Normal development of the immature brain undergoes physiologic pruning mediated by neuronal apoptosis [22]. The developing central nervous system is exquisitely sensitive in its internal milieu. Peak synaptogenesis occurs between the third and seventh postnatal weeks in rats [23]. This is equivalent to the period between 25 gestational week and 1 year of age in humans. However, neurogenesis and context-dependent modulation of neural plasticity continues throughout life from the perinatal period to adulthood. In fact, the rate of neurogenesis peaks in different brain regions in an age-dependent fashion, with a majority of this process occurring primarily during the perinatal period and less during adulthood. It appears that newly born neurons are most vulnerable to the neuroapoptotic effect of anesthetic and sedative drugs [24, 25]. Therefore, nonphysiologic exposure to various drugs and stressors (painful stimuli, maternal deprivation, hypoglycemia, hypoxia, and ischemia) during this critical window may induce neurodegeneration. These findings beg the question of whether other confounding variables are involved in this process. The potential contribution of coexisting medical conditions and undiagnosed genetic syndromes to neurodevelopment has to be considered in light of the potential neurotoxic effects of drugs used for sedation.

The molecular mechanisms that produce immobility, analgesia, and amnesia are still unknown. This fundamental gap hinders the ability of investigators to identify the specific mechanisms that impact the developing central nervous system. Although the anesthetic mechanisms of NMDA antagonists and GABA agonists are divergent, both clearly induce neurodegenerative and neurocognitive changes in animal models [26]. Transient pharmacological blockade of the NMDA receptor with the noncompetitive pharmacological antagonist MK801, phencyclidine, or ketamine induced developmental stage-dependent widespread apoptosis in the developing brain [27]. The initial response from the scientific community was that anesthetic and anticonvulsant drugs and ethanol accelerate this normal “pruning” or apoptotic process. However, this notion was dismissed by a report that commonly used anesthetics (midazolam, isoflurane, and nitrous oxide) induced neuroapoptosis and subsequent derangements in long-term potentiation (an electrophysiological correlate of learning) and neurobehavioral performance [28]. A large number of experimental data from several research groups have confirmed these results [7]. Of note, the proapoptotic effect depends on the developmental stage: being most pronounced at postnatal day 7 and inexistent in 15-day-old rodents. Chloral hydrate has been shown to induce neuroapoptosis in neonatal rats, and lithium protects against this neurotoxic reaction [29]. These preclinical reports clearly demonstrate that drugs that are routinely utilized to sedate pediatric patients have neurotoxic properties.

Several lines of investigation have implicated other neuronal cell death mechanisms such as excitotoxicity, mitochondrial dysfunction, aberrant cell cycle reentry, trophic factor dysregulation, and disruption of cytoskeletal assembly [30–36]. A combination of these and other parallel neurodegenerative pathways likely mediate the neurotoxic effects of anesthetic drugs.

On the surface, the notion that sedative drugs can be excitotoxic can be a contradiction. However, GABA agonists stimulate immature neurons due to a developmental variation of the chloride channels [37]. Subsequent reports on the mechanism of GABAergic-induced seizures in newborn rats revealed that the NKCC1 chloride channel blocker, bumetanide, attenuated the both neuroapoptosis and epileptiform activity [31]. Prolonged exposure to a NMDA antagonist such as ketamine leads to an upregulation of the NMDA receptor, leading to an increased accumulation of excitotoxic intracellular calcium [30]. Excitotoxic insults are also linked to mitochondrial dysfunction in neurons, and prolonged exposure to anesthetic drugs incites a comparable response [32].

Sedative drugs induce neuronal apoptosis in fetal and neonatal rhesus monkeys in a dose- and duration-dependent fashion [38–40]. A 3-h-long exposure to ketamine did not seem to affect cell death, while a 5-h-long exposure has been shown to induce apoptosis both in the fetal and early postnatal brains. This experimental paradigm resulted in persistent cognitive deficits assessed by an operant test battery [8]. Monkeys receiving a 24-h-long ketamine anesthesia at postnatal day 5 showed impaired motivation and learning but no problems with short-term memory when tested up to 3.5 years post-exposure. Propofol anesthesia for 5 h resulted in apoptosis of neurons and oligodendrocytes in fetal and neonatal nonhuman primates [40].

Exposure to sedative drugs during brain development not only induces neuronal cell death but can also impair neurogenesis and synaptogenesis in an age-dependent manner. Postnatal rat pups had decreased neuronal progenitor proliferation and persistent deficits of hippocampal function, while older rats increased progenitor proliferation and neuronal differentiation, and this was correlated with improved memory function [41]. Propofol impairs the survival and maturation of adult-born hippocampal neurons in a developmental stage-dependent manner by inducing a significant decrease in dendritic maturation and survival of newly born neurons that were 17 days old but not at 11 days [25]. The developmental stage-dependent effects of anesthesia exposure during brain maturation are also true in the context of synaptogenesis. Exposure of 7-day-old pups to different kinds of anesthetics and sedatives consistently lead to a rapid and permanent decrease in the number of synapses in the hippocampus and the cerebral cortex. In contrast, when these drugs were administered at later stages of the brain growth spurt, neuronal viability is preserved with a significant increase in synaptic density [11, 42].

