Unanswered questions of anesthesia neurotoxicity in the developing brain

Abstract

Purpose of review

This article reviews recent advances and controversies of developmental anesthesia neurotoxicity research with a special focus on the unanswered questions in the field both from clinical and preclinical perspectives.

Recent findings

Observational cohort studies of prenatal and early childhood exposure to anesthesia have reported mixed evidence of an association with impaired neurodevelopment. Meta-analyses of currently available studies of early childhood exposure to anesthesia suggest that, while limited to no change in general intelligence can be detected, more subtle deficits in specific neurodevelopmental domains including behavior and executive function may be seen. Several studies have evaluated intraoperative blood pressure values and neurocognitive outcomes and have not found an association. Although many animal studies have been performed, taking into consideration other peri-operative exposures such as pain and inflammation may help with translation of results from animal models to humans.

Summary

Advances have been made in the field of developmental anesthetic neurotoxicity over the past few years, including the recognition that anesthetic exposure is associated with deficits in certain cognitive domains but not others. Although the most important question of whether anesthetic agents actually cause long-term neurodevelopmental effects in children has still not been answered, results from recent studies will guide further studies necessary to inform clinical decision-making in children.

INTRODUCTION

The quest to understand if and to what extent early-life anesthesia exposure impacts brain development occupies an important and controversial place in the research agenda of pediatric perioperative care. The biological plausibility of developmental anesthesia neurotoxicity is supported by extensive preclinical evidence. Results from clinical studies, however, do not align with existing preclinical data. One plausible explanation to this translational discrepancy stems from the fundamentally different experimental designs and outcome measures applied in laboratory versus clinical investigations. Indeed, while most of the preclinical work specifically focuses on the ‘drug effect’, human studies address the impact of the perioperative period, including not only drug exposure but also surgery and related changes in systemic homeostasis, on brain development. These major conceptual differences raise several important and yet incompletely answered questions that should be further explored if we want to advance our understanding on how early-life anesthesia and surgery (or the perioperative period as a global entity) may influence the immature brain. By focusing on both clinical and preclinical science, the goal of this review is to provide insights into some of these unanswered questions. 

Box 1.

no caption available

DOES ANESTHESIA CAUSE NEUROTOXICITY IN HUMANS?

The most important unanswered clinical question is whether anesthetic agents cause long-term neurodevelopmental effects in children. There are two main types of studies, which have been performed to try to address this question. The first is randomized controlled trials (RCT) in which children are exposed to general anesthesia vs. regional anesthesia, which by their design isolate the effect of a general anesthetic exposure. The second is observational studies, which primarily compare children who have had anesthesia and surgery with those who have not. The challenge with observational studies is that they cannot distinguish the effects of the anesthesia from the effects of the surgery. Also, there is likely to be some level of unmeasured confounding, most importantly, confounding by indication, or the concept that children who need surgery have a higher likelihood of baseline problems that predispose them to an increased risk of neurodevelopmental deficits.

In trying to determine causality, RCTs remain the gold standard. However, because of the difficulties in performing RCTs to evaluate neurotoxicity in children, only one large-scale trial has been performed [1,2]. A major challenge in performing RCTs and other prospective studies on this topic is the latency between anesthetic exposure and outcome assessment. It is thought that young children, possibly under age 3 years, are most at risk for anesthetic exposure [3]. Given that functional neurodevelopmental outcome assessment is challenging in young children, accurate assessment may need to occur in children who are of school age or greater, resulting in this long latency period and the need to follow children for several years. If objective intermediate outcomes could be evaluated shortly after anesthesia exposure, they would be valuable in facilitating clinical studies. The use of biomarkers as intermediate outcomes has been suggested, but serum biomarkers commonly lack specificity to the nervous system [4], and difficulty obtaining CSF in children preclude the use of CSF biomarkers. The use of imaging studies as a biomarker has, therefore, become an appealing alternative. One recent study evaluated MRI assessments as well as neuropsychological outcomes in a cohort of 102 children, of which 24 had been exposed to anesthesia [5^▪]. The anesthetic exposed children were found to have lower levels of gray matter volume in the right interior frontal gyrus as well as worse emotional control scores. Similar volumetric gray matter reductions have also been reported in a separate study [6], but as both are observational studies, limitations because of confounding are present. These types of studies, however, may help identify MRI findings as a viable intermediate outcome to be used in future prospective studies and clinical trials. It should be noted that in order to use MRI findings as intermediate outcomes, the relationship between these findings and functional neuropsychological outcomes also needs to be firmly established, which has not yet been done.

