Functional connectivity reveals inefficient working memory systems in post-traumatic stress disorder

Abstract 

We applied a covariance-based multivariate analysis to functional magnetic resonance imaging (fMRI) data to investigate abnormalities in working memory (WM) systems in patients with post-traumatic stress disorder (PTSD). Patients (n=13) and matched controls (n=12) were scanned with fMRI while updating or maintaining trauma-neutral verbal stimuli in WM. A multivariate statistical analysis was used to investigate large-scale brain networks associated with these experimental tasks. For the control group, the first network reflected brain activity associated with WM updating and principally involved bilateral prefrontal and bilateral parietal cortex. Controls' second network was associated with WM maintenance and involved regions typically activated during storage and rehearsal of verbal material, including lateral premotor and inferior parietal cortex. In contrast, PTSD patients appeared to activate a single fronto-parietal network for both updating and maintenance tasks. This is indicative of abnormally elevated activity during WM maintenance and suggests inefficient allocation of resources for differential task demands. A second network in PTSD, which was not activated in controls, showed regions differentially activated between WM tasks, including the anterior cingulate, medial prefrontal cortex, fusiform and supplementary motor area. These activations may be linked to hyperarousal and abnormal reactivity, which are characteristic of PTSD.

Keywords: Post-traumatic stress disorder, Working memory, Functional MRI, Multivariate analysis, Brain mapping, Image processing