Controversies in Anesthesia-Induced Developmental Neurotoxicity

Abstract

Advances in the field of pediatric anesthesiology have enabled the performance of complex and life-saving procedures with minimal patient discomfort. However, preclinical studies over the past two decades have been reporting substantial neurotoxic potential of general anesthetics in young brain, thus challenging the safety of these agents in pediatric anesthesiology practice. Notwithstanding the overwhelming preclinical evidence, the translatability of these findings has proven inconsistent in human observational studies. The significant degree of anxiety and apprehension surrounding the uncertainty of long-term developmental outcomes following early exposure to anesthesia have prompted numerous studies around the world to investigate the putative mechanisms and translatability of preclinical findings regarding anesthesia-induced developmental neurotoxicity. Guided by the vast preclinical evidence, we aim to highlight relevant human findings presented in the currently available clinical literature.

Introduction

Conquering painful sensations has been of pivotal interest in medicine and surgery for centuries. The invention of anesthesia helped to overcome an insurmountable barrier in how modern medicine is practiced. By controlling the nociceptive sensory systems triggered by the adverse sensations during painful procedures, complex, but potentially life-saving procedures once nonfeasable, became limited only by the available technology and operator’s skills.

Although traditionally considered very innocuous, recent concerns over the safety of anesthetics in select vulnerable patient populations had brought on some concerns regarding the long-term outcomes from anesthesia exposure. Notwithstanding the concerns regarding potential long lasting effects of anesthesia in elderly patients, our review will focus on rapidly accumulating evidence in animal studies and emerging findings in human studies suggesting that early-life exposure to general anesthetics may pose some risks to cognitive and socio-affective development of young individuals [1–5]. Having said that, the incomplete translatability of animal models combined with the ethical and practical conundrums of designing clinical studies creates uncertainty as to the most sound approach on how to move forward. When considered in view of readily observable and undeniable benefits of life-saving and/or life-altering surgery, the cancellation or even postponement of neonatal surgical procedures is often very problematic [6]. In this review, we provide an overview of currently available animal evidence, then proceed to discuss a select group of clinical studies looking at the relevance and magnitude of anesthesia-induced developmental neurotoxicity.

Overview of Mechanisms of Action and Neurotoxicity of General Anesthetics

Although general anesthetics are known to induce lack of consciousness and insensitivity to pain, the complete understanding of their mechanism(s) of action remains unclear. It appears that their action at the anatomical level involves a disruption of ascending signaling pathways from the spinal cord, via the thalamus to the higher cortical centers [7, 8]. At a molecular level, a so-called “lipid hypothesis” proposes the mechanism of action via partitioning of anesthetics within the lipid bilayer of the cell membrane, leading to changes in membrane fluidity and consequential interference with function of numerous proteins anchored in the cellular membrane [9]. Although somewhat forgotten in recent years, a newly published study revived the importance of lipid hypothesis as a mechanism of action of inhaled anesthetics, such as isoflurane and chloroform, through disruption of GM1 lipid rafts and phospholipase D2-dependent mechanism [10].

A second, more widely accepted hypothesis posits that general anesthetics exert their effect primarily via different proteins in cell membranes, more specifically ion channels [11]. Most general anesthetics used in daily clinical practice modulate the excitatory glutamatergic N-methyl-D-aspartate (NMDA) and/or inhibitory gamma-aminobutyric acid type A (GABA_A) receptors [11], although their promiscuous nature has been shown to engage additional cellular targets such as potassium [12] and T-type calcium channels [13, 14]. In concert with glutamatergic and GABA-ergic putative targets, these cellular targets contribute to the observed end-point effects of general anesthetics.

Considering that these cellular targets are crucially important for the formation of fundamental neuronal circuitries and neuronal communication [15–18] during normal brain development, a seemingly simple question was posed: Could a substantial modulation of these targets have potentially harmful effects on neuronal development, maturation and migration? Based on numerous animal reports over the past two decades, it is becoming clear that anesthesia exposure during critical points of brain circuitries formation may cause substantial damage that seems to be largely dependent on the timing of exposure [19–21]. The period of peak vulnerability correlates with the period of intense synaptogenesis which is species-specific. In humans, it starts in the last trimester of in utero life and goes on for the first couple of years of postnatal life. In rats and mice, it is largely a postnatal phenomenon and it occurs during the first 2-3 weeks of postnatal life [22]. During this early developmental period, trillions of synapses are being formed and remodeled; neurons are directed to their final destination, connections are reinforced and neurons are wired; those that fail to form meaningful functional synapses are considered redundant and, as such, are apoptotically deleted [23–25].

