Abstracts

This is a Late Breaking abstract

Rationale: Anxiety is a debilitating psychiatric comorbidity affecting ~50% of epilepsy patients and increases the risk of suicide in epilepsy more than 3-fold. While it is the sixth leading contributor to global disability, up to 30% of patients with anxiety do not respond to currently available therapies. Thus, studies dedicated to understanding the neural circuits involved in anxiety hold therapeutic promise for epilepsy patients and the broader populations who suffer from anxiety. One of the most debilitating symptoms of anxiety is dysregulation in threat processing, leading to avoidance and hyper-arousal. While fMRI studies have shown activation of the salience network, amygdala (AMY), insula (INS) and anterior cingulate (ACC), during threat and anxiety processing, the spatio-temporal dynamics and mechanisms of inter-regional salience network communication are poorly understood.

Methods: We recorded the activity of 335 single neurons from salience network regions (AMY, INS and ACC) in seven subjects undergoing intracranial EEG for seizure onset localization while they performed an anxiety-inducing spatial navigation threat avoidance task. This task involves ~240 trials (127-327 trials) of navigating a spaceship while avoiding collision with an asteroid belt. We recorded neuronal responses to 3 events: threat detection (asteroid appearance), aversive outcomes following threat (hit) and successful threat avoidance (miss). We examined whether neurons exhibited a significant change in activity secondary to any of the measured events by comparing firing rates during events to a 500-ms baseline window preceding the asteroid appearance.

Results: Of the 45 recorded neurons from the ACC, 22 neurons responded to the task. Of these, 12 neurons exhibited a transient, well-timed 2-fold increase in firing rates following threat appearance. 15 ACC neurons transiently increased their firing rate 2.5-fold only following successful threat avoidance while showing no change in firing rates following aversive outcomes. Of the 60 recorded neurons from the AMY, 5 neurons had sustained increase in firing rate from 11 to 14 spikes/s preceding and following aversive outcomes, while showing no change in firing rates in trials with successful threat avoidance. Of the 80 single units recorded from the INS, 11 INS neurons transiently increased their firing rate 2.5-times during aversive events but not during threat avoidance; while 9 INS neurons increased their firing rate only during threat avoidance, but not following aversive outcomes.

Conclusions: Our study revealed neuronal substrates for threat detection in the ACC and AMY and a sub-population of ACC neurons that effectively discriminated successful threat avoidance from aversive events. We also observed neurons in the INS that responded to threat detection, successful threat avoidance, and aversive outcomes. These dynamic interactions may uncover threat prediction, prediction error signals, and reinforcement learning within the salience network. Our mechanistic study of threat processing thus lays the foundation for the development of circuit therapeutics for anxiety disorders in epilepsy.

Funding: NIH KL2 TR001862, NeuroNEXT U24NS107136 and ARDC P30AG066508 to ED