Maternal History of Childhood Maltreatment and Brain Responses to Infant Cues Across the Postpartum Period

Abstract

Adults with histories of childhood maltreatment (CM) are more likely to display problematic parenting behaviors. The goal of this study was to examine changes in maternal brain activation to negative infant cues over the early postpartum period among new mothers with and without histories of CM, as this is a period of immense neuroplasticity in the maternal brain. CM was measured using the Adverse Childhood Experiences Scale. Functional magnetic resonance imaging (fMRI) conducted at approximately 5 and 13 weeks postpartum measured brain responses to own and unfamiliar infant cues in primiparous women. Women with histories of CM displayed increasing activation in the anterior cingulate cortex, and greater increases in anterior cingulate cortex activation was associated with maternal reports of less regulatory capacity in their infants. Preliminary results suggest that new mothers with CM histories display greater brain responses to negative infant cues compared to new mothers without CM histories. Women with CM histories may benefit from additional supports during the transition to parenthood.

Childhood maltreatment (CM) is a behavior towards a child, including physical abuse, sexual abuse, emotional abuse, or neglect that poses substantial risk for physical or emotional harm (World Health Organization, 2020). Approximately one in four pregnant women experienced maltreatment in their own childhoods (Plant et al., 2017), and adults who were victims of CM are 2–3 times more likely to become perpetrators of maltreatment as parents than non-victimized adults (Widom et al., 2015). Parents with CM histories are also more likely to display problematic parenting behaviors including permissive parenting, hostile caregiving, and greater use of harsh physical discipline (Banyard, 1997; de Paul et al., 1995; DiLillo & Damashek, 2003; Madigan et al., 2015).

The neural mechanisms linking CM and the emergence of problematic parenting behaviors are not well understood. Given evidence that CM history is associated with altered neural systems that overlap with parenting behaviors (Dannlowski et al., 2012), understanding associations between CM history and brain responses to infant cues among new mothers may yield valuable insights into intergenerational transmission of risk for CM. A small number of studies have investigated neural responses to infant cues in parents with CM histories, and results are mixed. For example, some studies show that women with CM histories display lower amygdala responses to negative infant faces (Kim et al., 2014; Olsavsky et al., 2019), while others report greater amygdala response to negative infant cues in women with CM histories (Neukel et al., 2018; Olsavsky et al., 2021). Greater amygdala activation has been associated with less maternal intrusiveness, suggesting that greater activation to distress may be adaptive for maternal behavior (Olsavsky et al., 2021). Inconsistencies in past results may be due, in part, to different types of stimuli and task designs used in prior fMRI paradigms, and differing assessment time points across studies.

The maternal brain undergoes dramatic changes over the postpartum period. Brain responses in areas associated with emotion regulation, attachment, and sensitivity (including the insula, anterior cingulate cortex (ACC), and amygdala) increase in response to infant cues over gestation (Pawluski et al., 2022). In the postpartum period, maternal brain activation increases between 48 hours and 4–6 weeks’ postpartum (Gingnell et al., 2015), and continues to increase over the first postpartum year in a dose-dependent manner (Parsons et al., 2017). To address the dynamic nature of maternal brain activation over time, in this pilot study we examined changes in postpartum brain activation to negative infant cues in the insula, ACC, and amygdala among women with and without histories of CM. We hypothesized that women with histories of CM would display greater activation to images of infant distress given that CM history likely increases the salience of negative emotional cues. We also hypothesized that greater activation in these regions would be associated with maternal reports of difficult infant temperament as child behavior problems may serve as an indirect pathway from CM history to problematic parenting (Michl-Petzing et al., 2019).

Methods

Study Design

Participants were eligible to participate if they were primiparous, with singleton pregnancies, between ages 18–40, English-speaking, and free of implanted metal or another MRI contraindication. Participants were recruited through flyers and announcements placed in an obstetric practice in the Northeast region of the United States and were enrolled in third trimester of pregnancy. At the prenatal session, participants completed an interview and questionnaires to evaluate CM and obstetric history. At approximately 5 (SD = 1, range: 3–7 weeks) and 13 weeks after delivery (SD = 2, range: 10–18 weeks), participants completed fMRI sessions to measure neural responses to the infant faces task (see below). At 13 weeks postpartum, participants completed self-report measures of depressive symptoms and infant temperament. Participants were compensated for their participation. All participants completed written informed consent prior to participation. This study was approved by the Rhode Island Hospital and Brown University institutional review boards.

