Consequences of General Anesthesia in Infancy on Behavior and Brain Structure : Anesthesia & Analgesia

Secondary Logo

Journal Logo

Featured Articles: Original Clinical Research Report

Consequences of General Anesthesia in Infancy on Behavior and Brain Structure

Salaün, Jean-Philippe MD, PhD*,†; Chagnot, Audrey PhD*; Cachia, Arnaud PhD‡,§; Poirel, Nicolas PhD‡,§,∥; Datin-Dorrière, Valérie MD, PhD‡,∥,¶; Dujarrier, Cléo MSc*; Lemarchand, Eloïse PhD*; Rolland, Marine MD*,†; Delalande, Lisa PhD; Gressens, Pierre MD, PhD#; Guillois, Bernard MD§; Houdé, Olivier PhD‡,§,∥; Levard, Damien MSc*; Gakuba, Clément MD, PhD*,†; Moyon, Marine PhD; Naveau, Mikael PhD**; Orliac, François MD, PhD‡,∥; Orliaguet, Gilles MD, PhD††; Hanouz, Jean-Luc MD, PhD†,‡‡; Agin, Véronique PhD*; Borst, Grégoire PhD‡,§; Vivien, Denis PhD*,§§

Author Information
Anesthesia & Analgesia 136(2):p 240-250, February 2023. | DOI: 10.1213/ANE.0000000000006233

Abstract

KEY POINTS

  • Question: Does early exposure to general anesthesia impact long-term behavior and brain structure in mice and humans?
  • Findings: This is the first translational study that suggests lasting effects of early life exposure to anesthetics on later emotional control behaviors and brain structures, both in rodents and humans.
  • Meaning: These considerations should stimulate the development of less brain-toxic anesthetic agents and to avoid early exposure to general anesthesia whenever possible.

Increasing preclinical studies suggest that general anesthesia (GA) has negative effects on the developing brain.1 Exposures to gamma-aminobutyric acid (GABA) agonists or N-methyl-D-aspartate (NMDA) antagonists resulted in structural and functional brain impairments, including behavioral changes.2 The issue of the clinical relevance of general anesthetics on brain toxicity in children was highlighted in the last decade, but gathered less consistent proof. Some cohort studies reported an association between GA and impaired neurodevelopment,3 while others do not.4 On one hand, preclinical studies are mandatory, even though species differences may preclude any clinical translation. On the other hand, it is practically difficult to design a randomized clinical study on the effects of GA on the developing brain. Human cohort analysis remains an interesting way to examine the question raised. Although the morphologies of rodent and human brains are different, neural circuits involved in cognitive and emotional functions are beginning to be better understood.5,6

We used a translational approach. We present the result of: (1) an experimental study focused on behavioral and anatomical changes that occur after postnatal exposure to GA in mice, and (2) the analysis of data recorded in an ongoing French cohort study of developing children (“APprentissages EXécutifs et cerveau chez les enfants d’âge scolaire” [APEX]). The authors had access to behavior and brain magnetic resonance imaging (MRI) of children 9 to 10 years of age who had been exposed or not exposed to GA for minor surgery during infancy.

The main objective of the study was to examine behavioral and structural brain changes in mice and children at a distance from the exposure of the developing brain to GA.

METHODS

The preclinical study was in accordance with the French ethical laws (Decree 2013-118, approval No. 8962) and the European Communities Council guidelines (2010/63/EU).

The clinical study was approved by a French national ethics committee (institutional review board 2015-A00383-46, May 21, 2015) and performed in line with the declaration of Helsinki (1964). Parents or legal guardians gave written consent, and children agreed to participate.

The main objective of the preclinical study was to examine behavioral and structural brain changes in mice at a distance from exposure to postnatal GA.

F1
Figure 1.:
Characteristics of the children exposed to anesthesia. A, Ages of the anesthetized children. B, Reasons for anesthesia.
Table 1. - Number and Age of Mice Used in the Study
Behavioral experiments Non-AG mice (6 wk) AG mice (6 wk)
Behavioral tests/regimen of exposure
 Body weight n = 30 n = 29
 Actimetry n = 22 n = 17
 Fear conditioning n = 22 n = 22
 Y-maze n = 24 n = 24
Imaging experiments
 Deformation-based morphometry MRI Non-AG mice (8 wk) AG mice (8 wk)
n = 10 n = 7
Abbreviations: AG, anesthesia-exposed mice; MRI, magnetic resonance imaging; Non-AG, control mice.

The main objective of the APEX data analysis was to examine long-term behavioral and structural brain imaging according to the exposure or nonexposure to GA in infancy. The secondary objectives of the APEX data analysis were to examine the relationship between: (1) behavioral changes and structural brain abnormalities, and (2) the age at exposure to anesthesia and the magnitude of behavioral and structural brain changes.

Preclinical Investigations

Animals

Experiments were performed on 6- to 8-week Swiss male mice (Janvier Labs, Le Genest-Saint-Isle, France). The animal care facilities were the Centre Universitaire de Ressources Biologiques (CURB; Caen Normandy University, France; approval No. A14118015).

