Dopamine reverses reward insensitivity in apathy following globus pallidus lesions

Abstract

Apathy is a complex, behavioural disorder associated with reduced spontaneous initiation of actions. Although present in mild forms in some healthy people, it is a pathological state in conditions such as Alzheimer's and Parkinson's disease where it can have profoundly devastating effects. Understanding the mechanisms underlying apathy is therefore of urgent concern but this has proven difficult because widespread brain changes in neurodegenerative diseases make interpretation difficult and there is no good animal model. Here we present a very rare case with profound apathy following bilateral, focal lesions of the basal ganglia, with globus pallidus regions that connect with orbitofrontal (OFC) and ventromedial prefrontal cortex (VMPFC) particularly affected. Using two measures of oculomotor decision-making we show that apathy in this individual was associated with reward insensitivity. However, reward sensitivity could be established partially with levodopa and more effectively with a dopamine receptor agonist. Concomitantly, there was an improvement in the patient's clinical state, with reduced apathy, greater motivation and increased social interactions. These findings provide a model system to study a key neuropsychiatric disorder. They demonstrate that reward insensitivity associated with basal ganglia dysfunction might be an important component of apathy that can be reversed by dopaminergic modulation.

Fig. 1

Sections demonstrating the extent of basal ganglia lesions. KD's GPi lesion was larger on the left than on the right. The lesions are projected onto boundaries of the GPi (orange), GPe (yellow), putamen (green) and caudate (purple). The bottom left coronal section is a close up at the level of the anterior commissure.

Fig. 2

Lesions and cortical connections. (A) For DTI analysis, three cortical sites are shown: LOFC (yellow), VMPFC (red) and M1 (blue). (B) Regression coefficients (betas) extracted from the voxel of maximum intensity within the lesion on the left (L) and right (R) for the three tracts. High values indicate that the tract passes through the lesion with a high probability.

Fig. 3

Traffic lights task (TLT). (A) Subjects fixated a circle which successively turned red, amber and green. They were required not to move their eyes until the onset of the green light, otherwise they receive a small (constant) fine or punishment. To maximize reward, participants had to make a saccade to the contralateral target as quickly as possible after green light onset. (B) Amber durations were selected at random from a normal distribution (mean = 750 msec, SD = 125 msec). (C) Reward was calculated with a hyperbolically decaying function with a maximum value of 150 pence (£1.50) when SRT was zero. Thus to maximize reward subjects should program an eye movement to coincide with green light onset. However, amber durations were not constant and therefore they either had to take a risk (high reward or punishment) or wait for the green light before programming a saccade (low reward).

Fig. 4

Directional saccadic reward task. Participants attended a central fixation spot which was extinguished after 1000 msec of fixation. They then made a saccade as fast as possible to a target presented either to the left or right (50% each side). One side was rewarded while the other received no reward. The rewarded side (RS) remained constant for an unpredictable number of trials before switching to the other side.

Fig. 5

Traffic lights task (TLT): saccadic distributions. (A) Saccades for age-matched controls (n = 13) performing the TLT two distinct distributions: an early, anticipatory distribution and a later, reactive one made in response to green light onset. Early responses were divided into errors (saccades before the green light came on) and correct anticipations (saccades with < 200 msec latency after the green light). The plot here is for a total of 6500 saccades. (B) Pre-treatment, KD made mostly reactive saccades (461/500 trials [92.2%]) with a median latency of 248 msec. He made very few anticipatory saccades. (C) After treatment with l-DOPA 100 mg (Madopar CR 125 mg) three times a day for 12 weeks, there was a dramatic increase in early responding in KD. (D) After 12 weeks treatment with a dopamine agonist (ropinirole XL, 4 mg once a day), KD's distribution of saccades looks most similar to that of control subjects.

Fig. 6

Percentage early responses on traffic lights task (TLT) over time. (A) Over the course of the first session, healthy controls showed increased early responses but KD did not. (B) In the second session, an hour later, controls showed no further change but KD 1 h after receiving l-dopa showed escalating early responses. (C) During the drug holiday period (off l-dopa), KD's early responses reverted to pre-treatment levels.

Fig. 7

Results from the directional saccadic reward task. The control group (n = 12, arrows to side) showed a preference for the rewarded target locations, with significantly shorter SRTs. KD showed no reward preference at baseline, before treatment (Session 1). In Session 2 he was given a single dose (100 mg) of levodopa which led to a significant reward preference. This was maintained throughout chronic dopaminergic therapy (Sessions 3 Madopar 125 mg three times daily for 4 weeks, Session 4 Madopar CR 125 mg three times daily for 12 weeks). Following a treatment holiday (4 weeks), this reward preference was absent (Session 5). However, with subsequent treatment on the dopamine agonist ropinirole (1 mg three times a day), there was both a re-establishment of reward preference and significant decrease in latency to both rewarded and unrewarded targets. Error bars are +/− 1 SEM (standard error of the mean).