Maternal childhood adversity associates with frontoamygdala connectivity in neonates

Abstract

Background:

It is well established that exposure to adversity, especially during sensitive periods of development such as childhood, has both behavioral (e.g., increasing one’s risk for psychiatric illnesses) and neurobiological consequences. But could these effects of early life exposure to adversity also be transmitted across generations? Here we directly address this question, investigating the associations between maternal exposure to adversity during her own childhood and neural connectivity in her neonate.

Methods:

A sample of Black mother-neonate dyads (n=48) – a group that is disproportionately affected by early life adversity – completed questionnaires assessing their current distress (i.e., a composite measure of anxiety, depression, and perceived stress) during the first and third trimesters of pregnancy and retrospectively reported on their own childhood experiences of abuse and neglect. At one-month postpartum, neonatal offspring of these women completed a resting-state fMRI scan during natural sleep.

Results:

Greater maternal exposure to emotional neglect during her own childhood correlated with stronger functional connectivity of two different frontoamygdala circuits in these neonates, as early as one month after birth. This effect was specific to early experiences of emotional neglect and was not explained by maternal exposure to other forms of childhood maltreatment or maternal distress during pregnancy.

Conclusions:

These results provide novel evidence that the absence of emotional support early in a mother’s life, and years before conception, are associated with neural changes – namely, in functional connectivity between amygdala and medial prefrontal regions – in her offspring shortly after birth.

Early life adversity is linked to a number of negative health outcomes in one’s own life, including increased risk for neuropsychiatric illnesses such as depression and anxiety (1). This risk is conferred in part via adversity-induced alterations in the hypothalamic pituitary adrenal (HPA) axis, one of the body’s main stress response systems. Although some of these alterations are adaptive in the context of an acute stressor, persistent HPA axis alterations can have detrimental effects long term, including increased risk for psychiatric illness (2). Moreover, HPA axis alterations may have particularly potent effects if they occur early in life, for instance as a consequence of exposure to adverse experiences or parental deprivation during childhood (3).

Recent behavioral work suggests that adversity-associated risk may additionally be transmitted across generations, even if the stress occurs years before a child is born (4). For example, maternal experiences of physical and sexual abuse during her own childhood increase her child’s risk for anxiety (5). In addition, maternal exposure to adversity early in life further predicts altered HPA axis functioning both in abused mothers and in their infants (6). Together, this work suggests that adversity occurring years before a child is born can influence that child’s neuropsychiatric risk from very early in life. Thus, to create effective preventative interventions, it is of the utmost importance to understand how adversity becomes biologically embedded to increase risk for neuropsychiatric disorders across generations.

The fetal stage of development has been proposed as a time when the intergenerational transmission of early life adversity in the mother may be particularly likely to occur (7,8). Early life adversity may lead to long-term alterations in an individual’s HPA axis, immune functioning, and epigenome, all of which may have cascading effects on the eventual uterine environment in which a fetus develops during gestation. These biological alterations may lead to increased fetal exposure to circulating glucocorticoids, the final hormonal output of the HPA axis. In turn, fetal brain development, which occurs rapidly across gestation, may be impacted, particularly in regions that are rich in glucocorticoid receptors (e.g., the amygdala and medial prefrontal cortex). Although this theoretical model clearly identifies the intrauterine environment as a developmental stage when early life adversity in mothers may influence offspring development, empirical work is needed to provide support for this model of transmission in humans and to identify the biological sequelae of inheriting maternal adversity.

Of particular importance, it has been difficult to tease apart the influence of adversity that occurred during a mother’s childhood from the effects of prenatal adversity or adversity that occurs during the child’s lifetime. Preventative interventions depend upon our understanding of how timing and type of maternal stress exposure impact child risk; similarly, it is necessary to determine how early in development we can observe the effects of maternal adversity on child stress regulation and emotional development given that these are early precursors of neuropsychiatric health (9). Fortunately, recent technological advances in functional magnetic resonance imaging (fMRI), such as resting-state fMRI (rsfMRI) can aid this effort by enabling us to noninvasively examine brain functioning very early in life. The use of this methodology with neonates who have limited postnatal exposure to their mothers additionally allows us to better parse apart the respective roles of maternal preconception, prenatal, and postnatal stress in shaping child development.

