Behavioral and Cognitive Outcomes of Rhesus Macaques Following Neonatal Exposure to Antiseizure Medications

Abstract

Objective:

Exposure of neonatal macaques to the antiseizure medications phenobarbital and midazolam (PbM) causes widespread apoptotic death of neurons and oligodendrocytes. We studied behavior and neurocognitive performance in 12–24mo-old macaques treated as neonates with PbM.

Methods:

Fourteen monkeys received phenobarbital and midazolam over 24hrs under normothermia (n=8) or mild hypothermia (n=6). Controls (n=8) received no treatment. Animals underwent testing in the human intruder paradigm at ages 12 and 18mo, and a 3-step stimulus discrimination task (SDT) at ages 12, 18, and 24mo.

Results:

Animals treated with PbM displayed lower scores for environmental exploration, and higher scores for locomotion and vocalizations compared to controls. Combined PbM and hypothermia resulted in lower scores for aggression and vigilance at 12mo compared to controls and normothermic PbM animals.

A mixed-effect generalized linear model was used to test for differences in neurocognitive performance between the control and PbM groups in the first step of the SDT battery [Shape Center Baited (SCB) to Shape Center non-Baited (SCNB)]. The odds of passing this step differed by group (p=0.044). At any given age, the odds of passing for a control animal were 9.53 (95% CI:1.06–85) times the odds for a PbM animal. There was also evidence suggesting a higher learning rate in the SCNB for the control relative to the PbM group (Cox model; HR=2.13; 95% CI:1.02–4.43, p=0.044).

Interpretation:

These findings demonstrate that a 24-hour-long neonatal treatment with a clinically relevant combination of antiseizure medications can have long-lasting effects on behavior and cognition in nonhuman primates.

Introduction

Over the past two decades, we and many other investigators reported that positive modulators of GABA_A receptors, including barbiturates, benzodiazepines, alcohol, and volatile anesthetics, trigger apoptosis of neurons and oligodendrocytes in the rodent and nonhuman primate brain,^1–5 suppress neurogenesis,⁶ inhibit normal synapse development and maturation,^7,8 and cause long-lasting behavioral and cognitive impairments, when exposure occurs during the period of rapid brain growth.^7,9–11 In humans, this period starts in the third trimester of gestation and extends to the third year of life.¹²

Prenatal or early postnatal exposure to anesthetics may contribute to adverse neurodevelopmental outcomes in humans.^13–23 Lower relative white matter volumes at term and lower cognitive and motor composite scores at 2 years of age were reported in preterm infants exposed to anesthesia.²² The General Anesthesia compared to Spinal anesthesia trial (GAS; prospective randomized, controlled), the Pediatric Anesthesia Neurodevelopment Assessment (PANDA; retrospective), and Mayo Anesthesia Safety in Kids (MASK; retrospective) studies assessed intelligence in children exposed to anesthesia during infancy.^21,23,24 General intelligence did not differ between groups. However, analysis of behavior and executive function revealed significant differences, with increases in behavioral problems being statistically significant after a single exposure to general anesthesia.¹⁷

Severity of brain injury resulting from early life treatment with sedatives, anesthetics, and antiseizure medications depends on dose, duration, and timing of treatment during the phase of rapid brain growth.^1,3,16,25,26 Notably, in the MAS, GAS, and PANDA clinical studies, anesthesia was short. However, treatment of refractory seizures and status epilepticus with antiseizure medications requires that these be administered at high doses and in combinations for days, weeks or even months. The severity of the resulting brain injury in humans is expected to depend on the type of medications and combinations used, plasma medication levels, duration of treatment, and age at the time of treatment. It is conceivable that, whereas short anesthesia in late infancy and early childhood may cause no or very subtle long-term effects on behavior and cognition, a robust antiseizure regimen administered over days to neonates may have more impactful consequences. Combinations of Pb, benzodiazepines, and other antiseizure medications constitute the recommended standard of care for the management of recurrent seizures and status epilepticus in infancy.^27,28

We had previously reported that the PbM combination causes degeneration of neurons and oligodendrocytes in neonatal NHPs.⁵ Neurodegeneration was substantially more severe after a 24-hour compared to an 8-hour-long exposure.⁵ Mild hypothermia, which we tested as a potential protective treatment, failed to ameliorate histological injury in the setting of the 24-hour-long treatment with PbM.²⁹ This failure of HT to protect does not eliminate the possibility of long-term beneficial effects on behavior and cognition.

Here we investigated the impact of a 24hrs neonatal treatment with PbM on the behavior and cognitive performance of rhesus macaques in the first two years of life. We included animals treated under normothermic and hypothermic conditions, aiming to determine whether HT might still influence neurobehavioral outcomes, despite its inability to prevent neuropathological injury.

Materials and Methods

Animals and treatment

Animal procedures were approved by the Wisconsin National Primate Research Center and the University of Wisconsin - Institutional Animal Care and Use Committee (IACUC) and were conducted in full accordance with the Public Health Service Policy on Humane Care and Use of Laboratory Animals. The experimental outline is presented in Fig. 1.

