Impaired reward prediction error encoding and striatal-midbrain connectivity in depression

Abstract

Anhedonia (hyposensitivity to rewards) and negative bias (hypersensitivity to punishments) are core features of major depressive disorder (MDD), which could stem from abnormal reinforcement learning. Emerging evidence highlights blunted reward learning and reward prediction error (RPE) signaling in the striatum in MDD, although inconsistencies exist. Preclinical studies have clarified that ventral tegmental area (VTA) neurons encode RPE and habenular neurons encode punishment prediction error (PPE), which are then transmitted to the striatum and cortex to guide goal-directed behavior. However, few studies have probed striatal activation, and functional connectivity between VTA-striatum and VTA-habenula during reward and punishment learning respectively, in unmedicated MDD. To fill this gap, we acquired fMRI data from 25 unmedicated MDD and 26 healthy individuals during a monetary instrumental learning task and utilized a computational modeling approach to characterize underlying neural correlates of RPE and PPE. Relative to controls, MDD individuals showed impaired reward learning, blunted RPE signal in the striatum and overall reduced VTA-striatal connectivity to feedback. Critically, striatal RPE signal was increasingly blunted with more major depressive episodes (MDEs). No group differences emerged in PPE signals in the habenula and VTA or in connectivity between these regions. However, PPE signals in the habenula correlated positively with number of MDEs. These results highlight impaired reward learning, disrupted RPE signaling in the striatum (particularly among individuals with more lifetime MDEs) as well as reduced VTA-striatal connectivity in MDD. Collectively, these findings highlight reward-related learning deficits in MDD and their underlying pathophysiology.

Fig. 1

Behavioral results: Observed behavioral choices for reward [green—healthy controls; Red—MDD] and punishments [orange—healthy controls; purple—MDD] in controls vs. MDD. The learning curves depict, trial-by-trial, averaged proportion of “correct” stimulus (associated with a probability of 0.8 of winning money) in the gain condition (upper graph), and the “incorrect” stimulus (associated with a probability of 0.8 of losing money) in the loss condition (lower graph) chosen by participants. Error bars represent standard error.

Fig. 2

ROI results: Parameter estimates reflecting reward and punishment prediction errors extracted from the right (a) and left (b) striatum, habenula (c), right insula (d) and ventral tegmental area (VTA) (e) in healthy controls and MDD. Error bars represent standard error. *p < 0.05

Fig. 3

Correlation between number of depressive episodes, and (a) reward prediction error in the right striatum and (b) punishment prediction error in the habenula, in the MDD group. [Note: To evaluate the effect of number of MDEs on PE without potential confounds, these analyses were conducted while adjusting for length of current episode. Unstandardized residuals are shown in the figure; correlation plots with raw scores (with no covariates) are shown in the Supplementary Fig S6]. These correlations were significant even after controlling for both length of current episode and current depression severity (BDI scores) [right striatum: r = −0.60; p = 0.011 and habenula: r = 0.56; p = 0.018]. Information about number of episodes was missing for seven MDD individuals, so the sample size for this correlational analysis was N = 18. PE - prediction error.

Fig. 4

VTA-Right Striatum (a), VTA-Left Striatum (b), VTA-Habenula (c), and VTA-Right Insula (d) connectivity values in healthy controls and MDD. Error bars represent standard error. ^§p < 0.1