At least one recent study found that maltreatment from a mother’s childhood correlated with global volumetric differences in the brains of newborn babies, even after controlling for prenatal experiences of stress and depression (10). This work raises the possibility that experiencing adversity during a sensitive period of development (i.e., a mother’s own childhood) could have potent intergenerational effects that are not fully explained by continued stress into the perinatal period or by the child’s postnatal exposure to maltreatment-induced parenting alterations. Additional studies are needed to replicate this effect and, importantly, to determine whether maternal early life adversity impacts brain function into the next generation, especially since certain disorders may result from impaired functional circuitry rather than gross volumetric differences (11). It is further necessary to determine whether any observable effects are driven by particular types of childhood adversity (e.g., neglect versus abuse) in order to better understand potential moderators of risk transmission.

Here, using a prospective longitudinal design and rsfMRI in neonates as young as one-month old, we asked just these questions, examining whether and how different maternal experiences of adversity during childhood yield intergenerational effects on neonatal frontoamygdala connectivity, a circuit that is often disrupted in the context of depression and anxiety. Specifically, we examined amygdala connectivity with two distinct regions within the mPFC – the dorsal anterior cingulate cortex (dACC) and the ventromedial prefrontal cortex (vmPFC). We selected these regions for two primary reasons. First, they share differential connections to the amygdala (12) and may be differentially impacted by maternal prenatal distress (13). For instance, higher levels of maternal prenatal depression have been linked to stronger amygdala-dACC connectivity, but weaker amygdala-vmPFC connectivity in sleeping neonates (13). Second, these mPFC regions are functionally distinct (14) and likely support different aspects of stress processing. As an example, functional connectivity between the amygdala and dACC increases during fear acquisition in adults (15), while enhanced amygdala-vmPFC connectivity is linked to inhibited fear learning (16). Whether maternal childhood adversity differentially associates with this neural circuitry in neonates has not yet been tested – the very question asked here.

Methods and Materials

Participants.

Mothers were recruited from an ongoing longitudinal study that follows Black women through pregnancy (R01NR014800) (17) and across the first 18 months postpartum (1R01MD009746) (18). Women were recruited for the larger ongoing study during the first trimester of pregnancy (M=11.5 weeks, SD=2.5) from a public hospital and a private hospital in the Atlanta area, resulting in a socioeconomically diverse sample (see Table 1). A majority of women (>85%) additionally completed a second prenatal visit during the late second/early third trimester (M=26.4 weeks, SD=2.7). Forty-eight mother-neonate dyads (n=27 female infants) enrolled in the present study, which involved the neonate completing a 30-minute MRI scan at approximately 1 month postpartum (M=40 days, SD=15). Seven enrolled dyads were excluded from final analyses due to unusable data (see Figure 1), resulting in a final sample size of 41 neonates. As shown in Table 1, there were no differences in demographics or on measures of maternal adversity between dyads who were and were not included in final analyses.

Table 1.

Demographics of sample and differences between included and excluded dyads.

Variable	Included in final sample (n=41)	Excluded+ (n=7)
Demographics
Gestational age at birth, M (SD)	38.7 (1.4) weeks	38.5 (2.1)
Preterm, N (%)	3 (7.3) yes	1 (14.3)
Infant gestational age at scan, M (SD)	44.5 (2.8) weeks	44.2 (0.9)
Infant age at scan, M (SD)	41.2 (17.1) days	40.9 (15.8)
Infant sex, N (%)	23 (56.1) female	5 (71.4)
Cohabitating with partner, N (%)	16 (39.0) cohabitating	2 (28.6)
Mom ethnicity, N (%)	41 (100) Black	7 (100)
Mom education, N (%)	13 (31.7) some college or more	2 (28.6)
Insurance type, N (%)	8 (19.5) low-income Medicaid	0 (0)
Maternal Childhood Adversity
CTQ emotional neglect, M (SD)	9.9 (5.5)	9.4 (5.2)
CTQ physical neglect, M (SD)	8.2 (3.6)	6.4 (2.7)
CTQ sexual abuse, M (SD)	6.9 (4.3)	11.7 (9.2)
CTQ physical abuse, M (SD)	6.8 (2.8)	7.4 (1.8)
CTQ emotional abuse, M (SD)	6.8 (4.2)	9.9 (6.9)
CTQ overall score, M (SD	38.9 (15.6)	44.9 (23.4)
CTQ any moderate to severe abuse or neglect, N (%)	20 (48.8) yes	3 (42.9)
CTQ any sexual, physical, or emotional abuse, N (%)	15 (36.6) yes	4 (57.1)
Maternal Prenatal Distress
1st Trimester EDS, M (SD)	7.1 (5.8)	5.6 (5.4)
1st Trimester PSS, M (SD)	23.0 (7.8)	19.4 (6.2)
1st Trimester STAI, M (SD)	33.4 (11.1)	30.3 (10.8)
3rd Trimester EDS, M (SD)	7.6 (5.6)	6.2 (8.7)
3rd Trimester PSS, M (SD)	22.9 (7.0)	21.6 (10.7)
3rd Trimester STAI, M (SD)	34.3 (10.7)	33.2 (11.3)