Figure 1:

Illustration of study design.

Neonatal rhesus macaques (2–15 days old; n=14) received an intravenous loading dose of phenobarbital (Pb; 40mg/kg) over 30min followed by an intravenous infusion of midazolam (M; loading dose 0.5mg/kg followed by 2–5mg/kg and hour for 23.5hrs). The midazolam infusion rate was adjusted to maintain moderate sedation. Phenobarbital maintenance doses (2.5mg/kg) were given at 6 and 18hrs, to maintain a therapeutically relevant level over 24hrs.

Normothermia (NT; T>36.5°C-37.5°C; n=8) and hypothermia (HT; 35–36.5°C; n=6) were induced/maintained with a servo-controlled infant hypothermia unit (Blanketrol III) and an esophageal/rectal temperature probe (Cincinnati Sub-Zero, OH, USA). Target hypothermic temperature was 35–36.5°C for 36hrs. Continuous pulse oximetry and a cardiorespiratory monitor were used. Gases, glucose, electrolytes, lactate, and hemoglobin were monitored on venous blood (femoral vein) at 0–24hrs. Hypothermic animals were slowly rewarmed starting at 36hrs by 0.5°C/hr and placed in the nursery once awake and able to feed. This became necessary because the first two infants were rejected by their mothers after 2–3 days of separation for the experiment and their recovery.

Control macaques were separated from their mothers at 2 weeks of age.

Experiments for all groups were performed in parallel. Details on housing are provided in the supplementary materials.

Human Intruder (HI) paradigm

Behavioral reactivity was tested at 12 and 18mo. The HI paradigm³⁰ evokes strong and distinct behavioral responses to novel and threatening stimuli and has been used to examine the role of limbic regions in emotional regulation.^31,32 It consists of four consecutive ten-minute phases: (1) alone condition (A1); (2) profile condition (NEC; No Eye Contact); (3) stare (ST) condition and (4) alone condition (A2).

To begin a session, animals were relocated to a novel room, placed in a testing cage, and left alone for 10 minutes (A1). After 10min, an unfamiliar human intruder (blinded to treatment) entered the room, sat down 2.5 meters away from the front of the cage, and presented their profile for 10min (NEC). Next, the intruder exited the room for 3min, reentered the room, and faced the animal maintaining a neutral face focusing on a fixation point beyond the test cage for 10 minutes (ST). The intruder exited and the animal was left alone for 10 minutes (A2). Animals were then returned to home housing. All sessions were remotely recorded for later analysis. The human intruder (same individual for all sessions) was male and unfamiliar to all animals. Behavior was scored by two experienced blinded observers, using a computer equipped software (The Observer v.11.0^®, Noldus, Inc., The Netherlands). Initial concordance between observers was 0.80 and inter-rater reliability of Cohen’s K was 0.72. Scoring above 0.7 is considered “substantial agreement”.³³

The frequencies and durations of mutually exclusive categories of behavior were recorded (Table 1). If a behavior was not expressed by an individual animal, the frequency of occurrence for the behavior would be “zero”. Duration calculations for these “zero” or non-expressed behaviors were returned as missing values and not included in the analyses.

Table 1.

Macaque behaviors studied in the Human Intruder paradigm.

Behavior	How measured	Definition
Vocalizations	Frequency	Shriek, bark, coo, girn
Individual behaviors	Frequency	Jump, locomote, huddle, stereotypy, fall, other, inactive, environmental exposure, self-direct, tooth gnash, lip smack, freeze, open mouth threat, yawn, display, eyebrow flick, forward orient
Environmental exploration
Environmental exploration average duration	Duration	Total time Exploring Environment/Occurrences of Environmental Exploration
Environmental exploration total time	Duration	Total session time spent exploring the environment
Latency to environmental exploration	Latency	Time from the session start to when they first explore
Vigilance
Latency to forward orient	Latency	Time from the session start to when animals orient with both eyes to facing the intruder
Forward orient average duration	Duration	Total time Forward Orient /Occurrences of Forward Orienting
Forward orient total time	Duration	Total time forward orienting
Vigilance Composite	Frequency	Average frequency of the occurrence of Freeze and Forward Orient
Locomotion
Latency to locomoting	Latency	Time from session start to first movement
Locomoting total time	Duration	Total session time spent locomoting
Locomoting average duration	Duration	Total time locomoting/Occurrences of locomoting/
Vocalizations
Latency to first vocal	Latency	Time from session start to any vocalization
All vocal	Frequency	The average frequency of shriek, bark, coo, girn
All vocal total time	Duration	Total session time spent vocalizing
Aggression
Aggression Composite	Frequency	The average frequency of bark, open mouth threat, display, ear flick, and forward orient
Anxiety
Generalized anxiety Composite	Frequency	The average frequency of tooth gnash, lip smack, freeze, fear grimace