Body weight was measured before behavioral evaluations. All behavioral tests were performed between 8 am and 5 pm in a room with dim illumination (6 l×) by an investigator blinded to the allocation group.

Mice were randomly allocated to 1 of these 2 groups:

  • 1. The GA-exposed group: mice were exposed to 1.3% isoflurane (corresponding to 1 minimum alveolar concentration in mice) in 100% O2 during 90 minutes each day from 4 to 10 days of life).
  • 2. The control group: mice were exposed to 100% O2, during 90 minutes each day from day 4 to day 10.

The exposition occurred in a plastic anesthesia chamber of 19 × 19 × 11.5 cm with a fresh gas inlet (3 L·min−1) on its lid, an aspiration on one of its sides, and a humidification system. A warm pad was placed in the anesthesia chamber to maintain the rectal temperature of the neonatal mice to 37°C during the procedure. A gas analyzer was applied to monitor gas concentrations (Capnomac Ultima, Datex Engstrom). Hypnosis for mice placed under GA was established by the control of loss of righting reflex, absence of limb withdrawal when applying pressure to the paw, and monitoring of heart and respiratory rates. Monitoring of heart and respiratory rates (Small Animal Monitoring and Gating System from SA Instruments Inc) was performed during each period of exposure to anesthetics and during each MRI procedure. Awakening of the anesthetized mice after purging volatile anesthetic from the ventilation system was assessed by paraclinical (plethysmography) and clinical observation (responses to tactile stimulations). Finally, these pups were smeared with their own bedding and sent back to their dams. At 6 to 8 weeks old, behavioral experiments and brain imaging were performed (Table 1).

Behavioral Assessment

The following behavioral tests were used for motor, emotional, and cognitive function assessment: actimetry,7 fear conditioning,8 and Y-maze.9

Actimetry.

Spontaneous locomotor activity was quantified by using a rack of 8 activity cages equipped with horizontal infrared beams located across the long axis of the cage (Imetronic). Mice were placed in individual acrylic chambers (30 × 20 × 20 cm) for 60 minutes. The number of horizontal and vertical movements was determined by breaks in movement-sensitive photobeams that were then converted into locomotor activity counts. This test evaluates the hypoactivity or hyperactivity.

Contextual Fear Conditioning.

Contextual fear memory was investigated in a conditioning chamber with white walls (67 × 53 × 55 cm, BIOSEB). A signal generated by mice movement was recorded and analyzed through a high-sensitivity weight transducer system. The analogical signal was transmitted to the Freezing software module through the load cell unit for recording purposes and posterior analysis in terms of activity/immobility (Freezing). On training day, animals were first individually placed into the conditioning chamber during a 4-minute acclimatizing session. During the subsequent training (acquisition) session, 2 shocks (0.4 mA, 2-s duration, 2-min intertrial interval) were administered in the mice’s feet. Thirty seconds after the final shock, the mice were returned to their home cages. Twenty-four hours or 1 month after the training session, contextual fear memory was assessed by returning the mice to the conditioning chamber and measuring freezing behavior during a 4-minute recall test. The measurement of fear was performed by considering the freezing time, defined as immobility (ie, the absence of all movements with the exception of those related to respiration) for at least a period of 500 µs.

Y-maze.

Mice were submitted to a 2-trial place recognition task in a Y-maze (34 × 8 × 15 cm; BMP Chaudronnerie) based on a novelty free-choice exploration paradigm previously developed to study spatial cognition processes in rodents. Extramaze distal cues were suspended on the black curtains of the room walls at a distance of 123 cm from the maze (from the top of the cues to the center of the maze). During the first trial (acquisition session), 1 of the 3 arms of the maze was closed. Each mouse was then placed at the end of the starting arm (arm 1) and was allowed to visit the accessible portions of the maze (arm 1/arm 2) for 5 minutes. The mouse was then returned to its cage for 3 hours, before being tested for a second trial (recall session), in which they had free access to all 3 arms (arm 1/arm 2/new arm) for 5 minutes. Spatial performance was assessed through the comparison of the number of visits in each arm. It has been shown that control mice spend more time in the new arm than in the familiar arms (arm 1/arm 2), therefore indicating processing of remote spatial cues.

Brain Imaging

We used 7T preclinical MRI (Bruker, Pharmascan, Advance III) to acquire a 3D isotropic T2-weighted image of the brain of each mouse (repetition time [TR] = 13.5 s, echo time [TE] = 48 ms, RARE (Rapid Acquisition With Refocused Echoes) acceleration factor = 8, matrix = 90 × 90 × 90, and isotropic resolution = 0.2 mm3). The acquisitions were realized for mice anesthetized with isoflurane (1% to 2% in 100% O2) over 20 minutes. Deformation-based morphometry (DBM),10 and not voxel-based morphometry (VBM), was used to analyze mouse brain anatomy, as DBM overcomes the VBM brain segmentation step, which is tedious with mice.

