Objective: To investigate the effect of depression and dopaminergic medication on reinforcement learning (RL) in Parkinson’s disease.

Background: Dopaminergic medication in PD is known to improve learning from rewards, while impairing learning from punishment. However, recent clinical studies have demonstrated considerable heterogeneity, with greater medication-related changes in PD patients with more severe depressive symptoms.

Method: To investigate RL performance, we assessed accuracy and win-stay-lose-shift (WSLS) behaviour in 160 PD patients ON medication, 55 PD patients OFF medication, and 59 healthy controls. All subjects performed a probabilistic instrumental RL paradigm during fMRI scanning. The paradigm consisted of visual cues that were probabilistically rewarded (75% vs 25% on the positive outcome). Two conditions were separated: the gain condition, where the reward was €0 or €10m and a loss condition, where punishment was -€10 or €0. Bayesian mixed effects models were used to assess whether accuracy and WSLS-behaviour was modulated by medication status (ON vs OFF), depression (Beck Depression Inventory), valence (GAIN vs LOSS) and outcome (rewarded or punished). Parametric regressors for the expected value and prediction error signals were correlated with brain activity.

Results: We found that accuracy was modulated by valence, medication, and depression status (3-way interaction, brms: 95%CI = [-0.28 -0.01]). Medication increased accuracy for trials in the gain condition, but only in depressed patients. In contrast, non-depressed patients ON medication show reduced accuracy in GAIN trials. Similarly, WSLS-behaviour was modulated by valence, medication, depression, and outcome on the previous trial (4-way interaction, brms: 95%CI = [-0.13 -0.01]). Post-hoc analyses showed that this effect was driven by depressed patients ON medication, showing increased WSLS behaviour in GAIN trials compared with LOSS trials. fMRI data are currently being analysed and will be presented during the conference.

Conclusion: Depressed PD patients who take dopaminergic medication are more sensitive to reward learning than non-depressed and/or unmedicated PD patients, potentially due to alterations in the cortico-striatal reward circuit. Clinically, this finding may explain why depressed PD patients are at risk of developing impulse-control disorders after dopaminergic treatment.

« Back to MDS Virtual Congress 2021

MDS Abstracts - https://www.mdsabstracts.org/abstract/effects-of-dopaminergic-medication-on-reinforcement-learning-in-parkinsons-disease-depend-on-depressive-symptoms/