Yon et al. examined a dynamic nature of anesthesia-induced neurotoxicity in rats by studying the severity of developmental neuroapoptosis over the course of the first two weeks of postnatal life [21]. The neural injury was shown to be low at birth, it peaked at postnatal day 7 (PND7) in multiple cortical, subcortical and thalamic brain regions, and by PND14 was indistinguishable from controls [21]. The exquisite vulnerability of PND7 rats correlated with the peak of their synaptogenesis and was associated with anesthesia-induced downregulation of anti-apoptotic protein, Bcl-x_L, triggering release of mitochondrial cytochrome c and activation of executioner caspases. In contrast, PND14 rats responded via upregulation of Bcl-x_L, leading to stabilization of mitochondrial membrane. Furthermore, a programmed shift in the expression of chloride channels evidenced by the reversal of NKCC1/KCC2 ratio after week 2 of rodent life [15] was shown to be delayed beyond the physiologically-appropriate age range following neonatal sevoflurane administration [26]. Although GABA excitatory signal may be necessary for neuronal survival during early development [15], the hyperexcitability resulting from abnormal NKCC1/KCC2 ratio was shown to facilitate seizure activity in immature rats [27], and might play a role in disrupting the formation of neuronal networks of critical importance for neurocognitive and behavioral functioning [15].

Despite a seemingly brief and strictly defined window of neuroapoptotic vulnerability, early-life anesthesia-induced deviation from normal developmental trajectory may persist for weeks or months. Elimination of redundant synaptic connections via the process of pruning may be permanently altered by neonatal ketamine administration [28, 29]. As noted in these studies, the disruption of normal pruning of infrapyramidal bundle in mouse hippocampus was accompanied by the hyperexcitability of CA1 pyramidal neurons at timepoints several weeks after anesthesia exposure and long after the drug is completely eliminated from the system [28, 29].

From Animals to Humans – Long-Term Behavioral Outcomes Following Neonatal Anesthesia Exposure

While the abundance of data in preclinical studies has provided evidence for behavioral phenotype via long-term follow-up of neonatally anesthetized animals [1–5], the long-term consequences of early-life anesthesia in humans remains an open question. The inconsistencies in clinical findings are heavily influenced by the studies with complex findings such as significant impairment in one behavioral domain but not the other [30, 31]; significant findings following multiple exposures but perhaps not a single one [32–34] and, significant impairments noted only during a specific age window which do not always follow the predicted vulnerability window, i.e., paradoxically occurring in somewhat older children but not the younger ones [35, 36]. Therefore, the recommendations about the safety of these drugs in light of clinical evidence, although steadily accumulating over the past decade or so, are exceptionally difficult to formulate if the goal is to provide proof beyond reasonable doubt.

In an attempt to bridge the gap between convincing preclinical findings and inconsistency of clinical reports, we will now explore a select set of factors which may account, at least in part, for some of the variabilities in clinical reports, and provide opinion on future directions.

Timing of Anesthesia Exposure

An early attempt to answer the question whether the phenomenon of anesthetic neurotoxicity is relevant to humans was published in 2009 [37]. The study was conducted in a retrospective fashion focusing on children who had undergone urologic surgery during the first six years of life. Understanding the critical importance of age, the authors stratified the children in several age groups according to the age at the time of the procedure (0-6 months, 6-12 months, 12-24 months and >24 months) and administered a questionnaire to the parents about the child’s well-being. The purpose of the questionnaire was to screen for school performance and to capture a range of externalizing and internalizing behavioral outcomes. Although the results of the study suggested a relationship between early anesthesia and increased risk of “deviant behavior”, huge confidence intervals dictated by the small sample size meant that no definitive answer could be provided thus calling for large-scale studies to be conducted in the future.

Two subsequent retrospective studies launched by unrelated research groups to address similar questions looked at a much larger cohort of children exposed to anesthesia during the first years of life that were matched to unexposed controls. Both studies [35, 36] reported a small, but statistically significant, increase in the rate of developmental disorders in children exposed to anesthesia compared to unexposed controls. Interestingly, the authors reported a higher incidence in children older than two years, but not in younger ones when compared to their age-matched controls. This seems to be in contrast with the reports by DiMaggio et al. [38] and Stratmann et al. [30] who independently observed increased risk when the exposure occurred prior to the age of three and one, respectively. This could be, at least in part, due to the fact that children diagnosed with developmental disorders prior to 5 years of age, regardless of exposure status, [35] were excluded which could have played a significant role in the lack of association reported in children exposed to anesthesia before the age of two.

