What Initiates Neuronal Injury?

At the most fundamental level, injury requires a period of insufficient delivery of oxygen and substrates such as glucose (and other substrates in the fetus) such that neurons and glia cannot maintain supplies of high-energy metabolites and thus cannot sustain homeostasis. When this happens, the energy-dependent mechanisms of intracellular homeostasis, such as the sodium/potassium adenosine triphosphate–dependent pump, fail, leading to neuronal depolarization. This creates an osmotic and electrochemical gradient that in turn favors cation and water entry, leading to cell swelling (cytotoxic edema). If sufficiently severe, this may lead to immediate lysis. Importantly, these edematous neurons may still recover, at least temporarily, if the hypoxic insult is reversed or the environment is manipulated. Multiple factors can cause cell injury during and after depolarization. These include the extracellular accumulation of excitatory amino acid neurotransmitters due to impairment of energy-dependent reuptake, which promotes further receptor-mediated cell swelling, excessive intracellular calcium entry, and the generation of oxygen free radicals and inflammatory cytokines. These excitatory factors are balanced by a disproportionate release of inhibitory neurotransmitters such as gamma-aminobutyric acid and adenosine. ^, These inhibitory factors suppress the metabolic rate (termed adaptive hypometabolism) and protect the brain by delaying the onset of cell depolarization. The duration of neuronal depolarization in turn critically determines the severity of neural injury.

Cerebral Injury Evolves Over Time

The central concept that enabled modern studies of neuroprotection is that although brain cell death can occur during sufficiently severe hypoxia-ischemia (termed the “primary” phase of injury), even after surprisingly severe events, many cells can initially reestablish oxidative metabolism in a so-called latent phase, followed by progressive secondary failure of oxidative metabolism and cell death over hours to days. ^, Studies using magnetic resonance spectroscopy demonstrated that many infants with evidence of moderate to severe asphyxia show initial, transient recovery of cerebral oxidative metabolism after birth, followed by secondary deterioration as shown by delayed cerebral energy failure from 6 to 15 hours after birth. The severity of the secondary deterioration is closely correlated with neurodevelopmental outcome at 1 and 4 years of age. Conversely, infants with HIE who did not show initial recovery of cerebral oxidative metabolism had extremely poor outcomes.

An identical pattern of initial recovery of cerebral oxidative metabolism followed by delayed (secondary) energy failure was confirmed after hypoxia-ischemia in piglets, rats, and fetal sheep and is closely correlated to the severity of neuronal injury. ^{, ,} The timing of energy failure after hypoxia-ischemia is closely linked to the appearance of cell death on brain histology. Continuous measurements of cytochrome oxidase, the terminal electron acceptor in the mitochondrial transport chain, using near-infrared spectroscopy demonstrated that after severe asphyxia in fetal sheep, there was initial recovery of cytochrome oxidase to sham control values, followed by a progressive fall that started after approximately 3 to 4 hours and continued until approximately 48 to 72 hours after asphyxia. Delayed loss of mitochondrial activity was associated with a marked increase in relative intracerebral oxygenation, strongly indicating impaired ability to use oxygen. This evidence of a “latent” phase of transient recovery during oxidative recovery offered the tantalizing possibility that therapeutic intervention after hypoxia-ischemia might be possible.

The timing and physiologic events during these phases of injury are now well described in preclinical studies. After restoration of circulation and oxygenation the initial hypoxic depolarization-induced suppression of cerebral oxidative metabolism, cytotoxic edema and accumulation of excitatory amino acids resolve over approximately 30 to 60 minutes. ^, Despite recovery of oxidative cerebral energy metabolism and mitochondrial activity, electroencephalographic (EEG) activity remains depressed. Cerebral blood flow initially recovers but typically shows a delayed, transient reduction below control values within hours after reoxygenation. ^, During the secondary deterioration from approximately 6 to 15 hours after moderate to severe hypoxia-ischemia, delayed seizures develop and then continue for several days, ^, accompanied by secondary cytotoxic edema, accumulation of excitotoxins, failure of cerebral mitochondrial activity, ^, and ultimately, cell death. Secondary edema and seizures are not seen after milder insults that do not cause cortical injury. By contrast, more severe hypoxia-ischemia typically accelerates the evolution of neuronal loss.