One of the main critiques of observational studies evaluating neurotoxicity is the presence of confounding by indication. One way to attenuate this type of confounding is by evaluating prenatal exposures in children, where anesthetic exposure is due to maternal need for surgery during pregnancy and not any underlying medical issue in the child. As a result, prenatally exposed children should not have more underlying medical issues that put them at an elevated risk of a neurodevelopmental deficit. Two recent studies have evaluated prenatal exposures, the first evaluated 22 prenatally exposed children, finding more externalizing behavioral problems compared with unexposed children [7^▪]. Bleeser et al.[8^▪] prospectively surveyed 129 mothers with general or regional anesthesia during pregnancy, and found that the 111 children exposed to general anesthesia had BRIEF executive function scores that were significantly worse than the unexposed children. CBCL Total behavioral problem scores were evaluated as a secondary outcome, but were not found to be significantly worse. These studies are novel because they are able to address the critique that all differences are due to confounding by indication in children. Although these differences may still be caused by underlying medical issues in the mother or inflammation or physiological disturbances during the maternal surgery, that the reported differences in behavior and executive function are consistent with findings from a meta-analysis of the GAS, MASK, and PANDA studies [9^▪▪], may suggest similarities between prenatal and postnatal exposures.

One of the possible explanations for differences in neurodevelopmental outcomes in children exposed to surgery and anesthesia may be intraoperative physiological changes. This has been explored recently in a study evaluating intraoperative blood pressures in children undergoing a range of ambulatory surgeries; however, in that study, intraoperative mean arterial pressure (MAP) was not associated with mental disorder diagnoses [10^▪]. A similar finding was reported in a secondary analysis of the MASK study [11], which also found no difference in learning deficit and attention-deficit hyperactivity disorder (ADHD) based on intraoperative blood pressure readings [12^▪]. Another study evaluated perioperative characteristics in children undergoing esophageal atresia repair and found that children with higher blood pressures had slightly better outcomes in motor function [13^▪]. Overall, these studies have limitations, and while the impact of intraoperative physiological factors cannot be ruled out, a robust association between intraoperative blood pressure and neurodevelopmental outcomes has not yet been reported.

WHAT ARE THE APPROPRIATE OUTCOMES TO EVALUATE?

One of the challenges with evaluating neurodevelopmental outcomes in children exposed to anesthesia is that if an effect exists, it is likely to be small and difficult to recognize on physical examination. As neurotoxicity was initially discovered in animals and not in humans, the phenotype of injury in humans remains unclear. As a result, over 100 studies have evaluated children who have been exposed to surgery and anesthesia, using over 400 different neuropsychological outcomes and scores [14^▪▪]. This heterogeneity has made it difficult to interpret the published studies and to determine which neuropsychological domains differ in children who have had surgery and anesthesia. An attempt to address this issue was made by classifying all reported outcomes from published studies into neurodevelopmental domains and performing a domain-specific meta-analysis. Although anesthetic-exposed children had worse scores in all neurodevelopmental domains in this meta-analysis, larger deficits were identified in domains such as executive function and behavior, with the smallest observed differences in cognition.

Two studies have recently studied autism in following anesthetic exposure, but the results are conflicting. In one, a birth cohort of children from Minnesota was evaluated with general anesthetic exposure not found to be significantly associated with autism [15^▪], whereas in the other, a population cohort of children born in Sweden were evaluated and an association was found [16^▪]. As more studies are published, a clearer understanding of which neurodevelopmental domains differ in anesthetic-exposed children will emerge. By distilling the information from the published studies, meta-analyses may help identify the appropriate neurodevelopmental outcomes to be evaluated in clinical studies of neurotoxicity.

WHICH SUBGROUPS OF CHILDREN ARE VULNERABLE?

It is hypothesized that younger children may be most vulnerable to anesthetic agents. Moser et al. evaluated 731 premature infants exposed to sedatives, anesthetic agents, and opioids between birth and 45 weeks of postconceptual age. The premature infants who required exposure to anesthetic agents had full scale intelligence quotient (FSIQ) scores that were lower than infants who were not exposed, but no difference in FSIQ was seen based on opioid exposure. The results from this study should be interpreted with caution because of a significant risk of confounding by indication, but the lack of an association between opioids and FSIQ is an interesting finding and is consistent with other studies [17,18].

Differences based on child sex have been hypothesized with findings from animal studies that males and females may have differential vulnerability to anesthetic agents [19]. This idea, however, has not been well evaluated in clinical studies. Epigenetic effects of anesthetics, have also been hypothesized based on animal studies. This idea that anesthetic agents could change gene expression and result in a transgenerational effect has been explored in a study identifying mothers who had surgery in early childhood or puberty, and evaluating their children [20^▪]. This study, however, did not show a consistent difference in cognition and educational outcomes in children of exposed mothers.

BEYOND DRUGS: WHAT IS THE CONTRIBUTION OF SURGERY AND OTHER PERIOPERATIVE FACTORS TO NEUROTOXICITY?

