Chapter 10: Glucose and perinatal brain injury: Questions and controversies

Sarbattama Sen, Jane E. Harding

Introduction

Neonatal hypoglycemia is the most common biochemical abnormality of the newborn, affecting 5%–15% of all infants and 50% of those with risk factors. Although severe hypoglycemia has been incontrovertibly linked with brain injury and impairment, the glucose thresholds below which injury occurs have not been clearly established, and in fact, may differ based on various host factors. In the absence of clear evidence of a causative link between mild or asymptomatic neonatal hypoglycemia and brain injury, committees have recommended that infants who are at high risk based on maternal or fetal physiology are screened and treated if their blood glucose drops below “operational thresholds,” or concentration of blood glucose at which clinicians should consider intervention. These thresholds have been determined based on consensus opinion, incorporating a combination of norms in healthy infants, neuroglycopenic thresholds, and associations between early glycemia and later development from observational studies.

In this chapter, we will review key metabolic transitions that occur soon after birth that, when dysregulated, can result in neonatal hypoglycemia, summarize mechanisms that underlie the association between neonatal hypoglycemia and later outcomes, and review the animal and human studies that have investigated the link between hypoglycemia and neurodevelopment. We will also summarize current glycemic assessment modalities and prevention and treatment recommendations. Throughout, we will highlight unanswered questions and areas of controversy that urgently require further inquiry.

Perinatal glucose regulation

The primary source of energy for the fetus is glucose, which the fetus receives from its mother down a concentration gradient.1–3 Insulin is secreted at lower glucose concentrations in utero compared to postnatally as its function during this time is primarily related to fetal growth rather than glucose regulation.⁴ Just before delivery, maternal and fetal glucose concentrations increase⁵ until clamping of the umbilical cord, when maternal glucose supply to the infant is interrupted and neonatal glucose concentrations decrease, reaching a nadir between 1 and 3 hours after birth. In response to falling glucose concentrations, the infant’s insulin secretion should decrease while secretion of glucagon and catecholamines increase, stimulating glucose production through gluconeogenesis and glycogenolysis. Typically, these physiologic transitions, which involve “resetting” the threshold for glucose-stimulated insulin secretion by the pancreatic beta cell to the higher, adult glucose concentration range,⁶ occur over the first 48 to 72 hours after birth, resulting in infants achieving normoglycemia by adult definitions by this time point.⁷^,⁸

Established risk factors for neonatal hypoglycemia: Maternal, fetal, and neonatal conditions that predispose infants to a delayed or exaggerated glycemic transition fall into three broad categories: high in utero fetal insulin secretion, decreased insulin sensitivity, and decreased stores of glycogen. Current guidelines recommend screening infants for hypoglycemia based on the following risk factors that together result in approximately one-third of infants warranting screening for hypoglycemia, of whom approximately 50% experience hypoglycemia based on current definitions.⁹ The risk factors that are identified include:

