Challenges to Maternal Wellbeing during Pregnancy Impact Temperament, Attention, and Neuromotor Responses in the Infant Rhesus Monkey

Abstract

The relative maturity, alertness, and reactivity of an infant at birth are sensitive indices of the neonate’s health, the quality of the pregnancy, and the mother’s wellbeing. Even when fetal growth and gestation length have been normal, the maturing fetus can still be adversely impacted by both physical events and psychological challenges to the mother during the prenatal period. The following research evaluated 413 rhesus monkeys from 7 different types of pregnancies to determine which conditions significantly influenced the behavioral responsiveness and state of the young infant. A standardized test battery modeled after the Neonatal Behavioral Assessment Scale for human newborns was employed. The largest impairments in orientation and increases in infant emotional reactivity were seen when female monkeys drank alcohol, even though consumed at only moderate levels during part of the pregnancy. The infants’ ability to focus and attend to visual and auditory cues was also affected when the gravid female’s adrenal hormones were transiently elevated for 2 weeks by ACTH administration. In addition, responses to tactile and vestibular stimulation were altered by both this ACTH treatment and psychological disturbance during gestation. Conversely, a 2-day course of antenatal corticosteroids 1 month before term resulted in infants with lower motor activity and reactivity. These findings highlight several pregnancy conditions that can affect a young infant’s neurobehavioral status, even when otherwise healthy, and demonstrate that alterations or deficits are specific to the type of insult experienced by the mother and fetus.

INTRODUCTION

Although the behavioral repertoire of humans and most infant animals is limited at birth, and many of the neonate’s responses are constrained by species-specific requirements for survival, there is often quite a range of individual variation in alertness and responsiveness to the environment. Some differences are attributable to the infant’s relative maturity, most notable when comparing premature to full-term infants (Lundqvist & Sabel, 2000). Another influential factor is the duration and difficulty of the labor, especially if the neonate still shows sedative effects from anesthesia employed for an assisted obstetrical delivery (Gunnar, Isensee, & Fust, 1987; Lester, Als, & Brazelton, 1982). In addition, variation in the infant’s behavior can reflect the quality of the pregnancy, including the adverse impact of maternal illness or teratogen exposure during gestation (Oyemade et al., 1994; Tronick, 1987; Yager & Ashwal, 2009). Almost 40 years ago, the pediatrician T. Berry Brazelton developed a standardized battery of tests for rating these normative reflexes, responses, and arousal states (Brazelton, 1973). This Neonatal Behavioral Assessment Scale (NBAS) has been employed widely in clinical practice to evaluate the health status and maturity of human babies at birth (Als, Tronick, Lester, & Brazelton, 1977). It has also been used to examine heritable aspects of sensory and emotional reactivity, which are intrinsic attributes that will become formative components of temperament and personality (Davis & Emory, 1995; de Bruijn, van Bakel, Wijnen, Pop, & van Baar, 2009).

The NBAS was subsequently modified for nonhuman primates, specifically for the rhesus monkey. This primate version of the instrument has been used to demonstrate that maternal stress, alcohol consumption during pregnancy, and many other teratogens can perturb fetal development (Golub & Gerswhin, 1984; Sackett, Ruppenthal, Hewitson, Simerly, & Schatten, 2006; Schneider, Coe, & Lubach, 1992; Schneider, Moore, Suomi, & Champoux, 1991; Schneider, Roughton, & Lubach, 1997). In addition, other research on monkey neonates showed that infant behavior at birth already reflects genetic differences between populations and individuals. Neonatal test scores were found to vary in offspring born to rhesus monkeys derived either from India or China (Champoux, Suomi, & Schneider, 1994). Within a single breeding colony, they were also associated with an allele polymorphism related to serotonin biology (Champoux et al., 2002).

The current analyses extended these evaluations of young monkeys in several important ways. The original testing involved an examination of infant rhesus macaques raised by humans in a nursery (Schneider et al., 1991). Thus, it was important to continue verifying the validity and efficacy of the test items on a larger number of infants reared by their biological mothers under standardized laboratory conditions. The postnatal growth and early behavioral development of a bottle-fed monkey in a nursery is often more precocious than that of mother-reared infants, which would affect performance on several of the critical test measures (Champoux & Suomi, 1988).