Note. There were no significant differences between mother-infant dyads who were and were not included in the final analyses. Dyads were excluded from the final sample because they did not sleep through the MRI scan (n=2), had too much motion during the scan (>1mm FD, n=2), data was lost due to technical malfunction (n=1), or there was an incidental finding that indicated non-normative neural development (n=2). CTQ=Childhood Trauma Questionnaire. EDS=Edinburgh Depression Scale. PSS=Perceived Stress Scale. STAI=State-Trait Anxiety Inventory. Preterm birth defined as less than 37 weeks gestational age.

Figure 1.

Recruitment flow chart.

Measures.

Maternal Childhood Adversity.

Mothers’ experiences of childhood adversity were assessed via retrospective report during the first trimester of pregnancy using the Childhood Trauma Questionnaire-Short Form (CTQ) (19). The CTQ is comprised of five subscales: Sexual Abuse, Physical Abuse, Emotional Abuse, Physical Neglect, and Emotional Neglect. High scores are indicative of more severe neglect and abuse.

Nearly forty-five percent (n=21) of women in the sample experienced at least one form of moderate to severe abuse or neglect during childhood, which is consistent with national prevalence rates of childhood trauma exposure among Black women (20). Although experiences of neglect were common in this sample, few women reported childhood physical, emotional, or sexual abuse, resulting in restricted range for these variables (see Supplemental Figure 1). As such, these variables were dichotomized and combined into a single variable that reflected whether a mother was the victim of any sexual, physical, or emotional abuse as a child (21). In contrast, physical neglect and emotional neglect showed a broad range in our sample and were therefore used in their continuous form to examine potential dosage effects.

Maternal Prenatal Stress.

Maternal prenatal distress was measured via multiple self-report measures that were completed during the first trimester of pregnancy and again during the third trimester. The Perceived Stress Scale (PSS) assesses the degree of stress an individual perceives in their current life, and has demonstrated construct validity (22–24). Mothers also completed the Spielberger State Trait Anxiety Inventory (STAI), a 20-item measure that assesses state and trait-like anxiety and stress (25,26). Finally, mothers completed the Edinburgh Depression Scale (EDS) to assess current depressive symptoms (27). A composite prenatal distress variable was created from these three measures of prenatal adversity by standardizing and averaging the total score of the PSS, STAI, and EDS. The creation of this composite score was based on a principle components analysis completed with the full prenatal study cohort (n>500) (17) and visual inspection of correlations between the PSS, EDS, and STAI (Table 2). A number of our infants (n=33) were not included in the PCA, so we validated our composite score in the larger prenatal cohort by correlating our composite score with the composite scores produced by the PCA. The two composite scores were highly correlated (r>0.99).

Table 2.

Intercorrelations between primary measures of adversity.

