Impact of early vs. late childhood early life stress on brain morphometrics

Abstract

Previous studies of early life trauma suggest that in addition to its emotional impact, exposure to early life stress (ELS) is associated with alterations in brain structure. However, little attention has been devoted to the relationship between emotional processing and brain integrity as a function of age of ELS onset. In the present study we examined whether ELS onset in older ages of youth rather than younger ages is associated with smaller limbic and basal ganglia volumes as measured by magnetic resonance imaging (MRI). We hypothesized that later age of manifestation during youth is associated with smaller volumetric morphology in limbic and basal ganglia volumes in adulthood. A total of 173 individuals were divided into three groups based on the age of self-reported ELS. The three groups included individuals only experiencing early childhood ELS (1 month–7 years, n = 38), those only experiencing later childhood ELS (8 years –17 years, n = 59), and those who have not experienced ELS (n = 76). Anterior cingulate cortex (ACC), hippocampus, amygdala, insula and caudate volumes were measured using a T1-weighted MRI. Analyses confirmed that later childhood ELS was associated with volumetric reductions in the ACC and insula volumes, while ELS experienced between the ages of 1 month and 7 years was not associated with lower brain volumes in these regions. The results may reflect the influence of more fully developed emotional processing of ELS on the developing brain and reinforce a body of research implicating both the ACC and insula in neuropsychiatric disorders and emotional regulation.

Introduction

Early life stress (ELS) refers to the exposure to a single or multiple events during childhood that exceeds the child’s coping resources, leading to prolonged phases of stress (Pechtel and Pizzagalli 2001). Common early life stressors include physical abuse, sexual abuse, emotional abuse, verbal abuse, neglect, social deprivation, disaster, household dysfunctions, parental separation, or illness. A number of studies have focused on adult outcomes following childhood abuse and neglect. Recent research had revealed that a significant percentage of the population have experienced emotional, physical, and sexual abuse, 10 %, 26 %, and 21 %, respectively; while emotional and physical neglect are less common, 15 % and 10 %, respectively (Dong et al. 2004). Further studies have concluded that a variety of stressful childhood events, in addition to abuse and neglect, have a detrimental impact on individual well-being (Koenen et al. 2010). Extensive research has revealed a strong relationship between various forms of early life stress and the development of psychiatric disorders such as depression and anxiety (Infrasca 2003; Levitan et al. 2003; Penza et al. 2003; Young et al. 1997). Detection of these relationships has been described in qualitative longitudinal studies demonstrating not only profound influences on psychopathology, but also impact on physical health, and emotional intelligence in adulthood (Brown et al. 2009). Additional research has shown that stress changes brain structure, morphology, and function in key regions of the limbic system. Stress affects gray matter volume through a number of possible mechanisms, including loss of neurons, decreased dendritic branching, spine density, and/or decreased neurogenesis (Lupien et al. 2009).

A recent study examining the relationship between reported ELS and brain morphology, using magnetic resonance imaging (MRI), in a large sample of adults with no history of psychopathology revealed that people who have experienced significant ELS have volumetric differences in brain structure compared to people who have experienced minimal ELS. These reductions were apparent in the anterior cingulate cortex and caudate nucleus, but not the hippocampus or amygdala (Cohen et al. 2006). Further, MRI studies have suggested that adults subjected to ELS have lower than average hippocampus volumes (Bremner et al. 1997; Bremner and Narayan 1998; Bremner 2002, 2003). Collectively these results indicate that ELS may have detrimental effects on brain integrity. Although the relationships described above demonstrate an impact of ELS on brain structure, it is unknown whether this relationship varies with the age in which the ELS occurred.

The prefrontal cortex increases in size slowly until age 8, followed by a rapid surge of growth throughout adolescence (Wessing et al. 2011). Studies have also shown that the prefrontal cortex is extensively connected to subcortical limbic structures known to regulate stress and emotional arousal, and is important for higher cognitive ability (Gray et al. 2002). Based on these findings, we hypothesized that ELS after age 7 may have a more profound influence on brain development within limbic structures than ELS events prior to age 7. Additionally, adolescence is a critical period for maturation of neurobiological processes that underlie higher cognitive functions and emotional behavior. It is possible that ELS experienced later in childhood and adolescence has a greater impact on brain integrity than ELS experienced in very early childhood when emotional and cognitive processing of environmental factors is less developed (De Bellis et al. 2001).

