ISSN 0439-755X
CN 11-1911/B

Acta Psychologica Sinica ›› 2023, Vol. 55 ›› Issue (4): 572-587.doi: 10.3724/SP.J.1041.2023.00572

• Reports of Empirical Studies • Previous Articles     Next Articles

The relationship between frontotemporal regions and early life stress in children aged 9 to 12: Evidence from multimodal fMRI

LI Wei1, BIAN Ziming1, CHEN Ximei1, WANG Junjie1, LUO Yijun1, LIU Yong1,2, SONG Shiqing1, GAO Xiao1,2, CHEN Hong1,2()   

  1. 1Faculty of Psychology, Southwest University
    2Key Laboratory of Cognition and Personality (SWU), Ministry of Education, Chongqing 400715, China
  • Published:2023-04-25 Online:2022-12-30
  • Contact: CHEN Hong


Early life stress (ELS) has been used to describe a broad spectrum of adverse and stressful events, including childhood trauma occurring during neonatal life, early and late childhood, and adolescence. Childhood is a vulnerable time point for stressful events due to an immature brain, which increases the risk of psychopathology in later life. However, to date, studies have focused almost exclusively on adolescents and adults, and little is known about the relationship between ELS and the structural and functional brain changes in children. Here, we adopted a multimodal approach combining voxel-based morphometry (VBM) and functional connectivity (FC) to examine the neural substrates of ELS in children aged 9~12 years.

A total of 139 children were recruited for this study. For each participant, the ELS level was assessed and an 8-minute rs-fMRI scan was performed using a 3T Trio scanner. Participants with unqualified data were excluded, resulting in a final sample of 78 participants (39 females; mean age = 10.18) (see Table 1). For statistical analysis, we used the gray matter volume (GMV) and FC to explore the brain structural and functional correlates of children’s ELS and then used a machine learning method to investigate whether and how structural connectivity profiles in predefined brain networks can predict ELS levels. Additionally, exploratory analyses were performed to investigate potential sex differences and age characteristics in GMV and FC associated with children’s ELS.

VBM analysis showed that greater ELS was associated with a larger GMV in the left medial orbitofrontal cortex, right insular cortex, left superior temporal gyrus, and left supplementary motor area. Subsequently, we used these clusters as seed regions to analyze the correlation between FC and stress in children. We found that greater ELS was associated with lower insular-inferior parietal lobule (IPL) connectivity. The results were not influenced by sex, age, total intracranial volume, or head motion (see Table 2, Figure 1). Furthermore, the predictive analysis of machine learning reported that the sensorimotor, frontoparietal, salience, visual, and cerebellar networks could marginally predict ELS scores (see Figure 2). Finally, exploratory analyses showed that there were no significant sex differences in the GMV or FC associated with ELS and that significant correlations of ELS with the GMV of the inferior occipital gyrus were mainly manifested in 9-year-old children (coordinates: x = −53, y = −66, z = −11; cluster size = 249; t = −4.08) (see Figure 3).

Using VBM and FC analyses, we detected structural and functional brain alterations associated with ELS in children aged 9~12 years. Specifically, the VBM analysis mainly reflected that children with high ELS may have abnormal emotional and cognitive functions, such as hypersensitivity to emotional stimuli and over-monitoring of their own behavior. In addition, FC analysis indicated that aberrant interaction of internal and external information may contribute to high ELS in childhood. This study not only provides unique insights into the neural substrates of ELS but may also help identify children who are susceptible to ELS within the general population, which may be advantageous for early prevention strategies and interventions for children.

Key words: early life stress, children, gray matter volume, resting-state functional connectivity, machine learning, structural network