There is controversial epidemiologic evidence to suggest that children who were exposed to anaesthesia as neonates or infants carry an elevated risk of adverse long-term behavioral or neurodevelopmental outcomes [1-6].

Intravenous anaesthetics that block N-methyl-D-aspartate (NMDA) glutamate receptors, such as ketamine, or are γ-amino butyric acid-A (GABA_A) receptor agonists, e.g. benzodiazepines, are known to cause apoptotic death of neurons in the brains of newborn animals [7, 8], associated with long-term cognitive deficits [9]. Induction of general anaesthesia in infants mostly involves volatile anaesthetics. In newborn rodents, pigs or non-human primates, an increase of neuronal cell death has been observed following isoflurane exposure [10-17], with rather subtle long-term cognitive dysfunction occurring in a partially sex-disparate fashion [4, 18, 19]. Isoflurane-induced neuronal apoptosis appears to occur only at a certain stage of neuronal development, which is dependent on the age of the neuron but independent of the age of the animal [15]. Isoflurane, however, appears to have protective properties as well. Survival of cortical neurons subjected to transient oxygen-glucose deprivation has been found to improve with isoflurane post-conditioning [20]. Moreover, there appears to be reduced brain damage in mice exposed to isoflurane before or after ischemic injury [21, 22] or in rats who received isoflurane after induction of germinal matrix hemorrhage [23].

Exposure to isoflurane, which affects various cell types outside and inside the brain [8, 24], causes a panoply of cardiorespiratory and metabolic side effects [10]. Its use in newborn rodents is associated with considerable mortality [25]. Therefore, we aimed to characterize the immediate action of isoflurane on isolated immature and mature neurons.

All experiments were carried out in accordance with ethical principles and guidelines for experiments on animals and were approved by the animal welfare committees of the Berlin State Office for Health and Social Affairs (LaGeSo), Germany (Reg 0163/03).

Primary neuronal cells were prepared from the cortices of Wistar rat embryos at day 18 (E18) (Forschungseinrichtung für Experimentelle Medizin, Charité - Universitätsmedizin Berlin).

The results of the present study indicate that isoflurane damages a proportion of freshly isolated immature neurons of embryonic rats cultured for 5 days but not mature neurons cultured for 15 days. Isoflurane-induced cell death is inhibited by the pan-caspase inhibitor, zVAD-fmk, pointing to apoptosis as its principal mode. The toxic action of isoflurane can be partially inhibited by bicuculline or picrotoxin, suggesting involvement of the GABA_A receptor pathway. Viability of immature neurons is not affected by increasing doses of fentanyl.

The results obtained with isolated neuronal cells are in line with what is known from experiments involving newborn animals being exposed to isoflurane. In newborn mice, general anaesthesia with 1.5 Vol.% isoflurane has been found to cause a 4-11 fold increase of neuronal death in various brain regions, as compared to controls.[10, 28, 29] A 6-hour isoflurane exposure eliminates approximately 2% of cortical neurons [14]. Apoptosis appears to be the predominant mode of isoflurane-induced death, as evidenced by caspase-3 cleavage [10]. Notably, mice genetically deficient in the apoptosis-triggering cell surface receptors Fas (CD95) or FasL (CD95L) show reduced isoflurane-induced neuronal cell death [30]. Isoflurane susceptibility vanishes in juvenile and adult animals except for brain areas displaying continued neurogenesis [28].

Fig. (3). MTT reduction and LDH release of primary neuronal cells (5 DIV) after 24 hours treatment with increasing concentrations of fentanyl (F1-F9, 0.8, 1.56, 3.125, 6.25, 12.5, 25, 50, 100 and 200 ng/ml, black) compared to controls (white). Data are means ± SEM; n = 3; one-way ANOVA test with Bonferroni post hoc test.

The neuronal apoptosis following neonatal isoflurane exposure is linked to altered behaviour and memory functions, with greater long-term cognitive impairment after multiple neonatal exposures to isoflurane than after a single exposure [18]. In addition to directly driving neurons into apoptotic cell death, isoflurane also impairs the capacity of astrocytes to support neuronal development [31] and evokes errors in axon targeting and growth cone guidance in the developing neocortex in a GABA_A receptor-mediated fashion [32]. Our experiments using bicuculline, a competitive GABA_A receptor antagonist, and picrotoxin, a non-competitive GABA_A receptor antagonist, demonstrate the involvement of the GABA_A receptor pathway in mediating the apoptosis-inducing action of isoflurane. The differing effects of bicuculline and picrotoxin might be explained by the modulation of the GABA_A receptor by isoflurane by increasing the affinity and that there seems to by a blocking action at or near the picrotoxin site [33].

