Threatening and Interpersonally Proximal Experiences are Associated With Increased Structural Network Connectivity in Young Children

• In Event: The Prenatal Environment and Early Experiences: Neurodevelopmental Consequences of Stress in Infants and Young Children

Abstract

Background: Early life stress (ELS) is associated with altered neurodevelopment. However, the neurodevelopmental trajectories in early childhood associated with stress are not well-understood. This may be, in part, due to variability in measurement. In terms of stress, previous work has commonly measured multiple diverse forms of stress cumulatively (e.g., Evans et al., 2013; Felitti et al., 1998 [ACES]), resulting in qualitatively different stressors being weighted equally. Previous studies have focused on specific regions of interest (see McLaughlin, Weissman, Bitran, 2019). It is unclear, however, how ELS may influence the broader development of cognitive networks. Our pre-registered study investigates the association between parent-reported threatening life experiences (e.g., witnessed domestic violence, involved in a serious accident, attacked by an animal) and child brain structural network connectivity using a multivariate approach. Specifically, we tested the predictive qualities of stressful life experiences at different interpersonal proximities to the child.

Methods: The sample included data from 96 children ages 5-8 years (M±SD age=6.94±0.67 years; 56% male). Threatening life experiences were measured using a parent-reported questionnaire and then summed to derive a total threatening life experiences (TTE) score. Endorsed events were also categorized based on interpersonal proximity to the child: self (3), family or friend (2), or in the neighborhood or broader community (1). These experienced events, weighted by proximity, were summed to derive proximity-weighted total event (PWTE) scores. Structural networks were indexed using diffusion spectrum imaging. Data were rigorously pre-processed, and fiber tracking was conducted (Yeh, 2011). Parcel-to-parcel (Glasser, 2016) connectivity was indexed as mean quantitative anisotropy (QA; Yeh, 2013) along the fibers between parcels. Connections were then averaged according the network membership (Yeo, 2011), and these network-level maps were entered into an Elastic Net analysis (with leave one subject out cross-validation) predicting TTE and PWTE scores. Elastic Net modeling performs well with small samples compared to other machine learning approaches and favors sparse solutions allowing for better interpretability. Finally, significant predictors were entered into a partial-eta model to test if events interpersonally proximal to the child explained more variance in the models. Specifically, events from each sphere were systematically removed, and the contribution of those events was quantified as the change in model R2.

Results: We found a positive association between each frontoparietal-somatomotor internetwork connectivity (FPN-SM) and visual intranetwork connectivity (VIS) and both TTE and PWTE scores. These associations held even after covarying for sex and age and correcting for multiple comparisons (qs<0.05). Partial-eta modeling revealed that events occurring directly to the child explained the most variance in each of the models (VIS: self=0.020, family=0.015, community=0.002; FPN-SM: self=0.028, family=0.014, community=0.008).

Conclusion: Findings suggest that threatening events that occur directly to the child may influence the development of primary sensory and higher cognitive networks in a manner consistent with accelerated maturation. These structural changes may underlie the differences in attentional behaviors observed in children experiencing ELS. Further work should examine if events occurring at specific ages explain variance in brain structure, or if these findings are associated with individual differences in behavioral or emotional outcomes.

Authors

• Jenna H. Chin, University of Denver
  Presenting Author

• M. Catalina Camacho, Washington University in St. Louis
  Non-Presenting Author

• Kathryn L. Humphreys, Vanderbilt University
  Non-Presenting Author

• Caroline P. Hoyniak, Washington University in St. Louis
  Non-Presenting Author

• Susan B. Perlman, Washington University in St. Louis
  Non-Presenting Author