Isoflurane Anesthesia Has Long-term Consequences on Motor and Behavioral Development in Infant Rhesus Macaques

A LARGE body of experimental evidence associates anesthetic exposure with neural cell injury and death in the developing brain. These exposures have been associated with impaired behavioral and cognitive development in young animals.^1–5  Several retrospective human studies of neurocognitive and behavioral disorders after childhood exposure to anesthesia suggest a similar association; results indicate that the effects may be more profound with repeated general anesthetic events, longer anesthesia duration,^6–9  or cumulative exposures to multiple types of anesthetics, independent of comorbidities.^10,11  Moreover, repeated anesthetic exposure has been associated with an increased risk of developing attention deficit hyperactivity disorder later in life.¹² 

It is less clear whether a single anesthetic exposure, particularly of short duration, is enough to trigger long-term sequelae in young children. While some retrospective studies identified an increased risk of developmental or behavioral disorder diagnosis even after a single short general anesthetic event,^13,14  other studies produced ambiguous results¹⁵  or observed no increased risk.¹⁶  In fact, one recent study suggests no risk at all, not even after multiple anesthetic exposures.¹⁷ 

Most studies, both preclinical and clinical, have focused exclusively on cognitive function and learning disabilities as the primary outcome after anesthesia exposure at a young age. Very few have examined other functional outcomes, such as early motor development or socioemotional development. Since anesthesia exposure early in life has neural consequences similar to fetal alcohol exposure, which is known to be associated with problems with social function (including anxiety level, curiosity, and motor skills¹⁸ ), it is theorized that anesthesia exposure might have similar consequences.

We examined the effect of isoflurane exposure utilizing a nonhuman primate (NHP) model. Brain development of rhesus macaques corresponds closely to humans (neuro-development of rhesus monkeys at birth corresponds to that of a 6-month-old human^14,19 ); thus, they are a highly relevant model for social functioning, complex behavior, and emotional state. So far, only three other studies have explored behavioral and/or cognitive outcomes after anesthesia exposure in young NHPs.^20–22  In our study, we examined multiple behavioral outcomes after both single and multiple anesthesia exposures. Our primary interest was whether isoflurane exposure affected socioemotional behavior as measured by standardized tests (in which the subject is presented with novel and/or potentially threatening stimuli) and behavior in the home cage. We hypothesized that exposure to isoflurane would cause an increase in the display of anxious/inhibited behavior in response to some provoked tests and alteration in behavior in the home environment compared to controls. In addition, because anxiety can be difficult to reliably measure during the first month of life, we examined whether anesthesia affected reflexes and general reactivity during this time. The study compares single 5-h isoflurane exposure with three consecutive isoflurane exposures over the course of the first 2 weeks of life to model the clinically relevant human condition of single (one extensive surgery) versus repeated anesthesia exposures of human infants. This allowed us to specifically test whether multiple exposures to anesthesia result in behavioral deficits in the NHP compared to single exposures.

The subjects for this study were 24 male and female rhesus macaques (Macaca mulatta) reared at the Oregon National Primate Research Center (ONPRC). Infants were born in breeding groups containing approximately 30 to 50 individuals. They were housed in one of eleven 130-m² outdoor enclosures. Along with their dams, the infants were temporarily removed from the group on three separate occasions during the first 12 months of life for experimental reasons (table 1). In each case, the pair was removed the day before testing and housed in a standard monkey cage overnight. When the subjects were approximately 1 yr of age (mean = 358.7 ± 12.6 days), they were weaned from their dam and moved into one of four indoor pens with 5 to 6 study subjects (table 2). The pens were approximately 3.7 × 2.1 × 2.1 m and were in rooms containing up to 32 cage-housed monkeys. Each pen was located in a different room.