The abundance of animal studies convincingly shows that anesthesia exposure during the peak of brain growth spurt triggers widespread apoptosis of neurons in a variety of animal species [1, 21, 39–47] with persisting behavioral deficits. However, it should be noted that the peak of synaptogenesis differs not only between the species, but also between various brain regions in each species. For example, although subiculum, cortex and thalamic nuclei reach the peak around PND7 in rodents [21], other brain regions such as cerebellum and olfactory bulbs peak several weeks later around PND21 [48]. Therefore, it is possible that later exposures could result in markedly different behavioral outcomes due to differential susceptibility of various brain regions responsible for respective behavioral domains.

Indeed in well-designed retrospective analysis, Ing et al. examined neurodevelopmental outcomes following anesthesia exposure between 3 and 10 years of age [49]. Using a range of neuropsychological tests, authors noted that deficits in children exposed after their third birthday substantially differed from those in children under the age of three at the time of exposure [50]. Unlike in younger children, in children older than three years at the time of exposure, significant deficits were noted in both gross and fine motor skills; however, unlike younger children, their language development and abstract cognition were intact. These findings suggest that certain functional domains may exhibit a different window of vulnerability. Interestingly, motor functions, presumably under cerebellar control, were preferentially affected by the later exposure to anesthesia which could be explained by delayed maturation of the cerebellum compared to other brain regions (as previously shown in animal studies) [48].

It is clear that although the age at exposure appears to be a major factor in developmental disorders resulting from early-life anesthesia, there is still much unknown about the exact relationship between the timing of anesthesia and the specific behavioral, cognitive and socio-affective domains that could preferentially be affected. This may help to explain the variety of clinical outcomes reported thus far and some of the controversies regarding the post-anesthesia developmental outcomes.

Cumulative Dose and Repeated Exposure to General Anesthesia

Animal studies have suggested that the cumulative dose of administered anesthetics, a function of not only the dose and duration of anesthesia but also the total number of exposures, could be directly correlated with the severity of patho-morphological changes and long-term functional impairments. For example, it has been reported that 6h-administration of 3% sevoflurane during the first week of life reliably and reproducibly triggers widespread neuroapoptosis and is associated with learning and memory deficits in mice later in life [39, 40]. Furthermore, when 3% sevoflurane was administered 2h daily for 3 consecutive days to PND6-8 mice, this anesthesia regimen similarly induced cognitive impairments [51]. Conversely, if the same regimen was administered to 60-day old mice or if 3% sevoflurane was administered as a single 2h-exposure to PND6 mice, no cognitive impairments were observed upon further testing [51] indicating that not only the timing of exposure (as discussed earlier), but the cumulative dose is also a crucial determinant of vulnerability to anesthetic neurotoxicity. It is noteworthy that although both regimens caused significant synaptic loss compared to controls, three shorter exposures (two hours each) to sevoflurane appeared to cause greater loss of synapses compared to a single longer exposure (6 hours) [52].

The implication that multiple anesthesia exposures may result in greater impairment than a single one seems to be relevant in human studies as well. In a population-based cohort study of children born in Olmsted County, Minnesota, two reports almost a decade apart have concluded that children exposed to multiple, but not a single general anesthesia prior to their fourth birthday were nearly twice as likely to exhibit learning disabilities prior to 19 years of age[32, 33]. Similarly, DiMaggio et al. [34] assessed developmental outcomes after a single and/or multiple anesthesia exposures and concluded that the risk of being diagnosed with cognitive and/or socio-affective disorders was significantly increased after two (hazard ratio = 2.8) and three or more (hazard ratio = 4.0), but not single exposure to anesthesia (hazard ratio = 1.1).

