The Desegregation of Neural Networks During Worry Induction in Late Life: An Effective Connectivity Analysis

Severe worry is a core component of anxiety and depressive disorders and is independently associated with significant morbidity and mortality. However, the neural basis of worry is poorly understood. We investigated effective connectivity (EC) using functional magnetic resonance imaging of a naturalistic worry induction and reappraisal task in late life.

A total of 112 participants age >50 years with varying worry severity completed a personalized, in-scanner worry induction and reappraisal task. We calculated voxelwise EC in neutral, worry, and reappraisal conditions with generalized psychophysiological interactions using seeds in the subgenual anterior cingulate cortex (ACC), dorsal ACC, and left and right amygdalae and used paired t tests to compare conditions. We assessed clusters for association with in-scanner worry severity using linear regression.

During the worry condition, EC increased between the subgenual ACC and the default mode network (DMN) and major hubs of the executive control and salience networks. Left amygdala EC to the posterior cingulate also increased during worry, and dorsal ACC connectivity to primary sensory and motor regions decreased. Reappraisal reduced subgenual and dorsal ACC EC observed during worry and the EC between the left amygdala and regions of the dorsal attention network. Broadly, left amygdala EC was robustly associated with in-scanner worry severity.

Worry induction robustly engaged the DMN and increased connectivity with other high-order associative networks, potentially subsuming cortical resources. Reappraisal reduced these connectivities and disengaged the amygdala from areas associated with top-down attention. These findings could inform targets for neuromodulatory treatment of severe worry in older adults.