Computational Phenotyping of Electroconvulsive Therapy Outcomes in Treatment-Resistant Depression

ABSTRACT IMPORTANCE Electroconvulsive therapy (ECT) is an effective medical procedure for patients with treatment-resistant depression. However, quantitative neural and behavioral measures that characterize how patients respond to ECT treatment are largely lacking. OBJECTIVE Determine whether neurocomputational models that integrate information about adaptive learning behavior and associated affective experiences can characterize neurobehavioral changes in patients whose depression improves following ECT treatment. DESIGN This observational study included two research visits from 2020-2023 that occurred before and after standard-of-care ECT for treatment-resistant depression. This report focuses on “visit 2”, which occurred after patients received their initial ECT treatment series. SETTING Wake Forest University School of Medicine; Atrium Health Wake Forest Baptist Psychiatric Outpatient Center; Atrium Health Wake Forest Hospital. PARTICIPANTS Participants who received ECT for treatment-resistant depression (“ECT”), and participants not receiving ECT but with depression (“non-ECT”) or without depression (“no-depression”) were recruited from the Psychiatric Outpatient Center and community, respectively. EXPOSURES Computerized delivery of a Probabilistic Reward and Punishment with Subjective Rating task with functional magnetic resonance imaging. MAIN OUTCOMES AND MEASURES Computational modeling of choice behavior provided parameters that characterized learning dynamics and associated affect dynamics expressed through intermittent Likert scale self-reports. Multivariate statistical analyses relating model parameters, neurobehavioral responses, and clinical assessments. RESULTS ECT (N=21; 47.6% female), non-ECT (N=36; 69.4% female), and no-depression (N=38; 65.8% female) participants. Parameters derived from computational models fit to behavior elicited during learning and the expression of affective experiences for all groups reveled specific changes in patients who responded favorably to ECT. ECT-responders demonstrated increased rates of learning from rewarding trials, normalized affective response to punishments, and an increase in the influence of counterfactual ‘missed opportunities’ on affective behavior. Additionally, ECT-responders’ showed changes in BOLD activity regions specific to each of these parameters. ECT-responders’ BOLD-responses to surprising punishments and counterfactual missed opportunities were altered from visit 1 to visit 2 in the inferior frontal operculum, Rolandic operculum, precentral gyrus, and caudate. CONCLUSIONS AND RELEVANCE Computational models of neurobehavioral dynamics associated with learning and affect can describe specific hypotheses about neurocomputational-mechanisms underlying favorable responses to treatment-resistant depression. Our results suggest computational estimates of learning and affective dynamics may aid in identifying depression phenotypes and treatment outcomes in psychiatric medicine where objective measures are largely lacking. KEY POINTS QUESTION How does ECT treatment change neurocomputational measures of learning and affective experiences in patients with treatment-resistant depression? FINDINGS In this observational study, computational models were used to quantify the behavioral dynamics of learning and associated changes in subjective feelings in patients who underwent ECT treatment for treatment resistant depression and controls. In ECT-responders we observed increases in reward-based learning, normalized affective responses to surprising positive and negative outcomes, and associated changes in fMRI-measured BOLD-responses. MEANING Computational phenotyping of task behavior and associated brain responses provides quantification of complex neurobehavioral dynamics and provides specific insight into the neurobehavioral mechanisms underlying successful ECT treatment.