Introduction

At the beginning of March 2020, a multitude of public health measures were implemented in Canada to combat the spread of the severe acute respiratory syndrome coronavirus 2 (SARS‑CoV‑2), including changes to prenatal care [1,2,3]. Pregnant individuals were faced with uncertainties and challenges, leading to higher perinatal anxiety and depression symptoms during the pandemic compared to pre-pandemic levels [4,5,6,7]. This is concerning given that prenatal maternal stress (PNMS) and prenatal mental health is associated with negative health outcomes for both pregnant individuals and their future children [8, 9]. PNMS is linked to poorer neurodevelopment, cognitive development, temperament, and behavioral problems in childhood as well as negative birth outcomes [10, 11].

PNMS due to an adverse event can be broken down into two components: objective hardship (i.e., level of exposure to an external stressor) and a more subjective psychological component of perceived stress (e.g., worries regarding infection and its impact on the fetus). In addition to stress indicators, distress indicators also exist. Pregnant individuals are at risk for psychological distress, including prenatal anxiety and depression, especially in conjunction with stressful life experiences [12]. While prenatal anxiety and depression can often be comorbid with each other, they are also both distinct constructs [12, 13]. Depression measures the presence of sad, empty, or irritable mood, loss of interest in pleasurable activities, feelings of worthlessness, and thoughts of death. On the other hand, anxiety assesses an individual’s level of worries, nervousness, and trouble relaxing [13]. Anxiety and depression symptoms vary depending on the individual, leading to different trajectories: anxiety dominant, depression dominant, or anxiety and depression comorbidity [14]. Prenatal maternal stress, anxiety, and depression can all impact social-emotional development in infants and children. Prenatal anxiety and depression have been associated with both higher negative affectivity (i.e., sadness, fearfulness, distress) and difficult temperament (i.e., excessive crying, irritability, slow adaptability) in infants and toddlers [11, 15, 16]. However, both constructs may impact child development differently, with comorbid anxiety and depression potentially having the greatest effects [11, 14, 17, 18]. In the 1998 Quebec Ice storm study, higher prenatal subjective stress and maternal illness/infection (e.g., flu, fever, pre-eclampsia) were associated with negative aspects of temperament (e.g., fussy/difficult, dullness, needs attention) in 6-month infants [19]. Similarly, higher prenatal maternal objective hardship due to the 2011 Queensland floods was associated with lower problem solving skills, more difficult temperament (moderated by infant sex and gestational age at the time of the flood), and marginally associated with lower personal-social skills in infants at 6-months of age [20, 21]. Additionally, greater levels of both forms of PNMS have been associated with higher levels of internalizing, externalizing, and other psychiatric problems in childhood, mostly independent of postnatal mood [11, 22].

Stressful maternal experiences during pregnancy influence the uterine environment and alter the development of the fetal central nervous system, which can then “set probabilistic parameters for future brain and behavior development” in infancy and childhood [8]. Particularly, PNMS has been associated with differences in amygdala as well as prefrontal cortex (PFC) volumes in children [9]. Both the amygdala and PFC, as well as the white matter fibers connecting the two brain regions, play a key role in emotional regulation [23,24,25]. The amygdala is a bilateral structure of the limbic system. Psychological distress in mid-pregnancy was negatively related to left amygdala volumes in newborn males [26]. At the same time, higher levels of depression and maternal cortisol (a biomarker of physiological stress) in the second trimester have been associated larger right amygdala volumes in girls during childhood, where amygdala volumes partially mediated the effect of high cortisol levels on affective problems in girls [27, 28]. Furthermore, Jones et al., [29] found that larger amygdala volumes mediated the association between greater objective hardship due to the 1998 Quebec ice storm and externalizing symptoms in 11-year-old girls [29]. One study, however, found no association between prenatal depression and amygdala volumes in neonates [30]. Smaller amygdala volumes have been associated with greater levels of internalizing problems in children [31, 32], although other studies reported the opposite association [33, 34].