The APEX Cohort

Participants

The data for the clinical study were derived from the APEX study (2015–2019), which aimed at testing in healthy school-aged children (9–10 years of age) the behavioral and neural changes induced by an inhibitory control training or a metacognitive mindfulness meditation practice, compared to an active control training (details of APEX recruitment can be found in a previous report).11 Only baseline data, namely before intervention, were included in the present study. We included all participants (n = 102) of the APEX cohort, who were right-handed as determined by the Edinburgh Handedness Inventory,12 born at full term, with normal or corrected-to-normal vision, and no history of neurological diseases or cerebral abnormalities. Thus, 24 children who were exposed to GA once for nonurgent minor surgery between 1 and 10 years of age (Figure 1) constituted the GA-exposed group. Exposure to GA was evaluated by a standardized data form completed by the children’s parents. We did not have access to the drugs and dosage used during GA. The 78 children who had not been exposed to GA constituted the control group.

Behavioral Assessment

The following behavioral tests were used to assess working memory ability, nonverbal intelligence, and emotional control.

Working Memory.

Children’s working memory abilities were assessed thanks to the Backward Digit Span task.6 In this task, the children listened to series of discrete digits and were asked to recall the series of digits in the reverse order of presentation. The children first performed 2 series of 2 digits. The series of digits were incrementally increased by 1 digit every 2 trials. The task was stopped when the children failed to recall 2 trials with the same number of digits. The working memory score was defined as the number of series correctly recalled. Working memory was also investigated thanks to the Behavior Rating Inventory of Executive Function (BRIEF) parent form13 to investigate daily working memory abilities.

Nonverbal Intelligence.

We used Raven’s progressive matrices14 to assess nonverbal intelligence. This test is a nonverbal task typically used to measure abstract reasoning and estimate fluid intelligence. In each test item, children are asked to identify the missing element that completes a pattern.

Emotional Control.

Emotional control was assessed using the standard BRIEF parent form.13

Brain Imaging

The APEX study included brain MRI in children 9 to 10 years of age. Before the scans, children were familiarized with the MRI session noise in a fake MRI and were trained not to move during the acquisition. T1 MRI acquisitions were performed while they passively watched a mute cartoon on an MRI-compatible screen to reduce motion, provide a positive experience, and decrease waiting times.15 We used the APEX cohort high-resolution isotropic 3T sagittal structural T1 MRI data (TE = 3.3; TR = 7.2; and flip angle = 9; 1 × 1 × 1 mm) acquired at the Cyceron biomedical imaging platform (Archieva, Philips Medical System). Local gray matter volumes were automatically assessed on the whole brain based on the standard VBM approach using the Statistical Parametric Mapping software (SPM12) (http://www.fil.ion.ucl.ac.uk/spm/) and the Computational Anatomy Toolbox (CAT12) (http://www.neuro.uni-jena.de/cat/), a fully automatic toolbox for performing robust and accurate VBM analyses.16 We used the CAT12 (CAT12.6) toolbox, which is optimized for the segmentation of brains of children (robust initial affine registration and sample-specific template based on a diffeomorphic image registration algorithm [DARTEL]).17

Statistical Analysis

Data were analyzed using GraphPad Prism software (GraphPad Software). All analyses were 2-tailed.

Preclinical Behavioral Data

Analysis of the data sets via parametric approaches turned out to be inappropriate in some cases due to violation of normality (Shapiro-Wilk tests). Therefore, parametric or nonparametric approaches were used whenever appropriate. Behavioral performance was compared between groups using unpaired Student t tests (actimetry, fear conditioning, and body weight). Behavioral performance was compared within groups using paired Student t tests (Y-maze: acquisition session). Friedman’s tests were used for multiple comparisons (Y-maze: recall session). When an appropriate significant effect was detected by Friedman’s tests, we performed post hoc analysis using Dunn’s multiple comparison tests.

Preclinical Imaging Data

Using advanced normalization tools (ANTs), an anatomical template of the control mice was built. Estimating the volumetric differences between the animals, the Jacobian determinant of the deformation field needed to warp the individuals to the template that was used. Using SPM12, a second level of analysis was performed of these Jacobian determinant images to test morphological differences between groups. This second-level analysis consists of voxelwise t tests corrected for the familywise error (FWE) rate.

Clinical Behavioral Data

Shapiro-Wilk analysis of the whole data set revealed that the data followed a Gaussian distribution, thus enabling the use of standard univariate analysis of covariance (ANCOVA) for each clinical behavioral data. The models included the covariables age of the child during testing and monthly parental income as a proxy of socioeconomic status.

Clinical Imaging Data

Voxelwise statistical analyses were performed on the whole brain using SPM12 software, within the general linear model (GLM) framework. Analyses were performed on the smoothed (8 mm) and modulated gray matter maps with sex, age, and parental income as a proxy of socioeconomic status, and weight at birth as matching variables. All analyses were performed using a voxelwise threshold at P < .05, corrected for multiple testing using SPM12 FWE method. We further investigated some possible associations of gray matter deficits with behavioral scores using small volume correction (SVC), with a 10 × 10 × 10 mm search volume centered at the VBM cluster max voxel.

Sample Size Justifications

We decided the number of animals per group for neuroanatomy and behavior tasks based on our previous published studies and ethical recommendations. Because the clinical study was a secondary analysis of the APEX cohort, it was not possible to calculate statistical power a priori. All statistical powers were calculated a posteriori using G × Power.

