Role of serotonin in modulation of decision-making in Parkinson's disease

Abstract

Background: Dysfunction of dopaminergic pathways has been considered to play a pivotal role in Parkinson's disease (PD), affecting the processing of emotional and rewarding information, and potentially leading to symptoms of depression or apathy. However, some aspects of motivation in PD might be affected by non-dopaminergic mechanisms.

Aim and method: The objective of this experimental medicine study was to investigate the contribution of serotonergic modulation via administration of citalopram (20 mg) for 7 days on motivated decision-making in twenty PD patients, measured using several different computerised tasks and clinical questionnaires that probe different aspects of decision-making. Twenty healthy controls were additionally tested without medication to assess any baseline differences between the two groups.

Results: Results indicated that PD patients were overall less motivated than controls on an effort- and reward-based decision-making task. Citalopram increased or decreased willingness to exert effort for reward, depending on whether baseline motivation was high or low, respectively. A task assessing decision-making under risk revealed higher levels of risk aversion for potential losses in PD patients, which neither serotonin nor the patient's regular dopaminergic medication seemed to restore. However, citalopram in PD was associated with more risk-seeking choices for gains, although patients and controls did not differ on this at baseline.

Conclusion: The results provide evidence for a role of the serotonergic system in influencing some aspects of motivated decision-making in PD processes.

Keywords: Decision-making; Parkinson’s disease; serotonin.

Figure 1.

Schematic outline of study design. Patients were invited for a screening visit, and if enrolled, returned for a first dose visit and a research visit for the placebo and citalopram phase each. Placebo and citalopram orders were counterbalanced. Healthy controls were invited for a combined screening and research visit.

Figure 2.

Effort-based and risky decision-making task designs. (a) Effort-based decision-making: The task was divided into a decision and a work phase. In the decision phase, participants were asked to give yes or no responses to offers with different reward and effort levels on a keyboard. Out of 125 decisions in this phase, 10 decisions had to be acted upon in the following work phase, depending on whether the offer was accepted or rejected. Everyone received the same 10 trials to follow through during the decision phase. If an offer was accepted, squeezing the bar to the associated effort level resulted in a gain of apples that translated into money. If an offer was rejected, no action was required. (b) Risky decision-making: This binary choice task represented a risky gamble to increase a win or avoid a loss of money. In the Gain condition, the risky option could result in a larger monetary win than the certain option or result in no win (left). In the Loss condition, the risky option could prevent participants from losing money or make them lose more money than the certain option (right).

Figure 3.

Effects of PD and citalopram on decision-making. (a) Significant difference in the proportion of offers accepted between healthy controls and PD on placebo. There were no mean differences between PD on placebo and citalopram. (b) Significant correlation between change in accepted offers (citalopram–placebo) and AMI scores, reflecting a Medication × Apathy interaction. (c) Visualisation of Apathy × Medication interaction by binary mean split of apathy groups. Citalopram increased the proportion of accepted offers for patients who scored low on apathy, but not for patients scoring high on apathy at baseline. (d) Change in accepted offers (citalopram–placebo) by reward levels, illustrating strongest Citalopram × Apathy interaction for low rewards. This may suggest that citalopram affected processing or sensitivity of low rewards specifically, but in opposing directions for apathetic and non-apathetic patients. (e) Change in accepted offers (citalopram–placebo) by effort levels, showing strongest Citalopram × Apathy interaction for medium effort levels. (f) Intrinsic motivation, calculated with each patient’s random intercept during the citalopram and placebo phases, correlated negatively with level of apathy. Higher levels of apathy were associated with a decrease in intrinsic motivation on citalopram, and vice versa. (g) No significant correlation between change in reward sensitivity and AMI scores. (h) No significant correlation between change in effort sensitivity and AMI scores. Error bars and shaded area represent standard error.

Figure 4.

Choice heatmaps for risky decisions. Numbers on X- and Y-axes represent monetary values for risky and certain options, respectively. Grey squares represent missing combinations of values, where the certain option is larger or equal to the risky option, except for some catch trials. In the Gain condition, participants were more likely to choose the risky option when its expected value (= money to win) was larger than the certain offer. In the Loss condition, they were more likely to choose the risky option when the expected value (= money to lose) was lower than the certain offer.

Figure 5.

Effect of medication and PD on risky decisions. (a) PD patients on citalopram were less risk averse for gains than during placebo phase. PD patients were generally more risk averse for losses than controls. The upper horizontal line represents the border for risk neutrality in the Gain condition, the dashed horizontal line represents the border for risk neutrality in the Loss condition. (b) Patients on placebo accepted slightly fewer risky options for offers with expected values between 1 and 5 (i.e. difficult decisions) than patients on citalopram. (c) Healthy controls accepted more risky offers whenever the expected loss was lower than in the certain offer, reflecting more rational decision-making than patients on placebo. Error bars represent standard error.