i. Prematurity: 10.1% of infants in the United States are born preterm (<37 weeks gestation).¹⁰ Glycogen and adipose tissue accumulation accelerates in the third trimester, stimulated by relative maternal insulin resistance and hyperinsulinemia. Thus, in general, preterm infants have less glycogen and adipose tissue stores at birth than those born at term. In addition, the concentrations of the lynchpin enzyme for glycogenolysis, glucose-6-phosphatase, are lower in preterm infants than in term infants, rendering preterm infants less able to utilize available glycogen stores.¹¹ Lastly, preterm infants produce lower concentrations of ketones than term infants in the setting of hypoglycemia.¹²^,¹³
ii. Growth restriction: 8% of infants are considered growth restricted.¹⁴ Growth-restricted fetuses have increased production of catecholamines resulting in suppressed insulin secretion in utero, and postnatal compensations that result in inappropriately high insulin secretion.¹⁵ In addition, growth-restricted fetuses have reduced glycogen stores related to limited nutrient supply across the placenta, and limited capacity for oxidation of free fatty acids.¹⁶ Growth-restricted infants often have relative “head sparing,” so that brain size is disproportionately large for their birthweight, further compounding the mismatch between glucose demand and supply.
iii. Infants born to women with diabetes: Diabetes affects approximately 10% of pregnancies in the United States and disproportionately impacts pregnant women with overweight or obesity.¹⁷ Rates of diabetes in pregnancy have increased more than 50% in the past 10 years.¹⁷ Throughout pregnancy, women experience a gradual decline in insulin sensitivity, influenced by factors including obesity, placental hormones, structural changes and dysfunction of the placenta, and changes in cytokine levels. This decreased insulin sensitivity is accompanied by an increase in insulin production by the pancreatic beta cells in pregnancies with typical glucose tolerance.¹⁸ However, in pregnancies affected by diabetes, there is inadequate insulin production to accommodate the decreased insulin sensitivity, resulting in maternal and thus fetal hyperglycemia and leading to increased fetal insulin secretion.^19–21 Persistence of this increased insulin secretion after birth can lead to neonatal hypoglycemia, particularly in infants born to women with diabetes that is poorly controlled.
iv. Large-for-gestational-age infants: Infants who are large-for-gestational age but not born to women with diabetes are currently screened due to the theoretical concern that the large fetal size is a result of excess insulin production. In these infants, maternal hyperglycemia may not reach the threshold of clinical diabetes diagnosis, but the fetal exposure to excessive glucose supply and hence fetal insulin secretion predisposes them to impaired metabolic adaptation. It is reported that 16%–39% of large-for-gestational-age infants who were not born to women with diabetes developed hypoglycemia,⁹ depending on the study population.²²
v. Maternal β blocker exposure: Hypertensive disorders complicate 5%–10% of pregnancies and β blockers are commonly used in their treatment.²³ β Blockers cross the placenta and lead to fetal sympathetic blockade and increased insulin production. The risk of neonatal hypoglycemia in β blocker–exposed neonates was 4.3% versus 1.2% in unexposed infants.²⁴
vi. Family history of hypoglycemic disorder, inborn error of metabolism, or genetic syndrome linked to hypoglycemia (Table 10.1): Infants who fall into these categories are at risk based on abnormalities of insulin secretion (hyperinsulinism), decreased production of cortisol and/or growth hormone, and inborn errors of metabolism that limit glucose production.

TABLE 10.1

Genetic Disorders That Can Present as Neonatal Hypoglycemia and Examples or Candidate Genetic Defects

Congenital hyperinsulinism	ABCC8, KCNJ11, UCP2 mutation
Endocrinopathies	Panhypopituitarism
	Growth hormone deficiency
	Cortisol deficiency
	Hypothyroidism
Inborn errors of metabolism	Carbohydrate metabolism (e.g., galactosemia, glucose-6-phosphate deficiency)
	Glycogen storage disorders
	Organic acid metabolism (e.g., propionic acidemia, methylmalonic acidemia)
	Pyruvate carboxylase deficiency, Fatty acid oxidation disorders, mitochondrial disorders
Genetic syndrome	Beckwith-Wiedemann, Prader-Willi, Kabuki, Turner

Mechanisms underlying hypoglycemia-induced brain injury

Brain energy metabolism (Fig. 10.1): Glucose is an essential metabolic fuel for the brain. In the newborn, the disproportionately large brain for body size requires approximately 2 to 3 times the rate of glucose consumption relative to body weight compared to an adult²⁵^,²⁶ and utilizes more than 30% of hepatic glucose output.²⁷ Glucose uptake from blood to the brain occurs via facilitated diffusion through energy-independent glucose transporters.^28–31 Twelve glucose transporters have been identified and labeled as GLUT 1 through 12.³² Within the brain, GLUT 1 and 3 are predominant. All brain endothelial GLUT proteins are low during the first week of life and increase in the second and third postnatal weeks.³¹ There is a regional developmental progression in cerebral glucose utilization. Animal and human studies have reported that in early development, the brain stem utilizes the highest proportion of glucose. Over the first year after birth, glucose utilization progressively increases and expands through the sensorimotor cortex, thalamus, parietal, temporal, and occipital cortices and lastly in the frontal cortex.³³ Unsurprisingly but of developmental importance, electroencephalographic studies have reported that the functional development of these regions coincides with increases in glucose utilization.³⁴

Fig. 10.1Schematic of cerebral adaptations to hypoglycemia over time and postulated mechanisms that contribute to hypoglycemia-associated brain injury.

