Brain dynamics during acute psychosocial stress and its association with physiology and childhood adversity

Introduction

Stress response is underlined by neural, autonomic, and endocrine processes that unfold dynamically after encountering a stressor. Neuroimaging studies reported that networks including default mode and attention networks alter their activity patterns during stress induction. However, most studies have examined acute stress response using static activation or connectivity measures, which do not capture temporal patterns of neural responses. Hidden Markov Model (HMM) can be used to identify brain states and measure their temporal dynamics. Understanding how these states alter throughout the phases of stress response, i.e., anticipation phase, acute stress and recovery phase can help in understanding individual differences in stress reactivity. Moreover, how these dynamics relate to physiological markers and childhood adversity may aid in understanding additional vulnerabilities.

Methods

We leveraged two independent samples of healthy men who completed the ScanSTRESS paradigm and two (pre and post task) resting state functional magnetic resonance imaging scans (rs-fMRI) at 3 Tesla (Study 1: n= 39, mean age= 43.7, standard deviation= 9.8; Study 2: mean age= 25.9, standard deviation= 3.3). HMM was fitted to the task data to identify brain states and estimate fractional occupancy (FO). The task-derived states were also fitted to the pre- and post-rs-fMRI data. We employed a sliding-window approach to examine the slope of FO changes across stress and control conditions during the task, as well as during the pre- and post-rs-fMRI data. Cortisol and subjective stress were assessed throughout the measurement. In addition, childhood emotional neglect was measured. Differences between conditions and rs-fMRIs were compared with paired-sample t-tests. Moderation effects were tested with linear mixed models with fixed and interaction effects. Significant effects were followed by paired-sample t-tests as post-hoc comparisons (FDR corrected). Thresholds were p< .05. Statistical analyses were done in MATLAB and Python.

Results

A four‑state HMM solution provided the best fit. During stress, FO increased for a state involving somatomotor and ventral attention networks (p< .04) and decreased for a limbic-default mode state (p< .016). Task‑derived states were expressed during resting-state scans, showing reversed FO slopes from anticipation to recovery (e.g., state involving somatomotor and ventral attention networks p< .008). Cortisol responses modulated resting-state dynamics in Study 1 (p< .025), where higher cortisol response was associated with lower differences in brain dynamics between conditions. However, this was not replicated in Study 2. Emotional neglect was associated with reduced differences between pre- and post-rs-fMRI (e.g., across attention‑related states p< .037).

Conclusions

Results underscore the dynamic features of the brain response during acute stress response phases. These dynamics are furthermore shaped by endocrine responses and childhood emotional neglect. The reduced discrimination between stressful and non-stressful contexts may represent a maladaptive pattern with potential clinical relevance. The findings highlight the value of dynamic modelling for understanding individual variability in stress regulation and may inform future research on stress‑related vulnerability.

References

‘No references’.