Stimulus discrimination task

Animals underwent touchscreen-based cognitive testing at ~11–12, 17–18, and 23–24mo of age. Each round lasted approximately 2 months and each animal was tested ~5 days/week for up to ~1 hour/day in the morning prior to feeding. Animals were food restricted after 3 pm the prior day. Monkeys were behaviorally trained and tested by a single individual using a computer-controlled touchscreen apparatus. They were transferred to a cage (24” wide × 58” tall × 30.5” deep) positioned in front of a touchscreen monitor within a controlled testing room. The touchscreen was interfaced with a computer that controls all stimulus and reward presentations, and data collection. Food was offered as a reward available from a centrally located food cup. Utilizing touchscreen software each animal was trained and tested on the following tasks:

Chamber acclimation:

On day 1 the animal was placed outside the chamber and given food rewards to teach them the location of the food receptacle (2–5min). After successful retrieval, the cage was wheeled into the chamber.

Shape Center Baited (SCB):

The touchscreen was baited (karo syrup and marshmallows) to encourage touching a center image and building the stimulus-reward association. When successfully touched, a reward and a positive sound were provided. If unsuccessful, the screen turned blue and a negative sound was presented. The animal performed 100 trials per day. Once successful for 100 trials in a day, animals moved on.

Shape Center not baited (SCNB):

Animals progressed to touching a center object without bait. Once successful for 100 trials in a day, animals moved on.

Shape Random (SR):

Animals were trained to touch an image regardless of location. The random component eliminates spatial location as a reward contingency. Each session included 100 trials. After successfully completing trials for 3–5 days animals moved on.

Discrimination:

The challenging stimulus discrimination task tests the ability to discriminate between images and the acquisition and use of a cognitive rule. The correct object’s shape and color remain consistent but the location changes randomly. Successful completion requires correct responses on ≥90% of trials.

Wisconsin card sorting task (WCST) tests general executive function through concept formation, rule acquisition, maintenance of a cognitive set, cognitive flexibility, set switching, and preservation of set. The animal must learn a cognitive rule and recognize when the rule shifts. It is presented with 3 objects and must learn the rule to choose the correct response. After 10 correct responses the rule shifts, and a new rule must be learned. The animal must complete 10 correct at each of the 4 levels to complete the WCST.

Color Delayed response span task (DRST) tests recognition memory span by requiring the animal to choose the newly added image (based on color). A session lasts for 10 trial sets (9 images/trial). The trial ends when an incorrect image is chosen.

Five training levels are defined as follows: SCB-SCNB, SCNB-SR, SR-Discrimination, Discrimination-WCST and WCST-DRST. To be tested at a higher level, the animal must pass the previous level.

Statistical analysis:

Two-Way-ANOVA with Geisser-Greenhouse correction and Tukey’s posthoc test for multiple comparisons were used for group comparison. Fisher’s exact test was used to compare pass-rates in the SDT.

Significance levels were set at P<0.05.

Analysis of chances (odds) of passing in SDT:

A mixed-effect generalized linear model (mixed-effect logistic regression) was used to test for differences between groups while accounting for repeat testing at 12, 18, and 24mo. The model treated group and time as fixed effects with individual animals serving as a random factor.

Analysis of time (days of training) to pass in SDT:

The number of days needed to master the SCB task at each of the three ages (12, 18, or 24mo) was recorded and each animal was given up to 22 training days. Animals who failed or could not be trained within that time were censored for 22 days. A Cox model was used to test whether the learning rate for the task differed according to group. Models included age and group as explanatory factors and treated each collection of three measurements per animal as a cluster, with standard errors computed by jackknifing the clusters. Analyses were conducted using Stata (v 17.0; StataCorp LLC, College Station, TX) with statistical significance set at P<0.05.

Results

Treatment with PbM was tolerated well. Vital signs and laboratory measures remained within physiological levels. The mean phenobarbital plasma level at 24hrs in the hypothermia group was significantly higher than in the normothermia group (Table 2; Student’s t-test).

Table 2:

Physiologic variables in infant rhesus monkeys subjected to PbM for 24hrs under normothermia (PbMNT; n=8) or hypothermia (PbMHT; n=6). Measurements were taken at 0.5, 2, 6, 12, and 24hrs after initiation of Pb injection and represent means ± SEM. Phenobarbital and midazolam blood levels were measured at 2 and 24 hrs.