It was also of interest to conduct a new factor analysis for grouping the test items, which are usually combined into subscales. The NBAS for humans has seven subscales (Habituation, Orientation, Motor Performance, Range of State, Regulation of State, Autonomic Regulation, Reflexes), whereas the primate version appeared to be best resolved into four primary subscales (labeled originally as State Control, Orientation, Motor Maturity, Motor Activity) (Brazelton, 1978; Schneider et al., 1991, respectively). Further, because of the labor-intensive nature of primate studies, most research articles typically summarize the results from only a single experimental manipulation with a relatively small sample size. It was now possible to compile the test results from a much larger number of animals spanning many types of pregnancies across two decades of research. This large sample size had the added statistical benefit of facilitating the generation of more stable factors and reliable subscales. In addition, while many effects of prenatal alcohol exposure, infection, stress, or hormone administration had been described previously in both rodents and primates (Schneider et al., 1997; Weinstock, 1997), the comparative impact of these different paradigms are only rarely assessed directly (e.g., Fortier, Luheshi, & Boska, 2007).

These analyses in monkeys build upon a vast corpus of research, including the pioneering studies of the late Seymour Levine and others in rodents going back over 50 years, which first demonstrated the importance of both prenatal and early rearing conditions in shaping infant development (e.g., Levine & King, 1965; Milkovic et al., 1974; Weiner & Levine, 1978). Our inclusion criteria for using monkey data from a previously published experiment or from an unpublished neonatal test required that all aspects of the husbandry be comparable, other than the specific experimental manipulation or challenge during pregnancy. The current report on seven pregnancy conditions also includes a new comparison, contrasting the behavioral responses of totally undisturbed infants to offspring generated as “handling controls” for the inadvertent disturbance associated with experimental procedures. This second group of control infants was born after pregnancies involving saline injections, occasional blood sampling, or sham infections (e.g., Schneider et al., 1992).

Finally, the analyses also include new data on infants born to females infected with influenza virus during pregnancy or induced to mount transient inflammatory responses by administration of endotoxin. There has been a long-standing obstetrical concern about the risks that bacterial and viral infections pose for miscarriage and premature delivery during early gestation. But even when the pregnancy is viable and full-term, maternal illness during gestation can be a risk factor for neurodevelopmental disorders in children or for the later onset of psychopathology in adolescents (Mednick, Machon, Huttunen, & Bonnet, 1988; Yolken & Torrey, 1997). Despite clear effects on the brain and behavior in rodent models, it has been difficult to definitively establish how much of a danger a mild maternal infection is for the developing human fetus (Lowe, Luheshi, & Williams, 2008; Shi, Fatemi, Sidwell, & Patterson, 2003). Small differences in emotional reactivity and behavioral maturity at birth can potentially be of significance because they immediately begin to influence maternal responses, as well as govern how receptive the infant is to incorporating salient information from the rearing environment.

METHODS

Subjects

Mother and infant rhesus monkeys (Macaca mulatta) from a large, long-established breeding colony at the Harlow Primate Laboratory were evaluated. The adult females were laboratory reared, multiparous adults, 5–20 years of age, and each was bred with one male for 4–7 days to verify paternity and date of conception. All were descendents of rhesus monkeys imported from India 4–7 generations earlier in the 1950s and 1960s (for more details, see Price, Hyde, & Coe, 1999). After pregnancy was confirmed by the cessation of menstruation, the females were housed individually under standardized conditions until the birth of their infants. The monkeys were continuously in visual and auditory communication with other animals, and an enrichment husbandry program provided each mother with daily stimulation using foraging devices and other manipulanda. However, with the exception of cage cleaning, they were not disturbed by other human interventions beyond those involved in the specific experimental manipulations described below. All 413 infants were born naturally at term; caesarian delivery and prematurity were exclusion criteria.

Pregnancy Conditions

The infants were generated from seven different pregnancy conditions across two decades of research projects. To increase the sample size for animals designated as undisturbed controls, and because the husbandry procedures were identical, infants from this condition were combined across several previously published studies. In some cases, two similar experimental conditions from the original project were collapsed (e.g., maternal stress early and late during gestation; dexamethasone [Dex] and betamethasone were combined as antenatal corticosteroid treatments). A few conditions from the original studies were not used in order to facilitate the clarity of the current analyses (e.g., the current alcohol condition includes only infants from mothers that consumed alcohol, whereas the original project had a combined alcohol-and-stress condition). The seven pregnancy conditions were:

1. Undisturbed Controls: 177 infants (91 male, 86 female) born to adult females that were not manipulated experimentally and lived in stable housing conditions throughout the pregnancy.

2. Handled Controls: 35 infants (17 male, 18 female) from pregnancy conditions that served as a disturbance control for the handling procedures involved in injections (2–10 saline injections), blood collections (3–4 samples), or temporary rehousing for a sham infection during pregnancy (2- to 3-week relocation to a quarantine room).