From a translational perspective, the major challenge in preclinical research is to develop experimental models addressing the overall impact of the perioperative period on neurodevelopmental outcome. Indeed, questions about whether and how perioperative factors, other than drug exposure-induced specific pharmacodynamic effects on neural circuitry, may influence central nervous system development are incompletely explored. For example, we understand very little about the effects of surgery and the related systemic inflammatory cascades on the immature brain or how administration of general anesthetics in conjunction with the surgical insults may affect brain development. Although some observations suggest that ketamine administration may have neuroprotective effects following inflammatory pain in newborn rats [21–23], other preclinical data demonstrates additive neurotoxic effects of general anesthesia and noxious stimuli [24]. Likewise, the limited amount of laboratory data on the interaction between systemic inflammation and exposure to general anesthesia in terms of neonatal brain toxicity also provides us with contradicting results [25,26^▪]. The lack of experimental models and data addressing the effects of neonatal surgery on brain development is primarily related to the fact that surgery in neonatal rodents, weighing just a few grams, is technically highly challenging if not impossible [27]. A related and translationally important issue is to understand the impact of temporary changes in systemic homeostasis during the perioperative period on brain development. Although some but not all human data suggest some associations between various factors such as low blood pressure, hypoglycemia/hyperglycemia, hypoxia/hyperoxia and impaired postoperative neurological outcome in the very young, mechanistic studies aimed to decipher potential causalities are lacking. Nor do we know how transient perturbations of systemic homeostasis may interact with anesthesia exposure and/or surgery to influence functional outcome. Unfortunately, very small laboratory rodents, where the majority of developmental anesthesia neurotoxicity research has been performed so far, may be unsuitable experimental models to fully study these questions. In fact, exposure of neonatal rodent pups to general anesthesia can rapidly trigger important respiratory and metabolic changes that are difficult to control and, thereby, are important confounding factors of the toxicity paradigm [28,29,30^▪]. Larger animals, where anesthesia and surgery can be performed under controlled conditions comparable to human clinical practice, are therefore necessary to study these questions. Despite important ethical and financial concerns [31,32], the possibility of developing such approaches should be considered to better explore the issue of developmental perioperative anesthesia neurotoxicity with translational relevance to clinical practice.

ARE THERE NONTOXIC GENERAL ANESTHETICS?

In light of preclinical animal data, suggesting that most currently used anesthetics may exert neurotoxicity, the three lingering questions are whether the extent of toxicity differs between anesthetics; if there are anesthetics without neurotoxic potential; and if we need to develop new drugs. There are no easy answers to these questions. Although some laboratory studies addressed the comparative neurotoxic potential of general anesthetics such as isoflurane, sevoflurane or propofol [33–35], results stemming from these studies should be interpreted by caution as equipotency between different drugs were not thoroughly evaluated in these investigations [36]. At present, there is no solid preclinical evidence indicating that any of the commonly general anesthetics would have higher neurotoxic potential than the others. Unlike most other anesthetics, opioids and the α2 receptor agonist dexmedetomidine do not principally operate via the modulation of GABAergic or glutamatergic signaling pathways and, thereby, have been hypothesized to lack neurotoxic or even exert neuroprotective effects [37,38]. However, preclinical data provides us with mixed evidence on these assumptions pending on the experimental model or even on the laboratory in which investigations were performed [39,40]. Xenon, while not commonly used in anesthetic practice, has also been hypothesized as lacking neurotoxic properties, but also has mixed results in preclinical data [41]. The inability of preclinical investigations to provide us with a drug devoid of neurotoxic potential in the young leads us to the question if we need to develop new drugs. There are currently several anesthetic drugs in the making but many of them have not been tested for neurotoxicity potential in preclinical models [42^▪]. In this regard, the major question is: what criteria should be fulfilled to declare drug safety as far as neurotoxicity is concerned? [43] Given the aforementioned translational limitations of existing preclinical models, the challenge of identifying a well defined and reproducible phenotype of human developmental anesthesia neurotoxicity, and the difficulty with performing randomized controlled trials, the definitive studies necessary to answer this question remain unclear.

CONCLUSION

Several advances have been made in recent years in the field of developmental anesthesia neurotoxicity research. The recognition that anesthetic exposure is associated with deficits in certain cognitive domains but not others will help guide the choice of outcomes in future studies. If an intermediate outcome can be identified, it will help address one of the major challenges in performing prospective studies. Translational relevance in currently available animal models is limited by challenges in maintaining systemic homeostasis and the lack of consideration of systemic inflammation. As a result, future work is needed to elucidate how general anesthetics influence neurodevelopment in the presence of surgery. Although it remains unclear which children are most vulnerable, and whether certain anesthetic regimens are safer than others, studies are underway to answer these questions. One such study is the TREX study, which is an RCT evaluating the use of dexmedetomidine vs. sevoflurane in children [44]. The most important question, however, is whether anesthetic agents cause neurotoxicity in children. While potential pathways to address this question have been described both from the preclinical and the clinical perspectives [45^▪▪], further research is required to find the answer.

Acknowledgements

None.

Financial support and sponsorship

None.

Conflicts of interest

There are no conflicts of interest.

REFERENCES AND RECOMMENDED READING

Papers of particular interest, published within the annual period of review, have been highlighted as:

• ▪ of special interest

• ▪▪ of outstanding interest