Subjects were chosen based on the dominance status of their mothers. Because they needed to be removed from and returned to the group at various time points, we chose mid-ranking mothers with at least one other surviving offspring. The dams had not received any isoflurane during pregnancy although most received low doses of sedative for physical exams (ketamine hydrochloride [HCl], 5 to 10 mg/kg; 1 to 2 times) as part of the normal husbandry practice at the ONPRC.

Monkeys were fed commercial monkey chow (Purina, USA) and were given supplemental enrichment such as grain or produce daily. Water was provided ad libitum. The light cycle in the indoor pens was 12:12 (lights on at 7 am).

This study was approved by the ONPRC Institutional Animal Care and Use Committee (Beaverton, Oregon) and was in compliance with all federal regulations, as well as the guidelines set forth in the Guide for the Care and Use of Laboratory Animals.²³  The ONPRC is accredited by the Association for the Assessment and Accreditation of Laboratory Animal Care, International (Frederick, Maryland).

At approximately five postnatal days of age (P5), dams and their neonates were transferred from group housing to an adequately sized single cage next to the surgical facilities. Animals were typically removed within 15 min with minimal distress to the group. On the morning of the experiment, the dam was mildly sedated, and the neonate was transferred to the operating room.

After taking baseline measurements (e.g., heart rate, respiration, temperature, blood pressure, weight, blood gas, and metabolic parameters), the neonate was gently hand restrained, while the anesthetic agent (isoflurane) was administered via face mask and spontaneous ventilation. This previously described induction technique models the method that is routinely used for human infants.^24–26  After the animal had reached a state of tolerance, the trachea was intubated. The lungs were mechanically ventilated, and general anesthesia was maintained for 5 h (end-tidal isoflurane = 0.7 to 1.5 vol%). The rationale for choosing 5 h of isoflurane exposure was two-fold: this clinically relevant anesthesia duration is often needed for more extensive surgical operations (there are no studies specifically addressing this issue; however, data from our institution indicate that more than 30% of all infant anesthesia exposures are longer than 3 h; unpublished data from medical records; H.Gries, M.D., Ph.D., unpublished data, 2015); and we have a substantial database documenting the pattern and extent of neuropathology and glial pathology that occurs in the perinatal rhesus brain after this exposure. Isoflurane was continuously titrated to maintain a surgical plane of anesthesia and verified every 30 min as no movement and not more than a 10% increase in heart rate or blood pressure in response to a profound mosquito-clamp pinch at hand and foot. General anesthesia was provided using the same setup as in a human operating room environment and comparable equipment and methods as previously reported.^24–27  In brief, each animal was monitored and controlled using full physiologic monitoring.^24–27  At the end of the 5 h, isoflurane was stopped, and animals were extubated once recovered (after about 10 min) in all cases. Animals were then kept for 1 to 2 h in an NHP incubator and formula fed as tolerated. Once fully recovered from anesthesia, neonates were returned to their dams and kept in a caged environment until their first behavioral assessment on P14.

Animals randomized to receive multiple episodes of isoflurane, the ISO-3 group (n = 8), underwent the same procedures as described in Exposure to General Anesthesia (first two paragraphs) on three separate postnatal days (approximately P6, P9, and P12) with 48 to 96 h of recovery between each episode. Animals in the control group (no isoflurane; n = 8) underwent a sham exposure, which consisted of a similar procedure on the same postnatal days (P6, P9, and P12), including removal from the dam, an IV cannula, physiologic measurements, and a period of handling to simulate the environment that the other animals experienced before and after general isoflurane anesthesia. Animals in the ISO-1 group (one exposure to isoflurane; n = 8) received one general anesthesia as described in Exposure to General Anesthesia (first two paragraphs) on P6 followed by two sham exposures on P9 and P12. The sex distribution was balanced in all three groups (four males and four females). This sample size was determined to be the minimum necessary to detect differences in cognitive abilities across groups, based on previous experience.