In the present study we examined the relationship between brain morphology and age of ELS manifestation in a sample of healthy individuals with no history of psychopathology or brain disorder. In order to determine if the timing of ELS is associated with functional impact, we examined current symptoms of depression, anxiety, and stress. Past research had indicated that volumetric reductions in the limbic system are associated with greater emotional dysregulation (Cohen et al. 2006). Areas that were examined included those regions implicated in post-traumatic stress disorder (PTSD) and other emotional behaviors including the hippocampus, amygdala, anterior cingulate cortex, insula, and caudate nuclei. We hypothesized that later developmental stage of ELS onset is associated with greater volumetric reductions in these regions.

Methods and materials

Participants

A total of 173 subjects were extracted from a sample of 1,966 individuals as part of the Brain Resource International Database (BRID). Of this sample, 173 had brain magnetic resonance imaging (MRI) within 2 weeks of their behavioral assessment. The BRID is an international database containing demographic, psychiatric, health, neurocognitive, and brain imaging data on healthy adults (Gordon 2003). Individuals were excluded from the study if they reported suffering from major affective or anxiety disorders, schizophrenia, substance abuse, neurological disease, or major medical conditions that could affect cognition (e.g., stroke, heart disease, or cancer) currently or in the past. Participants completed the Somatic and Psychological Health Report (SPHERE) to exclude any individuals with a history of any psychopathology for purposes of satisfying the exclusion criteria (Hickie et al. 2001). The average age was 34.72 ± 17 (range from 8–79) and the sample was 52 % male. Participants were categorized into three groups; 76 with no ELS, 38 with early childhood ELS, and 59 with later childhood ELS groups (childhood ELS, ages 1 month–7 years; late childhood ELS, ages 8–17 years) using self-reported age of ELS. See Table 1.

Table 1.

Demographic information and current psychological state

	No ELS	Early childhood ELS (Ages 1 month-7years)	Later childhood ELS (Ages 8–17 years)
n	76	38	59
Age	33.69 ± 19.17	40.34 ± 18.84	31.36 ± 12.28
% Male	58.2	55.3	44.3
DASS depression	.71 ± .91	.80 ± .84	1.14 ±.95
*DASS stress	.96 ± .90	1.32 ± .99	1.50 ± .87
*DASS anxiety	.32 ± .57	.53 ± .68	.71 ± .72

Mean ± Standard deviation shown in table,

^*Between-group differences (p < .017)

All participants provided signed informed consent before participation in the study, and were financially compensated for participation.

Measures

Neuroimaging acquisition

MRI scanning was performed using a standard Siemens Sonata or Vision Systems 1.5 Tesla scanner at Westmead Hospital (Sydney, Australia) and Wakefield Imaging (Adelaide, Australia). Acquisition of isotropic TI-weighted images were in a sagittal orientation using a magnetization prepared rapid gradient echo (MPRAGE) sequence (repetition time [TR]: 9.7 msec; time-to-echo [TE]: 4 msec; Echo train: 7; Flip Angle: 12°; inversion time [TI]: 200 msec; NEX= 1). One hundred eighty contiguous 1 mm slices were acquired with an in-plane matrix of 256 × 256 at a resolution of 1 mm × 1 mm (Cohen et al. 2006).

Magnetic resonance image analysis

SPM2 running on MATLAB 6.5 was used for post processing and MR analysis. Imaging was normalized to a Brain Resource International Database specific TI-weighted template and standard T1 templates of segmented images provided by SPM were used to create customized template images. Images were segmented into gray matter (GM), white matter (WM), cerebrospinal fluid (CSF), and nonbrain tissues based on a cluster analysis method that separates pixels based on distribution of intensities and a priori knowledge of spatial tissues distribution patterns in normal subjects (Fristron et al. 1996). A correction was made to preserve quantitative tissue volumes following the normalization procedure (Ashburner et al. 1998, 2000; Ashburner and Friston 2000; Salmond et al. 2002).

ROI volumetric methods

Volumes for the five ROIs were determined based on the Automated Anatomical Labeling (AAL) anatomic atlas that is defined in MNI space (Tzourio-Mazoyer et al. 2002)–the frame of reference upon which the brains are “normalized.” Global volumes were obtained by summing the segmented volumes from GM, WM and CSF segmented images and total brain volume was attained by summing the three components (Cohen et al. 2006). Bilateral ROI volumes were determined for each participant for the ACC, the hippocampus, the amygdala, the caudate nucleus, and the insula. Right and left volumes were obtained for each ROI, in mL (cm³). The specific boundaries used to determine each ROI have been described in detail previously (Tzourio-Mazoyer et al. 2002).