The PFC is a region of the cerebral cortex that is heavily involved in social-emotional and cognitive processes, which emerge in early infancy [35, 36]. Psychological distress has also been associated with reductions in gray matter volume of the frontal lobes in children [9]. For example, Buss et al. [37] found that higher pregnancy anxiety at 19 weeks of gestation was associated with gray matter volume reductions in the prefrontal cortex of 6–9 year old children, independent of postnatal stress [37]. Maternal depression at 25 gestational weeks was associated with cortical thinning of the entire cortex, and particularly the frontal lobes in children. Cortical thinning of the PFC mediated the association between prenatal depression and child externalizing behaviors [38]. However, by studying volume, we can gain a better understand of the impact of PNMS on the PFC more broadly. Volume is a more comprehensive measure of structural integrity than thickness, as it also takes into account surface area, thickness, gyrification, and overall size [39, 40]. At birth, cortical thickness is more developed than surface area. While cortical thickness reaches adult levels at around 1 year of age, surface area continues to grow and accounts for most of the cortical volume growth after 1 years old [41]. Furthermore, research has shown that PFC volume strongly correlates with differences in child and young adult outcomes, including executive functioning and mental health (e.g., depression, schizophrenia, suicidality) [42,43,44,45,46].

A variety of mechanisms have been proposed to explain the associations of PNMS with brain development and behavior, including maternal inflammation, altered placental functioning, and greater fetal cortisol exposure [8, 47]. Abnormal or inappropriate levels of maternal cortisol due to hypothalamic-pituitary-adrenal (HPA) axis activation caused by high stress can have a neurotoxic effect on the fetus’ developing brain [8, 9]. The amygdala in the fetus is rich with cortisol receptors and may be vulnerable to maternal activation of the HPA axis [48]. PNMS mid-pregnancy may have the strongest impact on infant neuroanatomy [26]. However, a positive postnatal environment, such as a sensitive mother-infant attachment, may alter neurodevelopmental changes associated with PNMS [9].

Recent studies have examined infant neurodevelopment during the COVID-19 pandemic. More severe pandemic-related prenatal maternal stress has been linked to greater negative affect and changes to amygdala-prefrontal functional connectivity in infancy [49, 50]. Greater prenatal maternal distress during the pandemic was associated with delayed social-emotional development in infants at 2-months [51]. Further, infants born during the pandemic have been shown to have higher levels of negative affectivity and a higher risk of communication and personal-social impairment compared to pre-pandemic [52,53,54].

It is of utmost importance to study the impact of PNMS on infant development. However, there is a dearth of studies examining the effect of pandemic-related PNMS and psychological distress on infant brain development. To fill this gap, we investigated the association between PNMS, psychological distress, infant temperament, and brain development, with a particular focus on the amygdala and PFC. Our study had two main objectives; first, to investigate the association among PNMS (prenatal objective hardship and perceived stress), mental health (prenatal depression and anxiety), and infant amygdala/PFC volumes. The second aim was to determine whether amygdala and PFC volumes are associated with temperament in infants.

We hypothesized that greater prenatal stress (objective hardship and perceived stress) and higher levels of maternal mental health symptomatology (depression and anxiety) would predict larger amygdala volumes and smaller PFC volumes. Secondly, we hypothesized that larger amygdala volumes and smaller PFC volumes will be associated with more challenging temperaments in infants.

Methods and materials

Procedure & participants

Data were collected as part of the pan-Canadian Pregnancy During the COVID-19 Pandemic (PdP) Study. The PdP study was a prospective longitudinal cohort study that included multiple follow-up surveys completed during the prenatal and postpartum period. Recruitment was conducted between April 5th 2020 and April 30th 2021. The initial online survey was advertised on social media platforms. To be eligible for the study, individuals must have been pregnant, ≥17 years of age, living in Canada, able to answer questions in English or French, and ≤35 weeks of gestation at recruitment [55].

The initial survey was completed by participants during pregnancy using REDCap (Research Electronic Data Capture; [56]). It collected their demographic, socioeconomic, and obstetric characteristics (e.g., age, ethnicity, household income, health before and during pregnancy), in addition to their mental health, pandemic-related hardships, and other measures. Participants were asked to complete a follow-up survey once a month for the first three months, then every other month during their prenatal period. In total, participants could have completed a maximum of five prenatal follow-up surveys, which assessed their experiences since the last survey. Once a participant gave birth, they reported on their pregnancy outcomes, such as their mode of delivery as well as their infant’s birthweight, length, and gestational age. During the postpartum period, participants were sent follow-up surveys when their child reached 3, 6, and 12 months of age.