	Childhood Maternal Adversity		1st Trimester Maternal Distress				3rd Trimester Maternal Distress
	Emo. Neg.	Phys. Neg.	PSS	STAI	EDS	Composite Distress	PSS	STAI	EDS
Maternal childhood adversity
Phys. Neg.	r=0.69, p<0.001	--	--	--	--	--	--	--	--
1st trimester maternal prenatal distress
PSS	r=0.50, p=0.01	r=0.39, p=0.01	--	--	--	--	--	--	--
STAI	r=0.74, p<0.001	r=0.58, p<0.001	r=0.55, p<0.001	--	--	--	--	--	--
EDS	r=0.52, p=0.001	r=0.42, p=0.008	r=0.64, p<0.001	r=0.59, p<0.001	--	--	--	--	--
Comp Stress	r=0.69, p<0.001	r=0.54, p<0.001	r=0.86, p<0.001	r=0.83, p<0.001	r=0.89, p<0.001	--	--	--	--
3rd trimester maternal prenatal distress
PSS	r=0.29, p=0.08	r=0.35, p=0.03	r=0.63, p<0.001	r=0.55, p<0.001	r=0.48, p=0.002	r=0.64, p<0.001	--	--	--
STAI	r=0.53, p=0.001	r=0.51, p<0.01	r=0.51, p=0.001	r=0.74, p<0.001	r=0.60, p<0.001	r=0.70, p<0.001	r=0.66, p<0.001	--	--
EDS	r=0.24, p=0.14	r=0.44, p=0.003	r=0.45, p=0.004	r=0.55, p<0.001	r=0.69, p<0.001	r=0.64, p<0.001	r=0.58, p<0.001	r=0.66, p<0.001	--
Comp Stress	r=0.41, p=0.01	r=0.50, p=0.01	r=0.61, p<0.001	r=0.71, p<0.001	r=0.68, p<0.001	r=0.76, p<0.001	r=0.85, p<0.001	r=0.89, p<0.001	r=0.86, p<0.001

Note. All measures of maternal prenatal distress (i.e., the PSS, STAI, and EDS) were positively correlated with each other in our final sample (n=41), supporting the creation of two composite measures of maternal prenatal distress: prenatal distress during the first trimester and prenatal distress during the third trimester. CTQ=Childhood Trauma Questionnaire. PSS=Perceived Stress Scale. STAI=State Trait Anxiety Inventory. EDS=Edinburgh Depression Scale. Comp. Distress=Composite prenatal distress (i.e., the average of PSS, STAI, and EDS standardized scores).

Infant MRI.

All scanning procedures were completed by two trained research assistants and an MRI technician (see Supplemental methods). While the infant was in natural sleep, the following scans were collected using a standard 32-channel head matrix coil: at least one echo planar imaging (EPI) resting state scan (TR=1000ms, TE=31.4, acquisition matrix=72 × 72 × 39, and voxel size=2.5 × 2.5 × 2.5 mm³, 300 volumes), a T1 structural scan (TR=1900ms, TE = 2.28ms, 192 slices, FOV=192, voxel size=1×1×1 mm³ ), and/or a T2w structural scan (TR=2000ms, TE = 376ms, 176 slices, FOV=192, voxel size=1×1×1 mm³).

Preprocessing and data analysis.

FSL (5.0.11) was used to analyze functional data. Preprocessing included motion correction, detrending, slice time correction, intensity normalization, and spatial smoothing using an 8mm FWHM Gaussian kernel. The first 20 volumes were discarded to account for scanner start, and images were bandpass filtered (0.01–0.08 Hz) to retain low-frequency signal. Signal from six motion parameters, signal from a 3mm sphere surrounding a voxel that was manually placed within the ventricle, signal from a 3mm white matter sphere, and the mean global signal were included as nuisance regressors.

Resting-state correlation.

Resting-state functional correlations were operationally defined as the correlation of the time-varying BOLD signal between two ROIs. After preprocessing, the continuous time series for each voxel within an ROI was extracted and averaged together to create an average time series. The average time series for one ROI was then correlated with the average time series from another ROI. The resulting correlation coefficient (r) was transformed to Gaussian-distributed z-scores using Fishers transformation, and these z-scores were used as our measure of functional connectivity (FC) in further analyses (28,29). We examined ipsilateral left and right hemispheric connections separately to determine whether any associations with maternal adversity were lateralized, given that previous research has found stronger associations between maternal stress and left hemispheric alterations in offspring (13,30). Exploratory contralateral analyses are included in Supplemental Materials.

Data analysis plan.

