Data used in the preparation of this article were obtained from the Adolescent Brain Cognitive Development (ABCD) Study (https://abcdstudy.org), which is stored in the NIMH Data Archive (NDA). A list of participating cities and a complete list of study investigators can be found at This manuscript reflects the views of the authors and may not reflect the opinions or views of the NIH or ABCD consortium investigators.
Psychiatry and this material has been adapted for this thesis with the permission of the publisher and my co-authors.
Introduction
Therefore, identifying the effect of childhood maltreatment on brain structures may increase our understanding of the risk factors associated with psychopathology. Additionally, a binary approach in which individuals are categorized into an “exposed” group versus a “non-exposed” control group does not capture the range of severity of exposure to child abuse. Given the high co-occurrence of different types of child abuse, it is possible that child abuse is better represented by a hierarchical structure; that is, a.
The broader goal of the master's thesis is to investigate the relationship between childhood maltreatment and brain structure, measured by cortical thickness and gray matter volume (GMV) using latent factors of childhood maltreatment.
Study 1
Methods
- Participants
- Trauma Measure
- Image Acquisition, Quality Assurance, and Processing
- Statistical Analyses
A summary of demographic data based on the final sample is provided in Supplementary Material 2. A description of image acquisition, quality assurance and processing procedures for the ABCD study is detailed elsewhere (Hagler et al., 2019). Post-stratification weights based on propensity scores for age, sex, race/ethnicity, family income, family type and parental employment, household size, and participant's region of US origin were used for all analyzes to account for stratifying the sample across data collection sites.
Consequently, the model for testing the association between trauma exposure and brain structure was as follows: brain regions = β*age + β*sex + β*race/ethnicity + β*MRI scanner model + β*mean cortical thickness or total GMV + β* latent trauma factor, where i = 1..68 (i.e., number of brain regions) for cortical thickness and GMV and i = 1..19 for subcortical GMV analysis.
Results
For cortical thickness, sensitivity findings were largely consistent with the primary results when controlling for family income and highest parental education level as additional covariates. Greater trauma exposure remained negatively associated with the bilateral superior frontal gyri and the right caudal middle frontal gyrus, and positively associated with the left posterior cingulate. Although trauma exposure was associated with thicker cortices in the left isthmus cingulate at uncorrected levels, this did not survive FDR correction during.
When controlling for family income and parental education for the volume sensitivity analyses, there was no significant association between subcortical volume and trauma exposure (Supplementary Material 8).
Discussion
The results of Study 1 showed that trauma exposure was associated with variation in regional cortical thickness and subcortical GMV. associated with thinner cortices in the bilateral superior frontal gyrus and right caudal middle frontal gyrus and thicker cortices in the left posterior cingulate and left isthmus cingulate. In contrast to the thinner frontal cortices, thicker cortices were found in the posterior and isthmus cingulate cortices. These results suggest that the observed relationship between trauma exposure and GMV in subcortical regions may be explained by low SES, which is closely related to children's vulnerability to maltreatment (Paxson & Waldfogel, 2003).
Although Study 1 reveals important structural aberrations associated with trauma exposure, it may be useful to expand the definition of childhood adversity to include environmental stressors or different types of childhood adversity that may affect brain development.
Study 2
Methods
- Participants
- Stressor Measures
- Statistical Analyses
- Exploratory Structural Equation Modeling
- Bifactor Modeling
- Structural Equation Modeling
Gross housing density (ie, housing units per acre) was obtained from the Environmental Protection Agency (EPA) based on the child's zip code. Population density was obtained from the National Aeronautics and Space Administration (NASA) Socioeconomic Data and Applications Center (SEDAC) based on the 2010 census. A confirmatory bifactor analysis was performed with 2,875 hold-out participants based on the results of the ESEM.
GMV analyzes were performed with 68 cortical structures based on the Desikan-Killiany atlas and 19 subcortical structures.
Results
- ESEM
- Bifactor Modeling
- Structural Equation Modeling
- Cortical Thickness
- Cortical and Subcortical GMV
In general, the items that most affected Factor 3 reflected the disadvantage of the child's neighborhood environment. In order to extract a general factor that can explain the similarities between environmental stressors, bifactor modeling was conducted with items obtained from the ESEM. In addition, the intercorrelations in our model were relatively low, which could suggest the possibility that there is no general factor.
However, previous studies suggest that although the interfactor correlations are low, the general factor can still be a valid measure when at least some items have strong loadings on the general factor and criterion validity can be demonstrated where the general factor predicts relevant outcomes (Moore, 2020 ). Therefore, analyzes testing whether the general factor and specific factors are associated with brain structures were further explored. After FDR correction for multiple comparisons, the common stressor factor was associated with cortical thinning across cortices (see Figure 5; Supplementary Material 12).
Specifically, high general factor scores were associated with cortical thinning in bilateral parahippocampal gyri, lateral orbitofrontal cortex, cuneus cortex, pericalcarine cortex, lingual gyri and lateral occipital cortex, and left precentral gyrus (pfdr values £ 0.042). Specifically, greater neighborhood deprivation was associated with cortical thinning in bilateral rostral middle frontal gyri, bilateral postcentral gyri, left caudal middle frontal gyrus, left superior frontal gyrus, left precentral gyrus, left paracentral lobule, left superior temporal gyrus, left superior parietal gyrus, left supramarginal gyrus, and right medial orbitofrontal gyrus (pfdr values £ 0.037). Specifically, greater urbanicity was associated with thicker cortices in bilateral rostral anterior cingulate cortex, bilateral insulae, right rostral middle frontal cortex, right lateral orbitofrontal cortex, right medial orbitofrontal cortex, and right entorhinal cortex (pfdr values £0.030).