Subjects were assessed at P14 and P30 with a modification of the Infant Behavior Assessment Scale (IBAS), a widely utilized test battery designed to assess the development of early reflexes and general reactivity in infants.²⁸  The IBAS is modeled after the Neonatal Behavioral Assessment Scale²⁹  and has been validated in NHPs.

On the day of testing, the mother was lightly sedated with ketamine HCl (5 to 10 mg/kg) to allow the removal of the infant, which was transferred in a small transport box to the testing room. Upon arrival, the infant was removed from the box, swaddled in a towel, and held by the observer. The observer then rated the infant on several behavioral categories known to have specific developmental milestones, including basic motor reflexes (e.g., grasping objects, labyrinthine righting, and placing hands and/or feet on the countertop), sensory reflexes (e.g., orientation to objects and sounds), and survival reflexes (e.g., startle response and suckling responses). See figure, Supplemental Digital Content 1 (https://links.lww.com/ALN/B323), for specific details about the IBAS protocol. At the end of this testing battery, which took approximately 20 min, the infant was transferred to another room for a visual acuity test (not reported here) and was then returned to its dam. The infant–dam pair was returned to the social group later that day once the dam had fully recovered from the ketamine sedation.

Subjects were given a score of 1 for each measure of the protocol in which they performed the behavior at the highest expected level of development and a score of 0 if they were below this level. For example, it is expected that infants will have a strong grasp reflex in each limb by age P14. Subjects that had a strong grasp on this part of the test received a score of 1, while subjects with a weak grasping reflex received a score of 0. Measures related to each category were grouped and totaled so that each animal received a single score for each category. Thus, these scores were directly correlated to the number of areas in which the subject reached the defined criterion.

We used two methods to assess anxiety in our study: provoked response tests (e.g., free play, human intruder, and novel object tests, described in the next three subsections) and home environment assessments. Anxiety and related traits are multicausal and can be expressed in many different contexts³⁰ ; animals can be anxious in response to a social stimulus but have no anxiety in other situations. Thus, this kind of combined testing paradigm provides a more comprehensive picture of the emotional states of the animals than does a solitary assessment method. Provoked response tests assess unconditioned response to various threatening or potentially threatening stimuli, while home environment assessments examine the response to everyday naturalistic events such as interactions with conspecifics.³⁰  We performed provoked response tests at both 3 months and 1 yr of age. Home environment assessments were only performed on the animals after they had been weaned and moved into their new housing groups (table 1). In all tests, the observer was blind to the experimental treatment of the infants. The Observer software (Noldus Information Technology, The Netherlands) was used to score behavior (both live and from video; table 3, ethogram of behavior).

This test was designed to assess the infant’s propensity to explore an unfamiliar environment and intrinsic level of activity. A similar test is performed on children^31,32  and nonprimate animals (open field tests³³ ). Before testing, the mother was lightly sedated with ketamine HCl (5 to 10 mg/kg), after which both mother and infant (who always clung to the mother) were brought to the testing room (2.4 × 3.0 m), which contained a novel climbing structure and toys. A one-way window allowed observation and videotaping. The dam and infant were placed in an infant car seat located in the corner of the room, and the dam was given a second dose of ketamine. The mother was present to avoid the confounding factor of separation anxiety but sedated to prevent her from interfering with her infant’s behavior. The infant was videotaped for 5 min to provide a measure of initial reaction to the new situation and again at 25 to 30 min to assess acclimation to the testing condition.

The human intruder test (HIT)^34,35  is one of the most commonly used tests to assess anxiety and the related temperamental trait of behavioral inhibition in young macaques.³⁰  This test was used to measure behavioral responses in three situations: being alone in a novel cage, being in the presence of a human stranger whose gaze was diverted (a potentially threatening stimulus), and being in the presence of a human stranger making direct eye contact (an overt threat).