The PdP study also included neuroimaging in the Calgary area. Infants were scanned at Alberta Children’s Hospital using magnetic resonance imaging (MRI) at 3 months of age. Only infants born at full-term (>36 weeks) were eligible for neuroimaging. Infants were excluded if they had a major birth complication or were diagnosed with any severe genetic or neurologic conditions. 506 families/infants were invited for imaging at 3 months. 140 scans were scheduled. 119 infants attempted imaging; however, 18 scans were unsuccessful as the infants woke up, were crying, and/or were not sleeping.

All pregnant individuals who participated in the study provided informed consent. All participants signed an electronic informed consent form before starting the first questionnaire; at the time of the MRI, parents provided consent for infants. The study procedures were conducted in alignment with the Declaration of Helsinki.

Participants who completed the prenatal surveys were entered in a monthly draw for a gift card. Individuals who also participated in the follow-up surveys and imaging part of study received a small stipend. The data were manually checked for bots and invalid responses. This project received ethical approval from the University of Calgary Conjoint Health Research Ethics Board (REB20-0500).

Measures

Objective hardship

The PdP study team developed the Pandemic Objective Hardship Index (POHI) to measure participants’ level of objective hardship due to the pandemic over the duration of pregnancy [57]. This measure has four subscales: Scope, Loss, Threat, and Change, each with a maximum score of 50 points. Scope refers to the duration and intensity of the hardship. Loss (i.e., financial, social or physical loss), Threat (i.e., physical and health-related consequences of the pandemic), and Change (i.e., changes to daily routines, prenatal care, work, and social interactions) were also measured. The sum of all four subscales creates the total score, with a maximum of 200 points.

Perceived stress

At final follow-up, the Perceived Stress Scale (PSS-10) was used to measure participants’ subjective level of psychological stress during the previous month. Participants are asked 10 questions regarding how often they appraised their life as “unpredictable, uncontrollable, and overloaded” during the first prenatal follow-up questionnaire [58]. Each item is scored from 0 (never) to 4 (often), with total scores ranging from 0–40. Higher scores indicate a greater level of perceived stress. The PSS-10 has high internal consistency (coefficient alpha reliability = 0.84–0.86) and validity. This scale is strongly correlated with other mental health measures, such as Beck Depression (r = 0.67) and Anxiety Inventory (r = 0.58) [58, 59].

Depression

At recruitment, the Edinburgh Postpartum Depression Scale (EPDS) measured prenatal depression symptoms over the past week [60]. The EPDS contains 10 items, each scored on a scale from 0–3. Total scores range from 0–30, where a higher score indicates a greater level of depressive symptoms. Individuals who score above the clinical cut-off score of ≥13 are at risk for depressive disorder. The EPDS has good reliability, with a split-half reliability of 0.88 and a standardized alpha coefficient of 0.87 [61, 62].

Anxiety

The Patient-Reported Outcomes Measurement Information System (PROMIS) Anxiety–Adult Short Form at recruitment measured participants’ general anxiety symptoms within the past week through a 7-item questionnaire [63]. Raw scores were converted to T-scores using the US general population norms. Possible T-score values ranged from 36.3–82.7 with a mean of 50 (SD = 10). T-scores between 60–69.9 indicate moderate anxiety levels, while scores ≥70 indicate severely elevated anxiety levels [64].

Infant temperament

Infant temperament was measured using the Infant Behavior Questionnaire–Revised Very Short Form (IBQ-R) at 6-months of age [65, 66]. There are 37 items total, each asking parents to report on specific infant behaviors over the past week. The IBQ-R assesses three dimensions of temperament: negative affectivity, positive affectivity/surgency, and regulatory capacity/orienting. Parents score each item on a 7-point Likert scale ranging from 0 (never) to 7 (always), where higher scores indicate higher levels of each dimension of temperament. This scale has strong psychometric properties, with good overall test-retest reliability (r = 0.54–0.93), interparent agreement averaging r = 0.41, and high internal consistency (Cronbach alpha > 0.70), with specific scale internal consistencies of 0.81 (negative emotionality), 0.80 (positive affectivity), and 0.74 (Orienting/Regulatory Capacity) [66].