As well as evolving over time, cell death spreads outward from the most severely affected regions toward less severely affected regions. In piglets exposed to transient hypoxia-ischemia, the cerebral apparent diffusion coefficients normalized almost completely by 2 hours after resuscitation, followed by a fall in apparent diffusion coefficients beginning in the parasagittal cortex and then spreading through the brain. This pattern likely reflects both severity of injury and active mechanisms such as the opening of astrocytic connexin hemichannels on the cell surface, which can facilitate waves of spreading depression that can trigger cell death in less-injured tissues. The secondary phase resolves over approximately 3 days after severe hypoxia-ischemia into a tertiary phase of ongoing injury, involving chronic inflammation and epigenetic changes affecting repair and reorganization that may last weeks to months and even years.

These concepts—that an acute, global period of hypoxia-ischemia can trigger evolving cell death and that characteristic events are seen at different times after the insult—are central to understanding the causes and treatment of HIE. The initial triggers of the delayed death cascade during exposure to hypoxia-ischemia, including exposure to oxygen free radical toxicity, excessive levels of excitatory amino acids, and intracellular calcium accumulation down the concentration gradient due to failure of energy-dependent pumps during hypoxia and opening of channels linked to the excitatory neurotransmitters. However, these events rapidly resolve during reperfusion from the insult and thus cannot readily be related to the effects of postinsult interventions such as cooling. It is striking that in vitro neuronal degeneration can be prevented by cooling initiated well after exposure to an insult. Thus the key therapeutic targets must involve secondary consequences of hypoxia-ischemia, such as the intracellular progression of programmed cell death (apoptosis), the inflammatory reaction, and abnormal receptor activity.

Intracellular Mediators of Delayed Cell Death

Multiple factors are involved in delayed development of cell death despite initial recovery of oxidative metabolism after hypoxia-ischemia, including activation of cell death pathways, withdrawal of trophic factors, and secondary inflammation ( Fig. 46.1 ). The cell death pathways are stimulated by entry of calcium during anoxic depolarization, exposure to reactive oxidative species during reperfusion, and likely other factors.

Flow Chart Illustrating the Relationship Between the Mechanisms Active in the Pathophysiologically Defined Phases of Cerebral Injury After Moderate to Severe Hypoxia-Ischemia.

During the immediate reperfusion period, lasting approximately 30 to 60 minutes, cellular energy metabolism is restored, with resolution of hypoxic depolarization and cell swelling. This is followed by a latent phase, with near-normal oxidative cerebral energy metabolism as measured by magnetic resonance spectroscopy but with depressed electroencephalographic activity and often a delayed fall in cerebral blood flow. The latent phase is associated with the intracellular components of the apoptotic cascade. This may be followed by secondary deterioration with delayed seizures and cytotoxic edema, extracellular accumulation of potential cytotoxins (such as the excitatory neurotransmitters), and 4 to 15 hours after the asphyxia, failure of oxidative metabolism and damage. The changes in the secondary phase may take 3 days or more to resolve. Effective neuroprotection requires that therapeutic hypothermia is started as soon as possible in the latent phase, before the onset of the secondary deterioration. EAAs , excitatory amino acids; NO , nitric oxide; OFRs , oxygen free radicals.

There is good histologic evidence that activation of preexisting programmed cell death pathways contributes to posthypoxic cell death in the developing human brain. The pattern of cell death is not purely apoptotic but rather includes elements of apoptotic and necrotic processes, with one or the other being most prominent depending on factors such as maturity and the severity of insult. Consistent with the hypothesis that apoptotic processes are a key therapeutic target, postinsult hypothermia started after severe hypoxia-ischemia was reported to reduce apoptotic cell death but not necrotic cell death in the piglet. Similarly, protection with posthypoxic-ischemic hypothermia in fetal sheep has been closely linked with suppression of activated caspase-3.