The HIT began with a 12-min acclimation period in a novel monkey cage with no human present (alone 1). After this period, a human intruder, who had never been in contact with the monkey, entered the room and stood approximately 0.3 m from the cage for 2 min. The intruder was always female, to avoid the confounding factor of gender, and always wore the same protective clothing including a mask, transparent plastic face shield, and gloves. She stood with her facial profile presented to the monkey, taking care not to make eye contact with the infant (profile). The infant was again alone for the third 2-min time period (alone 2). In the fourth period (stare), the human intruder reentered the room and made continuous direct eye contact with the infant for 2 min. The intruder then left for another 2-min period (alone 3) after which the novel object test commenced.

The novel object test was designed to test the infant’s inclination to explore ethologically relevant novel objects (i.e., pieces of unfamiliar fruit and novel items) that were both nonthreatening and potentially threatening. After the alone 3 period of the HIT, the intruder reentered the room and put various novel objects in the cage, each for 5 min. Except for the novel food, all items were removed before new objects were introduced. The novel objects, in order, were a piece of kiwi (novel food), brightly colored bird toy, Mr. Potato Head (only eyes and feet attached), and a rubber snake with a piece of apple (highly desirable item) on top. Mr. Potato Head was chosen as a potentially threatening stimulus because of the large eyes. For the 1-yr time point, we also included a black box hung on the outside of the cage. The box was a piece of polyvinyl chloride tubing with a cover at one end, so that the subject could not see what was in the box. At the end of the test, the infant was either returned to its mother and transferred back to the social group later in the day (3 months) or returned directly to its social group (1 yr). Behaviors coded from this test included the latency to inspect (approached within 3 cm), touch (intentional contact), and manipulate each item.

Home-cage behavioral assessments were performed to examine how the subjects adapted to the stress of being weaned from their mothers and put into a new social group. A highly trained and experienced observer (N.D.R.) took 10-min continuous focal observations³⁶  on the monkeys 2 to 3 times per week for 2 to 3 weeks (for a total of 60 min of observations per individual) beginning 1 to 2 days after the subjects were introduced into their new housing. The observer, who was familiar to the monkeys, entered the room and stood next to the pen for 10 min to allow the animals to acclimate to her presence. The monkeys were used to having people in their rooms and typically ignored her after a few minutes. She then began to record behavior of each individual directly onto a laptop computer for 10 min each. To limit the effects of the time of day, observations were taken between 12:00 and 3:00 pm. Behaviors coded (table 3) focused on social behavior, aggression, and aspects of emotionality such as anxiety behaviors and appeasement/submissive behavior.

We used ANOVA to examine differences among our three groups. For all variables, assumptions of normality and homoscedasticity were tested. Data were normalized with arcsine square root or other transformations. When no transformations normalized the data, nonparametric analyses (Kruskal–Wallis test) were utilized. Chi-square analyses were used for categorical data. α values (two tailed) were set at 0.05.

We did not detect sex differences in any of the variables. To address the multiple hypothesis testing, the false discovery rate (expected proportion of type 1 errors among all significant findings) was controlled using the Benjamini–Hochberg procedure.³⁷  Using an ordered list of P values generated from the individual tests, the Benjamini–Hochberg procedure determines which tests should be rejected in order to maintain a predetermined false discovery rate. We set the false discovery rate at 5% and report the raw P values from the ANOVA or Kruskal–Wallis test. For raw P < 0.05, we report the Benjamini–Hochberg-adjusted P value (q) as well. Because an individual’s response to one stimulus does not necessarily predict how it will respond to other stimuli,³⁰  we utilized the Benjamini–Hochberg procedure within each testing paradigm (i.e., IBAS testing, response to provocative tests, and home-cage testing). Post hoc comparisons (using the Tukey or its nonparametric equivalent, the Nemenyi test) were carried out on significant findings.

SYSTAT 11 software (SYSTAT Software, Inc., USA) was used for all analyses. Benjamini–Hochberg-adjusted P values and Nemenyi Q values were calculated by hand.³⁷  Data are presented as mean ± SD unless otherwise stated.