Image acquisition & analysis

100 infants were scanned at 3 months of age using a GE 3T MR750w MRI with a 32-channel head coil at the Alberta Children’s Hospital to acquire brain imaging data. All infants were scanned while asleep atop an inflatable MedVac infant scanning bed. T1-weighted images were obtained (repetition time = 5200 ms, echo time = 2200 ms, inversion time = 540 ms, field of view = 1900 mm, matrix = 512 × 512, bandwidth = 41.67, voxel 1 × 1 × 1 mm3, flip angle = 12°, 136 slices, total time = 3:32). The brain images were rotated according to the standard atlas and subsequently segmented using infant Freesurfer (https://surfer.nmr.mgh.harvard.edu/). This program uses the intensity of each voxel to estimate the probability of whether it belongs to a particular brain structure [67]. 97 brain scans were successfully processed by FreeSurfer. One participant’s MRI data was removed from the study due to extremely high levels of motion in the image, which was determined through visual inspection. No quantitative motion metrics were used. All amygdala segmentations were visually inspected for segmentation quality by a trained expert. Amygdala volumes that were identified as outliers, where their masks were smaller or larger than they should have been, were subsequently edited for segmentation errors. If FreeSurfer over/underestimated the boundaries of the amygdala, the edges of the masks were outlined or erased using manual segmentation tools. The masks were reviewed from axial, sagittal, and coronal views to ensure that they did not overlap with other structures. The PFC volumes were manually segmented using ITK-SNAP. The cross-hairs were aligned by the corresponding section that, in a brain aligned in Talairach space, would be y = 26, as described by Rajkowska & Goldman-Rakic [68]. Only certain brain regions were selected to be visible (e.g., frontal pole, caudal middle frontal, lateral/medial orbitofrontal, pars opercularis/triangularis/orbitalis). The area anterior to the cross-hairs were selected as the PFC volume. Two PFC segmentations were removed from the study due to extremely large holes (missing sections) in their segmentations.

Covariates

Covariates included maternal education (i.e., highest education obtained), as well as infant sex, gestational age at birth, and total cerebral volume [TCV]. Age at scan and ethnicity were added as covariates to models where these variables were correlated with the dependent variable. Ethnicity is an important covariate to include in temperament models. Research has shown that temperament varies between cultures and countries. For example, negative affectivity is reportedly higher in Asian and South American countries, who often value collectivism, compared to European countries [69].

Data analysis

Statistical analyses were conducted using IBM SPSS (v26, Statistics for the Social Sciences) and the PROCESS Macro [70].

To address our first aim, generalized linear models (GLM), using identity link functions, were used to determine the association between PNMS (objective hardship and perceived stress) and mental health (anxiety and depression) during pregnancy and infant brain volumes (left/right amygdala and total PFC volumes) at 3 months of age. In a basic model, infant brain volumes were the dependent variables and objective hardship as well as perceived stress were the independent variables, adjusting for biological sex, parent education, gestational age at birth, and TCV. In a subsequent extended model, maternal anxiety and depression and their interaction were added to the models to determine their moderating effects of maternal mental health on infant brain volume and PNMS. Models were run for the left and right amygdala, and PFC separately. As two brain regions were assessed, the models were Bonferroni corrected for multiple comparisons (p = 0.05/2 or 0.025). Missing data were handled through listwise deletion.

To address our second aim, GLMs were used to determine the association between infant brain volumes (i.e., amygdala and PFC) at 3 months and each of the three subscales (negative affectivity, positive affectivity/surgency, and regulatory capacity/orienting) of temperament at 6 months of age assessed with the IBQ-R, adjusting for biological sex, maternal education, gestational age at birth, maternal ethnicity, and TCV. As we assessed three subscales on the IBQ-R the models were corrected for multiple comparisons using the Bonferroni method (p = 0.05/3 or 0.0125).

Results

Participants

A total of 100 mother-infant dyads (mean maternal age = 33.3 years old, SD = 4.33) were recruited from the Calgary area during the COVID-19 pandemic. Participants completed the recruitment questionnaires at an average of 18.3 weeks (SD = 16.38) after the initial Canadian country-wide shut down period on March 11, 2020. Approximately one third (28.6% [n = 28]) of individuals were enrolled in the study in the first trimester of pregnancy, 45.9% (n = 45) were in the second trimester, and 25.5% (n = 25) were in the third trimester. The majority of our participants were white (Caucasian), married or living with their partner, had a university education, and had an annual household income >$70,000 CAD. Approximately half of participants (53.5% [n = 53]) did not have any other children. At intake, 18.8% (n = 19) of individuals exhibited clinically elevated prenatal depression symptoms above the EPDS cut-off score of 13, while 30.1% (n = 28) exhibited clinically elevated prenatal anxiety levels compared to a general US population reference sample. Descriptive maternal characteristics are reported in Table 1. Participants gave birth at an average of 39.6 weeks gestation (SD = 1.28). Infants were 57% (n = 57) male and 43% (n = 43) female. Infant sample characteristics are presented in Table 2.