Descriptives and measures of variability were examined for all variables. Physical neglect and emotional neglect subscales from the CTQ were log-transformed to correct for positive skew (physical skew=0.76; emotional skew=1.14), which was effective at improving distribution normality (log-transformed physical skew=0.43; emotional=0.63). We also repeated our analyses using the non-transformed neglect variables and our results were unchanged. Hierarchical linear regressions were used to test study hypotheses, with relevant covariates (see Supplemental Methods) entered into the first step, and the primary predictor variable entered in the second step. Cohen’s f² was calculated as an additional measure of effect size (31), with cutoffs for small (f²=0.02), medium (f²=0.15), and large (f²=0.35) effect sizes (32). Linear regression assumptions were assessed in several ways. Unstandardized residuals were visually examined using histograms to determine normality and residuals were plotted against predicted values to ensure homoscedasticity. Finally, Cook’s D was used to identify potential outliers since it considers both leverage and discrepancy. Univariate ANCOVAs were used for analyses examining childhood abuse history as a predictor of neonatal connectivity.

Results

First, given that motion is a known confound of rsfMRI, we limited the effects of motion in several ways. In addition to correcting for motion during preprocessing as described above and in Supplemental Materials, we confirmed the number of frames included in analysis was not correlated with amygdala-dACC (left: r=−0.13, p=0.40; right: r=0.03, p=0.86) or amygdala-vmPFC connectivity (left: r=−0.12, p=0.45; right: r<0.01, p=0.99). We additionally confirmed that frame to frame motion displacement was not associated with amygdala-dACC FC (left: r=−0.22, p=0.16; right: r=0.13, p=0.41) or amygdala-vmPFC FC (left: r=−0.01, p=0.95; right: r=0.02, p=0.92) in our final sample. Neonatal framewise displacement was also unrelated to our measures of maternal adversity (p’s>0.14).

Next, we examined our question of whether maternal experiences of neglect and abuse from her own childhood correlated with frontoamygdala circuitry in neonates (see Tables 3 and 4). After controlling for relevant covariates and maternal distress during the first trimester of pregnancy, emotional neglect from the mother’s childhood robustly correlated with amygdala-dACC FC in neonates (ΔR²=0.20, f²=0.25; see Table 3). That is, the more emotional neglect mothers experienced during their own childhood, the stronger the functional coupling was between the amygdala and dACC in her neonate across both the left and right hemispheres (Figure 2). Intriguingly, this association was only significant after controlling for maternal prenatal distress (see Supplemental Materials). Neither physical neglect nor abuse from the mother’s childhood were associated with amygdala-dACC FC in neonates after controlling for relevant covariates and maternal prenatal distress (Table 3).

Table 3.

Correlates of neonatal of ipsilateral connectivity between the amygdala and dACC.

Outcome: Left neonatal amygdala-dACC FC
	β	t	b 95% CI	p	R2	ΔF	ΔR2
Maternal Childhood Emotional Neglect	0.68	3.40	0.01, 0.04	0.002	0.53	11.57	0.20
Maternal Childhood Physical Neglect	0.14	0.62	−0.36, 0.67	0.54	0.40	0.39	0.01
Maternal Any Childhood Abuse	0.22	1.23	−0.07, 0.26	0.23	0.32	1.52	0.04
Outcome: Right neonatal amygdala-dACC FC
	β	t	b 95% CI	p	R2	ΔF	ΔR2
Maternal Childhood Emotional Neglect	0.46	2.24	0.001, 0.04	0.04	0.56	5.0	0.09
Maternal Childhood Physical Neglect	0.09	0.44	−0.41, 0.62	0.67	0.47	0.19	<0.01
Maternal Any Childhood Abuse	0.17	0.98	−0.09, 0.25	0.34	0.36	0.97	0.03

Note. We examined different types of maternal adversity as predictors of neonatal amygdala-dACC connectivity separately for ipsilateral connections within the left and right hemispheres. Mothers’ experiences of emotional neglect from her childhood correlated with stronger positive functional coupling of the amygdala and dACC in neonates on average one month after birth. This effect emerged only after controlling for maternal distress during the 1^st trimester of pregnancy. Results are similar when controlling for maternal distress during the 3^rd trimester instead of during the 1^st trimester. All analyses also control for number of included frames, framewise displacement, neonatal age at scan, neonate sex, neonate head circumference at scan, birth weight, gestational age at birth, maternal education, private insurance vs. low-income Medicaid, maternal age, cohabitation with partner, and number of people living in the home in the first step of the model.