Early Life Stress Questionnaire (ELSQ)

All individuals were asked to complete the ELSQ during behavioral assessment. The ELSQ is based on the Child Abuse and Trauma Scale (McFarlane et al. 2005; Paul et al. 2005), and has shown strong internal consistency, test-retest reliability and validity (Sanders and Becker-Lausen 1995). The questionnaire consisted of 19 items that asked whether or not the individual had experienced physical, emotional, or sexual abuse as well as other traumatic experiences such as bullying, poverty, divorce, illness, or domestic violence. The 19 items have been shown to be perceived as traumatic in previous studies (Sanders and Becker-Lausen 1995; Harrison et al. 1997; De Bellis et al. 2001; McGee et al. 1995). The participants were asked to respond yes or no, and when ELS was present, the participants were asked to identify the age of onset (0–17 years of age).

Depression, Anxiety, and Stress Scale (DASS)

The DASS (Lovibond and Lovibond 2005) was administered to characterize the current psychological health of the sample, specifically the manifestation of symptoms of depression, stress, and anxiety. This test has strong reliability and validity (coefficient alpha = .92; Lovibond and Lovibond 2005). Each participant rated their degree of agreement with the questions using a 4-point Likert scale. Scores were then generated from the three subscales (i.e., depression, anxiety, and stress).

Statistical analysis

T tests were used to ensure that there were no differences between age or cognitive status between our group of 173 and the remaining 1,793 participants in the dataset. A series of analyses were conducted to examine the relationship between age of manifestation of ELS and brain morphometry. Frequency distributions were generated for ages of ELS manifestation. The participants were dichotomized based on self-reported age of ELS into two groups (ages 1 month–7 years, ages 8 years–17 years). Individuals who did not self-report ELS were used as healthy controls. Chi squared analysis was used to determine if gender differed across the three groups. Descriptive statistics were also obtained for the DASS in order to check for normality. To test our primary hypothesis, a multivariate analysis of co-variance (MANCOVA) was performed. Age of ELS manifestation served as the independent variable and ROI volumes as a whole and laterally of the insula, caudate, ACC, hippocampus, and amygdala served as dependent variables. Current age was covaried a priori. A priori contrast analysis was performed to determine specific differences in brain volume between the group that did not experience ELS and each ELS group.

Univariate between group comparisons for each group of ELS manifestation were conducted to determine whether differences existed laterally across volumes of the five ROIs. To normalize the data, a square root transformation was performed on all three of the DASS variables (depression, anxiety, and stress). A MANCOVA, with current age serving as a covariate, was used to characterize the relationship between ELS age of manifestation and current emotional distress.

Results

No significant differences were observed in age or cognitive status between our group of 173 and the remaining 1,793 participants in the dataset. Results of the chi- squared goodness-of-fit test revealed no significant differences in gender across the three ELS groups, X² _{(2, 173)} = 2.81, p=0.245. MANCOVA examining the effect of age group on brain regions revealed that the three groups differed with respect to brain volumes across all ROIs F_(12,330) =1.95, Wilks Lambda = .872, p = 0.028, with smaller volumes in the later childhood ELS group. When right and left hemi-spheres were considered laterally, significant differences existed overall across all ROI’s F _{(20, 322)} = 1.69, Wilks Lambda = .819, p = 0.034, also with smaller brain volume in the later childhood/adolescent group. See Table 2. Results of a priori contrast analysis with a Sidak correction revealed differences in the right and left ACC and left insula between the later ELS group and individuals with no ELS. Trends were observed between the two groups in the left and right caudate. See Table 3. When hemisphere specificity for each ROI was considered separately, results revealed significant between group differences at the Bonferroni-corrected threshold of α= .025, in the left insula (p = 0.012), and right (p = 0.011) and left ACC (p = 0.023), with smaller volumes in the right hemisphere across each ELS group.

Table 2.

Mean volume of brain structures as a function of early life stress

Brain structure	No ELS	Early childhood ELS (Ages 1 month-7 years)	Later childhood ELS (Ages 8–17 years)	P value
ACC	5304.98 ±691.41	5088.16 ±642.01	5046.27 ± 594.07	.061
*Right	5139.32 ± 677.85	4941.91 ±618.97	4894.01 ± 576.51	.018
*Left	5470.64 ± 704.96	5234.41 ± 665.05	5198.53 ±611.63	.031
Amygdala	1219.12 ± 128.45	1194.16 ± 129.94	1209.53 ±115.87	.237
Right	1277.76 ± 138.46	1243.35 ± 137.79	1267.71 ± 125.45	.181
Left	1160.47 ± 124.80	1135.22 ± 117.61	1151.34 ± 111.82	.354
Caudate nucleus	3256.08 ± 336.47	3134.11 ±349.40	3169.67 ± 357. 69	.079
Right	3289.55 ± 356.77	3148.47 ± 352.34	3214.37 ± 386.48	.077
Left	3222.62 ± 337.46	3093.27 ± 330.11	3124.97 ± 339.38	.097
Hippocampus	3943.20 ± 454.42	3887.05 ±418.47	3843.47 ± 387.72	.558
Right	3860.04 ±451.57	3792.72 ± 409.40	3575.45 ± 380.63	.531
Left	4026.36 ± 473.78	3956.82 ± 428.71	3929.50 ± 405.76	.608
Insula	8286.63 ± 1032.23	8025.02 ±941.17	7890.26 ± 887.92	.053
Right	8212.37 ± 1084.66	7982.27 ± 884.89	7835.26 ± 903.88	.155
*Left	8360.55± 1004.84	7980.72 ± 909.42	7945.26 ± 897.19	.018