3. Dex-treated: 16 infants (7 male and 9 female) from pregnancies that included a 2-day treatment with either dexamethasone (.125 mg/kg at 12 h intervals) or betamethasone (.2 mg/kg at 24 h intervals) injected intramuscularly on Days 145–146 postconception, approximately 25 days before term (N = 10 and 6, respectively) (Coe & Lubach, 2005). The betamethasone condition was added after many clinicians switched to this alternative glucocorticoid drug, and the once per day dose followed prevailing clinical practice (Lee, Stoll, McDonald, & Higgins, 2006).

4. Flu/LPS: 33 infants (19 male, 14 female) generated from pregnancies involving either intranasal exposure to a H3N2 strain of influenza virus (A/Sydney/5/97) during Days 119–126 postconception or administration of a low intravenous dose of endotoxin injected twice at 0900 on Days 125 and 126 of gestation (lipopolysaccharide, 2–4 ng/kg derived from Escherichia coli 0:113) (Short et al., 2010).

5. Stress: 100 infants (46 male, 54 female) generated from mothers that experienced acute daily disturbance for 25% of their 24-week pregnancy, either between Days 50–92 or Days 105–147 postconception (Coe, Lubach, & Shirtcliff, 2007). This arousing manipulation involved a brief relocation to a darkened room in the mid-afternoon, during which an acoustical startle paradigm was utilized to transiently provoke and increase adrenal hormone secretion. A loud sound (110 dB) was broadcast briefly for <1 s at three unpredictable intervals during the 10 min relocation period.

6. ACTH-stimulated: 13 infants (8 male, 5 female) from pregnancies that included a 2-week period of markedly elevated adrenal hormone levels (Schneider et al., 1992). Across Days 120–133 postconception, the gravid females were administered ACTH daily (Acthar gel, 1 USP/kg IM). Seven received it at 0900 and 6 at 1600 to contrast the effect of time of administration because hypothalamic–pituitary–adrenal should normally decline from morning to evening. Cortisol levels were significantly elevated to an average 70 μg/dL at 4 h after injection, but returned to mean basal levels of 37 μg/dL by the next day.

7. Alcohol-exposed: 39 infants (18 male, 21 female) born to females provided with sweetened alcoholic beverages each afternoon for part of their pregnancy, equivalent to .6 g/kg when totally consumed (Schneider et al., 1997, 2005). The prenatal alcohol exposure spanned 2–4 months during early- and mid-gestation, from Days 0 to 50 (N = 12), Days 0 to 135 (N = 7), or Days 50 to 135 postconception (N = 9), and relied on voluntary consumption by the gravid female. This amount of drinking in monkeys results in a blood alcohol level of 20–50 mg/dL at 1 h postconsumption.

Administration of Infant Behavioral Assessment Scale (IBAS)

Each infant was evaluated with the standardized IBAS test battery by trained raters blinded to prenatal condition. This report focuses on tests performed at 2 weeks of age, which is a more stable time point than the first days right after delivery, and previously shown to be representative of the infant’s subsequent responses at 3 and 4 weeks of age (Schneider & Suomi, 1992). Infants were removed from their mothers for approximately 20–30 min between 0900 and 1100. To be consistent across the two decades of research, a total of 49 neuromotor reflexes, attention reactions to visual cues and sounds, responses to vestibular and tactile stimulation, and attributes of emotionality and arousal state were scored during the test, although prior factor analyses had narrowed the focus to a smaller subset of indices (Schneider et al., 1991). The current factor analysis reduced the list further to 29 informative items, grouped together in the Results Section within the new factors that emerged. Items that showed low variance across animals were excluded: when over 56% of the infants received the same score. Definitions for the items and scoring criteria have been described previously (Schneider et al., 1992).

Statistical Evaluations

A Principal Components Analysis (PCA) using varimax rotation was employed for 29 of the more reliable IBAS items. Avarimax rotation was selected to allow factor scores to be modestly correlated, while still maximizing the variance to be explained by the PCA. In addition, the varimax rotation frequently results in more interpretable factors because a completely orthogonal factor structure is less common with behavioral data. It involves changing the coordinates within the PCA and relaxing the constraint that the factors must be orthogonal. Initially, data from 177 infants were combined from all Undisturbed Control pregnancies and considered together so that the factor scores and scales could be normalized for an undisturbed pregnancy. Based on these analyses, comparable factor scores were then generated and saved for all 413 infants, including the experimentally manipulated pregnancies. The subsequent PCA based on the 413 infants verified that the first clustering of test items was also representative of the full sample of infants. To evaluate the effects of pregnancy treatments, analyses of variance (ANOVA) were conducted with Prenatal condition and Sex of Infant as between factors. The minimum N per prenatal condition was 13, and the maximum was 177. As a consequence of this difference in sample size, assumptions about homogeneity of variance for the four factors were verified with the Levene’s test. Despite the unequal N across pregnancy conditions, the variance of the four-factor scores did not differ significantly (p = .20, .27, .21, and .20 for Factors 1–4, respectively). Post hoc testing of the main effects in the ANOVA were based on planned orthogonal contrasts with the primary goal of determining if infants from manipulated pregnancies differed from undisturbed controls and then handled controls. Alpha was maintained at .05 given this finite number of planned comparisons.