In the brain, glucose is phosphorylated to glucose-6-phosphate by hexokinase. Glucose-6-phosphate can then be aerobically oxidized to generate adenosine triphosphate (ATP) through the citric acid cycle. Glucose-6-phosphate can also be stored as glycogen or shunted toward lipid production or nucleic acid synthesis. Low glucose concentrations are likely to result in inadequate brain energy delivery. Lactate provides a potential alternative fuel in the first 48 hours, as demonstrated by studies in the newborn dog reporting that 95% of the brain energy supply arises from glucose and 4% from lactate under normoglycemic conditions.³⁵ With hypoglycemia, the proportion of energy from lactate increases and there is evidence to suggest that the neonatal brain is better able to utilize lactate than an adult, although the energy produced is significantly lower than with glucose metabolism. Ketones can also provide an alternative to glucose for cerebral oxidative metabolism. However, ketone production remains low even under conditions of hypoglycemia, until 3 to 4 days after birth, likely related to persisting relatively hyperinsulinemic conditions.³⁶ With prolonged exposure to hypoglycemia, amino acid utilization also increases in the brain to preserve energy production.⁴⁷ A recent neonatal study reported that glucose contributed 72% to 84% of available potential ATP. Lactate contributed 25% of potential ATP on the first day and remained the largest potential source of ATP other than glucose throughout the first 5 days. Ketones were most available on days 2 to 3 but still only contributed 7% of potential ATP. Total potential ATP available from these fuels was 17% lower on days 1 to 2 than on days 4 to 5.³⁷

Cerebral adaptations and responses to hypoglycemia: Under conditions of hypoglycemia, brain energy delivery is partially preserved by recruitment of cerebral capillaries, resulting in increased cerebral blood flow. This adaptation was noted below a blood glucose threshold of 30 mg/dL in preterm infants³⁸ and was inversely correlated with cerebral regional oxygenation in term and preterm infants in another recent study.³⁹ Early data regarding the impact of hypoglycemia on cerebral blood flow was described from animal models and supported a compensatory increase in cerebral blood flow with low blood glucose concentrations.⁴⁰ Studies have since investigated the associations between glycemia and cerebral blood flow and oxygen saturations in infants. Recently, Matterberger et al. reported cross-sectional data in a cohort of 75 infants showing that one measured blood glucose concentration at 15 to 20 minutes after birth was negatively correlated with cerebral blood flow and oxygen saturations (r = −0.35 in term infants and −0.69 in preterm infants, both p <0.05).³⁹ Similar findings have been reported in a cohort of 11 preterm infants, pairing continuous glucose monitoring with near-infrared spectroscopy (NIRS).⁴¹ Data regarding the predictive value of high cerebral oxygen saturations has been limited to the population of infants with hypoxic-ischemic encephalopathy, where higher cerebral oxygen saturation is associated with brain injury on MRI in infants with hypoxic-ischemic encephalopathy.^42–44

Other cerebral compensatory mechanisms during hypoglycemia include glycogenolysis to mobilize glucose from available stores and increased lactate utilization for energy production, as previously described. Although these mechanisms together may fully compensate for decreased glucose availability in many infants, the increased cerebral perfusion, and resultant hyperoxia and oxidative stress, particularly if experienced for longer periods, may contribute to brain injury.

With severe, prolonged hypoglycemia in adult animals, a failure of the neuronal energy-dependent Na⁺/K⁺ pump results in accumulation of intracellular Na⁺, extracellular K⁺, and influx of intracellular Ca⁺⁺. Similar to neuronal injury induced by hypoxic conditions, Ca⁺⁺-mediated accumulation of excitatory neurotransmitters and free radicals contributes to the final pathway resulting in cell death.²⁸^,³⁴^,⁴⁵^,⁴⁶ In addition with prolonged exposure to hypoglycemia, ammonia levels increase markedly, likely related to use of amino acids as an energy source. A possible underlying cause of cellular injury in the newborn may, however, be related to the release of excitatory amino acids. In a preterm animal model of hyperinsulinemic hypoglycemia, decreasing blood glucose concentrations were strongly inversely correlated with concentrations of extracellular glutamate.⁴⁷ Similarly, in hypoglycemic newborns, lower blood glucose concentrations were associated with higher concentrations of glutamate and aspartate in cerebrospinal fluid.⁴⁸