	Age (days) (mean ± SEM)	0.5hrs (mean ± SEM)	2hrs (mean ± SEM)	6hrs (mean ± SEM)	12hrs (mean ± SEM)	24hrs (mean ± SEM)
PbMNT (n=8) 5M, 3F	5.38 ± 1.01
Temp (°C)		36.76 ± 0.20	36.55 ± 0.50	37.11 ± 0.18	37.68 ± 0.24	37.03 ± 0.49
pH (venous)		7.37 ± 0.02	7.39 ± 0.03	7.40 ± 0.03	7.38 ± 0.02	7.35 ± 0.03
HR (beats/min)		201.60 ± 10.39	187.10 ± 9.63	189.00 ± 12.05	180.30 ± 11.69	196.70 ± 13.35
pCO2 (mmHg)		39.58 ± 2.72	44.90 ± 2.34	40.39 ± 1.27	41.59 ± 1.53	40.08 ± 3.38
SaO2 (%)		95.57 ± 1.23	95.29 ± 1.11	96.43 ± 0.69	96.43 ± 0.78	96.29 ± 0.84
Lactate (mM)		3.10 ± 0.81	2.07 ± 0.42	1.22 ± 0.16	0.91 ± 0.07	1.84 ± 0.56
Hb (mg/dl)		13.43 ± 0.83	13.95 ± 1.17	11.69 ± 1.10	11.62 ± 1.08	11.86 ± 1.11
Glucose (mM)		78.75 ± 5.36	63.00 ± 4.23	62.38 ± 5.85	66.86 ± 15.75	57.29 ± 11.25
Phenobarbital level (mcg/ml)			32.69 ± 5.72			35.51 ± 2.19
Midazolam level (ng/ml)			2,086 ± 1,113			1,904 ± 411
PbMHT (n=6) 2M, 4F	9.8 ± 1.655
Temp (°C)		35.37 ± 0.37	34.87 ± 0.52	35.02 ± 0.48	35.08 ± 0.47	35.08 ± 0.47
pH (venous)		7.38 ± 0.02	7.37 ± 0.01	7.36 ± 0.02	7.31 ± 0.02	7.34 ± 0.03
HR (beats/min)		218.00 ± 9.13	205.30 ± 11.16	207.30 ± 13.98	200.20 ± 9.32	189.50 ± 18.06
pCO2 (mmHg)		42.47 ± 3.89	41.73 ± 1.77	40.35 ± 3.70	38.46 ± 2.73	40.70 ± 2.94
SaO2 (%)		95.50 ± 2.03	97.50 ± 0.72	98.33 ± 1.12	94.50 ± 1.88	97.33 ± 0.96
Lactate (mM)		2.80 ± 0.46	1.41 ± 0.15	1.26 ± 0.13	1.33 ± 0.19	1.36 ± 0.18
Hb (mg/dl)		13.37 ± 0.62	13.42 ± 0.6q	13.2 ± 0.59	11.76 ± 0.79	11.45 ± 0.80
Glucose (mM)		92.67 ± 12.79	55.83 ± 6.95	54.67 ± 6.67	48.40 ± 6.11	86.20 ± 32.36
Phenobarbital level (mcg/ml)			41.95 ± 1.63			44.46 ± 1.91*
Midazolam level (ng/ml)			4,639 ± 1,967			2,156 ± 785

^*P<0.05 compared to phenobarbital level at 24hrs in the normothermia group (Student’s t-test).

PbM: phenobarbital/midazolam; NT: normothermia; HT: hypothermia; M: males; F: females; Temp: rectal temperature; HR: heart rate; pCO₂: partial CO₂ pressure; SaO₂: oxygen saturation measured by pulse oximetry; Hb: hemoglobin.

There were no significant differences in mean weights between the three groups indicating comparable nutritional status (Table 3).

Table 3:

Sex distribution and weights at ages 6, 12, 18, and 24 months.

Treatment	Body Weight (kg), Mean ± SEM
	Gender, n	6 months	12 months	18 months	24 months
Controls	All, 8	1.51 ± 0.05	2.18 ± 0.07	2.81 ±0.11	3.24 ± 0.16
PbMNT	All, 8	1.46 ± 0.09	2.10 ± 0.15	2.67 ± 0.15	3.29 ± 0.229
PbMHT	All, 6	1.48 ± 0.11	2.09 ± 0.11	2.71 ± 0.20	3.22 ± 0.28
Controls	M, 5	1.56 ± 0.06	2.24 ± 0.05	2.97 ± 0.08	3.37 ± 0.20
Controls	F, 3	1.43 ± 0.05	2.08 ± 0.16	2.55 ± 0.18	3.04 ± 0.27
PbMNT	M, 5	1.45 ± 0.12	2.01 ± 0.20	2.59 ± 0.14	3.17 ± 0.24
PbMNT	F, 3	1.47 ± 0.18	2.23 ± 0.24	2.80 ± 0.0.35	3.49 ± 0.48
PbMHT	M, 2	1.58 ± 0.18	2.31 ± 0.22	2.77 ± 0.36	3.16 ± 0.58
PbMHT	F, 4	1.38 ± 0.12	1.87 ± 0.19	2.66 ± 0.25	3.28 ± 0.20

PbM: phenobarbital/midazolam; NT: normothermia; HT: hypothermia; M: males; F: females.