Having said that, it is noteworthy that based on the data collected from nearly three thousand children pooled from the Raine study cohort (The Western Australian Pregnancy Cohort), born in the period between 1989 and 1992, it appears that both single and multiple exposures to general anesthesia before the age of three could lead to increased relative risk of impairment in language and abstract cognitive difficulties when tested at the age of ten [50]. Importantly, the degree of impairment appeared to be similar between single- and multiple-exposed children. As the authors noticed in the discussion, one possible explanation could be that the tests used to examine the language and abstract cognitive capacities of children in this study might have greater sensitivity compared to the outcomes previously assessed [32–34]. On the other hand, the smaller sample size, especially in the anesthesia-exposed groups, might have precluded the observation of dose-response relationship between the number of exposures and the studied outcomes [50]. It should be noted that, although there was no direct access to surgical and anesthetic records after the perinatal period, the majority of children were most likely exposed to halothane based on its prevalence of use during the study period. Although the early halothane exposure was similarly associated with detrimental effects on synaptogenesis and learning and memory deficits in rats [53, 54], it is noteworthy that the preclinical evidence is not as well-established relative to the newer volatile and injectable anesthetics.

The Impact of Underlying Disease on Long-Term Behavioral Development Following Early-Life Anesthesia

Another potential, yet often overlooked, reason for the apparent disconnect between animal and human studies may be due to the presence of confounding factors associated with the disease which prompted anesthesia in the first place. An important difference between animal and human studies is that in humans, anesthesia is commonly administered in the setting of a disease process, whereas in animal studies anesthesia is the primary and often the sole injurious stimulus to the developing brains.

Several studies to date have attempted to shed light on the contribution of the underlying diseases on neuronal injury induced by neonatal anesthesia with mixed findings. To study the effects of concurrent noxious stimulation on the anesthesia-induced neuroapoptosis, Liu et al. performed intraplantar administration of complete Freund’s adjuvant (CFA), known to induce painful sensations and regional inflammation, simultaneously with 6h ketamine anesthesia in PND7 rats. The authors observed that concurrent anesthesia and nociceptive stimulation by CFA resulted in less apoptotic neuronal death compared to anesthesia alone which may be attributable to a decreased cell cycle reentry by countering the ketamine-induced cyclin D1 upregulation [55]. In contrast, another study by Shu et al. investigated neuronal death following either intraplantar formalin injection or surgical incision at the time of isoflurane anesthesia [56].In contrast to the other study, Shu et al. demonstrated significant increase in neuroapoptosis in the setting of formalin-induced inflammation and/or surgical incision when compared to isoflurane alone.

Several important points should be made when analyzing these two studies. Ketamine, as an NMDA receptor antagonist, has been previously shown to protect against cell death in response to inflammatory pain induced by repetitive formalin injection [57]. On the other hand, isoflurane, acting mainly via GABA-mimetic mechanism, may independently cause upregulation of proinflammatory cytokines in the brains of neonatally-exposed mice [58] which could have additive or even synergistic effects with the inflammatory response induced by surgical incision and/or formalin administration. Finally, it should be emphasized that in both studies the administration of noxious stimuli has occurred simultaneously with anesthesia.

This paradigm may not accurately reflect the reality of most clinical scenarios where inflammation and pain may precede anesthesia sometimes for a prolonged period of time. A more recent study looking at the effects of systemic inflammation caused by lipopolysaccharide injection 12h prior to onset of anesthesia reported a significantly worsened acute neuronal injury as well as long-term behavioral impairments in rats treated neonatally with sevoflurane [59]. The key difference in this study design is that the inflammatory mediators induced by lipopolysaccharide have been given time to exert a priming effect on the brain for an extended period of time prior to anesthesia which appeared to result in significantly worsened outcomes compared to either treatment alone with long-term, sex-specific behavioral deficits [59].

There is much uncertainty left surrounding the relative contribution of the multitude of perioperative factors such as inflammation and pain which could substantially modulate the effects of anesthesia on the young brain. They can either compound the injury further or provide some relief by providing additional stimulus to the dormant neurons. These factors are invariably present in nearly all clinical scenarios, where they may not be easily accounted for and, importantly, are seldom studied in well-controlled laboratory settings. It is our belief that attempting to include more variables in preclinical studies could help bridge the gap between preclinical and clinical studies and improve clinical relevance.

Prospective Evaluation of Early-Life Anesthesia Sequelae – MASK, PANDA and GAS clinical studies

The relevance of extensive animal evidence seems to be challenged by mixed and often inconclusive findings reported in clinical studies of anesthetic neurotoxicity in children. Due to the predominantly retrospective or ambivalent nature of these reports, several prospective clinical studies were launched and since have been completed with the intention of prospective assessment of neurocognitive and socio-affective development of children exposed to anesthesia prior to the age of 3.

