Reward pathways in Parkinson's disease: Clinical and theoretical implications


*Diana M.E. Torta, PsycholD, Department of Psychology University of Turin, Via Po 14 10124 Torino, Italy. Email:


Aims:  The mesolimbic and mesocortical circuits are particularly involved in reward-related behavior in humans. Because these systems may be in some way altered in Parkinson's disease (PD), it is likely that some psychiatric manifestations of PD, such as hedonistic homeostatic dysregulation and pathological gambling, as well as impulsive decision making, may be ascribed to their involvement. The aim of the current article was to review recent literature on this topic in order to analyze whether these disturbances share a common ground and whether actual theoretical frameworks on addiction prove a useful tool for their interpretation.

Methods:  Data were identified on searches of MEDLINE/PubMed databases from relevant articles published up until March 2007.

Results:  All clinical manifestations (hedonistic homeostatic dysregulation, pathological gambling and impulsive decision making) seem to share a common multifaceted ground in which factors related to antiparkinsonian therapy, premorbid personality and progression of disease interact. Theoretical interpretations and conclusions drawn from experimental studies may help to shed light on the underlying pathological mechanisms.

Conclusions:  Further studies are needed to analyze why, despite a common ground, only some patients develop those neuropsychiatric complications described here.

PARKINSON'S DISEASE (PD) is a progressive neurological disorder primarily considered a motor disease, but the high prevalence of psychiatric complications is increasingly suggesting that PD is more accurately conceptualized as a neuropsychiatric disease.1,2 For this reason in the last few years growing attention has been addressed to neuropsychiatric symptoms of PD such as depression,1–6 apathy,7,8 anxiety,9 and anhedonia.10 But other clinical manifestations such as pathological gambling and drug hoarding still remain less inspected and constitute the main topic of the present review. In this article we consider recent literature on the subject under both clinical and theoretical perspectives, analyzing risk factors and possible causes enhancing these disturbances. Furthermore we discuss whether experimental research on impulsive decision making and reward systems in PD may help to shed light on the mechanisms underlying neuropsychiatric complications such as pathological gambling and drug hoarding.

Cell loss has been demonstrated in the dopaminergic neuronal population in the mesolimbic and mesocortical networks of PD patients, which may contribute to numerous of the aforementioned symptoms;11 this variable, less pronounced, disruption may occur to the ventral tegmental area projecting to the nucleus accumbens, amygdala and prefrontal cortices (such as the orbitofrontal). Some studies have considered this partial loss as a consequence of disease progression,12,13 while others have hypothesized that it may be responsible for a neuropsychiatric onset of the pathology.14,15

Several imaging studies have recently addressed the question as to whether the prospect of a reward may lead to an activation of the mesolimbic pathway in PD, assuming its involvement as proof of its total or, at least, partial integrity. Martin-Soelch et al., using positron emission tomography, compared changes in the cerebral blood flood (CBF) of parkinsonian patients, smokers and healthy controls.16 Interestingly, in the PD group no striatal activation was found, although the involvement of other brain areas was superimposable. Despite this striatal hypoactivation the hypothesis has been put forward that the effectiveness of reward may be spared: Goerendt et al. underlined an acceleration of the performance with the prospect of receiving monetary rewards.17 In this case a second compensatory cerebellar circuit may be recruited.18–21 In particular, it has been hypothesized that the cerebellar pathway may play a crucial role when the patient is untreated.17

In contrast, psychophysiological studies investigating anticipation of reward in PD found a reduced amplitude of the stimulus preceding negativity (SPN), a brain potential known to reflect anticipation of motivationally significant events.22 Because SPN offers an electrophysiological index of activity within cortical portions of the reward pathway,23 these results suggest an impairment of the reward anticipatory processes in PD.22

In this article we will discuss some neuropsychiatric complications of PD in the theoretical framework of hedonistic homeostatic dysregulation (HHD). These disorders are of primary interest because, notwithstanding the fact that their pathogenesis is poorly understood and it is likely to be multifaceted, an involvement of the mesolimbic component of reward system could be hypothesized. Finally, we will consider experimental models of pathological gambling and failure to inhibit impulses; the results from these studies could provide useful tools to analyze mesocortical impairments in neuropsychiatric complications of PD.


Literature search strategy

Data for this review were identified on searches of MEDLINE and PubMed databases from relevant articles published up until March 2007.

For the HHD section we used the following key words: ‘hedonistic homeostatic dysregulation’, ‘dopamine dysregulation syndrome’, ‘dopamine replacement therapy and reward systems’. For the pathological gambling section we used as key words ‘pathological gambling’ in combination with ‘Parkinson’s disease' and ‘dopamine replacement therapy’. For the impulsive decision making paragraph we searched: ‘impulsive decision making’, ‘gambling task’, ‘ventromedial prefrontal cortex’ in combination with ‘Parkinson’s disease'.

The same literature research criteria used for each aforementioned section were valid for the corresponding summarizing tables (Tables 1–3). Moreover, references from relevant articles were used when necessary.

