(PS3-B29) Behavioral Activation Therapy for Adolescents with Anhedonia: Reward-related Neural and Behavioral Predictors of Treatment Outcome

Background: Behavioral Activation (BA) therapy has been shown to be an effective treatment for adults and adolescents with depression. Given the therapy’s emphasis on increasing exposure to and engagement with intrinsically rewarding activities, BA may be especially therapeutically beneficial for patients with elevated anhedonia, who have been shown to have blunted neural and behavioral responses to rewards. More specifically, individuals with relatively blunted reward system function may be especially likely to benefit from a treatment directly targeting their deficit (i.e., a compensatory model). Alternatively, it may be that individuals with relatively enhanced reward system function are better able to take advantage of an intervention (BA) that capitalizes on their pre-existing strengths (i.e., a capitalization model). The present study tested whether relatively low vs. high levels of reward system function predicts greater symptom change in BA for adolescents with elevated anhedonia.

Methods: Participants included 36 adolescents aged 13 – 18 years with elevated anhedonia who were offered 12 weekly sessions of BA therapy. An fMRI monetary reward task (Forbes et al., 2009) and behavioral measure of reward learning (Probabilistic Reward Task; PRT) were administered at pre-treatment (Pizzagalli et al., 2008). Multilevel models tested whether pre-treatment striatal (putamen, caudate and nucleus accumbens) response to rewards or losses (relative to neutral outcomes) or reward learning predicted anhedonia (SHAPS) symptom change from pre-treatment to post-treatment (Predictor x Time interactions) (Snaith, et al., 1995). Models controlled for baseline anhedonia (SHAPS), depression (CES-D), age, gender, and SSRI prescription.

Results: Intent-to-treat multilevel models revealed that anhedonia (SHAPS) scores decreased significantly over the course of treatment (b = -0.56, t(35.6) = -5.50; p < 0.001). Greater pre-treatment right caudate (t(38.1) = -2.26; p = 0.029) and right nucleus accumbens (t(34.5) = -2.19; p = 0.036) response to wins (relative to neutral outcomes) predicted greater reductions in anhedonia. In contrast, striatal response to losses (relative to neutral outcomes) did not predict changes in anhedonia (all ps > .24). Rank ordering the bivariate correlations between change in anhedonia and all striatal ROIs (6 total: right and left putamen, caudate and nucleus accumbens) for each contrast separately (reward vs. neutral and loss vs. neutral), activation in 5/6 of the ROIs to the reward vs. neutral contrast exhibited stronger correlations with changes in anhedonia than the activation of the ROIs to loss vs. neutral contrast. Pre-treatment reward learning did not predict anhedonia outcome (t(27.8) = -0.11; p = 0.91).

Conclusion: Consistent with a “capitalization” rather than “compensatory” model of change, participants who had relatively higher levels of neural (reward circuit) reactivity to rewards, controlling for anhedonia and depressive symptom severity, derived the greatest benefit from an intervention (BA) putatively targeting reward system function (Cheavens et al., 2012).