fMRI Approach

Prior to the scan, participants were screened for MRI contraindications and new pregnancy. All scans included: (1) 3 plane localizer scan for slice positioning; (2) high-resolution T1 anatomical scan for structural MRI analysis and spatial normalization (Acquisition time: 4.01 minutes, Voxel size 1x1x1 mm, TR 1900 milliseconds [ms] TE 3.02 ms, Slices 160); and (3) fMRI scan including an infant face paradigm (Acquisition time: 5.49 minutes/run, Voxel size 2x2x2 mm, TR 1500 ms, TE 30 ms, Slices 60). Hypothesized regions of interest (ROIs) were the insula, ACC, and amygdala. Regions were selected based on past literature on CM and neural systems, and regions that are consistently activated in response to infant cues in fMRI studies.

Infant Faces Task

The infant faces task is a standardized paradigm used in previous research to elicit maternal brain responses to infant cues (Rutherford et al., 2020; Zhang et al., 2020). See Barrett et al. for a detailed description of the task (Barrett et al., 2012). Participants provided study staff with pictures of their infant displaying joy, distress, neutral, and ambiguous faces. Participants’ pictures were converted to black and white and placed on a black background to standardize images across participants and standardize to control images of unfamiliar infants matched on age, race, and gender from the City Infant Faces database (Webb et al., 2018). Participants completed two runs consisting of 12 blocks of 4 infant pictures (joy, distress, neutral, and ambiguous) from either the participants’ infant or the control infant photo each for 4 seconds in random order followed by an 8 second fixation.

Self-Report Measures

CM history was measured using the Adverse Childhood Experiences (ACE) questionnaire (Dong et al., 2004). The ACE is a widely used 10-item self-report instrument evaluating experiences of adverse events prior to the age of 18 years old. Those who endorsed childhood physical, sexual, or emotional abuse or neglect were categorized as having CM histories. Those who did not endorse these were categorized as having no history of CM. The Infant Behavior Questionnaire (IBQ)-Revised Very Short form measures infant temperament in the past 7 days. The IBQ is scored into three subscales: positive affectivity, negative emotionality, and regulatory capacity. Participants completed the Edinburgh Postnatal Depression Scale (EPDS), a 10-item self-report measure in which women report the severity of depressive symptoms in the past 7 days (Cox et al., 1987). Cronbach’s alpha for the scales were: IBQ = .66, EPDS = .77, indicating acceptable reliability.

Statistical Approach

fMRI data were analyzed with Analysis of Functional NeuroImages (AFNI, NIMH/NIH). This included individual-level preprocessing with AFNI’s afini_proc.py: (1) discarding the first two volumes of each fMRI series, prior to the magnetization equilibrium being reached, when no data were collected, (2) spatially normalizing the participant’s high-resolution anatomical scan to the Talairach template, using that transformation to normalize their fMRI dataset in turn, (3) spatially smoothing the data with a 6-mm Gaussian kernel to reduce the impact of anatomical variability on individual statistical maps, (4) including 6 motion parameters in the regression and censoring motion >1 mm. We used the TT Daemon atlas regions of interest included in the AFNI program, resampled to the same 3x3x3mm space as our fMRI dataset (Talairach & Tournoux, 1988). We used AFNI to conduct a ROI analysis. We first extracted fMRI brain activation in the left and right amygdala, ACC, and insula for each individual at Scan 1 and Scan 2. Then we used SPSS (version 25, IBM SPSS) to conduct univariate analysis of variance (ANOVA) to examine group differences in activation at each MRI separately. Repeated measures ANOVA tested group differences in activation in ROIs between scans, adjusting for a priori covariates of maternal depressive symptoms and breastfeeding. Correlations evaluated associations between activation in ROIs that differed by group and infant temperament.