Preclinical Investigations. Post hoc power calculations were found equal at 0.59 for fear conditioning at 1-month recall session, 0.05 for actimetry horizontal activity, 0.05 for actimetry vertical activity, and 0.24 for Y-maze recall session. Based on our group sizes for GA-exposed mice and controls, we determined the effect size required for 1-β = 0.9 and α = 0.05 at d = 0.65 for fear conditioning at 1-month recall session, d = 0.04 for actimetry horizontal activity, d = 0.04 for actimetry vertical activity, and d = 0.37 for Y-maze recall session. Post hoc power was found equal at 0.99 for MRI cluster. Based on our group sizes for GA-exposed mice and controls, we determined the effect size required for 1-β = 0.9 and α = 0.05 at d = 3.86 for MRI cluster.

The APEX Cohort. Post hoc power calculations were found equal at 0.43 for BRIEF emotional control score, 0.13 for BRIEF working memory score, 0.05 for Raven’s progressive matrices, and 0.05 for Backward Digit Span. Based on our group sizes for GA-exposed children and controls, we determined the effect size required for 1-β = 0.9 and α = 0.05 at d = 0.42 for BRIEF emotional control score, d = 0.19 for BRIEF working memory score, d = 0.005 for Raven’s progressive matrices, and d = 0.03 for Backward Digit Span. Post hoc power was found equal at 0.99 for MRI cluster. Based on our group sizes for GA-exposed children and controls, we determined the effect size required for 1-β = 0.9 and α = 0.05 at d = 2.86 for MRI cluster.

RESULTS

Preclinical Investigations

No mortality was reported during experiments.

F2
Figure 2.:
Preclinical imaging. A, For illustration purpose and ease of visualization, representation of the cluster (uncorrected statistics), showing an area covering the left periaqueductal gray matter. B, Proportion of gray matter is lower in general anesthetized mice versus controls in the cluster. C–E, Orthogonal slices representing the position of the cluster (FWE-corrected). F–H, Orthogonal slices representing the position of the extended cluster (FWE-uncorrected). I–K, three-dimensional views of the extended cluster (FWE-uncorrected). FWE indicates familywise error.
F3
Figure 3.:
Preclinical behavioral testing. Actimetry: horizontal (A) and vertical (B) spontaneous locomotor activity. No differences in horizontal or vertical movements were observed between control and GA-subjected mice (22 control and 17 anesthesia). Boxplots show distributions‚ with black horizontal lines indicating the median and box margins denoting the lower and upper quartiles. Whiskers show the minimum and maximum values. Fear conditioning: acquisition (C) and recall (D) sessions. Freezing response at 30 d after acquisition is higher in GA-subjected mice (22 control and 22 anesthesia). Boxplots show distributions‚ with black horizontal lines indicating the median and box margins denoting the lower and upper quartiles. Whiskers show the minimum and maximum values. Y-maze: acquisition (E and F) and recall (G and H) sessions. Visit frequency in the new arm is higher than in familiar arms (1 and 2) at recall in both control and GA-subjected mice (24 control and 24 anesthesia). Boxplots show distributions‚ with black horizontal lines indicating the median and box margins denoting the lower and upper quartiles. Whiskers show the minimum and maximum values. GA indicates general anesthesia.
F4
Figure 4.:
Clinical imaging and behavioral testing. A, Trend to higher BRIEF emotional control score in children subjected to a single GA (P-corrected = .05; controlled for sex, age, parental income as a proxy of socioeconomic status, and weight at birth). Scores on the BRIEF working memory scale (B), RAVEN36 (C), and Backward Digit Span (D) did not differ between the groups. E and F, Lower local gray matter volume in children who were subjected to a single GA (P-FWE < .05; controlled for sex, age, parental income as a proxy of socioeconomic status, and weight at birth). G, Three-dimensional rendering of the cortical surface and VBM cluster in the posterior part of the right inferior frontal gyrus (BA9) (voxel maximum: Talairach’s x, y, z = [51; 3; 27]). For illustration purpose and ease of visualization, the cluster is represented with uncorrected statistics (P-uncorrected <.001). H, Gray matter volume in the VBM cluster located in the posterior part of the right inferior frontal gyrus is lower in children subjected to a single GA (P-FWE < 0.05; controlled for sex, age, parental income as a proxy of socioeconomic status, and weight at birth). GA indicates general anesthesia; VBM, voxel-based morphometry.
Table 2. - Human Population Characteristics
Anesthesia (n = 24) Control (n = 78) P value
Demography Age (y) 9.9 (0.6) 9.8 (.5) .74
Male/female ratio 9/15 35/43 .47
Birth weight (g) 3290 (377) 3369 (480) .41
Monthly parental income (€) 3375 (647) 3400 (593) .87
Age at anesthesia (y) 4.0 (2.3) NA /
Behavioral testing RAVEN36 31 (3) 31 (4) .95
Backward digit span 3.88 (1.12) 3.91 (1.02) .82
BRIEF working memory score 49 (7) 51 (10) .58
BRIEF emotional control score 58 (12) 53 (11) .06
Abbreviation: BRIEF, Behavior Rating Inventory of Executive Function

Control and GA-exposed mice showed similar body weight at 6 weeks of life (control mice: weight [g] = 39.5 ± 2.8; GA-exposed mice: weight = 39.7 ± 2.7).

