Portal de Periódicos da CAPES

PRO: Anesthesia-Induced Developmental Neuroapoptosis: Status of the Evidence

2008; Lippincott Williams & Wilkins; Volume: 106; Issue: 6 Linguagem: Inglês

10.1213/ane.0b013e3181731ff2

1526-7598

Vesna Jevtović‐Todorović, John W. Olney,

Neuroscience and Neuropharmacology Research

Eight years ago, Ikonomidou et al.1 reported that treatment of infant rats with drugs that block N-methyl-d-aspartate (NMDA) glutamate receptors causes widespread apoptotic neurodegeneration in the developing brain. In a series of follow-up studies,2–9 it was determined that a similar neuroapoptotic reaction is readily induced in either the infant rat or mouse brain by: (1) drugs that activate γ-aminobutyric acid type A (GABAA) receptors; (2) ethanol, which has both NMDA antagonist and GABAA agonist properties; (3) antiepileptic drugs, including those that activate GABAA receptors and those that block sodium ion channels; (4) exposure for 6 h to a cocktail of anesthetic drugs (midazolam, nitrous oxide, and isoflurane) having both NMDA antagonist and GABAmimetic properties, and; (5) subanesthetic (based on species-dependent anesthesia requirements) exposure to individual anesthetic drugs, including ketamine, midazolam, propofol, and isoflurane. The period of peak vulnerability to the apoptogenic action coincides with the developmental period of rapid synaptogenesis,1,2 also known as the brain growth spurt period, which in mice and rats occurs primarily in the early postnatal period, but in humans extends from about midgestation to several years after birth.10 The cell death process triggered by these drugs has been demonstrated by various histological methods, including silver, flurojade-B, and TUNEL staining, and has been shown by electron microscopy to have all of the classical morphological characteristics of apoptosis.3,4,6 The mechanism is selective for specific neuronal populations, but the vulnerable neurons are distributed widely throughout the forebrain, midbrain, cerebellum, brainstem, spinal cord, and retina,1–3,6,11–13 signifying that no region of the central nervous system is totally spared. Several different patterns of degeneration are observed, depending on whether drug exposure occurs in early, mid, or late synaptogenesis. Studies using Bax knockout mice and infant rats have revealed that the cell death process is Bax-dependent and involves down-regulation of bclxL, mitochondrial injury, and extramitochondrial leakage of cytochrome c, followed by a sequence of changes culminating in activation of caspase-3.14–18 In addition, the involvements of brain-deprived neurotrophic factor-dependent and death receptor-dependent pathways were recently documented.17–19 Findings using caspase-3 knockout mice suggest that commitment to cell death occurs before the caspase-3 activation step,15 which signifies that immunohistochemical detection of neurons positive for activated caspase-3 provides a reliable means of mapping and quantifying dying cells that have already progressed beyond the point of cell death commitment. Accordingly, activated caspase-3 immunohistochemistry has been used extensively for marking dying neurons in recent studies focusing on drug-induced developmental neuroapoptosis.4,8,9,14–16,20 The above findings have potential relevance in a public health context because there are many agents in the human environment that have NMDA antagonist or GABAmimetic properties. Such agents include drugs that may be abused by pregnant mothers (ethanol, phencyclidine, ketamine, nitrous oxide, barbiturates, and benzodiazepines), and many drugs used worldwide in obstetric and pediatric medicine as anticonvulsants, sedatives, and anesthetics. Evidence for anesthesia-induced neuroapoptosis in the developing animal brain has sparked a vigorous debate between those who have generated much of this evidence and others who have argued that the available human data suggest that anesthetic drugs pose very little risk of inducing neuroapoptosis in the developing human brain. In the following paragraphs, we will consider how the arguments that have been embraced comport with available evidence. The first argument pertains to the doses used. Ikonomidou et al.1 originally reported that repeated doses of ketamine (20 mg/kg sc) triggered neuroapoptosis in the infant rat brain. Others confirmed this finding,21,22 but reported that a single dose did not trigger neuroapoptosis in the infant rat brain. On the basis of these findings, Soriano et al.23,24 argued that, since ketamine is usually used in human anesthesia as a one-time treatment in doses lower than have been shown to be toxic in infant rats, ketamine can be considered safe as a pediatric anesthetic. As