RESULTS

Data from only the Undisturbed Control infants were considered first to generate an initial factor solution for the IBAS. Our a priori expectation was that there would be four factors based on the previously published analyses from Schneider et al. (1991). There were 10 potential factors with eigenvalues greater than 1, but examination of the Scree Plot indicated a marked increase in explained variance at four factors (see Tab. 1). These four factors accounted for 53.5% of the total variance in the 29 items. Addition of a fifth factor helped to explain only 5.1% more of the variance and it was not very interpretable. Conversely, a three-factor solution had poor discriminant validity (i.e., many items loaded on two or more factors) and it explained only 45.8% of the total variance. Thus, in keeping with the amount of variance explained and our a priori expectation, we selected the four-factor solution. Table 2 shows that the average correlation between the four new factors was just r =.27, indicating that there was not a high degree of multicollinearity.

Table 1.

Four Factors That Emerged From the Principal Components Analysis (PCA), and the 29 Individual Test Items From the IBAS That Loaded With These Factors

	Factor 1: State Control	Factor 2: Motor Activity	Factor 3: Orientation	Factor 4: Sensory Sensitivity
Predominant State	.88
Irritability	.86
Struggle During Test	.84	.21		.22
Response Intensity	.84
Lack of Soothability	.82		−.26
Inconsolability	.77		−.24	.26
Calming self	.62			−.45
Inversion	.44	−.23
Restrain	.45	.20		−.36
Tremulousness	.40
Coordinate		.86
Motor activity	.33	.72
Spontaneous Crawl		.74		.25
Maintenance Balance		.71
Active Power	.25	.62
Not Passive	.39	.59
Response Speed	.24	.43	.26
Visual Orient			.81	.23
Attention			.80
Duration of Looking			.75
Visual Follow			.74	.35
Distractible	.27		−.62	.27
Orient to Auditory	.27	−.28	.48
Rotation Test				.54
Parachute		.24		.51
One Min Vocalization	.45			−.51
Galants	−.22			−.46
Head Posture Prone		.25		.40
Tactile Response	−.24		.20	−.40

Factor Loadings greater than .60 are shown above in bold to highlight the more reliable items with larger loadings. Factor loadings greater than .40 are indicated in italicized font with lighter shading to denote items considered to be lesser components of that factor. Empty cells indicate that the items had factor loadings less than .20.

Table 2.

Factor Solution for the IBAS Generated by a Principal Components Analysis of Data From 177 Infant Rhesus Monkeys

	Four Factor Solution
	F1	F2	F3	F4
Current factors
F1. State Control
F2. Motor Activity	.28**
F3. Orientation	−.33**	.09
F4. Sensory Sensitivity	−.37**	.10*	.32**
Schneider et al. (1991) factors
S1. State Control	.96**	.30**	−.32**	−.29**
S2. Motor Activity	.33**	.96**	.02	.05
S3. Orientation	−.31**	.11*	.98**	.33**
S4. Motor Maturity	−.07	.58**	.41**	.39**

The upper table shows the relative independence of the four factors (i.e., by the relatively low association, despite reaching statistical significance). The relationship between the new factors and original factors described previously by Schneider et al. (1991) is shown in the lower part of the table. Asterisks demarcate p values (*<0.05, **<0.01).

Factor 1: State Control

The first factor explained 23.5% of the total variance in the 29 IBAS items (see Tab. 1). It contained high factor loadings for predominant state (PS loading = .88), irritability (IRR loading = .86), struggle during test (ST loading = .84), response intensity (RI loading = .84), difficulty to sooth (SOO loading = .82), and the temperament rating of inconsolability (TRC loading = .77). It also contained moderately high loadings for reaction to restraint (RE loading = .45), frequency of vocalization per minute (VO loading = .45), reaction to inversion (INV loading = .44), and tremulousness (TRM loading = .40). This factor was designated as State Control, and higher scores indicated Poorer State Control.

Factor 2: Motor Activity

The second factor explained 13.6% of the total variance in the IBAS items. It contained high factor loadings for coordination (COO loading = .86), spontaneous crawl (SPC loading = .74), motor activity (MA loading = .72), maintenance of balance (MB loading = .71), and active power (AP loading = .62). It also contained moderately high loadings for lack of passivity (PAS loading = .59) and response speed (RS loading = .43). The items with higher loadings in this second factor included several that overlapped with the original Motor Activity scale from Schneider et al. (1991). Thus, we have continued to designate it as Motor Activity.

