Mechanistic link between right prefrontal cortical activity and anxious arousal revealed using transcranial magnetic stimulation in healthy subjects

Abstract

Much of the mechanistic research on anxiety focuses on subcortical structures such as the amygdala; however, less is known about the distributed cortical circuit that also contributes to anxiety expression. One way to learn about this circuit is to probe candidate regions using transcranial magnetic stimulation (TMS). In this study, we tested the involvement of the dorsolateral prefrontal cortex (dlPFC), in anxiety expression using 10 Hz repetitive TMS (rTMS). In a within-subject, crossover experiment, the study measured anxiety in healthy subjects before and after a session of 10 Hz rTMS to the right dorsolateral prefrontal cortex (dlPFC). It used threat of predictable and unpredictable shock to induce anxiety and anxiety potentiated startle to assess anxiety. Counter to our hypotheses, results showed an increase in anxiety-potentiated startle following active but not sham rTMS. These results suggest a mechanistic link between right dlPFC activity and physiological anxiety expression. This result supports current models of prefrontal asymmetry in affect, and lays the groundwork for further exploration into the cortical mechanisms mediating anxiety, which may lead to novel anxiety treatments.

Fig. 1. Overall design of the experiment.

a Overall design of the study. We tested anxiety before and after rTMS using the Neutral, Predictable, and Unpredictable (NPU) threat task. b NPU design: during the neutral blocks, subjects were not at risk for receiving a shock. During the predictable blocks, subjects were at risk for receiving a shock only when cued by a shape presented on the screen. During the unpredictable blocks, subjects were at risk for receiving a shock throughout the entire block. Arrows indicate startle probes, lightning indicates shock presentations. c Sternberg Working Memory (WM) paradigm design: subjects were then presented sequentially with a series of either 5 or 8 letters. On sort trials (5 letters only), subjects were asked to rearrange the letters in alphabetical order. On maintain trials (low = 5 letters; high = 8 letters), subjects were asked to remember the letters without rearrangement. After a brief delay interval, subjects were presented with a letter and a number, and asked to indicate with a forced-choice button press whether the position of the letter in the series matched the number. TMS coils indicate timing of the TMS trains.

Fig. 2. fMRI data and Electric-field (E-field) models used for target localization.

a Pipeline for fMRI localization EPI maps are created for the sort > low contrast. These were sampled using a group-level mask. The peak within the mask was extracted and used as the TMS target (figure shows targets used for all subjects). BOLD activity was extracted from targets and averaged across subjects. Bar indicates Mean ± SEM. b E-field models were calculated at the target for each subject. This was repeated across 24 equally spaced coil handle orientations. The optimized target was defined as the coil handle orientation that generated the largest e-field at the target location. An orientation plot was generated showing the e-field amplitude plotted as a function of the coil handle orientation for each subject. Lines correspond to single-subject e-field model amplitudes.

Fig. 3. Anxiety ratings during cue and the intertrial interval (ITI) of the Neutral, Predictable, Unpredictable (NPU) threat task.

Subjects reported less anxiety after both active and sham stimulation. Scores are averaged across active (A) and sham (S) conditions. Bars indicate Mean ± SEM.

Fig. 4. Potentiated startle during the Neutral, Predictable, Unpredictable (NPU) threat task.

a Anxiety-potentiated startle was significantly increased following active rTMS. b Fear-potentiated startle was marginally increased following active rTMS. Bars indicate Mean ± SEM. *p < 0.05.