recently as July 2007, Anand reiterated this point of view and advised the readers of Anesthesiology25 that it requires "huge doses and prolonged exposures" for anesthetic drugs to trigger neuroapoptosis in animal brain. This assertion is contradicted by several recent studies. For example, Young et al.7 reported that a single dose of ketamine (20 mg/kg or higher) triggers a significant and dose-related neuroapoptosis response in the infant mouse brain. Anand overlooked the fact that the appropriate basis for cross-species comparisons is the dose required in each species to achieve a specified level of anesthesia/analgesia. The dose of ketamine required to anesthetize a human infant is in the range of 5 mg/kg and the dose required to anesthetize a mouse is 80 mg/kg.26 Thus, in the Young et al. study, the dose required to trigger neuroapoptosis was in a subanesthetic range (1/4 of an anesthetizing dose). Equally relevant is evidence that subanesthetic exposure to midazolam7 or propofol8 triggers neuroapoptosis in the infant mouse brain. Moreover, similar findings have been reported for the volatile anesthetic, isoflurane. Loepke et al.27 demonstrated that the minimum alveolar concentration (MAC) for anesthetizing infant mice with isoflurane is 2.26%. Johnson et al.9 administered isoflurane to infant mice at 2% for 1 h, 1.5% for 2 h, or 0.75% for 4 h, and found that each of these sub-MAC protocols induced significant neuroapoptosis. Consistent with these findings, Ma et al.28 and Jevtovic-Todorovic et al.6 have demonstrated that 0.75% isoflurane (0.33 MAC) triggers neuroapoptosis in the infant rat brain. When viewed in light of such evidence, the "huge doses and prolonged exposures" argument25 is untenable. Some have argued that anesthesia-induced neuroapoptosis is a species-specific phenomenon that can only be demonstrated in rats. For example, it has been reported that exposure of infant mice, rabbits, or piglets29 or near-term fetal sheep30 to isoflurane for 4 or 6 h, respectively, does not trigger neuroapoptosis. The most likely explanation for the negative findings in these several species is that the authors did not examine the brains at an appropriate posttreatment interval. The interval in one study was 48 h29 and in the other it was 6 days30 after drug exposure. Twenty-four hours is the latest time interval that can be expected to show a neuroapoptosis response, and only the most extreme response (induced by high-dose drug exposure) will be detectable at this late interval. To detect significant, but less extensive neuroapoptosis induced by less extreme exposure conditions, it is necessary to examine the brains much earlier while clear-cut signs of acute cell death are there to be detected. This neuropathological phenomenon is subject to a dose-response gradient whereby low doses kill the most sensitive neurons in the brain and higher doses kill both the most sensitive and more resistant populations. When only the most sensitive neurons are killed, it is a rapidly occurring event that transpires within several hours after initiation of drug exposure. Looking for signs of acutely degenerating neurons at 48 h or 6 days is futile; the most sensitive neurons have died and been phagocytozed within the first 6-12 h and many of the more resistant neurons are already undergoing phagocytosis by 18-24 h. Detection of the acute neurodegenerative process is also dependent on the histological method used. Silver staining is useful for mapping severe neurodegenerative reactions because it indelibly stains dying cells, including fragments of debris given off by such cells, and darkens the site of injury until all such components have been removed. Activated caspase-3 immunohistochemical staining is ideal for demonstrating early degeneration of the most sensitive neurons because early in the degenerative process, before silver impregnation occurs, activated caspase-3 is generated throughout the cell body and dendritic arbor. Thus, the cell body and all of its processes are displayed early in full detail by this stain. However, as the cell decomposes, it loses immunoreactivity and becomes invisible to this stain. Therefore, both the histological procedure and the posttreatment time interval are critically important for accurate assessment of the neuroapoptosis response. In studies paying close attention to the above principles, it has been shown that subanesthetic doses of many individual anesthetic drugs (isoflurane, ketamine, propofol, and midazolam) trigger neuroapoptosis in the infant mouse brain.7–9 Anand has suggested25 that this implies that mice, compared with humans or other species, are hypersensitive to this toxic mechanism. In our view, this is not a