Factor 3: Orientation

The third factor explained 8.7% of the total variance in the IBAS items. It contained high factor loadings for visual orient (VIO loading = .81), attention (ATT loading = .80), duration of looking (DL loading = .75), visual follow (VF loading = .74), lack of distractibility (DIS = .62), and a moderately high loading for orient to auditory stimuli (OA loading = .48). This third factor was labeled Orientation.

Factor 4: Sensory Sensitivity

The fourth factor explained 7.7% of the total variance in the IBAS items. It had the weakest factor loadings and incorporated two items (VO and CS) that cohered somewhat with the State Control factor. Nevertheless, there were five distinctive items that loaded exclusively on this fourth factor, which included reactions to tactile and vestibular stimulation. It included response to a rotation test (ROT loading = .54), limb extension during a “parachute” descent (PA loading = .51), the Galant reflex bias to flex laterally toward dorsal tactile stimulation (GA loading = .46), response to touching of extremities (TR loading = .40), as well as vocalizations per minute (VO loading = .51), and ability to calm self (CS loading = .45). Head and eye turning in the direction of rotation (ROT) was perhaps the single most important test item. This fourth factor was consequently labeled Sensory Sensitivity.

We then compared this new factor solution with the factor solution from Schneider et al. (1991) to determine whether a similar labeling of the factors was appropriate. The current factors were highly correlated with those original factors, with the exception of Sensory Sensitivity (see Tab. 2). While this latter factor appears to capture unique and important information about the infant, on the basis of the significant correlations, Sensory Sensitivity did overlap somewhat with certain attributes related to State Control, Motor Activity, and Orientation.

To determine whether the PCA solution generated from the Undisturbed Controls was also representative for all 413 infants, a PCA with varimax rotation was recalculated for the entire sample. Four distinct factors again emerged, explaining 53% of the total variance, with each factor explaining 6.5–23% of the total variance in the 29 IBAS items. The four factors maintained high loadings on items that were similar to ones comprising the factors for Undisturbed Controls. Table 3 shows that the factor solution generated from the entire sample of 413 infants performed comparably to the solution from the Undisturbed Controls, with an average r =.88 between the two concordant sets of scales.

Table 3.

Comparison of the first Factor Solution Determined for the 177 Infants From Undisturbed Control Pregnancies and Subsequently From the Full Sample of 413 Infants From 7 Pregnancy Conditions

	PCA based on all 413 cases
A1. State Control	.98**	.22**	−.22**	−.36**
A2. Motor Activity	.08	.96**	.08	.24**
A3. Orientation	−.15**	.05	.95**	.29**
A4. Sensory Sensitivity	−.04	−.05	.01	.62**

The high correlations evident in the diagonal indicate that the modeling and grouping of test items into four factors was appropriate for assessing the impact of the pregnancy conditions. Asterisks demarcate correlations that reached statistical significance.

Pregnancy Conditions

The influence of Pregnancy (seven prenatal conditions) and Infant Gender (male, female) on the four IBAS factors was then examined with a series of univariate ANOVAs. Effects of pregnancy condition are described first, followed by any salient differences for male and female infants. Across the seven prenatal manipulations, there was only an overall statistical trend for a difference in the infants’ State Control during the test, F(6,395) = 1.89, p <.08. Upon closer examination, however, this tendency was driven primarily by the alcohol exposure (Fig. 1A). Using a planned orthogonal contrast for post hoc testing, the alcohol-exposed infants were found to have poorer State Control than the Unhandled Controls (p = .007) and Handled Controls (p = .053). These signs of increased irritability and emotional reactivity during the IBAS test appeared specific to the prenatal alcohol exposure.

FIGURE 1.

Influence of the 7 pregnancy conditions on the four IBAS factors. Panel A: Poor State Control (upper left) was most evident in infants from the prenatal alcohol condition. Panel B: Low scores for Motor Activity were evinced by infants from the Dex-treatment, ACTH-stimulation, and Prenatal Stress conditions (upper right). Panel C: Orientation ratings were most impaired by Alcohol-exposure and ACTH-stimulation (lower left). Panel D: Similarly, these two groups had the most aberrant Sensory Sensitivity scores, although infants from the Dex-treatment and Prenatal Stress conditions were also impacted (lower right). In contrast, the prenatal infection paradigm resulted in trend for scores to be shifted in the opposite direction. Asterisk (*) indicates significant difference from Undisturbed Controls; dot (˙) indicates a difference from Handling Disturbance control pregnancies.