Table 1 Maternal sample characteristics.
Full size table
Table 2 Infant sample characteristics.
Full size table

Maternal stress, mental health, and infant brain volumes

Our first aim was to examine the relationship between prenatal objective hardship, perceived stress, and infant brain volumes at 3 months of age. None of the regions of interest (i.e., amygdala, PFC) were associated with either maternal stress variables (all, p > 0.05). However, when examining the moderating effects between maternal mental health (i.e., prenatal depression and anxiety) and infant brain volumes, greater prenatal anxiety was significantly associated with smaller left amygdala volumes (B = −5.919, p = 0.016, Fig. 1, Table 3), independent of prenatal stress.

Fig. 1: Marginal effects plot: Left amygdala volumes by prenatal anxiety T-scores.
figure 1

Grey area indicates 95% confidence.

Full size image
Table 3 General linear model of prenatal anxiety predicting infant left amygdala volumes at 3 months.
Full size table

PFC & amygdala volumes and emotional regulation

Our second aim was to examine the relationship between infant brain volumes at 3 months and temperament at 6 months of age. Left amygdala volumes negatively predicted infant negative affectivity (B = −0.003, p = 0.002, Fig. 2, Table 4); right amygdala volumes were not associated with temperament. A mediation analysis was run post hoc to determine whether infant left amygdala volumes mediated any relationship between prenatal anxiety and infant negative affectivity. The mediation was not significant. Despite prenatal anxiety predicting left amygdala volumes, and left amygdala volumes predicting negative affectivity, there was no direct effect between prenatal anxiety and negative affectivity.

Fig. 2: Marginal effects plot: Infant negative affectivity at 6 months by 3-month left amygdala volumes.
figure 2

Infants with larger left amygdala volumes were more likely to have lower levels of negative affectivity.

Full size image
Table 4 General linear model of infant brain volumes at 3 months predicting infant negative affectivity at 6-months.
Full size table

Discussion

As part of a larger longitudinal pan-Canadian study of pregnant individuals during the pandemic, we examined the association of prenatal maternal stress, anxiety, and depression with 3-month infant amygdala and PFC volumes. We showed that greater prenatal maternal anxiety was associated with smaller left amygdala volumes in infants at 3 months. In addition, we also determined that smaller left amygdala volumes were associated with greater negative affectivity at 6 months.

The negative relationship between maternal anxiety and infant amygdala volumes adds to the growing body of literature describing the vulnerability of the amygdala to early stress. Our finding is supported by previous research that found altered amygdala volumes and functional changes in association with maternal stress during the intrauterine period [9]. Higher prenatal depression [28] and pregnancy-related anxiety [71] symptoms have been associated with larger amygdala volumes in young girls. Conversely, Lehtola et al. [26] found that prenatal psychological distress, a combination of anxiety and depression symptoms, predicted smaller amygdala volumes in newborn males [26]. Importantly, the direction of the effect of PNMS on amygdala volumes may vary based on the age of the children studied. The growth of the amygdala varies with age, with “rapid increases in volumes at early ages [that] decline as youth enter adolescence” [72]. PNMS could potentially affect the developmental trajectory of amygdala growth throughout time. In animal models, prenatal stress has resulted in reduced amygdala volumes (including decreases in amygdala neurons and glial cells) in offspring early postnatally [73]. However, in later developmental stages, prenatal stress was related to greater amygdala volumes compared to controls [73]. In children, greater maternal depression during pregnancy was associated with more curvilinear left amygdala volume trajectories [74]. Infant sex, PNMS severity, timepoint in gestation, HPA-axis activation, and genetic influences could all affect the resulting offspring amygdala volumes [8, 75].