Subjects were exposed to isoflurane either one time (ISO-1) or three times (ISO-3) or were not exposed (controls). Infants tolerated the isoflurane exposure and remained physiologically stable throughout each 5-h general anesthesia sessions and postanesthesia recovery. All infants appeared in good health and had vital signs and metabolic values within normal limits during the second or third procedure. Animals that were not exposed to isoflurane on a given experimental day were also separated from their mothers and underwent IV catheterization and some handling after the same schedule and time of day as exposed infants. At the end of each procedure, study subjects were returned to their dams and, at P14 (after IBAS testing), were subsequently released back into their natal group.

Four infants across all interventional groups became ill (e.g., prolonged diarrhea) while in their natal groups and were moved indoors for clinical treatment. After the treatment was finished, two were housed in indoor cages with their dams until they were weaned at approximately 12 months. The cages were connected with a stainless steel tunnel, allowing the infants to have access to each other to simulate group peer interactions. The other two were housed together with a surrogate dam for 5 months. All seven of these animals (the four subjects, two dams, and one surrogate dam) were housed in the same room, with visual contact to one another. One subject was refractory to treatment and was not able to complete the study, and thus, 23 juvenile monkeys were housed in small groups indoors after weaning (table 1).

We saw no group difference in behavior on the IBAS when infants were tested at P14 (see table, Supplemental Digital Content 1, https://links.lww.com/ALN/B323, which provides statistics for all tests). In contrast, treatment modality did influence IBAS responses when infants were tested at P30. Basic motor reflexes differed across groups at this age (H = 7.89, P = 0.015, q = 0.045; fig. 1). Infants in the control group scored significantly higher in basic motor reflex compared to those in the ISO-3 group at this age (post hoc analysis: Nemenyi test comparison Q_{(ISO-3 vs. control)} = 2.81; P < 0.005). Their scores for this test were approximately two times those of ISO-3 monkeys. While not significantly different than controls, ISO-1 infants had intermediate scores (i.e., in between the ISO-3 and controls; Nemenyi test comparison Q_{(ISO-1 vs. control)} = 1.35; P > 0.50; fig. 1). No other group differences were observed at this time point (see table, Supplemental Digital Content 1, https://links.lww.com/ALN/B323).

Only seven infants moved from their mothers during this 30-min test. There were no treatment differences with respect to this latency or in any of the behaviors we measured during either the first or last 5 min of the test (see table, Supplemental Digital Content 1, https://links.lww.com/ALN/B323). However, there was a difference in the change of behavior across the test. We compared the amount of time infants spent passive (i.e., sleeping or remaining completely inactive) during the two time points, to determine how infants acclimated to the 30-min time period (i.e., whether animals became more or less active during the test). While control monkeys tended to show the same amount of passive behavior (within 5%) in both time points, infants in the ISO-1 treatment group showed more passive behavior at the end of the test (i.e., reduced activity over time), and ISO-3 infants showed more passive behavior during the first 5 min (i.e., increased activity over time; chi-square = 16.3, df = 4, P = 0.003, q = 0.03; table 4).

Contrary to our expectations, we did not see any group differences with respect to behaviors indicative of anxiety or behavioral inhibition in either the profile or stare periods (see table, Supplemental Digital Content 1, https://links.lww.com/ALN/B323, for statistics). Only two animals displayed anxiety behaviors (e.g., scratching and shaking) during this stimulation. There was a difference in the amount of time it took the infants to respond to the direct eye contact (i.e., latency to respond; H = 7.24, P = 0.03, q = 0.13) although this difference was not significant (NS) after Benjamini–Hochberg correction for multiple testing. Only 15 subjects (five in each of the three treatment groups) showed aggressive behavior toward the intruder. Examining only these 15 individuals, there was a difference between treatment groups in the amount of time subjects were aggressive toward the intruder (H = 7.62, P = 0.02, q = 0.13; NS), yet this difference did not reach statistical significance after correcting for multiple testing. Further, the frequency of grunting, a specific kind of vocalization, in the period immediately after the stare threat (alone 3) was different between groups although not statistically significant after correcting for multiple testing. While none of the control animals grunted during this time period, four of the ISO-3 animals and five of the ISO-1 group continued to react to the intruder by grunting (chi-square = 7.47,df = 2, P = 0.02, q = 0.13; NS) even though there was no longer any threat present.