Human intruder paradigm

At 12mo species-specific behaviors were present (Table 1). We quantified composite categories of behavior associated with vigilance, aggression and anxiety. Comparisons were first made between the PbMNT and PbMHT animals. When no significant differences were found, they were combined into one group (PbM; n=14) and compared with the controls (n=8). Significant differences were seen in several categories.

Neonatal exposure to PbM and HT affects aggressive behavior

The aggression composite was used to index aggression. When comparing the PbMNT and PbMHT groups, there was a significant effect of treatment (HT vs NT) at 12mo [F(1,12)=8.696, P=0.0122] and also at 18mo [F(1,12)=5.512, P=0.0369].

When comparing the three groups (controls, PbMNT, PbMHT), a significant effect of treatment (PbM) and task phase was found at 12mo (Fig. 2A). At 18mo there was a strong but non-significant trend suggesting treatment effect. There was significant effect of task phase at 18mo. Using post hoc Tukey’s multiple comparisons analysis significant differences were detected between the PbMNT and PbMHT groups during NEC and A2 at 12mo and between controls and PbMHT groups during NEC at 18mo (Figure 2A &B).

Figure 2:

(A&B) Effect of neonatal PbM exposure on aggression at 12 (A) and 18mo (B) of age in macaques. Columns represent the average (mean ± SEM; dots represent individual animal scores) frequency of behaviors indicative of aggression (Table 1) during each of the 4 phases of the HI paradigm (A1, NEC, ST and A2). There was a significant effect of treatment and task phase on aggressive behavior at 12mo and task phase at 18mo. The PbMHT group showed significantly lower average frequency of aggressive behaviors compared to the PbMNT group at 12mo during NEC and A2 and 18mo during NEC (F and P values are shown in the histograms).

(C&D) Effect of neonatal PbM exposure on vigilance at 12 (C) and 18mo (D) of age in macaques. Columns represent the average (mean ± SEM; dots represent individual animal scores) frequency of vigilance composite (Table 1) during each of the 4 phases of the HI paradigm (A1, NEC, ST and A2).

There was a significant effect of treatment and task phase on vigilance at 12mo and task phase at 18mo of age.

2-way-ANOVA with Geisser-Greenhouse correction was used to compare the three groups. Group comparisons for behaviors during each task phase were made using posthoc Tukey’s test for multiple comparisons.

F and P values are shown in the graphs (*P<0.05; **P<0.01; ***P<0.0001).

Controls (n=8); PbMNT (n=8); PbMHT (n=6). A: alone; NEC: no eye contact; ST: stare.

Neonatal exposure to PbM and HT affects vigilance

When comparing the PbMNT and PbMHT groups, there was a significant effect of treatment (HT vs NT) at 12mo [F(1,12)=7.791, P=0.0163 for treatment; F(2.510,30.11)=0.7691, P=0.4993 for task phase] but not at 18mo [F(1,12)=0.5158, P=0.4864 for treatment, F(2.811,33.73)=6.186, P=0.0022 for task phase].

When comparing the three groups (controls, PbMNT, PbMHT), a significant effect of treatment (PbM) but not task phase on vigilance composite was found at 12mo (Fig. 2C). A significant difference was found between the PbMNT and PbMHT groups during A2 at 12mo (Figure 2C). There was no significant treatment effect but a significant effect of task phase at 18mo (Fig. 2D).

Neonatal exposure to PbM affects environmental exploration

Three scores were used to measure environmental exploration, average duration, total time, and latency.

When comparing the PbMNT and HT groups there were no statistically significant treatment effects (NT vs HT) on any of the three scores at any age.

Environmental exploration average duration:

At 12mo, significant effects of treatment and task phase were observed. Using post hoc Tukey’s multiple comparisons analysis significant differences were detected between the control and PbM groups during A2 at 12mo (Fig. 3A). No significant effects were evident at 18mo.

Figure 3:

Effect of neonatal PbM exposure on environmental exploration at 12 and 18mo of age in macaques. Columns represent means ± SEM (dots represent individual animal values) of environmental exploration average duration at 12mo (A), environmental exploration total time at 12mo (B) and environmental exploration latency at 12mo during each of the 4 phases of the HI paradigm (A1, NEC, ST and A2). There was a significant effect of treatment on environmental exploration duration, exploration total time, and environmental exploration latency at 12mo. Effects were also dependent on task phase.

2-way-ANOVA with Geisser-Greenhouse correction was used to compare the two groups. Group comparisons for behaviors during each task phase were made using posthoc Tukey’s test for multiple comparisons.

F and P values are shown in the graphs (*P<0.05; **P<0.01; ***P<0.0001).

Controls (n=8); PbM (n=14). A: alone; NEC: no eye contact; ST: stare

Environmental exploration total time:

At 12mo, there was a significant effect of treatment and task phase with lower scores in the PbM group (Fig. 3B). No significant effects were seen at 18mo.

Environmental exploration latency:

At 12mo, there was a significant effect of treatment and task phase with higher scores in the PbM group (Fig. 3C). No significant effects were present at 18mo.