Mean ± Standard deviation shown in table, Measurements in mm³ for each structure bilaterally.

^*Between-group differences (p < .05)

Table 3.

A priori contrast analysis among ELS groups and brain volumetrics

Brain structure	Early childhood ELS (1 month-7 years) vs. No ELS	Later Childhood ELS (8–17 years) vs. No ELS
ACC
*Right	.768	.010
*Left	.705	.020
Amygdala
Right	.293	.325
Left	.560	.581
Caudate Nucleus
Right	.630	.077
Left	.657	.078
Hippocampus
Right	.981	.573
Left	.990	.664
Insula
Right	.823	.146
*Left	.578	.015

^*Between-group differences (p < .05)

Sidak Adjustment for multiple comparisons

ELS and current emotional state

An additional MANCOVA was conducted to examine potential group differences across current psychological state (depression, anxiety, and stress). As a group, individuals reported scores on the DASS that fell below the expected range for community samples (Crawford and Henry 2003). Results of the multivariate analysis revealed that depression, stress, and anxiety scores differed within the three age groups, F_{(3, 160)} = 102.08, Wilks Lambda= 0.657, p< 0.01, with higher mean scores in the later childhood ELS group. See Table 1. Univariate analysis of between group differences showed a significant difference in stress (p = 0.003) and anxiety (p = 0.004), but not depression scores (p=0.027) at the bonferroni corrected threshold of α= 0.017

Discussion

The novel and most important finding of the current study is that the impact of ELS on brain integrity is related to the age at which ELS occurred, with deleterious impact associated only with exposure later in childhood/adolescence. Analyses revealed that individuals who experienced ELS in later childhood as opposed to those with either early childhood ELS or no ELS have smaller ACC and insula volumes, with differences in laterality in the ACC and insula across both age groups as well. These results are important as the findings may partially account for unexplained variations in brain morphology across individuals with a history of ELS. These findings are especially noteworthy given that our sample represented a very healthy group of individuals, thus allowing us to determine that there may be lingering effects of ELS on adult emotional experience resulting in volumetric differences, even if they go on to be free of major psychiatric illness in adults. In contrast to our expectations, individuals who experienced ELS in later childhood/adolescence did not show a significant difference in volumes of the caudate, hippocampus and amygdala. The fact that the amygdala and hippocampus were not related to ELS status in the present study or when examined by Cohen et al. (2006) suggests that these regions may be less vulnerable to the effects of ELS than predicted from clinical populations.

Results from the current study reveal structural differences between the three groups and aids to reinforce the body of literature that the ACC and insula are implicated in not only psychological disorders, but emotional regulation (Adamec 1997; Davidson and Irwin 1999; Shea et al. 2005; Whittington et al. 2004). Structural and functional changes in the stress-related regions of the amygdala, hippocampus and insula influence reward processing and are part of the addiction circuitry (Goldstein et al 2009; Naqvi et al. 2007; Kalivas and Volkow 2005; Li and Sinha 2008). Therefore, cumulative stress can affect neurobiological processes that promote addictive behavior patterns and ultimately chronic disease risk (Sinha 2008). In addition, the insula is involved in social cognition and processing such that lower volume in this region may also adversely affect functions involved in regulating interpersonal relationships and negotiating social contexts (Heatherton 2011; Eisenberger et al. 2003). It may also mediate the effects of stress on risk for anxiety disorders (Paulus and Stein 2010; Stein and Paulus 2009). The relationship between ROI volumetric measures and emotional state can be inferred from the significant difference between the DASS scores of the three ELS groups and replicate findings from Cohen et al. 2006 showing that limbic structure morphometry is associated with a higher degree of current emotional dysregulation. These findings are important given evidence of the influence of emotional disturbance on brain structure, particularly in the absence of current psychiatric comorbidity. It can be inferred that ELS events later in childhood affect brain development in emotional regulation areas that can help determine psychological states in adulthood.