PFC volumes were not associated with prenatal maternal stress or mental health in this study. The lack of a significant relationship could be due to a number of factors. First, the PFC was treated as a broad region of interest made up of smaller subregions (e.g., frontal pole, orbitofrontal cortex, ventrolateral prefrontal cortex). Subregional variability could have masked effects. Second, PFC volumes are a function of both surface area and cortical thickness. Variability in both factors could have also masked effects. Future research can use surface-based analyses (e.g., vertex-wise cortical thickness via infant FreeSurfer) to tease apart the effect of PNMS on both factors. However, this type of analysis would be difficult in the current study due to various constraints (sample size, scanner time), which would limit that level of spatial granularity.

Overall, in our study, higher prenatal anxiety predicted smaller left amygdala volumes, which in turn predicted greater negative affectivity. Previous research has shown that PNMS is related to greater difficult/negative temperament, with more irritability and crying [11]. Pandemic-related stress during pregnancy has also been linked to greater negative affect in infants [49]. However, in this study, there was no association between prenatal anxiety and negative affectivity, and thus no significant mediation by left amygdala volumes. However, the association between amygdala volumes and temperament is mixed. Previous research found that smaller amygdala volumes were related to higher levels of infant negative emotionality, excessive crying and irritability [76, 77]. Filippi et al. [78] showed that infants with more negative temperament had slower growth in left amygdala volumes in childhood [78]. In children, smaller amygdalae were related to pediatric depression and anxiety [31, 79, 80]. The reduction of amygdalae volumes could be due to an excitotoxic process caused by PNMS (i.e., exposure to cortisol, inflammation) [8]. At the same time, some studies reported larger amygdalae volumes predict worse internalizing problems in children [27, 34] and greater infant fear [81]. The current study also provides some evidence towards lateralization of amygdala structure and function, where the left amygdala is more involved in negative emotions and local processing [82]. The left hemisphere also grows more quickly in early gestation [83], potentially making it more vulnerable to PNMS, similar to our results.

Genetics could have also played a moderating role in this study. Research has shown that prenatal maternal mental health (anxiety, depression) interacts with genotype to predict infant outcomes [84, 85]. First, the GUSTO study found that infants’ brain-derived neurotropic factor (BDNF) Val66Met gene (which effects synaptic plasticity) moderates the association between prenatal anxiety and DNA methylation of infants as well as the association between DNA methylation and amygdala volumes. Prenatal maternal anxiety had a greater effect on DNA methylation in the Met/Met polymorphism group compared to the Met/valine (Val) and Val/Val genotype groups [84]. The group also found that the impact of prenatal depression on right amygdala volume varied as a function of the infants’ genomic profile risk for major depressive disorder [86]. Furthermore, they determined that Genotype X Environmental models best predicted amygdala volumes compared to models which included genotype or environment (prenatal anxiety, depression, socio-economic status) variables alone [87]. Thus, the relationship between prenatal anxiety and amygdala volumes in this study could be similarly moderated by genetics. The group also studied the effect of prenatal anxiety on neonatal PFC cortical thickness. They found no main effect of prenatal anxiety on PFC cortical thickness. However, the association was moderated by single-nucleotide polymorphisms (SNPs) of the catechol-O-methyltransferase (COMT) gene (which regulates PFC catecholamine signaling). Cortical thickness in the right ventrolateral PFC decreased as prenatal anxiety increased for met homozygous infants. For val homozygous infants, prenatal anxiety was associated with an increase in cortical thickness [85]. Thus, the negative findings in our study could be due to a hidden moderating effect of infant genotype on PFC volumes, leading to cortical thinning in some and cortical thickening in others. Other studies have also shown that genotype also moderates the relationship between prenatal mental health and infant/child negative emotionality [11]. Together, these findings show that the associations between PNMS, infant brain volumes, and temperament in this study could have potentially been influenced by the infant’s genotype.

Possible limitations should be considered when interpreting our findings. Infant movements during scanning affected the quality of some scans. The scans were performed at 3 months of age, thus postnatal factors, such as infant feeding, sleep, and infection, could have also influenced infant brain development. Almost all the maternal data were collected using self-report questionnaires, potentially introducing reporting biases. Participants were recruited primarily from social media advertising, which could have contributed to selection biases. No genetic information was collected from parent or infant, limiting our understanding of the influences of genetics on the brain-behavior associations in this study. The moderating effects of genetics in relation to the impact of PNMS during the pandemic on infant outcomes should be studied in future research. Furthermore, in this study, we examined the interaction between prenatal anxiety and depression and its effect on infant brain volumes. The interaction was non-significant. However, given the study’s sample size, our analyses are likely underpowered to detect an interaction. Finally, participants were mostly white, financially stable, highly educated, and living in a country with a relatively contained outbreak and universal healthcare, both limiting the generalizability of our results and potentially underestimating the true effect of prenatal psychological distress on infant brain development in populations with greater vulnerabilities.