Table 4.

Correlates of neonatal ipsilateral connectivity between the amygdala and vmPFC.

Outcome: Left neonatal amygdala-vmPFC FC
	β	t	b 95% CI	p	R2	ΔF	ΔR2
Maternal Childhood Emotional Neglect	0.49	2.35	0.002, 0.03	0.03	0.54	5.53	0.11
Maternal Childhood Physical Neglect	0.19	0.91	−0.22, 0.55	0.37	0.46	0.82	0.02
Maternal Any Childhood Abuse	0.21	1.19	−0.05, 0.20	0.25	0.36	1.41	0.04
Outcome: Right neonatal amygdala-vmPFC FC
	β	t	b 95% CI	p	R2	ΔF	ΔR2
Maternal Childhood Emotional Neglect	0.38	1.71	−0.003, 0.03	0.10	0.47	2.91	0.06
Maternal Childhood Physical Neglect	0.06	0.27	−0.42, 0.55	0.79	0.41	0.07	<0.01
Maternal Any Childhood Abuse	0.03	0.18	−0.14, 0.17	0.86	0.33	0.03	<0.01

Note. We examined predictors of neonatal amygdala-vmPFC connectivity separately for ipsilateral connections within the left and right hemispheres. Mothers’ experiences of emotional neglect from childhood correlated with stronger positive functional coupling of the left amygdala and left vmPFC in neonates one month after birth. This effect emerged only after controlling for maternal distress during the 1^st trimester of pregnancy. Results are similar when controlling for maternal distress during the 3^rd trimester instead of during the 1^st trimester. All analyses also control for number of included frames, framewise displacement, neonatal age at scan, neonate sex, neonate head circumference at scan, birth weight, gestational age at birth, maternal education, private insurance vs. low-income Medicaid, maternal age, cohabitation with partner, and number of people living in the home in the first step of the model.

Figure 2.

The functional correlation between the amygdala and dACC in each neonate was calculated by averaging the BOLD time series of each voxel within the ROI mask and correlating the average time series of the amygdala with the average time series of the dACC. The same procedure was used to calculate the functional correlation (FC) between the amygdala and vmPFC, between the dACC and vmPFC, and between the amygdala and IFG. Next, maternal experiences of adversity were examined as predictors of these FCs in 1-month-old sleeping neonates. (a) Emotional neglect from the mother’s childhood was robustly associated with stronger positive amygdala-dACC FC in neonates (b) and with stronger positive amygdala-vmPFC FC. (c) Maternal childhood emotional neglect did not associate with neonatal vmPFC-dACC FC (d) or with neonatal amygdala-IFG FC.

Similarly, being emotionally neglected during a mother’s own childhood explained 11 percent of the variance in left hemisphere amygdala-vmPFC FC in neonates above and beyond relevant covariates and maternal prenatal distress (f²=0.11; see Figure 2). Again, the effect of emotional neglect only emerged after controlling for maternal prenatal distress. A similar pattern of results emerged for right neonatal amygdala-vmPFC connectivity, but this association was only marginally significant (Table 4). Moreover, the effect on left amygdala-vmPFC FC was also specific to emotional neglect from the mother’s childhood. Maternal experiences of physical neglect did not associate with amygdala-vmPFC FC in neonates nor did experiencing abuse during childhood. These results suggest strong specificity for an intergenerational effect of childhood emotional neglect on frontoamygdala circuitry into the next generation (see also Supplemental Results). Controlling for maternal distress during the third trimester of pregnancy also yielded similar results across both of these frontoamygdala circuits.

Finally, we conducted post hoc analyses to determine the specificity of associations with frontoamygdala circuitry. For these control analyses, we examined whether experiences of emotional neglect from the mother’s childhood correlated with alterations in neonatal FC between our two frontal regions of interest, and between the amygdala and a more lateral prefrontal region, the inferior frontal gyrus (IFG). As shown in Figure 2, mothers’ childhood emotional neglect did not associate with neonatal dACC-vmPFC FC (left: β=−0.17, ΔR²=0.01, f²=0.01, p=0.54; right: β=0.02, ΔR²<0.01, f²<0.01, p=0.94) or with neonatal amygdala-IFG FC (left: β=0.15, ΔR²=0.01, f²=0.01, p=0.53; right: β=0.25, ΔR²=0.03, f²=0.03, p=0.22) after controlling for relevant covariates and first trimester prenatal distress. Together these results suggest the intergenerational impact of early life emotional neglect is specific to communication between the amygdala and mPFC, at least at this early stage of development.