Animal studies of prenatal and postnatal stress have revealed alterations in brain development through mechanisms of accelerated neuronal loss (Sapolsky et al. 1990), delays in myelination (Dunlop et al. 1997), or abnormalities in synaptic pruning (Lauder 1988). Environmental stress activates the hypothalamic pituitary adrenal (HPA) axis resulting in an increase in cortisol secretion. Research demonstrates that cortisol hypersecretion in utero alters neuronal development in areas rich in glucocorticoid (GC) receptors (Lauder 1988). Thus, while cortisol exposure is necessary for normal HPA axis development, animal studies have indicated deleterious effects on brain morphology during prenatal and postnatal stages of substantial brain development. Results from these animal studies provide clear evidence that stress induced morphology is heightened during developmentally sensitive stages. Additional research has shown later childhood/adolescence represents a developmental period of heightened HPA axis activity, comparable to the level of hypersensitivity reported prenatally (Uno et al. 1994). These findings indicate that adolescence is a particularly sensitive period of vulnerability to brain development in the context of significant stress. This vulnerability in adolescence may explain the reductions in brain volume identified in numerous studies of adolescents (Romeo et al. 2006; Gunner et al. 2009; Stroud et al. 2009; McCormick and Mathews 2007) and explain the specific age-related abnormalities identified in the present study. Future work that incorporates real-time HPA-axis activity within the context of recent stressors would be valuable in determining this model of dysregulation.

Conclusions regarding the relationship between age of ELS manifestation and brain integrity should be tempered by several limitations of the current study. The assessment of ELS among adults inherently requires retrospective report of trauma experiences and previous research suggests a high rate of false negatives in the reporting of ELS (Hardt and Rutter 2004). However, it is worth noting that significant effects were identified in the current study and therefore the impact of false negatives, if any, would not alter the outcomes of the present study. Another limitation is the inability to identify the contributions of specific forms of ELS. Not all forms of ELS are equal in their functional significance (Cohen et al. 2006). For example, parental violence and abuse, in particular, are examples of ELS events that are known to have a more detrimental impact on emotional development than other stressors (Cicchetti 2004; Dong et al. 2004; Anda et al. 1999). In the present study many individuals experienced multiple forms of ELS, and therefore isolating the effects of specific stressor was not possible. As in previous studies that have examined this cohort (Cohen et al 2006), the frequency of multiple stressor types is common and, individuals that experienced one form of ELS often reported experiencing an additional form of ELS. This comorbidity of stressor type compromised the effort to identify the impact of specific stressor types. Furthermore, the assessment of ELS did not quantify the duration of stressful events. It is possible that people who have experienced chronic ELS exhibit greater volumetric differences as compared to those that experienced isolated stressful events, yet this information was not obtained from the ELSQ. In addition, specific age of ELS manifestation was not consistently recorded, rather subjects reported whether the event took place between bracketed age ranges. As such, we were unable to assess Tanner stage in the current study. Based on these limitations future research using alternate methods to obtain detailed reports of ELS would extend the current findings.

Future research is also needed to clarify the mechanisms of the relationships reported in the present study, particularly those examining the role of genetic factors. Gene by environment interactions are common factors of psychopathology among adults who have experienced childhood trauma. Without genetic information, it is difficult to determine whether the relationships found are due solely to age of manifestation or differences in genetic makeup. Examples of possible genetic influences include, 5HTTLPR (a low-expressing polymorphic variant of the serotonin transporter) which has been associated with emotional disorders, particularly in the vulnerability to depression and PTSD (Morey et al. 2011). Additionally, interactions have been seen in brain alterations following early life stress in individuals possessing the BDNF Val66Met polymorphism (Gatt et al. 2010, 2009). It is possible that differences in brain morphometry are due to genetic influences that in addition to the environment, manipulate the way people respond to stress and their brain structure and function (Gilbertson et al. 2002).

Collectively, results from the present study provide evidence that ELS in later childhood/adolescence is associated with smaller adult brain volumes in key regions associated with emotional regulation compared to those with early childhood ELS. The effect of ELS on the brain suggests that mediation of future stressful or demanding events may be more challenging for these individuals, particularly if the event requires effortful control, emotion regulation, or integrated social processing. Future research should identify mechanisms of resilience, which may mitigate the effects of ELS on brain volume in later childhood/adolescence. Given previous research relating ELS with changes in brain morphology, our findings have the potential to inform our understanding of volumetric differences of individuals in adulthood by noting changes based on timing of developmental trauma.