The ratings for Motor Activity were more clearly differentiated across the seven pregnancy conditions F(6,395) = 3.78, p <.001. Planned comparisons revealed that the offspring from Dex-treated and Maternal Stress pregnancies scored lower than the ones from Undisturbed Control pregnancies, p = .002 and p = .001, respectively. The control infants showed more active and mature profiles. In addition, the Motor Activity scores of infants from the ACTH-stimulated pregnancies were also below those of the Undisturbed Controls (p = .042). In the current analyses, they did not appear to differ from the Handled Control offspring (p = .14), but in the original experiment that included just a smaller group of saline-injected controls, the infants from ACTH pregnancies were found to exhibit significantly poorer coordination and immature muscle tonus (Schneider et al., 1992).

The Orientation factor also revealed a significant overall effect of pregnancy conditions, F(6,395) = 12.17, p <.0001. This effect on Orientation was relatively selective, mediated primarily by the prenatal alcohol exposure and by the ACTH-stimulation of the gravid female. The planned contrasts indicated that the alcohol-exposed animals had the lowest Orientation scores, significantly below both the Undisturbed Controls and Handled Controls (p’s <.0001). In addition, there was also a moderate impairment in Orientation evident for infants from the ACTH-stimulated pregnancies relative to both the Unhandled and Handled Controls (p = .005 and p = .013, respectively).

Finally, there was an overall main effect of Pregnancy Condition on the fourth factor, Sensory Sensitivity, F(6,398) = 14.38, p <.0001. This impact was again clear for the alcohol-exposed and ACTH-stimulated pregnancies, both of which differed from Undisturbed Controls and Handled Controls (p’s = .0001). However, infants from three of the other experimentally manipulated conditions also differed from the controls. The Prenatal Stress manipulation resulted in reduced Sensory Sensitivity scores relative to Undisturbed and Handled controls (p = .0001 and p = .025, respectively). Similarly infants from Dex-treated pregnancies had scores below the Undisturbed Controls (p = .042). For this subscale, it is also noteworthy that many infants from one condition, Flu/LPS pregnancies, were found to have higher Sensory Sensitivity ratings, affected in the opposite direction from the other groups. Their scores were above both Undisturbed Controls and Handled Controls (p = .035 and .049, respectively).

Thus, the prenatal effects were not just global, but rather specific changes associated with several pregnancy manipulations. It is also important to emphasize that the acute episodes of just moderate disturbance during pregnancy did not result in overt changes in neonatal behavior. The scores for the Handling Disturbance controls were mostly similar to the totally Undisturbed Control infants.

Sex Differences

For two factors, there was either an overall difference between males and females or the nature of the prenatal effect was moderated by gender. The most notable difference was found for the State Control factor, where there was a main effect of infant sex, F(1, 395) = 6.88, p <.009. Female infants, in general, were more emotionally reactive during the IBAS test than males (Fig. 2A). However, this difference between males and females did not change the effect of the pregnancy manipulations; the Condition × Sex interaction term was not significant (p >.50). In contrast, while male and female infants did not differ overall in their Motor Activity, this one factor revealed a differential influence of several pregnancy conditions (Fig. 2B). The Condition × Sex interaction, F(6,395) = 2.20, p = .042, was driven in part by a differential impact of prenatal alcohol exposure on males and females. Conversely, only males in the Flu/LPS condition had lower and less mature Motor Activity scores like those seen after Maternal Stress, ACTH, and Dex, whereas the female infants from the FLU/LPS condition actually had scores significantly above the Handled Controls.

FIGURE 2.

Differences between female and male infants on the IBAS test. Panel A: In general, females were more reactive than males, resulting in significantly higher scores on the State Control factor (left panel). Panel B: For several pregnancy conditions, the effect on Motor Activity was manifest differentially in male and female infants (right panel). Male infants appeared to be more affected by prenatal alcohol exposure. After the prenatal infection paradigms, female infants evinced higher and male infants showed lower Motor Activity scores. Asterisk (*) indicates significant difference from same-sex infants in the Undisturbed Control condition; dot (˙) indicates a difference from same-sex infants generated from the Handling Disturbance control pregnancies.

Similar ratings for the Orientation factor were found for male and female infants across pregnancy conditions. In addition, males and females did not differ systematically in their Sensory Sensitivity scores, nor were the effects of the prenatal manipulations on this factor manifest differently by either sex (the Condition × Sex interaction was not significant, p >.28).