MRI Deformation-Based Morphometry Analysis

There was an 11% (95% CI, 7.5–14.5) reduction in the volume of the periaqueductal gray matter in the GA-exposed group (Figure 2; peak level, t = 7.32, dof = 15; p-FWE = 0.046).

Actimetry

There was no significant difference between groups in horizontal locomotor activities (Figure 3A) and vertical activities (Figure 3B).

Contextual Fear Conditioning

There was no significant difference between groups in the percentage of freezing during the acclimatizing period (acquisition) session: all mice showed high levels of fear after the 2 aversive unconditioned stimuli without differences between groups (Figure 3C). One month after the training session, there was a higher freezing time (ft) in the GA-exposed mice (ft = 115 ± 73 s) compared to the control group (ft = 73 ± 56 s; 95% CI, 4–80; P = .03; Figure 3D).

Y-Maze

During the acquisition session, there was no significant difference between groups in the total number of visits in the 2 free access arms of the Y-maze (Figure 3E, F). During the recall session performed 3 hours later, the new arm of the maze was visited similarly in GA-exposed and control groups (Figure 3G, H).

The APEX Cohort

There was no difference between groups in main children’s characteristics (Table 2). In the GA-exposed children, mean age at anesthesia was 4.0 ± 2.3 years (29% of patients were ≤3 years of age, 54% were between 4 and 6 years, and 17% were >6 years of age).

BRIEF Emotional Control Score

Figure 4A suggests lower emotional control in the GA-exposed children (58 ± 12) compared to the control ones (53 ± 11; difference, 5; 95% CI of the difference, 0.33–9.10; P = .06).

BRIEF Working Memory Score

There was no significant difference between groups in the BRIEF working memory scores (Figure 4B).

Raven’s Progressive Matrices

There was no significant difference between groups in Raven’s progressive matrices (Figure 4C).

Backward Digit Span

There was no significant difference between groups in the Backward Digit Span performance (Figure 4D).

Multivariate analysis found no relationship between behavioral and memory scores and age at the time of anesthesia.

MRI Voxel-Based Morphometry Analysis

In the GA-exposed children, there was a 6.1% (95% CI, 4.3–7.8) lower gray matter volume in a single cluster corresponding to the posterior part of the right inferior frontal gyrus, BA9, “aal” atlas (MNI coordinates x, y, z = [51; 3; 27]; height threshold: Z = 4.85, p-FWE = 0.019; Figure 4E–H). There was no difference in local white matter volume between groups.

Furthermore, an SVC analysis centered at 51; 3; 27 revealed positive main effects of the BRIEF emotional control score (height threshold: Z = 2.65, p-FWE = 0.047) and of the age at GA (height threshold: Z = 2.94, p-FWE = 0.024) on the volume of gray matter in the posterior part of the right inferior frontal gyrus. Of note, the same analysis with the BRIEF working memory score did not yield any significant correlation.

DISCUSSION

Our preclinical study revealed exacerbated fear behavior associated with an atrophy of the periaqueductal gray matter in the adult mice exposed to postnatal GA. Our clinical study provides the first evidence that exposure to GA in childhood can lead to a gray matter atrophy in the right prefrontal gyrus. This atrophy is all the more pronounced if GA occurred early.

Our study revealed a periaqueductal gray matter atrophy in mice 6 to 8 weeks old after repeated postnatal anesthesia with 1 MAC of isoflurane (days 4–10). The periaqueductal gray matter is a key component in context fear discrimination in mice18 and is involved in anxiety and depression.19 These results are consistent with the literature.20 The lack of difference in locomotor activity and working memory between mice neonatally exposed to isoflurane is consistent with the literature. Repeated isoflurane anesthesia exposure of mice during development (days 7 to 9) does not impact locomotor activity.21 Moreover, performance in working memory was not impaired in adult mice exposed to 3 hours of 2.7% isoflurane (day 6).22 Human data obtained from the APEX cohort suggested that the GA-exposed children (age at anesthesia, 4.0 ± 2.3 years) presented lower gray matter volumes in the right inferior frontal gyrus, the more pronounced, the earlier the anesthesia. Reductions in the inferior frontal gyrus volume have been associated with impaired emotion functions and later depression.23,24 These neuroimaging results could be associated with impaired emotional control. Our results are consistent with a meta-analysis of the 3 main prospective clinical studies about anesthesia-induced brain toxicity, which showed an impaired BRIEF executive function global score (risk ratio, 1.68 [95% CI, 1.23–2.30]; P = .001) in 841 children exposed to GA.25

Strengths

These term-born children benefited from a single nonurgent minor surgery. Moreover, they had no history of neurological and cerebral abnormalities. Indeed, such an history could have influenced the interpretation of the results.26 The analyses were controlled for socioeconomic status. In our study, ear, nose, and throat surgeries (46.0%), digestive surgeries (20.8%), and urological surgeries (16.7%) were in decreasing order or frequency in the most represented surgical specialties. Similar patterns of the indications of surgery were reported in the literature.27 This clinical study provides the first evidence that exposure to GA in childhood (excluding the neonatal period) can lead to gray matter atrophy. It would be interesting to reevaluate the possible impact of GA on the parameters evaluated in the previous APEX-related studies such as the cortex anatomy11 and the baseline resting-state function, which may be intrinsically related to the local structure.28 Only one study had yet reported a gray matter decrease after GA in childhood.29 The correction for multiple comparisons was applied only to the voxels within the single region of interest and not for all the voxels of the brain as in our study. This statistical method could be associated with high false-positive rates. In our study, the gray matter atrophy persisted after FWE-correction for multiple tests on the whole brain.