Two Mayo Anesthesia Safety in Kids (MASK) prospective studies [60, 61] published in 2019 were guided by the report of cognitive deficits following ketamine administration in non-human primates [62] as evidenced by the Operant Test Battery (OTB) testing. The authors hypothesized that similar deficits might also be present in children exposed to anesthesia. Despite the efforts made by the authors to compare the human data to those of non-human primates, no differenceswere noted between unexposed, singly or multiply exposed children in any of the tested parameters [60]. Given the lack of association of IQ testing performance and general anesthesia in the primary report, these results may not come as a surprise since the OTB has been shown to correlate with the IQ [63]. On the other hand, the second follow-up aimed to further analyze the processing speed and fine motor skills with regard to early-life anesthesia exposure brought to our attention some interesting observations [61]. By applying the factor (grouping of tests based on the degree of correlation) and cluster (grouping of children based on the test scores) analyses, it was found that: 1) multiply-exposed children had performed significantly worse in the factor comprised of processing speed and motor ability tests, and 2) a disproportionately higher fraction of multiply-exposed children was found in the cluster with the lowest performance scores on most tests.

The Pediatric Anesthesia Neurodevelopment Assessment (PANDA) is another ambidirectional cohort study which investigated the long-term cognitive abilities of children when tested at 8-15 years of age who were exposed to anesthesia for surgical procedures prior to the age of three [64]. By employing sibling and half-sibling matching, the study concluded that anesthesia-exposed children did not differ significantly in IQ scores. Similar to the MASK study, the differences in the secondary outcomes of the study were noted particularly in domains of verbal fluency, externalizing, internalizing and adaptive behavior which was only partially accounted for after adjusting for sex and in same-sex sibling pairs [64].

Another prospective study, General Anesthesia Spinal (GAS) study [65], was a large multicentric randomized controlled trial which included children prior to the age of 3 undergoing inguinal herniorrhaphy who were randomized to either general anesthesia or awake-regional anesthesia. At 5-year follow up, the data analysis provided evidence of equivalence for the vast majority of outcomes including the IQ.

Several conclusions could be drawn about these studies. It would appear that early-life anesthesia, regardless of duration, had no association with gross changes in cognition as evidenced by the IQ scores later in life. However, these studies do raise important questions about the secondary outcomes such as motor skills and processing speed. Furthermore, it should be stated that all three studies provided parents or primary caregivers with the questionnaire such as Child Behavior Checklist (CBCL) and Behavior Rating Inventory of Executive Function preschool version (BRIEF-P). In all three cases, the parents reported worse scores in behavior, socio-affective and executive functions in anesthesia-exposed children.

While these prospective reports may challenge some of the findings reported in the earlier retrospective studies, a careful consideration should be made with regard to the study design. Namely, the mean anesthesia exposure times for single exposures were short - 45 min, 84 min and 54 min for MASK, PANDA and GAS studies, respectively. Although the duration of exposure to anesthesia reported in these studies may be relevant in patients with American Society of Anesthesiologists physical status (ASA-PS) scores 1 and 2; there is a small, yet significant percentage of children requiring much longer exposure times, up to and exceeding three hours [66, 67]. Furthermore, longer duration of exposures are often associated with higher ASA-PS scores [67] with significant comorbidities which can potentially influence the neurological injury as shown in animal studies [56, 59], yet these high-risk patients were not included in any of the three prospective reports. Importantly, the reports of significant deficiencies following multiple exposures in secondary outcomes such as fine motor tasks and speed of processing which require both sensory and motor outputs, raise further questions about the phenotype following early-life anesthesia that may also include socio-affective disorders such as ADHD as discussed below [68–71]. Finally, an overrepresentation of males in anesthesia-exposed groups [31, 64, 65], which is common in clinical studies of anesthetic neurotoxicity focused on short exposures and heathy children [38], makes it difficult, if not impossible, to reliably assess sex-differences associated with early-life anesthesia.

In summary, relatively short exposure times associated with minor risk surgeries such as inguinal herniorrhaphy together with the unequal sex representation and observed deficits in secondary, rather than primary outcomes of prospective clinical trials [31, 64, 65], may not be sufficient to extinguish concerns related to the potential harm caused by early-life general anesthesia.

