Abnormal dynamic functional connectivity during fear extinction learning in PTSD and anxiety disorders

Abstract

Examining the neural circuits of fear/threat extinction advanced our mechanistic understanding of several psychiatric disorders, including anxiety disorders (AX) and posttraumatic stress disorder (PTSD). More is needed to understand the interplay of large-scale neural networks during fear extinction in these disorders. We used dynamic functional connectivity (FC) to study how FC might be perturbed during conditioned fear extinction in individuals with AX or PTSD. We analyzed neuroimaging data from 338 individuals that underwent a two-day fear conditioning and extinction paradigm. The sample included healthy controls (HC), trauma-exposed non-PTSD controls, and patients diagnosed with AX or PTSD. Dynamic FC during extinction learning gradually increased in the HC group but not in patient groups. The lack of FC change in patients was predominantly observed within and between the default mode, frontoparietal control, and somatomotor networks. The AX and PTSD groups showed impairments in different, yet partially overlapping connections especially involving the dorsolateral prefrontal cortex. Extinction-induced FC predicted ventromedial prefrontal cortex activation and FC during extinction memory recall only in the HC group. FC impairments during extinction learning correlated with fear- and anxiety-related clinical measures. These findings suggest that relative to controls, individuals with AX or PTSD exhibited widespread abnormal FC in higher-order cognitive and attention networks during extinction learning and failed to establish a link between neural signatures during extinction learning and memory retrieval. This failure might underlie abnormal processes related to the conscious awareness, attention allocation, and sensory processes during extinction learning and retrieval in fear- and anxiety-related disorders.

Fig. 1. Abnormal dynamic connectivity in the AX and PTSD groups during extinction learning.

A Network components showing significant differences in dynamic functional connectivity (FC) between HC and AX groups across multiple networks. Each sphere represents a brain region; the color of the sphere represents its network assignment, and the size of the sphere presents the weighted number of abnormal connections. B Mean FC of the identified network components during extinction learning for the CS+ and CS−. C The proportion of significantly impaired edges within or between the 8 subnetworks. A darkly shaded cell indicates that the connections of that network pair (indexed from x- and y-axis) were extensively impaired. D Distribution of mean connectivity change (ΔFC, comparing late (last 4 trials) minus early (first 4 trials) extinction learning) with each of the 8 subnetworks during CS+ or CS− processing. E–H panels are similar to panels A–D, but for HC vs. PTSD analyses. AX anxiety group, HC healthy controls, PTSD posttraumatic stress disorder.

Fig. 2. Shared and distinct impaired edges of the AX and PTSD groups during extinction learning.

A There were 152 edges that were impaired in both AX and PTSD groups. Red circle highlights the dorsolateral prefrontal cortex (dlPFC) as an important hub with substantial overlapping impairments across the two disorders. B Separate display of the impaired FC between the dlPFC and the rest of the brain. Note substantial impairments in FC between dlPFC and key nodes of the ‘fear network’. dACC dorsal anterior cingulate cortex, vmPFC ventromedial prefrontal cortex, rACC rostral anterior cingulate cortex.

Fig. 3. Change in functional connectivity during extinction learning predicts brain activation during extinction memory recall.

The correlation between extinction-induced connectivity change and vmPFC activation during memory recall is significant within the HC, but not within the AX or PTSD group. Red circles in the AX and PTSD highlight the absence of any correlations from the same vmPFC location observed to be correlated within the HC. Scatter plots show correlations between change in FC and beta weights extracted from the vmPFC for each of the three groups.

Fig. 4. Change in functional connectivity during extinction learning correlates with functional connectivity during memory recall test.

A The FC change positively correlated with FC during extinction memory recall within the HC group. B The FC change negatively correlated with connectivity during memory recall within the AX group.

Fig. 5. The canonical correlation analysis between connectivity change during extinction learning and clinical measures.

A The correlation between connectivity variate and clinical variate is significant for the HC and AX cohort. B The canonical loadings of the clinical measures (top) and connectivity changes with each of the 8 subnetworks (bottom). C, D Same as in panels A and B, but for the TENC and PTSD cohort.