Vagus nerve stimulation for treatment-resistant depression: behavioral and neural effects on encoding negative material

Abstract

Objectives: Vagus nerve stimulation (VNS) can improve depression. Cognitive models of depression highlight an over-representation of negative thoughts and memories, with depressed individuals showing memory facilitation for negative material. We hypothesized that the antidepressant action of VNS may emerge through corrective influences on 'negativity bias' in memory. We therefore examined the impact of VNS on emotional memory and its underlying brain activity.

Methods: We tested a single patient undergoing VNS for treatment-resistant depression (TRD). Stimulation was set at a 30/66-second on/off cycle during three encoding blocks when the patient viewed randomly presented positive, negative, and neutral words. Following each block, VNS was switched off and the patient identified previously seen words from distractors in a subsequent recognition memory task. The patient was scanned using functional magnetic resonance imaging (fMRI) during the first encoding block.

Results: There was robust recall of negative material viewed during 'off' cycles of VNS but subsequent memory of negative words was attenuated during active VNS ('on' cycles). VNS did not influence memory for neutral and positive words. With neuroimaging, direct modulatory effects of VNS were observed in dorsomedial, dorsolateral, and orbital regions of the prefrontal cortex. Moreover, during encoding of negative words, compared with positive and neutral words, VNS also modulated activity within orbitofrontal, ventromedial and polar prefrontal cortices, midcingulate cortex, and brain stem.

Conclusions: Our observations show that VNS can interfere with memory of negative information, an effect that may contribute to its antidepressant role. Neuroimaging implicated regions including the ventral and medial prefrontal cortex as an underlying neural substrate.

Figure 1

A The panel illustrates the encoding block design. Vagus nerve stimulation (VNS) was set to a 30 second ‘on’ / 66 second ‘off’ cycle (red line at bottom of figure A) as the patient viewed randomised 1 second presentations of positive, negative and neutral words (illustrated as magenta, green and blue ‘events’. During the first of three encoding blocks, the patient was scanned using fMRI, signalling the onset of the VNS prior to scanning by a buttonpress. B Across later recognition memory blocks, performed 20 minutes after each of the three encoding blocks, the patient made significantly more recognition errors to words encoded when the stimulator active (i.e. in the 30 second ‘on’ period). This effect was almost entirely attributable to interference with the encoding of negative words (which, compared to normative data was relatively enhanced when the stimulator was ‘off’). The bar chart depicts the subject's recognition memory performance for each word category across all recognition blocks (filled = stimulator ‘on’ at encoding). C Activity in a discrete set of brain regions reflected the selective behavioural impact of VNS on the encoding of negative words. VNS enhanced the activity brainstem (pons) ventromedial prefrontal cortex, right anterior insula/lateral orbitofrontal (area 47) cortex and frontal pole during the processing of negative (but not positive or neutral) words that were subsequently recognized. Interestingly activity within these regions, particularly area 47 and frontal pole also predicted subsequent forgetting of words during the first encoding task, illustrated in a conjunction analysis depicted in 2 panels on the right. Significant activity relating to this interaction is plotted on a normalised T2*weighted echoplanar volume from the subject to illustrate fully the field of view. Data is presented at P<0.001, uncorrected for clusters of 10 or more contiguous voxels. Peak coordinates and magnitudes of the effects are given in Table 1.