhage within the white matter (left more than right, short arrow). (D) (T1-weighted transverse image at basal ganglia level) shows extensive abnormal low SI in the white matter in a pattern similar to the abnormal signal seen on the T2-weighted images. There is also some low SI in the thalamus on the left (arrow) and some high SI in the posterior basal ganglia. The PLICs are both myelinated. The diffusion tensor imaging (E) shows asymmetric extensive white matter injury and restricted diffusion within the posterior corpus callosum (long arrow), the occipital lobes (short arrow), the thalami and scattered small regions within the white matter and cortex. Note the thalamic SI is not as marked as in the white matter and is likely a secondary phenomenon. 20 (F) Diffuse white matter injury in a preterm infant with early hypoglycaemia and high glucose requirements (T2-weighted MR images in transverse plane at the upper ventricular level obtained 2 weeks after birth at 32 weeks GA). There are signal changes (long T1 and T2) throughout the white matter; these are most apparent posteriorly.

Figure 2
Algorithm for the management of hypoglycaemia in babies at high risk of brain injury.

Review
However when findings are limited to milder white matter injury there is still a very limited amount of data on school-age outcome. The imaging protocol should include: T1-weighted and T2-weighted axial images, T1-weighted sagittal images, diffusion tensor imaging and an MR sinus venogram; MR proton spectroscopy of white matter and basal ganglia/ thalami should be done if underlying metabolic problems are suspected.
The optimal timing for image acquisition is 5-14 days from the hypoglycaemic insult to allow assessment of subtle white matter injury; apparent diffusion coefficient values can be useful if the scans are done in the first 7 days. MR imaging helps in identifying important differential diagnoses such as hypoxic-ischaemic brain injury and cerebral sinus venous thrombosis. 24 25 TREATMENT The treatment algorithm (figure 2) focuses on the management of hypoglycaemia that is most likely to lead to cerebral injury; the management of mild to moderate hypoglycaemia (BG>1 mmol/l without acute neurological dysfunction) is beyond the scope of the algorithm, and local guidelines should be followed.

OUTCOMES
There is increasing evidence of neurodevelopmental sequelae following neonatal hypoglycaemia, usually not in the form of severe cerebral palsy, but with some milder motor problems, visual, learning and behavioural difficulties, poor head growth and later seizures. 9 16 18 24 26-31 Current data would support a formal neurodevelopmental assessment at 2 years and then longer-term follow-up if there is poor head growth or evidence of difficulty in any developmental domain. There is increasing evidence that outcome is worse for hypoxic-ischaemic encephalopathy (HIE) associated with hypoglycaemia and these children should be followed with extra care especially when head growth is suboptimal. 15 32 KEY PRINCIPLES ▸ Identify high risk groups based on expected inability to mobilise or use alternative cerebral fuels from glycogen and fat in the presence of low BG concentration, and/or increased likelihood of having impaired or immature counter-regulatory mechanisms. ▸ Promote early and frequent feeds. ▸ Monitor BG in high-risk infants using an accurate measurement system. ▸ Ensure energy provision to keep BG≥2.0 mmol/l (3.0 mmol/l in hyperinsulinism) ▸ Prompt treatment if BG<2.0 mmol/l on two consecutive pre-feed measurements or ≤1 mmol/l at anytime, or any infant with low BG and signs of acute neurological dysfunction. 2