Different Susceptibility of Analyzed Outcomes to Early-Life Anesthesia – A Directional Dilemma

Following the analysis of a number of retrospective [30, 32, 33, 35–38, 49, 50] and prospective [31, 64, 65] clinical trials, it appears that there is a dilemma in the field as to how to proceed although it is becoming clear that ignoring it would not be prudent. Multiple clinical studies have focused primarily on cognitive deficits, commonly assessed with IQ testing; however the probability of anesthesia-induced deficits in these domains seems small [31, 64, 65]. A recently published meta-analysis systematically reviewed reports of neurodevelopmental outcomes after a single or multiple anesthesia exposures with the intention to stratify the affected functional domains according to their vulnerability to anesthesia [72]. The findings strongly suggest that cognition was the least affected functional domain following either single or multiple exposures. Conversely, the highest degree of susceptibility was found in academics and in language domains especially after multiple exposures. Furthermore, a higher incidence of attention-deficit hyperactivity disorder (ADHD) was reported after any exposures with a 2-fold increase following multiple exposures [72]. These findings confirm the previous report of impaired executive function and higher risk of internalizing behavioral disorders even after a single exposure to anesthesia [73] which could be indicative of ADHD-like phenotype [74].

One report from the Mayo cohort found that multiple exposures before the age of 2 conferred significant risk of ADHD diagnosis later in life [68]. In this cohort, the prevalence of ADHD diagnoses at the age of 19 was 7.3%, 10.7% and 17.9% for unexposed, singly and multiply exposed children, respectively. These data suggest that children exposed to anesthesia more than once had greater than two-fold increase in the incidence of ADHD. The trend toward higher risk of ADHD was present even after a single exposure [68]. These findings were further confirmed in a nationwide study from Taiwan [69] showing a dose-response between cumulative anesthesia and increased risk of ADHD. The results of this study suggested that a cumulative anesthesia duration of >3h posed substantial risk to ADHD development. While it appears that there is strong evidence to suggest that multiple exposures, but not single, are associated with higher risk of ADHD [70], one retrospective study reported that the risk is increased even after a single minor exposure prior to the age of five [71]. Interestingly, it appears that the age at exposure might be of lesser importance as evidenced in this study; therefore, postponing the surgery might only be of limited value [71].

On the contrary, a retrospective study from Taiwan published in 2014 reported that the risk for ADHD diagnosis was not increased in children who were exposed to anesthesia prior to the age of three [75]. This study represents a nationwide assessment of insurance claims from a large number of randomly sampled children born between 2001 and 2005. The main advantage of this study cohort was a more recent anesthesia exposure, meaning that the anesthesia monitoring and the choice of anesthetics were more likely to reflect the current anesthesiology practices compared to some of the earlier study cohorts [32, 50, 68]. After adjusting for place of residence, parental occupation, perinatal conditions and congenital anomalies, it was determined that the cumulative number of anesthesia exposures prior to age of three was not a significant risk factor for ADHD later in life. Despite these reassuring findings, it was nonetheless observed that the increased risk for ADHD development in children with first exposure between age two and three, but not before, persisted even after adjustments [75]. Although the overall conclusion of the study was that general anesthesia during early life might not play a significant role in the development of ADHD, the existence of specific window of vulnerability between ages two and three still leaves unanswered questions about the safety of general anesthetics.

Unlike ADHD, the association between anesthesia and autism spectrum disorder (ASD) is more controversial at the moment and will require further analysis in the near future especially in view of newly developing preclinical evidence. A recent matched analysis of children by sex, maternal age and gestational age exposed to anesthesia prior to their third birthday indicated that anesthesia exposure was significantly associated with ASD regardless of the length of exposure; however, upon including further covariates (birth weight and health status at birth) the statistical significance was lost [76]. This appears to be in agreement with previously published results from a nationwide study in Taiwan [77] which observed no difference between anesthesia-exposed and unexposed children and the prevalence of ASD. It should however be noted that these studies pooled children from multiple types of surgical procedures; hence, the standardization of anesthesia protocol was not possible. Furthermore, the latter study [77] did not exclude children who had undergone anesthesia after the age of 2 in both groups which could have substantially underestimated the contribution of anesthesia as a risk factor for ASD.