correct interpretation of the data. The correct interpretation is that the most sensitive methods have been applied in the mouse studies. We do not know about humans because, for ethical reasons, the issue cannot be studied with the same sensitive methods in humans. It has also been hypothesized that either hypoxia/ischemia23,24 or hypoglycemia27 might be responsible for the neurodegenerative reaction associated with anesthetic drug exposure, but accumulating evidence contradicts these hypotheses. For example, it has been demonstrated repeatedly that the acute cell death response to hypoxia/ischemia has excitotoxic features that are distinctively different by ultrastructural analysis from the apoptotic features that characterize anesthesia-induced neuroapoptosis.3,31,32 It has also been shown that when infant mice are treated with a neuroapoptogenic dose of ketamine, or infant rats with an anesthetic combination containing isoflurane, midazolam, and nitrous oxide, blood gas values, including arterial oxygen saturation, remain normal throughout the posttreatment period while neurons are undergoing apoptosis.7,17 These findings were recently corroborated by Loepke et al.27 who found that when C57BL6 infant mice were exposed to isoflurane under conditions that produce neuroapoptosis,6,9 the mean arterial blood pressure remained stable and blood gas values did not vary markedly from unanesthetized controls. Although Loepke et al.27 did not find evidence for hypoxia/ischemia, they reported that exposure of infant mice to 1.8% isoflurane caused a decrease in blood glucose to a mean value (±sd) of 53 ± 22 mg/dL (n = 4). Therefore, they postulated that hypoglycemia may be a contributory factor for any neuroapoptosis that isoflurane might cause in infant mice. However, in recent studies involving a much larger number of animals, isoflurane was administered at three different concentrations and durations to infant mice,9 and a triple anesthetic cocktail (isoflurane, midazolam, and nitrous oxide) was administered to infant rats.19 In these studies, each exposure condition triggered neuroapoptosis whereas blood glucose values remained equal to or higher than control values. Finally, it has been argued that, because humans live longer than mice and have a more prolonged synaptogenesis period, it may require a much longer exposure to an anesthetic drug, perhaps as long as 2 wk, for a human neuron to succumb to the apoptogenic stimulus.23,24 This hypothesis is inconsistent with evidence33 that neurons in the fetal non-human primate brain (fascicularis monkey) undergo apoptosis very rapidly (within <8 h) after transplacental exposure to ethanol. It seems unlikely that if ethanol, a GABAmimetic agent, triggers neuroapoptosis in primate neurons within a brief period of hours, it would require 2 wk for primate neurons to show an apoptotic response to a GABAmimetic anesthetic drug. At a recent open public meeting of the Anesthetic and Life Support Drugs Advisory Committee of the United States Food and Drug Administration, Slikker et al. presented recently published evidence34 that an IV infusion of ketamine lasting 24 h triggered neuroapoptosis in the frontal cortex of the rhesus neonatal (P5) monkey brain, but they did not detect a significant neuroapoptosis response after ketamine infusion lasting 3 h. At the same meeting, Olney et al. presented preliminary evidence that administration of ketamine by IV infusion for 5.5 h to the pregnant fascicularis monkey at gestational age 120 days triggered neuroapoptosis that was detectable in several regions of the fetal brain 2 h after cessation of ketamine exposure. However, it was cautioned that no conclusions can be reached regarding sensitivity of the developing monkey brain until more extensive monkey data become available. Does the evidence for anesthesia-induced neuroapoptosis signify that neurons are permanently deleted from the brain and not replaced or otherwise compensated for? This is an important question, which deserves continued evaluation. It has been demonstrated by counting missing neurons weeks or months after drug exposure that a high dose of ethanol permanently deletes up to 68% of the neurons in certain brain regions; evidence for replacement of the missing neurons by neoneurogenesis was looked for but not found.4,35 Determining whether the neonatal neuropathological changes are associated with subsequent neurocognitive deficits is one way of addressing the permanency issue. Jevtovic-Todorovic et al.6 have demonstrated that exposure of infant rats to a clinically relevant cocktail of anesthetic drugs (midazolam, nitrous oxide, and isoflurane) for 6 h triggers widespread neurodegeneration in the developing brain