On the contrary, greater urbanicity was associated with thinner cortices in left supramarginal cortex (pfdr value < 0.001). After FDR correction for multiple comparisons, the common factor and urbanicity were associated with changes in GMV across broad regions (Supplementary Material 13).
Discussion
The similar global reductions in brain volume associated with the common factors of early life stressors and psychopathology suggest that early life adversity and environmental stressors are closely linked to psychopathology. The current findings are consistent with studies showing that low SES is associated with language deficits. The current study builds on previous work by using a large sample of children to demonstrate that structural abnormalities associated with neighborhood disadvantage manifest as early as ages 9 to 10 years.
In contrast to the general factor and neighborhood deprivation, urbanicity was associated with predominantly thicker cortices and globally greater cortical and subcortical GMV. Regarding mental health, previous studies have found that urban living is associated with an increased risk of psychopathology, especially psychosis (Krabbendam & Van Os, 2005; Myin-. Germeys, Delespaul, & Van Os, 2005; Van Os, Kenis, & Rutten, 2010). Furthermore, previous work on brain structure found that urban education was associated with reduced GMV in men with psychotic disorders (Frissen, van Os, Peeters, Gronenschild, & Marcelis, 2018), although no relationship was found with cortical thickness (Frissen, Van Os , Habets, Gronenschild, & Marcelis, 2017).
When interpreting the bifactor model, it is important to note that our urbanicity factor reflects the remaining variance after removing the common variance associated with the general factor. Thus, the finding that the general factor is associated with smaller brain volumes is consistent with previous studies showing the deleterious effects of living in high-density areas. In contrast, the urbanicity factor reflects the variance that remains after removing any adverse effects associated with the general factor.
By doing this we can now see that urbanity is associated with larger brain volumes. The bifactor model is advantageous in this study because it allows us to go beyond the obvious relationships and provides us with more nuanced information about the additional contribution of urbanity, on top of the general factor.
General Discussion
Overall, the results of the current thesis suggest that childhood trauma and environmental stressors may be risk factors for structural abnormalities in the developing brain. However, chronic trauma exposure (which may not yet be apparent in this young sample) may eventually lead to greater changes in the developing brain over time. Second, converging findings of globally smaller brain volumes associated with the general factor of stressors and the general factor of psychopathology in the same sample (Durham et al., 2021) substantiate a close relationship between.
Although analyzes with brain structures indicate that the general and specific factors found in the present study are useful measures in predicting abnormalities of brain structures, further investigation of the discriminant validity is needed as well as replication using longitudinal data sets (Lahey, Moore, Kaczkurkin, & Zald, 2020). Despite these limitations, the current thesis builds on and extends previous literature on the impact of early life stressors on structural abnormalities in the developing brain. Beating the brain on maltreatment: Empirical and meta-analytic studies of the relationship between maltreatment and hippocampal volume across childhood and adolescence.
Development of the cerebral cortex through adolescence: a multi-sample study of interrelated longitudinal changes in cortical volume, surface area and thickness. The impact of social support on the mental health of older adults: A case from Shaanxi Province, China. Assessment of culture and environment in the Adolescent Brain and Cognitive Development Study: Rationale, description of measures, and early data.
Missing: There were 3 participants excluded for missing propensity weight data (PS weight), 6 for missing Child Behavior Checklist (CBCL) data, 62 for missing a variable indicating the normality/abnormality of the structural MR images ("mrif_score "), 223 for missing data on an initial quality assurance variable ("iqc_t1_ok_ser"), 51 for missing data on an additional quality assurance variable ("fsqc_qc"), 2 for missing gender data, 13 for missing race-ethnicity data, 337 for missing elements regarding the K-SADS trauma events checklist and 1 for missing the whole brain variable (mean cortical thickness or total GMV). 1 A car accident in which your child or another person in the car was injured enough to require medical attention. 10 Beaten to the point of bruising by an adult in the home 11 A non-family member threatened to kill your child.
14 An adult in the home touched your child in his or her privates, let your child touch their privates, or did other sexual things with your child. Missing: There were 3 participants excluded for missing propensity weight data (PS-weight), 6 for missing Child Behavior Checklist (CBCL) data, 62 for missing a variable indicating the normality/abnormality of the structural MRI images (“mrif_score”), 223 for missing data on an initial quality assurance variable ("iqc_t1_ok_ser"), 51 for missing data on an additional quality assurance variable ("fsqc_qc"), 2 for missing gender data, 13 for missing race-ethnicity data, and 1 for missing the whole brain variable ( mean cortical thickness or total GMV). Items with standardized loadings in bold were retained for the next round of a confirmatory factor analysis in the second random sample (N = 2,875).
Exploratory factor analysis identifies a single latent factor of trauma exposure
Regions with significant associations between cortical thickness and latent trauma
Exploratory factor analysis identifies four factors of environmental stressors
A bifactor analysis delineates general and specific factors of environmental stressors
Regions with significant associations between cortical thickness and the general factor
Regions with significant associations between cortical thickness and neighborhood deprivation
Regions with significant associations between cortical thickness and urbanicity
Regions with significant associations between GMV and the general factor
Regions with significant associations between GMV and urbanicity