In vitro and animal models informing neuronal vulnerability to hypoglycemia: In vitro cell culture studies as well as animal models of pure hypoglycemia have been helpful in documenting the effects of significantly low blood glucose levels on brain pathology. In vitro nuclear magnetic resonance studies on energy metabolism of neurons and astroglia under various pathologic conditions have shown that hypoglycemia per se does not significantly alter the high-energy reserves of either neurons or glia.⁴⁹ Immature astrocytes exposed to a substrate-free medium (absence of glucose and amino acids) are able to survive for almost twice as long as the mature astrocytes, suggesting that immature neurons may have the ability to tolerate hypoglycemia better than adult neurons.⁵⁰ In adolescent primates with insulin-induced hypoglycemia, blood glucose concentrations <20 mg/dL for 2 hours or more led to neuronal necrosis throughout the cerebral cortices, with particular vulnerability in the parieto-occipital region, as well as the hippocampus, caudate, and putamen. Similar findings were present in primates exposed to prolonged (6 hours) hypoglycemia, with neuropathologic alterations occurring primarily in the basal ganglia, cerebral cortex, and hippocampus.⁵¹ The adult rat has been used as a model for defining the neuropathologic consequences of severe hypoglycemia, although differences between the immature and mature brains must be considered in interpreting these studies. Several important features of hypoglycemic brain damage have been described that distinguish it from typical patterns of ischemic injury, including a superficial to deep gradient of neuronal necrosis in the cerebral cortex, caudate putamen involvement near the white matter and near the angle of the lateral ventricle, and dense neuronal necrosis in the hippocampus at the crest of the dentate gyrus (which is always spared in ischemia).^52–54 Interestingly, white matter injury was not particularly addressed in these studies.

Taken together, the available data suggest that the newborn brain is capable of some compensatory mechanisms that help maintain brain fuel delivery under conditions of hypoglycemia. However, it is unclear which infants are or are not able to maintain adequate brain fuel delivery with these compensations, and how these compensations and their metabolic by-products, particularly if prolonged, can impact neural development during a time of extreme susceptibility.

Hypoglycemia and neurologic markers

A series of studies have identified a specific pattern of cerebral abnormality on neuroimaging after hypoglycemia involving the parieto-occipital cortex and underlying white matter. These imaging findings are consistent with the areas responsible for the observed neurodevelopmental associations (see below). Abnormal signal intensity with restricted diffusion is reported on MRI and is visualized better with diffusion-weighted imaging. Spectroscopy does not show significant elevations of lactate. About 10%–15% of these areas of restricted diffusion resolve, but some can result in volume loss of cortex and white matter. With severe hypoglycemia, a more diffuse pattern of cerebral cortical injury can occur.55–60 A recent study highlighted the persistence of cerebral imaging differences in childhood. Nine-year-old children who experienced hypoglycemia as neonates had smaller deep grey matter brain regions and thinner occipital lobe cortices than children who did not experience neonatal hypoglycemia.⁶¹

One early study (n = 5 newborns; 17 infants total) reported that blood glucose <47 mg/dL was associated with a prolongation of latency and abnormal sensory evoked potentials.⁶² However, a more recent study in neonates did not find that moderate hypoglycemia was associated with changes in amplitude-integrated EEG markers.⁶³ Severe and persistent hypoglycemia can also result in electroencephalographic and clinical seizures. A retrospective study of 36 children (age 6 months–15 years) with severe hypoglycemia and seizures in the neonatal period revealed posterior temporo-occipital (n = 23), multifocal or generalized spikes, polyspikes, and spike/wave discharges (n = 10) in the interictal period. Three patients had a normal initial EEG, and eight showed a hypsarrhythmic EEG associated with infantile spasms at seizure onset. Interictal EEG remained abnormal in 32 of 36 patients (88.8%). All of these infants had occipital brain injury on MRI in the neonatal period.⁶⁴