There were no differences across groups with respect to this test when tested at P90 (see table, Supplemental Digital Content 1, https://links.lww.com/ALN/B323).

Because one female study subject was removed from the study, sample sizes for the analyses at the 12-month time point were n = 8 for control, n = 8 for ISO-1, and n = 7 (four males and three females) for ISO-3 subjects.

In general, subjects exhibited low levels of submissive fear grimaces (0.02 ± 0.03 SD bouts/min), aggressive (0.1 ± 0.08 bouts/min), or anxious (0.3 ± 0.2 per minute) behaviors in their home group. There were group differences in lip-smack (H = 13.53, P = 0.001, q = 0.006; fig. 2) and anxiety (scratch and shake; H = 7.78, P = 0.015, q = 0.045; fig. 3). Interestingly, ISO-3–treated subjects were five times as likely to engage in lip-smack, a behavior which is often used in appeasement or affiliation, than control monkeys and 10 times more likely to lip-smack compared to ISO-1–treated monkeys. They also displayed the highest levels of anxiety behaviors, exhibiting this behavior more than 1.5 times as often as control monkeys. ISO-3 animals were more likely to yawn than the others although the difference did not reach statistically significance (chi-square = 5.0, df = 2, P = 0.08; NS). Only six animals yawned during the focal observations; four of these were in the ISO-3 treatment group. There were no other differences across treatments (see table, Supplemental Digital Content 1, https://links.lww.com/ALN/B323, for statistics).

One-yr-old subjects spent the majority of time (61.3 ± 25.7%) of the profile period freezing and spent little time in other behaviors. There were no treatment differences (see table, Supplemental Digital Content 1, https://links.lww.com/ALN/B323, for statistics).

Subjects showed more of a behavioral response to the intruder in the stare period than in the profile period although there were no differences across treatment groups (see table, Supplemental Digital Content 1, https://links.lww.com/ALN/B323, for statistics). In contrast to the testing at 3 months, none of the subjects grunted in the alone 3 period.

There were no treatment differences with respect to the novel objects (see table, Supplemental Digital Content 1, https://links.lww.com/ALN/B323, for statistics). It is worth noting that while none of the ISO-3 monkeys put their hand inside the black box, in contrast, 3 ISO-1 and 4 control animals did explore the box.

This study demonstrates long-term behavioral consequences after repeated isoflurane anesthesia exposure in infant NHPs. We examined the behavioral and emotional consequences of such exposure during the first year of life. Specifically, ISO-3 treatment resulted in motor reflex deficits when animals were tested at 30 days of age. Animals in the ISO-1 group also had lower motor reflex scores than controls although the difference did not reach statistical significance. These measures of early neuromotor activity are important as they have been associated with later cognitive functioning. Schneider et al.³⁸  found low neonatal motor scores correlated with poor performance on a nonmatching-to-sample cognitive test during adolescence in rhesus macaques. Furthermore, animals exposed to prenatal alcohol showed motor deficits similar to those in our ISO-3 subjects.³⁹  Our subjects underwent cognitive tests to determine the effect of isoflurane exposure on learning and memory (data currently being analyzed). Based on our findings of motor reflex deficits, we expect to see cognitive impairment in ISO-3 animals, while ISO-1 animals might be less impaired. Such results would be similar to recent retrospective studies in children showing that multiple, but not single, exposures of anesthesia early in development are associated with learning disabilities later in life.^6,7 

ISO-3 subjects were more likely than others to display anxiety behavior (e.g., scratch and/or shake) at 1 yr of age, after they were weaned from their natal group. These monkeys also engaged in more lip-smacking, a behavior that generally conveys affiliation or appeasement.⁴⁰  Appeasement behaviors have been associated with social anxiety in humans.⁴¹  Taken together, these results suggest that ISO-3 animals were more anxious than others after being moved to their new group.