Taken together, three factors appear to induce AIDN in laboratory models:

Most of the clinical reports that examine the effect of anesthetic exposure on neurocognitive development are based on retrospective observations of pediatric patients undergoing surgery and presumably general anesthesia. These reports do not specifically identify the classes of anesthetic and sedative drugs administered. Furthermore, these reports do not consider the direct effects of surgery and underlying comorbidities. Although most of the studies have attempted to control for obvious confounders, the retrospective nature of these investigations make it impossible to control for all the known and unknown confounders. Of note, there is no consensus from the published retrospective analyses of the behavioral sequelae after anesthesia and surgery during infancy and children.

Several retrospective reports demonstrate an association between surgery/anesthesia and learning and behavioral disorders. In a series of retrospective reports, the Mayo Clinic group examined a cohort born from 1976 to 1982 for learning disabilities. The patients who were exposed to surgery and anesthesia before the age of four had increased incidence of learning disabilities at age 19 years [43]. Risk factors included more than one anesthetic exposure and general anesthesia lasting longer than 2 h. A similar study, using matched cohorts, revealed that children under the age of 2 who had more than one anesthetic were almost twice as likely to have speech and language disabilities than those who had a single or no anesthetic exposure [44]. In contrast, a cohort study from a birth registry from Australia reported that even a single exposure to general anesthesia before age 3 years was related to decrease performance on receptive and expressive language and cognitive testing done at 10 years [45]. A similar retrospective report derived from Iowa revealed a negative correlation between the duration of surgery/anesthesia and scores on academic achievement tests [46]. Data analysis from the Medicaid database indicates that, even after adjustment for potential confounding factors, children who underwent hernia repair before the age of 3 years were twice as likely as children in the comparison group to be subsequently diagnosed with a developmental or behavioral disorder [47]. When this group was controlled for gender and birth weight, there was still a nearly twofold increase in these issues. A follow-up study that matched patients with non-anesthetic-exposed siblings found that the former had a 60 % greater association between exposure to anesthesia and later neurologic and developmental problems [48].

Meanwhile other investigators report no evidence of an association between exposure to general anesthesia at a young age and later school problems. An analysis of a twin–twin registry from the Netherlands comparing the educational achievements of identical twin pairs revealed that twin pairs exposed to general anesthesia had lower educational achievements than unexposed twin pairs [49]. However, when one twin was exposed and the other was not, there were no differences in educational achievements. These findings imply that exposure to general anesthesia was not associated with impaired educational performance. A Danish birth cohort compared average test scores at ninth grade in infants who had inguinal hernia study and reported no statistically significant differences from the naïve cohorts after adjusting for known confounders [50]. A similar analysis of infants undergoing pyloromyotomies revealed no difference in their educational performance to a surgery-naïve cohort [51]. Since these retrospective reports are based on patients undergoing surgery and presumably general anesthesia, they may not have been relevant in the setting of procedural sedation.

Several reports have been published on the effect of sedation on neurocognitive parameters in intensive care patients. In a review of premature neonates receiving sedation for mechanical ventilation, prolonged sedation was not associated with a poor neurological outcome [52]. A similar report examining the impact of perioperative administration of sedatives in pediatric cardiac surgery found no association between the dose and duration of these drugs and adverse neurodevelopmental outcome at 18–24 months [53]. A reevaluation of these children at kindergarten age demonstrated that the number of days on chloral hydrate was associated with lower-performance intelligence quotient, and the cumulative dose of benzodiazepines was associated with lower visual motor integration (VMI) scores [54]. The Beery-Buktenica VMI scores reflect the ability to integrate visual and motor abilities and screen for possible learning, neuropsychological, and behavioral problems [55]. These sedation studies in the intensive care unit may reveal a mild association between GABA agonists and neurodevelopmental deficits. However, the overwhelming impact of severe illness [56] and prolonged administration of the sedative drugs cannot be discounted. Dexmedetomidine is the only sedative that does not have overt neurotoxic properties in preclinical settings. Its use as a primary sedative for preterm and term neonates has been shown to be effective without major side effects [57]. A direct comparison between dexmedetomidine and lorazepam on septic adults revealed a significant reduction in brain dysfunction with the former [58]. Dexmedetomidine maintains cognitive function in adult intensive care patients requiring cooperative sedation [59]. The effect of dexmedetomidine on both short- and long-term cognitive domains in pediatric patients remains to be investigated.