DISCUSSION

These analyses on a large number of infants generated under rigorously controlled conditions have validated that the IBAS is a sensitive instrument to assess the behavioral status and maturity of the young rhesus monkey. The general methodology, which was translated from the NBAS used in humans, and standardized originally for nursery-reared monkeys, is also appropriate for the mother-reared infant (Schneider et al., 1991). While applied most commonly in rhesus monkeys, the IBAS has also been used with some other primate species, including the squirrel monkey (Schneider & Coe, 1993). However, the new factor analyses on this much larger population of infants modified the subscale breakdown in two important ways. Whereas the original factors included two subscales related to neuromotor responses (Motor Activity and Motor Maturity), the current analyses suggest that the test items cohere better as one factor related to Motor development. In addition, a new factor emerged, which has been labeled Sensory Sensitivity, because it included several elements that involve responses to tactile stimulation or vestibular reactions to movement of the whole body (e.g., response to rotation or rapid descent). This factor may be a useful addition to capture a propensity for sensory processing disorders, which are frequently characterized by an atypical hypo- or hyper-responsiveness to nonnoxious sensory stimulation (Ayres & Robbins, 1979; Schneider et al., 2009; Streissguth, Barr, & Martin, 1983).

Examination of infants from seven different pregnancy conditions also permitted us to quantitatively and qualitatively distinguish the impact of several manipulations on the infants’ behavioral responses and emotional reactivity. When considered in this manner, the most distinctive animals were the ones exposed prenatally to alcohol. Many effects of maternal alcohol consumption have been described previously (Guerri, Bazinet, & Riley, 2009; Jones & Smith, 1973), but it should be reiterated that the current effects were induced by only moderate drinking. The consumption was voluntary, equivalent to 1–2 drinks per day, and for a subset of the animals spanned just the first two months of pregnancy (Schneider et al., 1997). None of the infants evinced dysmorphic craniofacial features, which are the hallmark of the full-blown fetal alcohol syndrome (Clarren & Smith, 1978; Clarren, Astley, & Bowden, 1988; Sampson, Streissguth, & Bookstein, 1997). These monkeys have now continued to be assessed into adulthood, when long-lasting effects on their dopaminergic neurotransmitter systems were documented with PET scans (Schneider et al., 2005). The second pregnancy condition that resulted in a large impact on the infants involved ACTH administration to the gravid female. Determination of maternal cortisol levels at 4 and 24 h indicated that the female’s adrenal hormone release was maximally stimulated for more than 4 h each day (Schneider et al., 1992). Thus, the developing fetus would have been exposed to abnormally high amounts of cortisol via placental transfer every day for 2 weeks (Campbell & Murphy, 1977; Gitau, Cameron, Fisk, & Glover, 1998). There is an extensive literature in rodents, sheep, and primates documenting that elevated levels of corticosteroid hormones in the fetal compartment can slow growth, impair the normal maturation of the brain, and perturb many other physiological systems (Langley-Evans, 1997; Milkovic et al., 1974; Wadhwa et al., 2004; Williams, Hennessy, & Davis, 1995; Wintour et al., 2003). We have also previously reported that there are immune alterations in infant monkeys generated from an ACTH-stimulated pregnancy (Coe, Lubach, Karaszeski, & Ershler, 1996).

In this regard, it was of considerable interest that the antenatal corticosteroid treatment did not produce similar results. For many individual test items, the Dex-treated infants actually appeared the most quiescent, showing less motor activity and sensory sensitivity than both the Undisturbed Control and Handled Control infants. This differential response and the fact that the effects did not track the broader action of the ACTH-stimulation condition on Orientation responses is probably due to the low doses of Dex and betamethasone employed, which were administered only as a 2-day course. The particular protocol was chosen to mirror clinical treatments used commonly in obstetrical practice and in the Neonatal Intensive Care Unit (American Committee on Obstetrics and Gynecology, 2008; Rajadurai, 2003). However, even with short and low dose regimens, there has continued to be some lingering concern about the antenatal use of corticosteroids, despite the many consensus reports and known benefits for stimulating lung maturation and preventing intraventricular hemorrhage (Liggins, 1969; Liggins & Howie, 1972; Matthews, 2000). In the monkey, at least, low doses for a short duration do not appear to cause major health or developmental problems, although one can discern some lasting physiological effects, such as a reduced postnatal sensitivity of lymphocytes to cortisol (Coe & Lubach, 2005). At higher doses (10 times above that used in clinical practice) or when administered for longer durations (2 weeks or more), many studies have documented that there can be serious teratogenic effects in both primates and other animal species (French, Hagan, Evans, Godfrey, & Newnham, 1999; Uno et al., 1994).