General anesthesia is often employed in cesarean section emergencies in order to facilitate and expedite the delivery [78]. Despite the duration of anesthesia being very brief, poor neonatal outcomes such as lower Apgar scores, lower umbilical vein pH and higher admission rate to the neonatal intensive care unit were nonetheless noted in previous reports [79–81]. One report raised concerns about the higher risk of ASD in children born to cesarean section under general anesthesia [82]. This study indicated that the risk was even greater if general anesthesia and regional anesthesia were used concomitantly. Of note, the risk of developing ASD, particularly the severe form, was significantly higher in females. Although this study was later criticized on the basis of indication bias, non-reassuring fetal monitoring and variable length of gestation [83], it nonetheless provides potential consideration that anesthesia may be a risk factor for ASD if administered very early in life.

In a recently published study performed on over 1.5 million children [84], it was reported that, despite careful adjustment of the propensity scores, general anesthesia prior to the age of five was associated with almost a two-fold higher risk for developing ASD with the risk being greater with younger age at the time of exposure. To account for potential confounders such as diagnostic MRI for autism, numerous exclusion criteria encompassing a variety of neurological and behavioral conditions were included. Although early detection bias may have played a role in these findings, a study of such magnitude raises renewed concerns that there is a plausible association between early-life anesthesia and increased risk of ASD.

Concluding Remarks

Despite the inconsistencies and often contradictory findings reported in clinical studies thus far, the data nevertheless collectively seems to suggest that early-life anesthesia could be a significant risk factor for the development of concerning behavioral and socio-affective impairments (e.g., ADHD, ASD, the executive function, language domains, processing speed, motor coordination, disturbances in internalizing/externalizing domains, etc.). Importantly, their findings seem to correlate well with currently available pre-clinical studies, in particular, the non-human primate studies [5, 62, 85] that are crucial when it comes to creating tightly controlled experimental design in higher mammalian species close to a human infant.

Even after two decades of intense research, we are left with perhaps more questions than answers. Important differences in intraoperative monitoring standards and anesthesiology practices, including but not limited to decreased use of halothane in favor of novel anesthetics, makes comparison between the findings of newer randomized controlled trials and earlier retrospective reports somewhat difficult. Although findings from the recent randomized controlled trials [31, 64, 65] are reassuring, the absence of important risk factors in these reports, namely longer duration of exposures and pre-existing inflammation associated with the underlying disease, present outstanding issues which need to be addressed. In addition, since male sex is typically overrepresented in studies of anesthetic neurotoxicity, future studies with the emphasis of balanced distribution of sexes are needed in order to ascertain potential sex-specific differences in developmental outcomes.

To date, no viable and truly clinically-relevant neuroprotective strategy has been shown to reliably protect against the developmental neurotoxicity caused by currently used general anesthetics [86]. This would suggest a serious need for the development of a new line of anesthetics; an effort that has been stagnant for over 25 years. Some of the recent studies suggest that the development of safer general anesthetics [87] is a possibility that we believe should be embraced since general anesthesia is often a necessity that cannot be avoided when a child’s wellbeing is in danger. We owe it to our society to put forth our best effort in providing the best care to our youngest members.

Practice Points

• Rapidly accumulating animal evidence shows that early-life exposure to general anesthetics triggers widespread neuronal apoptosis with perstisting behavioral deficits.

• The period of peak vulnerability to neurotoxic effects of general anesthetics coincides with the period of intense synaptogenesis, which in humans begins in utero and continues during the first couple of postnatal life.

• Due to the ethical and practical considerations, the long-term consequences of early-life anesthesia in humans remains an open question.

• Multiple exposures during early life, with severe underlying pathology, may increase the likelihood of anesthesia-induced developmental disorders.

• Early-life anesthesia is unlikely to be associated with gross cognitive deficits later in life.

• Early-life anesthesia may be a significant risk factor for the development of behavioral and socio-affective impairments (internalizing/externalizing behavior, executive function, verbal fluency and processing speed).

Research Agenda

• Timing of anesthesia, cumulative anesthetic dose and number of exposures, impact of the underlying disease and analyzed outcomes are the key factors to be considered in clinical reports of anesthesia-induced neurotoxicity.

• Inclusion of higher-risk patients with longer exposure times, together with equal sex representation, are required in future prospective studies.

• The putative link between attention-deficit hyperactivity disorder and autism-spectrum disorder warrants further exploration.

• Effective neuroprotective strategies and novel anesthetics with improved safety profile are urgently needed due to the potentially serious repercussions on pediatric health.