and is associated with neurocognitive deficits that persist through adolescence into adulthood. Stratmann et al.36 have presented preliminary findings that infant rats exposed to 1.9% isoflurane anesthesia for 4 h displayed significant cognitive deficits when tested at 5 and 8 mo of age. Wozniak et al.35 found that a one-time exposure of infant mice to ethanol causes profound learning/memory deficits in adolescence, followed by partial functional recovery in adulthood. Thus, it appears that neuroapoptosis induced in the developing animal brain does give rise to subsequent neurocognitive disturbances, although some degree of recovery may be possible. It is logical to expect significant recovery of function because the pathological process occurs at a time of great neuroplasticity, and it does not ablate entire regions of the brain, but rather deletes some neurons from many regions, whereas sparing others that can take over the function of their lost neighbors. If significant functional recovery is a feature of this type of neuropathological syndrome, that would be very fortunate, but it would also substantially complicate our efforts to determine whether humans, as well as animals, are susceptible to anesthesia-induced neuroapoptosis and cognitive injury. Arguing for the human relevance of the animal findings is the well established fact that ethanol damages the developing human brain and causes a spectrum of neurobehavioral disturbances ranging from mild hyperactivity/attention deficit and learning disturbances to frank mental retardation (fetal alcohol syndrome).37 The most promising candidate mechanism for explaining this alcohol-induced neuropathological syndrome is the developmental neuroapoptosis mechanism that ethanol shares with numerous other drugs, including most, if not all, general anesthetics. It can be argued that there is no compelling evidence in the medical literature that exposure of the developing human brain to anesthetic drugs causes fetal alcohol syndrome-like neurodevelopmental pathology. However, this argument begs the sobering response that over the millennia ethanol has damaged millions of human fetal brains, but it was not until 30 yr ago that the medical community began to recognize a causal connection. Even more sobering is the fact that it was not neurocognitive disturbances that alerted physicians to ethanol's deleterious effects; it was gross craniofacial malformations38 that ethanol causes by an unknown mechanism in the first trimester of pregnancy. Neuroapoptosis induced by ethanol or by anesthetic drugs in late pregnancy or after birth would be expressed as neurocognitive disturbances in the absence of craniofacial malformations and, therefore, would be much more likely to go unrecognized. An additional reason why it might be much more difficult to detect the pathological impact of anesthetic drugs, compared with ethanol, is that a fetus whose mother is addicted to alcohol is likely to be exposed repeatedly during the period of vulnerability to sustained high blood concentrations of the apoptogenic agent, whereas exposure of human infants to anesthesia is often limited to a single relatively brief episode. Although it only requires a single exposure to an apoptogenic drug for neurons to be deleted from the brain, repeated exposures to sustained high blood levels would be much more likely to cause massive neuronal losses and correspondingly more severe and more easily documented neurocognitive disturbances. Is there any evidence that a single exposure to anesthesia can cause neurological injury? Isoflurane and related halogenated ethers are used very extensively worldwide for procedural sedation or for maintaining a surgical plane of anesthesia in pediatric patients for long periods of time, in some cases for up to 6-9 h,39 and more rarely for one or more days.40,41 In the studies just cited, acute neurological dysfunction was reported as a consequence of prolonged isoflurane exposure, but the patients recovered within 3 days. Rapid recovery of neurological function was considered a sign that the injury was reversible and benign; long-term follow-up studies were not conducted. When infant animals are exposed to anesthetic protocols that acutely delete neurons from the brain and cause long-term neurocognitive disturbances, there are no overt signs of neurological dysfunction during the immediate period of recovery from anesthesia.6 It follows that if obvious neurological impairments do occur in human infants after prolonged isoflurane anesthesia, and these impairments resolve within 3 days, as was the case in the studies cited above,40,41 this does not warrant the conclusion that the impact on the brain was totally benign.