The most likely mechanism to explain the atrophy detected after early GA exposure is the apoptosis of GABAergic and glutamatergic neurons.2 After exposing neural stem cells to 1 MAC of sevoflurane, cell viability decreased and cytotoxicity increased in a time-dependent way.30 This is all the more likely because periaqueductal gray matter contains GABAergic interneurons.31 Another possible mechanism is neuroinflammation.32 In monkeys, the interest of a noninvasive neuroimaging evaluation of glial activation after early sevoflurane exposition was pointed out by using positron emission tomography.33

Limitations and Perspectives

The first limitation is the lack of data about the drugs administered in children from the APEX study. It could be hypothesized that anesthesia involved sevoflurane, the most currently used anesthetics in pediatric anesthesia in France.34 Moreover, no perioperative anesthetic complications were reported in the health records. The second limitation is that the GA-exposed children were exposed at various ages. The third limitation is that mice underwent multiple exposures to approximate concentrations of isoflurane and that their mean arterial pressure during anesthesia could not be monitored for technical reasons. However, heart and respiratory rates were monitored, and there was no death in the GA-exposed mice. The fourth limitation is that if rodents are the most commonly used mammalian models for studying the neurotoxicity of anesthetics, their brain morphologies are dissimilar to humans.35 However, the affected regions by GA (1 in mice, and 2 in humans; periaqueductal gray matter and the right inferior frontal gyrus, respectively), although not identical, are both involved in the control of emotions. For instance, Rozeske’s study has identified a subpopulation of prefrontal-periaqueductal gray-projecting neurons that control switching between different emotional states during contextual fear discrimination in mice.18 Similarly, Blakemore’s study has highlighted a cortico-subcortical network involving right inferior frontal gyrus and periaqueductal gray matter also implicated in emotional control in humans.36 Moreover, rodent brain activity during daily life situations mainly rests upon fear processes, an adaptative surviving behavior that largely depends on the periaqueductal gray matter.37 Similarly, the human prefrontal part of the brain is a core region mainly necessary for daily life activity to accurately deal with emotional and behavioral regulation.23

Future preclinical and clinical studies should also investigate to what extent detrimental effects of GA could be driven by side effects occurring during GA, such as hypotension, hypocapnia, or hypoxia.38 The NECTARINE study recently highlighted difficulties to accurately monitor neonates and small infants.39 This is also difficult in newborn mice. Finally, intraoperative data are often missing in studies on anesthesia toxicity.13,40

CONCLUSIONS

We present the first translational study that analyzes the behavioral and structural brain impact of childhood exposure even to a single GA procedure, the result of a successful collaboration between preclinical and clinical researchers. It should stimulate the development of less brain-toxic anesthetic agents, better monitoring of anesthesia in children, and setting up long-term monitoring of our cohort to determine whether children are prone to develop psychiatric disorders.

ACKNOWLEDGMENTS

The authors thank the children who participated in the study and their families.

DISCLOSURES

Name: Jean-Philippe Salaün, MD, PhD.

Contribution: This author helped with conception and design of the study, acquisition and analysis of data, and drafting a significant portion of the manuscript or figures.

Name: Audrey Chagnot, PhD.

Contribution: This author helped with acquisition and analysis of data and drafting a significant portion of the manuscript or figures.

Name: Arnaud Cachia, PhD.

Contribution: This author helped with conception and design of the study and drafting a significant portion of the manuscript or figures.

Name: Nicolas Poirel, PhD.

Contribution: This author helped with conception and design of the study.

Name: Valérie Datin-Dorrière, MD, PhD.

Contribution: This author helped with acquisition and analysis of data.

Name: Cléo Dujarrier, MSc.

Contribution: This author helped with acquisition and analysis of data.

Name: Eloïse Lemarchand, PhD.

Contribution: This author helped with the revision process.

Name: Marine Rolland, MD.

Contribution: This author helped draft a significant portion of the manuscript or figures.

Name: Lisa Delalande, PhD.

Contribution: This author helped with acquisition and analysis of data.

Name: Pierre Gressens, MD, PhD.

Contribution: This author helped draft a significant portion of the manuscript or figures.

Name: Bernard Guillois, MD.

Contribution: This author helped with conception and design of the study.

Name: Olivier Houdé, PhD.

Contribution: This author helped with conception and design of the study.

Name: Damien Levard, MSc.

Contribution: This author helped with acquisition and analysis of data.

Name: Clément Gakuba, MD, PhD.

