Repeatability of Neural and Autonomic Responses to Acute Psychosocial Stress

Abstract

FMRI Montreal Imaging Stress Tasks (MIST) have been shown to activate endocrine and autonomic stress responses that are mediated by a prefrontal cortex (PFC)-hippocampus-amygdala circuit. However, the stability of the neurobehavioral responses over time and the ability to monitor response to clinical interventions has yet to be validated. The objective of this study was to compare the fMRI and physiologic responses to acute psychosocial stress in healthy volunteers during initial and follow-up visits approximately 13 weeks later, simulating a typical duration of clinical intervention. We hypothesized that responses to stress would remain highly conserved across the 2 visits in the absence of an intervention. 15 healthy volunteers completed a variant of control math task (CMT) and stress math task (SMT) conditions based on MIST. Neural responses were modeled using an event-related design with estimates for math performance and auditory feedback for each task condition. For each visit, measures of stress reactivity included differential fMRI and heart rate (SMT-CMT), as well as salivary alpha-amylase before and after scanning sessions. The results revealed that differential fMRI, as well as increased heart rate and salivary alpha-amylase from before and after scanning remained similar between visits. Intraclass correlation coefficient (ICC) values revealed areas of reliable task-dependent BOLD fMRI signal response across visits for peaks of clusters for the main effect of condition (SMT vs CMT) within dorsal anterior cingulate cortex (ACC), insula, and hippocampus regions during math performance and within subgenual ACC, posterior cingulate cortex, dorsolateral PFC regions during auditory feedback. Given that the neurobehavioral response to acute stress remained highly conserved across visits in the absence of an intervention, this study confirms the utility for MIST for assessing longitudinal changes in controlled trials that can identify underlying neurobiological mechanisms involved in mediating the efficacy of stress-reduction interventions.

Keywords: functional magnetic resonance imaging; intraclass correlation; psychophysiology; repeatability; stress.

FIGURE 1

Comparisons of visit 1 (V1), visit 2 (V2), and collapsed (visit 1 and 2 combined) assessments of (A) cardiac and (B) alpha-amylase (α-amylase; right) stress responses during V1 (top panel), V2 (middle panel), and collapsed across (bottom panel) MRI scanning visits. Heart rate (HR), measured in beats per minute (BPM) was increased during stressful math compared to control math, during V1, V2, and collapsed across visit (a). Salivary α-amylase (U/ml) was increased following stressful math at +30 min post-MRI compared to –60 min and –30 min pre-MRI collapsed across MRI scanning visits. Salivary α-amylase was increased at +30 min compared to –60 min and –30 min during V1, and at +30 min compared to –60 min during V2. Error bars reflect SEM after adjusting for between-subjects variance (Loftus and Masson, 1994). Asterisks indicate significant main effects of condition (SMT vs CMT) on mean HR and time point (–60 min vs +30 min; –30 min vs +30 min) on salivary α-amylase revealed by ANOVA, Bonferonni-corrected post-hoc, and planned contrast analyses, *p < 0.05; **p < 0.01.

FIGURE 2

Effects of MIST condition: Clusters (NN3) of significant activation for (a) the main effect of Condition (stress math task [SMT] vs control math task [CMT]) and (b) for the main effect of Condition (Negative Feedback [Neg] vs Positive Feedback [Pos]) that survived the volume-corrected threshold (uncorrected voxel-wise p < 0.001, corrected to α = 0.05). Voxel-wise intraclass correlation (ICC) values (≥0.4) for the subject factor during (c) math performance and during (d) auditory feedback resulting from 3dLME analyses.

FIGURE 3

Mean fMRI BOLD percent signal change (% signal) for cluster peaks identified by the analysis for the main effect of Condition with reliable activation across visits (i.e., ICC was ≥0.40, see Tables 2, 3) during (A–D) math performance (left panel) and (E–H) auditory feedback (right panel). Mean estimates of effects demonstrated consistency of directional differences in activation for cluster peak regions identified by the two main 3dLME analyses. ACC, anterior cingulate cortex; dACC, dorsal anterior cingulate cortex; dlPFC, dorsolateral prefrontal cortex; PCC, posterior cingulate cortex.