Discussion

This study is the first to examine associations between maternal experiences of early life adversity and neonatal neural connectivity. We found that maternal childhood experiences of emotional neglect robustly correlate with stronger functional coupling between the amygdala and vmPFC and between the amygdala and dACC in their infants after controlling for maternal prenatal distress. These effects were particularly robust with respect to amygdala-dACC connectivity. Importantly, changes in neonatal frontoamygdala connectivity were specific to emotional neglect. These novel findings illustrate that certain experiences from a mother’s own childhood are associated with differences in the development of frontoamygdala circuitry in the next generation as early as one month after birth.

Notably, being emotionally neglected during mothers’ own childhood correlated with strengthened functional coupling of the amygdala with medial prefrontal regions in their newborns. This enhanced positive coupling is especially interesting given that connectivity between these regions increases in the context of fear learning (33) and following an acute stressor (15). Amygdala-mPFC connectivity also strengthens across childhood (34), so presuming the same developmental pattern generalizes to infancy, the present results are consistent with the stress acceleration hypothesis (35), which posits that exposure to emotional neglect (e.g., parental deprivation) and other forms of early life stress may lead to the early maturation of frontoamygdala circuitry. This accelerated development may predispose children to more readily detect threat or to self-regulate in an environment that lacks the buffering influence of an involved caregiver. However, this early acceleration may also come at the cost of decreasing neural plasticity (16) and increasing risk for certain neuropsychiatric disorders long-term. Indeed, descriptive research shows stronger resting-state FC between the amygdala and mPFC among individuals with anxiety disorders (12), which is consistent with the observed neural phenotype we found in neonates of emotionally neglected mothers. Our findings contribute to extant research by showing that childhood experiences of emotional neglect may also strengthen frontoamygdala FC into the next generation. However, more work is needed to determine whether the present results represent accelerated development, transient increases in frontoamygdala connectivity, or persistently high functional correlations between frontoamygdala regions.

Interestingly, emotional neglect from the mother’s childhood only correlated with strengthened amygdala-dACC and strengthened amygdala-vmPFC connectivity after, but not before, controlling for prenatal distress. This finding suggests the influence of maternal childhood emotional neglect is separate and distinct from that of prenatal distress, and from the reporting and recall differences that can accompany heightened distress. The emotional neglect subscale of the CTQ also contains some of the most subjective items (e.g., “I felt loved”). Recent research suggests that subjective reports of childhood maltreatment are 1) separable from concurrent neuropsychiatric illness and 2) more strongly linked to future neuropsychiatric symptomology compared to objective measures of maltreatment severity (e.g., court records) (36). It is possible the CTQ emotional neglect subscale best captures subjective maltreatment experiences, which is why it yielded the strongest intergenerational impacts in the present study. Alternatively, emotional neglect may be a particularly potent form of maltreatment that uniquely shapes early development, which aligns with work showing the long-lasting impacts of early parental deprivation on offspring brain development (37).

It is important to note that the behavioral consequences of strengthened connectivity between the amygdala and mPFC during the neonatal period remain unclear. Indeed, there is evidence that certain regions do not develop functional selectivity before six months of age, but instead are refined after months, or even years, of postnatal experience (38). Although the functions of many brain regions during the neonatal period remain unknown, examining resting-state FC patterns still offers important insight into the developing brain given that connectivity between regions may precede the development of functional selectivity (39). Moreover, neural connectivity patterns early in life can be used to predict what a specific region will become selective for in the future (40) as well as future emotional functioning in children (41). In sum, the inputs of a given region (i.e., what it is connected to) may drive its specialization later in development, making infant resting-state connectivity patterns a promising tool for understanding the neural underpinnings of neuropsychiatric illness and risk transmission.