Effects of neonatal exposure to PbM on locomotion

Two scores were used to measure locomotion, latency and total time.

When comparing the PbMNT and HT groups there were no statistically significant effects of treatment (NT vs HT) or task phase on the scores at 12 or 18mo.

Latency to locomoting:

At 12mo and 18mo, there was a significant effect of treatment and task phase. A significant difference was detected between the control and PbM groups during NEC at 12mo (Fig. 4A&B). Latencies to locomoting were shorter in the PbM group.

Figure 4:

Effect of neonatal PbM exposure on locomotion at 12 and 18mo of age in macaques. Columns represent means ± SEM (dots represent individual animal values) of latency to begin locomoting (A, B) and locomoting total time (C, D) during each of the 4 phases of the HI paradigm (A1, NEC, ST and A2).

There was a significant effect of treatment and task phase on latency to locomoting at 12 and 18mo. No significant effect of treatment on locomoting total time was seen at 12 or 18mo.

2-way-ANOVA with Geisser-Greenhouse correction was used to compare the two groups. Group comparisons for behaviors during each task phase were made using posthoc Tukey’s test for multiple comparisons.

F and P values are shown in the graphs (*P<0.05; **P<0.01; ***P<0.0001).

Controls (n=8); PbM (n=14). A: alone; NEC: no eye contact; ST: stare.

Locomoting total time:

No statistically significant effect of treatment with PbM was present at 12 or 18mo but an effect of task phase was seen at 18mo (Fig. 4C & D).

Effects of neonatal exposure to PbM on vocalizations

Three measures were used to quantify vocalizations, latency, frequency and total time.

Comparison of the PbMNT and HT groups revealed no statistically significant treatment effects (NT vs HT) on any of the scores at 12 or 18mo.

Latency to first vocal:

At 12mo, there was a significant effect of treatment and task phase with shorter latencies in the PbM group (Fig. 5A). No significant effects were found at 18mo (Fig. 5B).

Figure 5:

Effect of neonatal PbM exposure on vocalizations at 12 and 18mo of age in macaques. Columns represent the means ± SEM (dots represent individual animal values) of latency to first vocalization (A and B), all vocal frequency (C), and all vocal total time (D) during each of the 4 phases of the HI paradigm (A1, NEC, ST and A2).

There was a significant effect of treatment and task phase on latency to first vocal at 12 but not at 18mo. There was a significant effect of task phase on all vocal frequency at 18mo (C) and a significant effect of treatment and task phase on all vocal total time at 18mo (D).

2-way-ANOVA with Geisser-Greenhouse correction was used to compare the two groups. Group comparisons for behaviors during each task phase were made using posthoc Tukey’s test for multiple comparisons.

F and P values are shown in the graphs (*P<0.05; **P<0.01; ***P<0.0001).

Controls (n=8); PbM (n=14). A: alone; NEC: no eye contact; ST: stare.

All vocal frequency:

No significant effects of treatment but a significant effect of task phase were present at 12mo. There was a strong but not significant trend for an effect of treatment and a significant effect of task phase at 18mo (Fig. 5C).

All vocal total time:

No significant effect of treatment but a significant effect of task phase were present at 12mo, and a significant effect of both treatment and task phase at 18mo with higher indices in the PbM group. A significant difference was detected between the control and PbM groups during A2 at 18mo (Fig. 5D).

Neonatal exposure to PbM does not affect anxiety

The generalized anxiety composite score was used to measure anxiety. When comparing the PbMNT vs HT and the PbM vs control groups there were no statistically significant treatment or task phase effects (NT vs HT) at 12 or 18mo.

Effect of gender

An exploratory analysis for effects of gender on performance in five HI paradigm domains was conducted (Supplemental Table 1). Significant differences were found in vigilance and all vocal frequency indices at 12mo suggesting that females may be more sensitive to PbM treatment.

Neurocognitive testing results in the SDT battery

Starting at 12mo, animals were able to master the SCB and SCNB tasks. Passing rates are shown in Supplemental Table 2 (A, B, D–H). There is an overall trend for the controls to perform better at each timepoint, but this difference did not reach significance using Fisher’s exact. When comparing training days at the SCB-SCNB level, there was no significant effect of treatment with PbM but there was a significant effect of age at testing on animal performance. Tukey’s post hoc analysis revealed that PbM-treated animals required significantly more training days at 18mo (Suppl. Table 2C).

Exploratory analysis for effects of gender on training days in the SCB-SCNB task among PbM-treated animals revealed a significant effect of gender. Females required significantly more training days than males (Suppl. Table 2C).

At the SCNB-SR training level, there was a significant effect of age but not treatment on mean training days (Suppl. Table 2E).

Chances (odds of passing) the SCB-SCNB task:

Supplemental Table 3 was used for this analysis. There was no evidence [X²(2df)=4.25, P=0.120] to suggest that NT vs control or HT vs control differed with respect to the odds of passing the test. Neither was there any indication [X²(1df)=0.08, P=0.776] that the odds of passing differed between the PbMNT and HT subgroups.