There was a difference in general activity during the free play test when infants were tested at 3 months. While there were no group differences with respect to the amount of time animals spent sitting passively (i.e., sleeping or sitting very still as opposed to being alert and attentive to their surroundings) at either the beginning (0 to 5 min) or end (25 to 30 min) of the test, there was a difference among groups with respect to change in this behavior over time. The control infants tended to remain relatively consistent across the time points; they spent about the same amount of time sitting passively at the beginning and end of the test. Interestingly, ISO-3 infants tended to become more active during the test, while ISO-1 infants tended to become more passive. The reasons for these changes are not clear. It is possible that the ISO-1 animals were initially in a heightened state of arousal, and later became drowsy and thus more passive. ISO-3 infants, on the other hand, may have become more aroused during the course of the test. It is important to note that there were no differences in the amount of time spent inactive during either time point; the ISO-3 infants were not less active than the other groups at the first time point. Rather, they were more likely than other groups to show a delayed onset and subsequent increase in activity and attention to the environment over time.

These results suggest that multiple exposures to isoflurane early in life may lead to behavioral differences later in life. Our results are particularly striking given our relatively low sample size. It is possible that we lacked sufficient statistical power to determine differences with respect to other tests. Alternatively, it is possible that infant exposure to isoflurane results in anxiety behaviors that manifest in a social but not in a nonsocial context. Mice exposed neonatally to sevoflurane, an anesthetic similar to isoflurane, had deficits in social interaction tests but not in response to novelty when tested as adults.¹  Similarly, Raper et al.²¹  recently found that infant rhesus macaques exposed to sevoflurane within the first month of life displayed more anxiety behavior on the HIT when tested 5 months later compared to controls. That study did not examine anxiety in a nonsocial setting. We did not see statistically significant group differences with respect to the behavior on the HIT, which could be due to several methodological differences between the study by Raper et al.,²¹  and our study. Still, both studies found that multiple early exposure to general anesthesia resulted in heightened levels of anxiety and emotionality^42,43  in rhesus macaques during their subsequent development, which underscores the relevance of these findings. Additional research is needed to examine (1) the morphologic basis for this altered phenotype; (2) the potential for other anesthetics resulting in the same behavioral effects; (3) potential modulators that could protect against these effects; and (4) the relationship between the observed behavioral effects and learning and memory performance at later stages of development.

There were some differences between the ISO-1 and other animals although they did not reach statistically significance after controlling for multiple comparisons. ISO-1 and, to a lesser extent, ISO-3 animals continued to respond to the intruder making eye contact by grunting, even after the threat had left the room, something that the control animals did not do. ISO-1 animals that showed aggression to the intruder during the stare period (n = 5) did so at a higher frequency than others. These results might indicate an increased tendency toward aggression in animals exposed one time compared to controls although studies are needed to explicitly examine this hypothesis. It is somewhat surprising that the data from the behavioral tests did not result in statistically significant differences between ISO-1 and control animals given that ISO-1 exposure reproducibly causes a 10- to 14-fold increase in apoptotic cell death in gray and white matter of neonatal rhesus macaques as we showed previously.^24,25  It is possible that the high degree of behavioral variability between individual study subjects, along with our relatively low sample size, led to a lack of statistical power to detect more moderate changes in behavioral phenotype between ISO-1 and control animals. While we did not see group differences in anxiety that reached statistical significance between ISO-1 and control animals, we do not conclude that a 5-h exposure to volatile anesthetics is safe for children as has been suggested recently based on a less rigorous exploration of 3- and 7-month-old cynomolgus macaques after infant sevoflurane exposure.²²  Rather, our data suggest a need for further systematic and well-designed work with larger sample sizes, longer than 12-month follow-up, and additional tests to determine the effects of a single neonatal exposure of anesthesia on behavioral outcomes in NHPs.