This study also had several major strengths. Its prospective longitudinal design allowed us to examine the associations between maternal stress variables, infant brain volumes at 3 months, and infant temperament at 6 months of age. We had a relatively large sample size of 100 mother-infant dyads, resulting in a unique and well-characterized cohort. Participant recruitment and data collection started at the early phases of the pandemic. Thus, our study was able to capture maternal stress and mental health levels related to the initial government mandated restrictions and rise of COVID-19 cases. The PdP study team developed a novel measure of Objective Hardship that captured participants level of loss, threat, and change due to the pandemic, as well as the duration and intensity of their hardship. Thus, we were able to disentangle the relative effects of objective hardship, perceived stress, and maternal mental health on infant development while they were exposed to a relatively independent stressor (the pandemic). In addition, the study team collected information on many potential confounding variables, and used measures that had strong psychometric properties, such as the EPDS and PROMIS.

Our findings have implications for perinatal policy and health care. The results suggest that screening for prenatal anxiety could help identify at-risk individuals during pregnancy. Anxiety reduction interventions could also mitigate its impact on infant brain development [88]. The brain remains quite plastic throughout infancy and childhood. Thus, offspring of at-risk individuals may also benefit from early evidence-based interventions, such as those targeting emotional regulation. Future research should use longitudinal scanning to help us determine the effect of prenatal anxiety on the trajectory of brain development throughout childhood. Long-term follow-ups with these children will allow us to examine the link between pandemic-related prenatal stress, brain volumes, and many measures of child development, such as motor, behavioral, and social-emotional development. Future studies should also investigate potential personal and environmental factors that could protect infant brain development from the negative effects of pandemic-related stress.

Going beyond the current study, examining the impact of PNMS during the pandemic on functional and structural connectivity can further offer additional insight that should be pursued in future work. In general, previous research shows that prenatal maternal stress and mental health is associated with infant brain connectivity. For example, greater PNMS has been associated with less positive resting state functional connectivity but greater structural connectivity between the medial PFC and the amygdala in 5-week-old infants [89]. Next, in terms of maternal mental health, greater prenatal depression symptoms have been linked to an increase in infant negative amygdala-dorsal PFC functional connectivity, a decrease in structural connectivity between the right ventral PFC and the right amygdala, as well as higher functional connectivity between the infants’ amygdala and other brain structures, including the left temporal cortex, insula, and anterior cingulate cortex [90, 91]. Prenatal maternal anxiety may also play a role. Research has shown that higher prenatal anxiety is associated with functional connectivity between infants’ amygdala and brain regions important for fear learning (e.g., lower amygdala-fusiform gyrus functional connectivity, higher amygdala-thalamus connectivity) as well as between young children’s amygdala and brain regions important for sensorimotor functioning (e.g., increased negative connectivity between the amygdala and bilateral parietal regions) [92, 93]. According to another PdP study, social support could moderate the association between prenatal distress during the pandemic and infant amygdala-prefrontal functional connectivity. Infants of participants with low social support had a significant relationship between higher prenatal distress and weaker amygdala-PFC functional connectivity, while participants with high social support showed no effect [50]. Finally, the impact of prenatal stress and mental health on brain connectivity could have subsequent impacts on children’s behavior. One study found that weaker amygdala-frontal structural connectivity, which was associated with greater prenatal depression, in turn predicted externalizing behavior in young boys [94]. All in all, structural and functional connectivity is another area of research that should be understood in the context of PNMS and the pandemic.

In conclusion, the present study described the relationship between prenatal stress, including symptoms anxiety and depression, during the COVID-19 pandemic, infant brain volumes, and temperament. We observed a negative association between prenatal anxiety and infant left amygdala volumes, and lower left amygdala volumes predicted higher negative affectivity. Infant temperament may be partially programmed in utero and influenced by the development of the amygdala. The current study supports the hypothesis that exposure to stress in utero influences infant brain and behavioral development.