Inflammatory Second Messengers

Brain injury also induces the inflammatory cascade with increased release of cytokines. These compounds are believed to exacerbate delayed injury, whether by direct neurotoxicity, triggering apoptosis, or promoting stimulation of leukocyte adhesion and infiltration into the ischemic brain. Experimentally, cooling can potently suppress this inflammatory reaction. For example, in vitro , hypothermia inhibits proliferation and superoxide and nitric oxide production by cultured microglia, and in adult rats, hypothermia suppresses the posttraumatic release of interleukin-1β and accumulation of polymorphonuclear leukocytes. Similarly, neuroprotection with postinsult hypothermia was associated with suppression of microglial activation in fetal sheep. ^{, ,}

Other receptor- and nonreceptor-mediated toxic factors are likely to contribute to neural injury in the latent phase. For example, there is some evidence from preterm fetal sheep of delayed production of oxygen free radicals after hypoxia-ischemia, which may be particularly associated with death of oligodendroglia.

Excitotoxicity After Hypoxia-Ischemia

In contrast with their role during the primary phase, the importance of excitotoxins after reperfusion is questionable given that extracellular levels rapidly return to baseline values. ^, Pathologically elevated levels of extracellular excitatory amino acids such as glutamate are seen in a biphasic pattern. The initial increase occurs during hypoxia-ischemia and resolves rapidly after reperfusion, to control values. Levels remain low throughout the latent phase and then secondarily rise many hours later, in association with delayed seizures and cytotoxic edema. ^, In near-term fetal sheep, for example, intense seizures are seen from approximately 9±2 hours to 30±3 hours after cerebral ischemia. It is important to appreciate that completely suppressing these large-amplitude seizures with a selective glutamate antagonist was associated with very limited reduction in neuronal damage in more mildly affected regions but with no effect on infarction of the parasagittal cortex and no improvement in recovery of EEG activity. Moreover, combined antiglutamate treatment and mild whole-body hypothermia after severe asphyxia did not show additive neuroprotection in preterm fetal sheep.

The preclinical studies of therapeutic hypothermia that involved subsequent clinical protocols were structured around these observations.

Timing of Starting Hypothermia: The Earlier the Better

There is extensive preclinical evidence that hypothermia must be started as early as possible within the latent phase for optimal benefit. In turn, this pattern is highly consistent with progressive mitochondrial failure during the latent phase, demonstrated by magnetic resonance spectroscopy after moderate to severe hypoxia-ischemia in human infants, ^, piglets, and near-infrared spectroscopy in preterm fetal sheep. Immediate initiation of hypothermia is protective across species and many paradigms of hypoxia-ischemia. For example, in anesthetized piglets exposed to either hypoxia with bilateral carotid ligation or to hypoxia with hypotension, either 12 hours of mild whole-body hypothermia (35°C) or 24 hours of head cooling with mild systemic hypothermia started immediately after hypoxia prevented delayed energy failure, reduced neuronal loss, and suppressed posthypoxic seizures. ^,

Increasing the duration of hypothermia until resolution of the secondary phase of injury after approximately 72 hours can enable neuroprotection despite delayed initiation. In near-term fetal sheep, cerebral hypothermia induced 90 minutes after reperfusion from a severe episode of cerebral ischemia, that is, in the early latent phase, and continued until 72 hours after ischemia prevented secondary cytotoxic edema and improved electroencephalographic recovery. There was a concomitant dramatic reduction in cortical infarction and improvement in neuronal loss scores in all regions. Comparable neuroprotection was seen when treatment was delayed until 3 hours after the end of ischemia. By contrast, when the start of hypothermia was delayed until just before the onset of secondary seizures (5.5 hours after reperfusion), only partial neuroprotection was seen. With further delay until after seizures were established (8.5 hours after reperfusion), there was no electrophysiological or overall histologic protection with cooling.

Similar results have been reported in other paradigms. In nonanesthetized 21-day-old rat pups, mild hypothermia (a 2°C to 3°C decrease in brain temperature) for 72 hours after hypoxia-ischemia prevented cortical infarction, whereas cooling delayed until 6 hours after the insult had an intermediate, nonsignificant effect. More recently, in 7-day-old rat pups, hypothermia induced either immediately or at 3 hours was neuroprotective after a “moderate” duration of hypoxia-ischemia (90 minutes). By contrast, even immediate hypothermia did not improve outcomes after very prolonged hypoxia-ischemia (150 minutes). This illustrates that the window of opportunity for therapeutic hypothermia is critically dependent on the severity of the primary period of hypoxia-ischemia.

How Cold Is Too Cold?

