Affective bias predicts changes in depression during deep brain stimulation therapy

Abstract

Introduction: Deep brain stimulation (DBS) is a promising treatment for refractory depression, utilizing surgically implanted electrodes to stimulate specific anatomical targets within the brain. However, limitations of patient-reported and clinician-administered mood assessments pose obstacles in evaluating DBS treatment efficacy. In this study, we investigated whether an affective bias task, which leverages the inherent negative interpretation bias seen in individuals with depression, could serve as a reliable measure of mood changes during DBS therapy in patients with treatment-resistant depression.

Methods: Two cohorts of patients (n = 8, n = 2) undergoing DBS for treatment-resistant depression at different academic medical centers completed an affective bias task at multiple time points before and after DBS implantation. The affective bias task involved rating the emotional content of a series of static photographic stimuli of facial expressions throughout their DBS treatment. Patients' ratings were compared with those of non-depressed controls to calculate affective bias scores. Linear mixed-effects modeling was used to assess changes in bias scores over time and their relationship with depression severity measured by the Hamilton Depression Rating Scale (HDRS-17).

Results: We observed significant improvements in total affective bias scores over the course of DBS treatment in both cohorts. Pre-DBS, patients exhibited a negative affective bias, which was nearly eliminated post-DBS, with total bias scores approaching those of non-depressed controls. Positive valence trials showed significant improvement post-DBS, while negative valence trials showed no notable change. A control analysis indicated that stimulation status did not significantly affect bias scores, and thus stimulation status was excluded from further modeling. Linear mixed-effects modeling revealed that more negative bias scores were associated with higher HDRS-17 scores, particularly for positive valence stimuli. Additionally, greater time elapsed since DBS implantation was associated with a decrease in HDRS-17 scores, indicating clinical improvement over time.

Discussion: Our findings demonstrate that the affective bias task leverages the inherent negative interpretation bias seen in individuals with depression, providing a standardized measure of how these biases change over time. Unlike traditional mood assessments, which rely on subjective introspection, the affective bias task consistently measures changes in mood, offering potential as a tool to monitor mood changes and evaluate the candidacy of DBS treatment in refractory depression.

Keywords: affective bias; deep brain stimulation; facial emotion; mood proxy; subcallosal cingulate; treatment-resistant depression; ventral capsule striatum.

Figure 1

Deep brain stimulation (DBS) timeline. (A) Emory E001. (B) Baylor B001 (COVID-19 disruptions). (C) Baylor B002 (patient paused research activities). Example Emory (A) and Baylor (B, C) patient deep brain stimulation timelines. B001 timeline (B) Depicts impact of COVID-19 disruptions on the study. B002 timeline (C) Depicts impact of study disruptions resulting from patient pausing research activities.

Figure 2

Deep brain stimulation electrode placement. (A) Emory. (B) Baylor. Sagittal and coronal views of DBS electrodes and regions of interest for Emory and Baylor patients. Emory patients (A) received DBS to the SCC. Baylor patients (B) received DBS to the SCC and VCVS. DBS, deep brain stimulation; SCC, subcallosal cingulate; VCVS, ventral capsule/ventral striatum.

Figure 3

Aective bias task information and visit structure. (A) Morphed faces. (B) Task slide. (C) Outpatient visit structure (Emory). (D) Outpatient visit structure (Baylor). Visualization of valence and intensity parameters for faces shown in the aective bias task (A). Faces were morphed from neutral to sad, and neutral to happy in steps of 10, 30, 50, and 100%. Faces were presented on a black background (B). Participants used a slider bar to rate the emotion of the face. Outpatient visit structure shows order of behavioral inventory tasks, and changes to stimulation parameters over the 1 h visit (C,D). Original images were taken from the NimStim facial expression stimulus set (Tottenham et al., 2009).

Figure 4

Pre- to post- DBS change in affective bias score. (A) Trial-level total bias change. (B) Permutation test distribution. (C) Trial-level bias change by valence. Total bias (across positive and negative trials and all intensities) becomes less negative on average after DBS compared to before DBS onset (A). Permutation tests demonstrate the robustness of total bias shift between pre- to post-DBS timepoints, where the true distribution difference is more extreme than shuffled pre- and post-DBS distribution differences (B). Total bias distributions are separated by valence (positive, negative), which shows a significant pre- to post-DBS bias change for the positive but not negative stimulus condition (C).