Contribution: This author helped draft a significant portion of the manuscript or figures.

Name: Marine Moyon, PhD.

Contribution: This author helped with acquisition and analysis of data.

Name: Mikael Naveau, PhD.

Contribution: This author helped with acquisition and analysis of data.

Name: François Orliac, MD, PhD.

Contribution: This author helped with acquisition and analysis of data.

Name: Gilles Orliaguet, MD, PhD.

Contribution: This author helped draft a significant portion of the manuscript or figures.

Name: Jean-Luc Hanouz, MD, PhD

Contribution: This author helped draft a significant portion of the manuscript or figures.

Name: Véronique Agin, PhD.

Contribution: This author helped with acquisition and analysis of data, and drafting a significant portion of the manuscript or figures.

Name: Grégoire Borst, PhD.

Contribution: This author helped with conception and design of the study.

Name: Denis Vivien, PhD.

Contribution: This author helped with conception and design of the study.

This manuscript was handled by: James A. DiNardo, MD, FAAP.

GLOSSARY

ANCOVA
analysis of covariance
ANT
advanced normalization tool
APEX
APprentissages EXécutifs et cerveau chez les enfants d’âge scolaire
BRIEF
Behavior Rating Inventory of Executive Function
CAT
Computational Anatomy Toolbox
CI
confidence interval
CURB
Centre Universitaire de Ressources Biologiques
DARTEL
Diffeomorphic Image Registration Algorithm
DBM
deformation-based morphometry
TE
echo time
FWE
familywise error
GA
general anesthesia
GABA
gamma-aminobutyric acid
GLM
general linear model
MRI
magnetic resonance imaging
NMDA
N-methyl-D-aspartate
RT
repetition time
SPM
Statistical Parametric Mapping
SVC
small volume correction
VBM
voxel-based morphometry