The critical depth of hypothermia required for protection may be affected by multiple factors such as the delay before initiation and the severity and nature of the insult. There is some evidence from studies of moderate to severe ischemia that when cooling is delayed until 6 hours in both adult rodents and fetal sheep, greater functional and histologic neuroprotection may be seen with a 5°C reduction in brain temperature than with a 3°C reduction. ^, By contrast, in 7-day-old rat pups, there was no additional protection with cooling after hypoxia-ischemia to rectal temperatures below 33.5°C. Similarly, in neonatal piglets exposed to global cerebral ischemia, whole-body hypothermia with a reduction in body temperature of either 3.5°C or 5°C was associated with significant (and highly similar) overall neuroprotection, whereas a reduction of 8°C was detrimental.

If Some Is Good, Is More Better?

There is now compelling preclinical evidence that continuing cooling for approximately 72 hours provides optimal neuroprotection. ^, Broadly, and critically for clinical practice, the greater the delay before starting cooling or the more severe the insult, the greater the duration of cooling required for protection. For example, in adult gerbils, when the delay before initiating a 24-hour period of cooling was increased from 1 to 4 hours, neoronal survival in the CA1 region of the hippocampus after 6 months of recovery fell from 70% to 12%. Subsequent studies demonstrated that protection could be restored by extending the duration of moderate (32°C to 34°C) hypothermia to 48 hours or more, even when the start of cooling was delayed until 6 hours after reperfusion. Similarly, in near-term fetal sheep, cooling that started after 5.5 hours and continued until 72 hours was still partially protective. Finally, in the same paradigm, although delayed cooling from 3 hours after ischemia until 48 hours was partially protective, it was substantially less effective for both recovery of EEG power and neuronal survival than cooling for 72 hours.

Given this compelling evidence that hypothermia must be continued for at least 72 hours in large animals and humans, there has been interest in whether further prolonging the duration of hypothermia may be associated with greater benefit. However, in near-term fetal sheep, when delayed hypothermia starting 3 hours after ischemia was continued for 5 days compared with 3 days, there was no further improvement in electrophysiological recovery or neuronal survival or further reduction in cortical microglial induction. Indeed, post-hoc analysis suggested that extended cooling was associated with a small reduction in neuronal survival in the parasagittal cortex and the dentate gyrus.

Evidence From Randomized Controlled Trials

A systematic meta-analysis of 11 randomized controlled trials of either selective head cooling or whole-body cooling initiated within 6 hours of birth and involving 1505 term and late preterm infants with moderate/severe HIE found consistent beneficial effects after hypothermia. Mild hypothermia was associated with reduced mortality or severe neurodevelopmental disability by 18 months of age (relative risk [RR], 0.75; 95% confidence interval [CI], 0.68–0.83). Cooling reduced mortality (RR, 0.75; 95% CI, 0.64–0.88; 11 studies, 1468 infants), with reduced neurodevelopmental disability in survivors (typical RR, 0.77; 95% CI, 0.63–0.94; 8 studies, 917 infants).

Long-term follow-up of these studies is ongoing; the available evidence suggests a similar improvement in outcomes in middle childhood after mild induced hypothermia for HIE. For example, the Total Body Hypothermia for Neonatal Encephalopathy Trial (TOBY) showed that significantly more children in the mild hypothermia group survived with an IQ score of 85 or more compared with the control group (52% versus 39%; RR, 1.31; P = .04) and that more children in the hypothermia group survived without neurologic abnormalities than in the control group (45% versus 28%; RR, 1.60; 95% CI, 1.15–2.22). Further, there was a significant reduction in the risk of cerebral palsy (21% versus 36%, P = .03) and of moderate or severe disability (22% versus 37%; P = .03). Moreover, recent cohort studies of infants cooled for HIE showed a lower incidence of epilepsy at 2 years of age compared with the cooling trials, as well as reduced severity of cerebral palsy. The reader should note that it is not possible to exclude the possibility that infants with less severe HIE were recruited once therapeutic hypothermia became standard care. Nevertheless, these studies used the same criteria for hypothermia as the original cooling trials, including amplitude integrated EEG (aEEG) monitoring.

Is More Better?