The association between maternal childhood emotional neglect and infant frontoamygdala connectivity was not explained by maternal prenatal distress in the present study. This result was surprising given that at least one study has reported that maternal prenatal depression (which was included in our composite prenatal distress measure) modulates amygdala-dACC and amygdala-vmPFC connectivity in neonates (13). Although we used similar analytic techniques, sampling differences may explain our inability to replicate this effect. Our sample is comprised entirely of Black mother-neonate dyads which is rare among neuroscientific research (e.g., in (14) Black dyads made up less than 13% of the study sample). It is possible that early life adversity impacts neonatal frontoamygdala connectivity more strongly than prenatal adversity specifically for Black mother-neonate dyads. Perhaps the buffering influence of sensitive caregiving is especially important in the context of unique race-related stressors that Black youth experience in the US (42). Indeed, recent work suggests that Black individuals are more impacted by early life adversity compared to their White counterparts and that these differences persist into adulthood (43). Such differences speak to the importance of increasing sample diversity in developmental neuroscience, and are consistent with findings that suggest sample ethnicity and socioeconomic status influence study findings related to brain development (44); again, here ethnicity and socioeconomic status are differentially associated with adversity across the lifespan, with Black race and low socioeconomic status being associated with different types (45) and intensities of adversity exposure (46), as well as differences in the stress response beginning very early in life (47). The present study makes an important contribution to the field by examining brain development in an underrepresented sample who may be disproportionately exposed to and affected by adversity, and suggests the need for further empirical investigations regarding race specific impacts of maternal stress.

There are likely mediating processes that were not measured in the present study. Evidence suggests that early adversity may induce lasting epigenetic changes – such as alterations to DNA methylation, which in turn controls gene expression (48) – that can influence the uterine environment during pregnancy and shape offspring neurodevelopment. For example, increased methylation (and decreased expression) of the 11β-HSD-2 gene has been demonstrated in women exposed to early adversity (49). 11β-HSD-2 regulates fetal exposure to glucocorticoids by converting cortisol to its benign form (i.e., cortisone) as it crosses the placental barrier. Therefore, offspring of mothers with reduced 11β-HSD-2 expression are likely exposed to elevated levels of cortisol in utero, which has been associated with an increased risk for depression and anxiety across the lifespan (50). Elevated glucocorticoid levels have also been shown to mediate the association between early maternal separation (an extreme form of childhood emotional neglect) and increased frontoamygdala connectivity in children (51).

Notably, reduced 11β-HSD-2 expression is most strongly associated with women’s exposure to early versus later life adversity (49). This aligns with the finding from the present study that maternal experiences of emotional neglect during her childhood were more closely associated with offspring frontoamygdala connectivity compared to maternal distress during pregnancy. Together, it is possible that emotional neglect during childhood enacts epigenetic alterations that lead to elevated fetal glucocorticoid exposure, which in turn strengthens frontoamygdala circuitry in neonates.

Although we identified effects on child neural connectivity strikingly early in development, these neonates were exposed to one month of parenting before completing the fMRI scan. This first month of parenting behavior, although limited, could mediate the relationship between maternal childhood emotional neglect and offspring neural development as mothers who were emotionally neglected during their own childhood may engage in fewer behaviors that buffer their neonate from the effects of stress (52). For instance, these mothers may be slower to pick up their newborn infant when he/she cries, which offers the infant more opportunities to self-soothe. These repeated opportunities to self-soothe may contribute to the strengthened frontoamygdala connectivity observed in this study. Mothers who were emotionally neglected as children may also continue to lack emotional support during pregnancy and early in the postnatal period. This lack of emotional support may manifest as the infant’s father being less involved and less available to offer sensitive parenting in the manner described above. Indeed, in our sample, only 37% of women co-habitated with a partner.

The findings from this novel study suggest that childhood experiences of emotional neglect may have potent intergenerational effects on infant neural connectivity and especially on left hemispheric frontoamygdala circuitry. This finding advances our knowledge of factors that shape the development of emotion-related neural circuits and highlights that certain maternal experiences that occur years before conception may shape child development in impactful ways. It will therefore be important to include measures of maternal early life adversity in future research examining the effects of prenatal stress on child development in order to delineate cumulative, interactive, or competing effects. In turn, this will lay the groundwork for creating a comprehensive model that explains the intergenerational transmission of adversity and neuropsychiatric risk.