There were several limitations to this study. First, we had a relatively small sample size, which, along with the multiple tests we used, affected our ability to interpret the results. A larger sample size would have allowed us to better detect other, potentially subtle changes in behavior, particularly after single exposure, and, if there were differences, determine whether they were due to individual sensitivity or systematic effects of the total duration of the isoflurane exposure (one vs. three times). Larger sample sizes would also have allowed us to determine whether there were sex differences in sensitivity to infant isoflurane exposure as was reported recently in rodents.⁴⁴  Second, we only tested isoflurane; it is uncertain whether exposure to other anesthetics known to be neurotoxic for the developing brain^27,45  would result in similar behavioral abnormalities in this model. Third, we tested the effects of isoflurane anesthesia in the absence of surgery, while most anesthetic exposures in humans occur in conjunction with a surgical procedure. Such studies are urgently needed since it is possible that the long-term effects of anesthesia and surgery combined may be worse than those of anesthesia alone. However, while such combined studies are important from a translational perspective, they have their own methodological limitations by introducing the specific pathophysiologic responses to surgery, which can confound the results. In contrast, our current study design allowed testing the effects of infant isoflurane exposure alone on subsequent behavioral development in the primate, which is an important step toward understanding whether anesthetics themselves cause toxicity that result in long-term behavioral deficits.

Our findings suggest that infant exposure to isoflurane has long-term consequences affecting both motor and socioemotional aspects of behavior in an NHP model. Our data suggest dose-dependent effects, with multiple exposures resulting in more profound changes in some behaviors, particularly long-term anxious behavior and heightened emotionality already appearing at 1 yr of age. While these results do not prove a direct link between early anesthesia exposure, most often provided to tolerate surgery, and long-term behavioral consequences in humans, they suggest that emotional consequences should be analyzed in future clinical studies that aim to identify long-term outcomes of infant anesthesia. Most recent studies, including the Pediatric Anesthesia Neurodevelopment Assessment and General Anaesthesia compared to Spinal anesthesia studies, focus on cognitive function and intellectual performance (language skills and intelligence quotient) as the primary outcome measures^46–49 ; few have examined other functional outcomes, such as socioemotional development. Further, future experimental research in primates should aim to identify the mechanisms responsible for the behavioral consequences of infant anesthesia exposure in order to develop safe anesthesia regimes or protective strategies for young children requiring surgical or diagnostic interventions.

The authors are grateful to the animal husbandry and clinical staff at the Oregon National Primate Research Center (ONPRC), Beaverton, Oregon, for their excellent care of the monkeys. The authors also thank the following for their help with the behavioral testing and/or manuscript preparation: Adriane Maier, M.S. (Division of Comparative Medicine, ONPRC, Beaverton, Oregon), Lisa A. Houser, B.S. (Division of Comparative Medicine, ONPRC, Beaverton, Oregon), Daniel H. Gottlieb, Ph.D. (Division of Comparative Medicine, ONPRC, Beaverton, Oregon), Michelle Bermudez (Division of Comparative Medicine, ONPRC, Beaverton, Oregon), and Katie Menzel Ellis, M.D. (Department of Anesthesiology and Perioperative Medicine, OHSU, Portland, Oregon).

Supported by Frontiers in Anesthesia Research Award 2012 to Dr. Brambrink (awarded by the International Anesthesia Research Society, San Francisco, California) and National Institutes of Health (Bethesda, Maryland; grant no. 8P51OD011092) to the Oregon National Primate Research Center, Beaverton, Oregon.

The authors declare no competing interests.