REFERENCES

1. Jevtovic-Todorovic V. Monkey business: the importance of mounting behavioural evidence for anaesthesia-induced developmental neurotoxicity. Br J Anaesth. 2018;120:617–619.
2. Jevtovic-Todorovic V. Exposure of developing brain to general anesthesia: what is the animal evidence? Anesthesiology. 2018;128:832–839.
3. Wilder RT, Flick RP, Sprung J, et al. Early exposure to anesthesia and learning disabilities in a population-based birth cohort. Anesthesiology. 2009;110:796–804.
4. McCann ME, de Graaff JC, Dorris L, et al.; GAS Consortium. Neurodevelopmental outcome at 5 years of age after general anaesthesia or awake-regional anaesthesia in infancy (GAS): an international, multicentre, randomised, controlled equivalence trial. Lancet. 2019;393:664–677.
5. Russell LE, Dalgleish HWP, Nutbrown R, et al. All-optical interrogation of neural circuits in behaving mice. Nat Protoc. 2022;17:1579–1620.
6. Poirel N, Borst G, Simon G, et al. Number conservation is related to children’s prefrontal inhibitory control: an fMRI study of a Piagetian task. PLoS One. 2012;7:e40802.
7. Mannara F, Radosevic M, Planagumà J, et al. Allosteric modulation of NMDA receptors prevents the antibody effects of patients with anti-NMDAR encephalitis. Brain. 2020;143:2709–2720.
8. Chatterjee S, Angelakos CC, Bahl E, et al. The CBP KIX domain regulates long-term memory and circadian activity. BMC Biol. 2020;18:155.
9. Dellu F, Contarino A, Simon H, et al. Genetic differences in response to novelty and spatial memory using a two-trial recognition task in mice. Neurobiol Learn Mem. 2000;73:31–48.
10. Chung MK, Worsley KJ, Paus T, et al. A unified statistical approach to deformation-based morphometry. Neuroimage. 2001;14:595–606.
11. Delalande L, Moyon M, Tissier C, et al. Complex and subtle structural changes in prefrontal cortex induced by inhibitory control training from childhood to adolescence. Dev Sci. 2020;23:e12898.
12. Oldfield RC. The assessment and analysis of handedness: the Edinburgh inventory. Neuropsychologia. 1971;9:97–113.
13. Baron IS. Behavior rating inventory of executive function. Child Neuropsychol. 2000;6:235–238.
14. Raven J, Raven JC, Court JH. Manual for Raven’s Progressive Matrices and Vocabulary Scales. Oxford Psychologists Press; 1998.
15. Lemaire C, Moran GR, Swan H. Impact of audio/visual systems on pediatric sedation in magnetic resonance imaging. J Magn Reson Imaging. 2009;30:649–655.
16. Ashburner J. Computational anatomy with the SPM software. Magn Reson Imaging. 2009;27:1163–1174.
17. Ashburner J. A fast diffeomorphic image registration algorithm. Neuroimage. 2007;38:95–113.
18. Rozeske RR, Jercog D, Karalis N, et al. Prefrontal-periaqueductal gray-projecting neurons mediate context fear discrimination. Neuron. 2018;97:898–910.e6.
19. Ho YC, Lin TB, Hsieh MC, et al. Periaqueductal gray glutamatergic transmission governs chronic stress-induced depression. Neuropsychopharmacology. 2018;43:302–312.
20. Satomoto M, Sun Z, Adachi YU, et al. Neonatal sevoflurane exposure induces adulthood fear-induced learning disability and decreases glutamatergic neurons in the basolateral amygdala. J Neurosurg Anesthesiol. 2018;30:59–64.
21. Rosenholm M, Paro E, Antila H, et al. Repeated brief isoflurane anesthesia during early postnatal development produces negligible changes on adult behavior in male mice. PLoS One. 2017;12:e0175258.
22. Kodama M, Satoh Y, Otsubo Y, et al. Neonatal desflurane exposure induces more robust neuroapoptosis than do isoflurane and sevoflurane and impairs working memory. Anesthesiology. 2011;115:979–991.
23. Zhang L, Li W, Wang L, et al. Altered functional connectivity of right inferior frontal gyrus subregions in bipolar disorder: a resting state fMRI study. J Affect Disord. 2020;272:58–65.
24. Luby JL, Barch D, Whalen D, et al. Association between early life adversity and risk for poor emotional and physical health in adolescence: a putative mechanistic neurodevelopmental pathway. JAMA Pediatr. 2017;171:1168–1175.
25. Ing C, Jackson WM, Zaccariello MJ, et al. Prospectively assessed neurodevelopmental outcomes in studies of anaesthetic neurotoxicity in children: a systematic review and meta-analysis. Br J Anaesth. 2021;126:433–444.
26. Andropoulos DB, Easley RB, Brady K, et al. Changing expectations for neurological outcomes after the neonatal arterial switch operation. Ann Thorac Surg. 2012;94:1250–1255.
27. Habre W, Disma N, Virag K, et al.; APRICOT Group of the European Society of Anaesthesiology Clinical Trial Network. Incidence of severe critical events in paediatric anaesthesia (APRICOT): a prospective multicentre observational study in 261 hospitals in Europe. Lancet Respir Med. 2017;5:412–425.
28. Salvia E, Tissier C, Charron S, et al. The local properties of bold signal fluctuations at rest monitor inhibitory control training in adolescents. Dev Cogn Neurosci. 2019;38:100664.
29. Backeljauw B, Holland SK, Altaye M, et al. Cognition and brain structure following early childhood surgery with anesthesia. Pediatrics. 2015;136:e1–12.
30. Qiu J, Shi P, Mao W, et al. Effect of apoptosis in neural stem cells treated with sevoflurane. BMC Anesthesiol. 2015;15:25.
31. Yin W, Mei L, Sun T, et al. A central amygdala-ventrolateral periaqueductal gray matter pathway for pain in a mouse model of depression-like behavior. Anesthesiology. 2020;132:1175–1196.
32. Baud O, Saint-Faust M. Neuroinflammation in the developing brain: risk factors, involvement of microglial cells, and implication for early anesthesia. Anesth Analg. 2019;128:718–725.
33. Zhang X, Liu S, Newport GD, et al. In vivo monitoring of sevoflurane-induced adverse effects in neonatal nonhuman primates using small-animal positron emission tomography. Anesthesiology. 2016;125:133–146.
34. Apai C, Shah R, Tran K, et al. Anesthesia and the developing brain: a review of sevoflurane-induced neurotoxicity in pediatric populations. Clin Ther. 2021;43:762–778.
35. Chinn GA, Pearn ML, Vutskits L, et al. Standards for preclinical research and publications in developmental anaesthetic neurotoxicity: expert opinion statement from the SmartTots preclinical working group. Br J Anaesth. 2020;124:585–593.
36. Blakemore RL, Rieger SW, Vuilleumier P. Negative emotions facilitate isometric force through activation of prefrontal cortex and periaqueductal gray. Neuroimage. 2016;124:627–640.
37. Deng H, Xiao X, Wang Z. Periaqueductal gray neuronal activities underlie different aspects of defensive behaviors. J Neurosci. 2016;36:7580–7588.
38. Weiss M, Vutskits L, Hansen TG, et al. Safe anesthesia for every tot—the SAFETOTS initiative. Curr Opin Anaesthesiol. 2015;28:302–307.
39. Disma N, Veyckemans F, Virag K, et al.; NECTARINE Group of the European Society of Anaesthesiology Clinical Trial Network; Austria; Belgium; Croatia; Czech Republic; Denmark; Estonia; Finland; France; Germany; Greece; Hungary; Ireland; Italy; Latvia; Lithuania; Luxembourg; Malta; Netherlands; Norway; Poland; Portugal; Romania; Serbia; Slovakia; Slovenia; Spain; Switzerland; Turkey; Ukraine; United Kingdom. Morbidity and mortality after anaesthesia in early life: results of the European prospective multicentre observational study, neonate and children audit of anaesthesia practice in Europe (NECTARINE). Br J Anaesth. 2021;126:1157–1172.
40. Warner DO, Zaccariello MJ, Katusic SK, et al. Neuropsychological and behavioral outcomes after exposure of young children to procedures requiring general anesthesia: the Mayo Anesthesia Safety in Kids (MASK) study. Anesthesiology. 2018;129:89–105.
Copyright © 2022 International Anesthesia Research Society