Measurement of blood glucose in neonates

There are three approaches to blood glucose measurement in the neonate: glucose oxidase-laboratory-based measurement, point-of-care whole blood glucose measurement, and enzymatic point-of-care analyzers. The gold standard for blood glucose measurement is the glucose oxidase method, typically performed in the laboratory. Due to red cell glycolysis, delays in processing blood samples and assaying glucose can reduce the glucose concentration by up to 6 mg/dL/h. Given the turn-around-time for these assays and the risk of delayed treatment, recommendations for the screening and treatment of neonatal hypoglycemia have commonly involved the use of point-of-care glucose meters for initial screening,⁶⁵ with a laboratory assay to confirm low point-of-care readings while treating the low glucose to prevent treatment delay.

Point-of-care devices were developed for adult diabetics and FDA guidance states that “95% of all values are within 12 mg/dL at glucose concentrations ≤75 mg/dL.”⁶⁶ The accuracy of different devices used to measure glucose concentrations can vary substantially and overall, these devices tend to be somewhat inaccurate at the low glucose concentrations characteristic of newborns.⁶⁷ In addition, whole blood glucose values are approximately 15% lower than plasma glucose concentrations.

A third approach to blood glucose measurement is enzymatic point-of-care analyzers with specific glucose cartridges. The drawbacks of using these devices are that they are not universally available and they require more blood collected in a specific collection device, although they may be associated with overall cost savings related to decreasing the need for laboratory glucose measurement.⁶⁸

A number of anticipated future advances in biochemical monitoring will likely improve care of the neonate at risk of neonatal hypoglycemia. These include the use of continuous interstitial glucose monitors, which have been shown to be safe and feasible in neonates⁶⁹^,⁷⁰ and provide information on direction and rate of change in glucose concentrations as well as duration of exposure to hypoglycemia. Despite their potential, these monitors are currently not approved for neonates, do not provide readings <40 mg/dL, and there are, as yet, few data on their clinical utility. Another much-needed advance is point-of-care devices that accurately measure glucose and alternative fuels, such as ketones and lactate, in very small samples, developed specifically for neonates. In the longer term, options for noninvasive glucose monitoring, including “pulse glucometers” analogous to pulse oximeters, are already in development.

Definitions of hypoglycemia

Identifying an outcomes-derived definition of hypoglycemia is challenging given that glycemic regulation is dynamic and host factors undoubtedly contribute, so that adverse outcomes may result in one infant but not in another at the same blood glucose concentration. Given this uncertainty, two approaches have been proposed to define hypoglycemia: the statistical definition based on population norms and a neurophysiologic definition based on glycemic thresholds below which a neurophysiologic marker is affected.

The statistical approach defines hypoglycemia as a blood or plasma glucose concentration lower than two standard deviations below the mean (<5%). In term appropriate-for-gestational age healthy newborns, blood glucose concentrations can range between 25 and 110 mg/dL within the first few hours after birth; however, by about 72 hours of age, glucose concentrations typically reach at least 60 to 100 mg/dL.⁴ In the GLOW study, the mean plasma glucose increased from a mean of 59 ± 11 mg/dL in the first 48 hours to 83 ± 14 mg/dL after 72 hours. The challenge with a statistical definition is that different populations of infants have different typical ranges that vary widely, making this difficult to translate to clinical practice.

The neurophysiologic definition has been used in informing current guidelines. Blood glucose concentrations <47 mg/dL have been associated with prolongation of neurosensory response latencies,⁶² although only five of the 17 infants studied were newborns. Cerebral blood flow has been shown to increase with blood glucose <30 mg/dL and transport of cerebral glucose is limited at <54 mg/dL.

As a bridge between the statistical and the neurophysiologic thresholds, operational thresholds have been proposed as an approach to recommend clinician interventions at specific thresholds. Different operational thresholds have been adopted by national and international organizations (Fig. 10.2) based on interpretations of population-based norms and neurodevelopmental outcome studies. However, none of these approaches incorporate thresholds for neuronal injury that may be modulated based on host susceptibility.