Experimentally, the toxic effect of isoflurane for the developing brain extends to non-human primates. Brains of both fetal [24] and neonatal rhesus macaques [8, 11] display signs of neuronal apoptosis upon exposure to isoflurane (0.7 -1.5 Vol.% for 5 hours). In humans, there is epidemiological evidence to suggest a modestly elevated risk of adverse behavioural or developmental outcomes in children who were exposed to anaesthesia during early childhood [1, 3, 4]. As all retrospective observational approaches, these cohort database analyses are tainted with confounding variables [34], but randomized controlled trials are about to yield answers [35]. Neurodevelopmental changes induced by anaesthetics are subtle and therefore require careful controls to be proven. Rather than causing gross or fine motor impairments which can be diagnosed by standardized examinations at 2 years of age, exposure to anaesthetics at very early age appears to preferentially provoke long-term deficits in cognition, language and executive functions [2, 5] which cannot be reliably assessed prior to school age.

Various aspects emerging from the epidemiological data correspond to the results obtained in the experiments with isolated immature neuronal cells. Susceptibility of the cells to isoflurane was seen only in immature cells, while cells cultured for 2 weeks became resistant to the action of isoflurane. There seems to be a critical period of cellular development during which neurons are susceptible to anaesthesia-induced apoptosis. Vulnerability towards anaesthetics reflects the age of the neurons rather than the age of the organism [15] and is linked to ongoing neurogenesis [28]. In human infants, susceptibility to anaesthetics appears to be restricted to early ages, with very preterm infants exposed before term being at greatest risk [36, 37]. Children exposed to surgery and general anaesthesia prior to 3 years of age had somewhat lower scores than their unexposed peers in receptive and expressive language, even with a single exposure to anaesthesia [2], as did healthy school-aged children and adolescents who had undergone surgery with anaesthesia before 4 years of age [5]. In contrast, exposure to anaesthesia at a later age had no measurable effects on language or cognitive function [38].

Brief exposure times (< 1h) to sevoflurane in vivo had shown to carry little or no risk in a randomized controlled trial involving human infants with a postmenstrual age < 60 weeks undergoing herniorrhaphy [35]. However, in term infants undergoing complex cardiac surgery, a positive association between volatile anaesthetic exposure (isoflurane), MRI-assessed brain injury, and lower neurodevelopmental outcome scores at 12 months of age were demonstrated [39]. In term infants undergoing major non-cardiac surgery, cognitive and motor developmental delay at 2 years of age was found in less than a quarter of the children assessed, with neurodevelopmental outcome scores on average 0.5 SD below the normative score of the healthy population [6].

We also studied fentanyl over a wide range of concentrations, initiated by conflicting reports about the effects of fentanyl in brains of adult rats [40-42]. Freshly isolated human lymphocytes may undergo apoptosis when incubated with fentanyl [43] but fentanyl did not affect cell viability of immature neuronal cells 5 DIV. We used concentrations from 0.8 to 200 ng/ml of fentanyl in cell cultures that correspond to plasma concentration levels of high-dose fentanyl application in adults [44-46] and neonates [47]. Our results are in line with experiments involving mechanically ventilated newborn piglets which showed no signs of neuronal apoptosis after receiving fentanyl administered at a continuous intravenous infusion over 24 hours at dosages of 50-75 µg/kg/h. In contrast, control animals receiving 2 Vol.% isoflurane showed widespread neuronal apoptosis [13]. Animal experiments using moderate doses of fentanyl in newborn rats support the neonatal use of fentanyl even in premature newborns [48]. In very low birth weight human infants, no association between cumulative fentanyl dosage and neurodevelopmental outcomes has been found after controlling for other variables [49].

In summary, the data presented here add weight to considerations to exert caution when using volatile anaesthetics in very preterm infants. Maturation of brain neuronal cells appears to confer resistance to the apoptosis-promoting action of isoflurane but establishing an age limit for safe use of volatile anaesthetics is awaiting further clinical investigations. The apparent lack of neurotoxicity of fentanyl encourages attempts to prioritize opioids over GABA_A agonists when designing anaesthesia protocols for preterm infants.