For the PbM group (n=14), data gave no evidence [X²(2df)=0.49, P=0.784] to suggest that the odds of passing changed over time as the animals aged, but there was evidence [X²(1df)=4.06, P=0.044] to suggest those odds differed by group. At any given age, the odds of passing the test for a control animal were 9.53 (95% CI: 1.06–85) times the odds that a PbM-treated animal would pass.

Times (days of training) to pass:

The number of days needed to master the SCB-SCNB task did not differ for either the NT vs control or HT vs control comparisons (F_2,21=2.15, P=0.141). We found no compelling difference between HT and NT groups with regard to time to master the SCB (F_1,21=0.17, P=0.686).

For the PbM group (n=14), there was no evidence (F_2,21=2.08, P=0.150) to indicate that the rate of learning varied by age; however, there was evidence suggesting a higher learning rate for the control group relative to PbM group (HR=2.13; 95% CI: 1.02–4.43, P=0.044).

Discussion

We report on the behavior and cognitive performance of juvenile NHPs after neonatal treatment with PbM at clinically relevant doses, blood levels, and duration under normothermic and hypothermic conditions. This treatment causes widespread apoptotic death of neurons and oligodendrocytes in NHPs.^5,29 The resulting functional alterations have not been characterized in primates.

Treatment with PbM impacted aggression, vigilance, exploration, locomotion, and vocalizations, and resulted in a slower learning curve in the SDT battery. Significant effects were observed in several behavioral domains in the HI paradigm. NT and HT animals behaved similarly, except in the aggression and vigilance domains where measures were significantly lower in the PbMHT but not the PbMNT group. In the remaining categories, no significant differences between PbMNT and HT groups were found. The results indicate lower scores for exploratory behavior and increased propensity to locomote and vocalize in the PbM group, whereas generalized anxiety measures did not differ from controls.

Lower environmental exploration indices suggest less curiosity and interest in novel environments. Neudecker and colleagues reported lower scores in the novel object test in NHPs treated with isoflurane as neonates which might partly result from behavioral inhibition.³⁴ Lower latencies to locomote in the PbM group with similar locomotor total times argue for increased impulsivity rather than a hypermotoric state.

The lack of significant differences in generalized anxiety contrasts reports by Raper and colleagues,^35,36 who described that the frequency of anxiety-related behaviors was significantly higher in 6–24mo monkeys exposed to neonatal anesthesia. It is possible that since PbM-induced brain injury is more pronounced compared to shorter anesthesia exposures, behavioral responses will differ. In addition, there are substantial methodological differences between the study by Raper and colleagues and ours.^35,36 Their intruder wore a mask which may affect the valence of the “anxiety provoking stimulus”. Importantly, the controls in the Raper et. al. study underwent repeated maternal separations, which affects stress responsivity and may increase the expression of anxiety-related behaviors.³⁸ In contrast, our animals were separated as neonates from their mothers because of anticipated rejection. Nursery and socially-reared NHPs show substantially lower levels of anxiety-related behaviors compared to mother-reared animals.³⁸ Possibly the overall lower anxiety levels in maternally separated NPHs prevented medication-related impact on this behavioral phenotype from becoming evident.

Raper and colleagues^35,36 described that hostile responses were not affected in NHPs who received sevoflurane anesthesia in infancy. This aligns with our finding that the PbMNT animals did not demonstrate different levels of aggression compared to controls.

Increased indices for locomotion and vocalizations in the PbM-treated group suggest hyperactivity and increased impulsivity, known features of attention deficit hyperactivity disorder (ADHD), while lower vigilance in the PbMHT animals might reflect lack of attention. These findings align with human studies which have identified ADHD as a disorder observed in children with a history of single or repeat anesthesia in early life.^18,21,23,39

When analyzing the collective performance of the PbM-treated and control animals, significantly higher odds of passing the SCB-SCNB task and a higher learning rate were found in the control group. This suggests inferior neurocognitive performance in the PbM-treated animals, reflecting lower executive function, attention, and possibly short-term memory. We were unable to test the performance of the groups in the WCST which would have allowed more robust conclusions about executive functions, memory, and attention.^40,41 Admittedly, the SDT battery has been designed for adults and has never before been tested in 12–24mo macaques. Hence, our results on the impact of neonatal PbM treatment on cognition remain limited, conclusions guarded, and will need to be complemented by testing in adolescence and adulthood.