Results

Demographic Characteristics

Twenty-five women enrolled in the study during pregnancy, and 13 had fMRI data from both postpartum sessions. Seven participants were lost to follow up after pregnancy, two missed one of two scans, and three had missing data due to technical reasons. Women were 30 years old (range: 21–39), 72% were partnered, and 80% were exclusively breastfeeding. Seventy-six percent were non-Hispanic White, 8% Hispanic White, 8% non-Hispanic Pacific Islander, and 8% did not specify. Average gestational age at delivery was 39 weeks (SD = 1). All women were primiparous. Twenty-four percent reported consuming some alcohol while pregnant and one participant reported tobacco use in pregnancy. Fifty percent of infants were male. Forty-six percent of the sample reported a CM history; 54% reported no CM.

CM groups did not significantly differ on maternal age (t = .52, p = .61), gestational age at delivery (t = −.042, p = .96), alcohol use (t = 1.043, p = .319), depressive symptoms (t = −.86, p = .41), or ethnicity (X² = .68, p = .41). Groups marginally differed by race (X² = 2.44, p = .12). Given the small sample size, we conservatively included race as a covariate.

fMRI Results

We observed significant group differences at Scan 1; women with CM histories displayed greater ACC activation to other infants in distress compared to women without CM (F = 15.82, p = .004). There were no significant group differences in activation in the insula or amygdala at either time point. Repeated measures analyses showed that those with CM histories had decreasing activation in the ACC when viewing other infants in distress; women without CM histories exhibited increasing ACC activation to other infants in distress (F = 5.46, p = .05). See Figure 1. There was no significant change in activation in the insula or amygdala when observing own or other infants in distress (p-values > .20). Finally, greater ACC activation to viewing other infants in distress at Scan 1 was associated with maternal reports of lower infant regulatory capacity (r = −.74, p = .004).

Figure 1.

Women with histories of childhood maltreatment displayed greater activation in the anterior cingulate cortex at T1, and decreasing activation in the anterior cingulate cortex over time, to an unfamiliar infant in distress. Left and right regions were averaged for analyses. T1 = Scan 1 at 5 weeks gestation. T2 = Scan 2 at 13 weeks gestation.

Discussion

Preliminary results from this study indicate that maternal history of CM is associated with differing patterns of brain activation to negative infant cues across the early postpartum period relative to women without a history of CM. Women with CM histories displayed greater initial activation in the ACC, and decreasing activation over time, when viewing other infants in distress, in line with past evidence on long-term neurological consequences of CM (Dannlowski et al., 2012). This may reflect decreasing emotional regulation to stressful stimuli. While decreasing reactivity to negative stimuli may be adaptive in adults with histories of CM, overall, this pattern of activation may increase risk for intergenerational transmission of risk when the negative stimuli is an infant in distress. However, decreasing ACC activation was only observed when women with CM histories observed “other” infants in distress and not their own infant.

Greater ACC activation was also associated with maternal reports of decreased infant regulatory capacity, which may reflect a neurological correlate of difficulty soothing and regulating distressed infants. Given that women with CM histories displayed elevated ACC activation, these results are consistent with findings that history of CM is indirectly associated with problematic parenting via child behavior problems (Michl-Petzing et al., 2019), and results from a systematic review reporting that withdrawal or emotional rejection from a child may be a mechanism linking CM history to problematic parenting (Greene et al., 2020). However, we do not have information on how mothers responded to infants displaying low regulatory capacity.

We did not observe differences in insula or amygdala activation among women with and without CM histories. Past studies have reported mixed activation in these regions among mothers with CM histories (Kim et al., 2014; Neukel et al., 2018; Olsavsky et al., 2019; 2021), and studies were conducted later in the postpartum period, thus differences in activation in these regions may emerge after 3 months postpartum.

Results from this study must be considered in the context of several limitations, including the small sample size, lack of objective measures of infant temperament, and the retrospective recall of childhood maltreatment (Baldwin et al., 2019). Strengths of this study include the repeated measures of brain activation in the early postpartum period among first-time mothers using their own infant as stimuli for the fMRI paradigm. Future, larger studies are needed to prospectively evaluate associations among brain activation to infant cues, using a variety of stimuli including infant cry, and observable measures of maternal and infant behavior across the first postpartum years, as these results will help to identify early targets for interventions during a period of immense neuroplasticity in both mothers and infants. Taken together, results from this study suggest that, possibly due to the neurobiological consequences of exposure to toxic stress early in life, women with histories of childhood maltreatment may benefit from interventions to support their adaptation to motherhood.