Neural responses to negative outcomes predict success in community-based substance use treatment.

BACKGROUND AND AIMS Patterns of brain activation have demonstrated promise as prognostic indicators in substance dependent individuals (SDIs) but have not yet been explored in SDIs typical of community-based treatment settings. DESIGN Prospective clinical outcome design, evaluating baseline functional magnetic resonance imaging data from the Balloon Analogue Risk Task (BART) as a predictor of 3-month substance use treatment outcomes. SETTING Community-based substance use programs in Bloomington, Indiana, USA. PARTICIPANTS Twenty-three SDIs (17 male, aged 18-43 years) in an intensive outpatient or residential treatment program; abstinent 1-4 weeks at baseline. MEASUREMENTS Event-related brain response, BART performance and self-report scores at treatment onset, substance use outcome measure (based on days of use). FINDINGS Using voxel-level predictive modeling and leave-one-out cross-validation, an elevated response to unexpected negative feedback in bilateral amygdala and anterior hippocampus (Amyg/aHipp) at baseline successfully predicted greater substance use during the 3-month study interval (P ≤ 0.006, cluster-corrected). This effect was robust to inclusion of significant non-brain-based covariates. A larger response to negative feedback in bilateral Amyg/aHipp was also associated with faster reward-seeking responses after negative feedback (r(23)  = -0.544, P = 0.007; r(23)  = -0.588, P = 0.003). A model including Amyg/aHipp activation, faster reward-seeking after negative feedback and significant self-report scores accounted for 45% of the variance in substance use outcomes in our sample. CONCLUSIONS An elevated response to unexpected negative feedback in bilateral amygdala and anterior hippocampus (Amyg/aHipp) appears to predict relapse to substance use in people attending community-based treatment.