Cognitive assessments have been performed previously in NHP following a single or 3 repeat short-duration neonatal 5hr-treatments with the anesthetic isoflurane.^35,42 Whereas stress and anxiety were more pronounced in the treated animals, cognitive functions during the first 2 years of life did not differ amongst groups.³⁶ However, Paule and colleagues¹¹ studied neurocognitive outcomes following longer (24hrs) neonatal treatment with ketamine, a known anesthetic and N-methyl-D-aspartate antagonist, which causes widespread apoptosis in the neonatal brain similar to that seen after PbM treatment. In the operant test battery, ketamine-exposed animals performed significantly worse than controls in learning, color and position discriminations, accuracy, and speed of task performance. This is highly relevant for seizure management, as ketamine is increasingly being used in the treatment of refractory seizures and status epilepticus.⁴³ Our results in the PbM-treated animals conform with those of Paule and colleagues and strongly argue that treatment-duration is critical for neurocognitive outcomes. Notably, we have reported that 5hrs sevoflurane- or 8hrs PbM-treatment cause significantly less brain injury than 24hrs of PbM-treatment.⁵ This might explain why short anesthesia treatments (3–5hrs) do not cause measurable cognitive deficits whereas the longer (24hrs) PbM-treatment does.

Our study has several limitations. The small group sizes (6–8) limit our ability to detect small alterations in behavior and cognition and sex-dependent differences within groups. Our exploratory analysis on effects of gender does provide some preliminary evidence that females may be more sensitive to PbM-treatment than males, but this will need to be explored and verified further. Testing was performed up to two years of age. For the behavioral assessments, but mostly for the neurocognitive battery chosen for this study, testing in adulthood would have enabled superior assessment of cognitive skills.

Also, we opted for maternal separation. This was not intended but unavoidable because PbM-treated animals were rejected, likely due to the long period of separation (48–72hrs). While this eliminates the influence of different qualities of parenting on infant development and allows for a more uniform social experience among all enrolled animals, maternal separation can impact behavior in primates.³⁸ Researchers reported that variation in infant behavioral responses to the HI paradigm among nursery- and mother-reared animals could be explained by four latent factors: “activity,” “emotionality,” “aggression,” and “displacement.” Socially reared animals had the lowest activity scores, the highest emotionality and the lowest aggression scores. Others have reported that nursery-rearing affects cortisol homeostasis with lower set-points in nursery-reared animals.⁴⁴ Environmental enrichment is critical for behavior and cognitive development of nursery-reared macaques.⁴⁵ While we made efforts to control environmental enrichment conditions for all animals, we are unable to assess whether neonatal PbM-exposure in maternally-reared animals would have impacted behavior and cognition differently.

One of our goals was to explore possible protective effect of hypothermia on PbM-induced apoptotic death and neurobehavior. Histological, behavioral, and cognitive data demonstrate lack of protective effect of HT. In fact, aggression and vigilance indices were significantly lower in the PbMHT but not the PbMNT animals. We speculate that PbMHT animals may have suffered more severe brain injury due to a “catch up” affect that occurred after rewarming (initiated at 36hrs). Although possible, this lacks confirmation, as histological analysis extends to 36hrs only.²⁹

Our study is limited to behaviors and cognition that can be assessed in NHPs between 1 and 2yr of age; whether neonatal PbM exposure continues to affect neurobehavior at later stages was not investigated. The study only tested one antiseizure medication combination and findings cannot be generalized to other compounds.

This study has several strengths. NHPs are the closest animal models to humans regarding genetics, physiology, and behavior and NHPs models are critical in translational research projects aimed at preventing, curing, or ameliorating human disease. NHPs are highly social animals, exhibit behaviors similar to humans, and are capable of performing cognitive tasks of higher complexity than rodents.⁴⁶

Importantly, we exposed the NHPs to conditions closely resembling those in the pediatric clinical setting. The medication levels measured were comparable to those in human infants. In the hypothermia group, mean phenobarbital levels were higher, which reflects slower clearance of the drug, similar to what has been described in humans.⁴⁷

Our study design allowed evaluation of the effects of infant PbM exposure without potential confounding effects from seizures or preexisting medical conditions. However, this is also a weakness, because antiseizure medications are administered to humans to treat neurologic disease and the combined impact on neurobehavior is what matters. Unfortunately, no suitable neonatal macaque models for seizures exist. Investigators have described chemically-induced seizures with pentylenetetrazol, penicillin or bicuculline, applied locally onto the cortex or hippocampus in older macaques.^48–49 These would need to be adopted and characterized in neonates before the combined impact of seizures and antiseizure medications on the developing primate brain can be assessed.

Conclusions

Our study in 12–24mo NHPs provides novel evidence for alterations in behavioral reactivity under challenge conditions of the HI test after PbM exposure during infancy. These NHPs model the developmental stage of children at 2–8yr of age.⁵⁰ Disturbances in interpreting and processing ambiguous environmental stimuli at this age could have serious long-term effects in children starting to attend school or other new environments and could result in difficulties in adapting to new social settings. In regards to the neurocognitive performance of the treated animals, results are limited but indicate that cognition is affected as well. Our findings strongly suggest that neonatal PbM exposure can result in behaviors that define attention deficit hyperactivity disorder.

Data availability

All main results of this study are available within the article. Further inquiries can be directed to the corresponding author.