A large, randomized clinical trial of 364 infants with moderate to severe HIE who were randomized either to prolonged duration (120 hours versus 72 hours), increased depth of therapeutic hypothermia (32°C versus 33.5°C), or both was abandoned due to lack of effect and safety concerns. ^, The adjusted risk ratio for death in the neonatal intensive care unit after cooling for 120 hours compared with 72 hours was 1.37 (95% CI, 0.92–2.04), and for cooling to 32°C compared with 33.5°C, it was 1.24 (95% CI, 0.69–2.25). Furthermore, there was no significant overall effect of longer or deeper cooling on death or disability at a mean age of 18 months. These consistent clinical and preclinical findings suggest that there is a relatively broad range of temperatures that are beneficial for the brain after hypoxia-ischemia and that, reassuringly, it should not be necessary to reduce core temperatures by more than approximately 3.5°C.

Is There Benefit From Cooling Started More Than 6 Hours After Birth?

The preclinical and clinical studies reviewed above consistently suggest that hypothermia should be started as early as possible in the first 6 hours of life to achieve optimal outcomes. However, some infants are unable to be started within this time window because of late diagnosis or being born in areas that cannot provide support for cooling. Even though it is not optimal, should these infants be offered therapeutic hypothermia after 6 hours of life? A recent randomized controlled trial conducted by the Neonatal Research Network centered in the United States compared 83 term infants who were cooled starting at 6 to 24 hours after birth (at a mean of 16±5 hours) with 85 noncooled infants (range, 36.5°C to 37.3°C). Death or moderate to severe disability were seen in 24.4% of cooled infants, compared with 27.9% of noncooled infants ( P = .23). Thus, although very delayed treatment was not harmful, this trial strongly suggest that it is critical to focus on initiating treatment as early as possible within the first 6 hours after birth.

Should We Cool Infants With “Mild” HIE?

The large, randomized controlled trials of therapeutic hypothermia excluded infants who had “mild” HIE in the first 6 hours of life in order to increase the rates of unfavorable outcome, so the potential benefit of treating these infants with therapeutic hypothermia is unknown. In cohort studies, some infants with mild HIE as defined using the trial criteria in the first 6 hours of life have material risk of disability. The exact results have been rather variable, likely because of variable criteria for “mild,” retrospective identification, less formal neurologic examinations than used in the prospective trials, or not using aEEG criteria. The definition of “mild” HIE varies markedly between studies. The CoolCap and TOBY studies simply specified the clinical Sarnat criteria on a gestalt basis but also required that infants had moderate to severe changes on an aEEG recording before randomization. ^, The NICHD trial used only clinical criteria and excluded infants who did not have three or more criteria for moderate/severe HIE ; thus infants with one or two moderate or severe criteria were defined as having “mild” HIE. This suggests that some of these infants actually had “moderate” HIE even though they were excluded from the trial and raises the important possibility that it may be possible to refine the clinical criteria to more reliably identify infants who are at risk of disability.

A meta-analysis of studies with well-defined HIE grading at birth and standardized neurodevelopmental assessment at 18 months or older suggested that 86 of 341 infants (25%) with “mild” HIE in the first 6 hours of life had an adverse outcome, defined as death, cerebral palsy, or neurodevelopmental test scores that were more than 1 standard deviation below the mean. Although most of these studies recruited infants solely on clinical criteria, a prospective cohort study of infants who were not treated with therapeutic hypothermia found that infants with mild HIE, determined by both early EEG and clinical examination, had adverse cognitive and neuromotor outcomes at 5 years of age compared with healthy controls. Although survival was much greater after mild than moderate or severe HIE, survivors with mild HIE showed no significant difference in cognitive outcomes compared with those who had had moderate HIE.

Given that this population of infants with “mild” HIE in the first 6 hours is heterogeneous, the balance of clinical risk and benefit is unclear. Treating all cases of “mild” HIE would lead to a considerable increase in the number of infants being separated from their parents for at least 3 days and receiving invasive treatments such as central lines, invasive respiratory support, sedation, and delayed oral feeding. It is reasonable to reflect that there is evidence from young rodents that therapeutic hypothermia seems to be more protective after milder hypoxia-ischemia. Taken as a whole, it is likely that established protocols for therapeutic hypothermia will also reduce cell loss in infants with milder clinical HIE in the first 6 hours of life. Given that there are roughly as many infants with mild HIE as there are with moderate to severe HIE, it is critical that the benefits of treatment for this group are now formally tested.