When compared to both Alcohol-exposure and ACTH-stimulation, the adverse effects of maternal stress on the IBAS ratings may appear to be relatively small. But it should be emphasized that we and others have demonstrated that there can be many other effects of maternal distress on infant behavior and neuroendocrine and immune physiology (Kaiser & Sachser, 2009; Koehl et al., 1999; Lemaire, Koehl, Le Moal, & Abrous, 2000; Welberg & Seckl, 2001). This maternal arousal paradigm in the rhesus monkey has been shown to reduce hippocampal volume, decrease neurogenesis in the dentate gyrus, increase the emotional stress reactivity of the older infant, and to affect many immune responses across the first year of life (Coe & Lubach, 2008; Coe et al., 2007). Thus, while a moderately impaired performance on two factors in the IBAS at 2 weeks of age may seem to be of limited consequence, it presages and is associated with more significant deficits that continue to be evident into adulthood. These findings in the monkey concur with concerns that have been expressed about the vulnerability of human infants when women experience stressful major life events during pregnancy (Cokkinides, Coker, Sanderson, Addy, & Bethea, 1999; Glynn, Wadhwa, Dunkel-Schetter, & Sandman, 2001; LaPlante, Brunet, Schmitz, Ciampi, & King, 2008; Yehuda et al., 2005). The results highlight a clear societal need for responsive prenatal services after environmental disasters, during periods of social upheaval and war, and when pregnant women suffer from domestic violence (Gazmararian et al., 1996; Ravelli, van der Meulen, Michels, & Osmond, 1998).

With respect to the particular issue of delivery and birth complications, it should be emphasized that despite the altered state of many of these monkey infants, all were full-term and delivered vaginally. Both prematurity and caesarian-section delivery were exclusion criteria, and thus all pregnancies were of normal gestational length. Each treatment was carefully calibrated to preclude premature birth and significant harm to the fetus. For example, in the case of the influenza infection, a mild, self-limiting strain of virus was selected. It was already known that more virulent strains can be life threatening, resulting in fetal loss and even maternal death if a secondary bacterial pneumonia follows the influenza infection (Barry et al., 2006; Dammann & Leviton, 1997; Harris, 1919). At 2 weeks of age, the behavior of the infants from these virus-exposed pregnancies appeared mostly normal. Subsequent neuroimaging later at 1 year of age revealed that the maturation of their cerebral cortex had been impacted by this prenatal infection, resulting in reduced gray matter and white matter in many hemispheric regions (Short et al., 2010). Similarly, the endotoxin-exposure condition also appeared to have minimal behavioral effects at first, but these infant monkeys became more stress-responsive over time and MRI scans revealed that the size of their cortical hemispheres was also different (Willette et al., in preparation). These examples of a developmentally delayed or latent effect would seem to concur with the situation frequently found in humans where the vulnerability to psychopathology, such as schizophrenia, first becomes unmasked and evident later in adolescence after puberty (Brown, 2006).

At this young age, the overall behavior and appearance of male and female rhesus monkeys appeared quite similar. In general, there were not major sex differences in how they performed on the IBAS. However, it is of interest that the female infants were on average more emotionally reactive and thus had higher scores on the State Control factor. In contrast, the male infants appeared to be more affected by several of the pregnancy manipulations, but only significantly so on the Motor Activity factor. Many other investigators have also found differential effects of prenatal insults and challenges on male and female offspring (e.g., Weinberg, Sliwowska, Lan, & Hellemans, 2008). While it is often said that male infants are more vulnerable (Geschwind & Galaburda, 1985), a closer examination of the findings indicates that it depends upon the experimental condition as well as the animal species (Morse, Beard, Azar, & Jones, 1999).

Assessment of the behavioral status and reactions of the young infant soon after birth serves two purposes for the basic scientist and in clinical practice. The first is to provide an indication of the normalcy of the infant’s prior development in utero and the quality of the pregnancy. As evident in our analyses of the young monkey, even when full term and in the normal weight range, it is possible to detect a lingering influence of perturbations that occurred months earlier during the prenatal period. Some abnormalities are overt enough to denote a major teratogenic insult and will likely be associated with a derailment of postnatal behavioral development and physical health (Barker, 1998; Bateson et al., 2004; Rees, Harding, & Walker, 2008; Weinstock, 1997). Other sensory or motor differences are subtle and reflect a more minor transformative influence that will remain within the tolerable realm of individual differences in temperament and reactivity, but can still affect stress responsiveness and later cognitive performance, (Sackett et al., 2006). The rearing environment may substantiate these initial propensities and vulnerabilities: reactive or behavioral inhibited infants can show an increasingly dysregulated physiology and poorer emotion regulation as they mature (Weinstein & Capitanio, 2008). Conversely, there is ample evidence to indicate that nurturing and supportive environments can move the developmental trajectory back toward a more benevolent course (Weinberg, Kim, & Yu, 1995). The NBAS for humans and the IBAS for the monkey provide invaluable assessment tools for characterizing the status of the young infant at this critical transition point between prenatal and postnatal life.