Table 1.  Hedonistic homeostatic dysregulation: Case reports of compulsive DRT and HHD in PD (Lawrence et al.24)
Referencen/SexPast or family history of psychiatric disorderPrevious history of substance abuseMean age of onset of PD (years)Mean age of onset of compulsive DRT (years)Daily DRT
  • DRT, dopamine replacement therapy; HHD, hedonistic homeostatic dysregulation; LEDD, levodopa equivalent daily dose; PD, Parkinson's disease.

  • , Maximum average daily levodopa dose;

  • ‡, maximum average daily Apomorphine dose;

  • ¶, maximum daily Dopamine agonist dose.

Quinn et al.251 M 4447≤4000 mg LEDD
Vogel and Schiffer261 MSuicide by mother4649≤2000 mg LEDD;
Priebe271 FNoNo4960≤5000 mg Levodopa
Nausieda285 MNo50 (41–55)54 (46–50)1900 mg (1500– 2500 mg) LEDD
Tack et al.291 FSchizophrenia1825≤2500 mg LEDD
Uitti et al.301 MNo3764750 mg LEDD
Soyka and Huppert311 MAlcoholismAlcoholism44≤1800 mg LEDD
Weinman and Ruskin321 MNoNo5363≥3500 mg LEDD
Spigset and von Scheele332 MNoNo55 (51–58)59 (52–66)1700 mg (1500– 2000 mg) levodopa
Courty et al.344 MDepressionAlcoholism50 (46–57)59 (50–65)≥1600 mg (1100– 1900 mg) LEDD; Apomorphine
Merims et al.351 MNo60782200 mg LEDD
Giovannoni et al.363 M,
1 F
Depression (3),
panic (1)
Alcohol (1)
nicotine (1)
39 (36–42): 3175 (2125– 5550 mg) (4) : 123.25 (75–170) (4); : Bromocriptine (2); Pergolide (1)
Gschwandtner et al.372 MDepressionNo51 (43–59)56 (50–62)1600 mg (1000– 2200 mg) levodopa (2); Ropinirole (1)
Serrano-Duenas383 M
1 F
No56 (49–62)66 (59–71)2600 mg (2400–2900); LEDD (4), Ropinirole (4)
Houeto et al.392 MBipolar disorder (1)Alcohol (1)
illegal drugs (1)
4559 (58–61)≤1400 mg Levodopa
Muller et al.401 FNoNicotine27351400 mg LEDD, Ropinirole
Bearn et al.416 M
4 F
Major depression (4)Alcohol and nicotine (5), illegal drugs (2)451916 (600–3200) mg LEDD
Pezzella et al.425 M
2 F
Mood disorder (6),
compulsive behavior (2)
 51.4960.5 ± 282.1 (therapeutic dose) mg LEDD
Table 2.  Pathological gambling and Impulse control disorder studies
ReferenceType of studyNo. cases and prevalenceSymptomsTherapy
  1. DA, dapamine agonist; LEDD, levodopa equivalent daily dose.

Molina et al.43Retrospective12 (out of 250); 4.8%Pathological gamblingLevodopa, no information about DA therapy
Seedat et al.44Case report1 patientPathological gambling
and hypomania
Pergolide, selegiline
Driver-Dunckley et al.45Retrospective9 (out of 1884); 0.47%Pathological gamblingLevodopa
8 pramipexole
1 pergolide
Kurlan46Case report2 patientsPathological gamblingLevodopa, pramipexole
Dodd et al.47Case report11 patientsPathological gamblingLevodopa
9 pramipexole
2 ropinirole
Imamura et al.48Case report6 (6 of 1411); 0.42%Pathological gamblingLevodopa (4) (average daily dose) 525 mg
Pramipexole (4) (average daily dose) 4.125 mg
Ropinirole (1) (average daily dose) 4.5 mg
Cabergoline (1)
Avanzi et al.49Epidemiological6 (out of 98); 6.1%Pathological gamblingLEDD 760 ± 207.36
3 levodopa and dopamine agonist
1 dopamine agonist
Weintraub et al.50Case report11 (out of 272)- active
ICD; 4.0% 18 (out
of 272)- ICD anytime
during PD; 6.1%
Compulsive sexual
behavior, pathological
gambling, compulsive buying
LEDD 925.5 ± 534.9
Table 3.  Decision making in Parkinson's disease: patients' clinical characteristics and performance in Gambling Tasks
ReferenceGambling taskPatient characteristicsOutcomeTherapy
  1. IGT, lowa Gambling Task; PD, Parkinson's disease; UPDRS, Unified Parkinson's Disease Rating Scale.

Czernecki et al.51Iowa
Gambling Task
n = 23
UPDRS on = 12.4 (2)
UPDRS off = 38.7 (2.8)
Age of onset = 42.6 (2.2)
Average disease
duration = 14.9 (1.2)
Deficit in the IGT not sensitive to
l-dopa treatment
Levodopa mg/day = 1115.3 (67.4)
Cools et al.52Cambridge
Gambling Task
n = 12
UPDRS on = 30.9 (6.8)
UPDRS off = 47.1 (7.7)
Average disease
duration = 6.5 (1.4)
Impulsivity in placing bets
when on medication
Levodopa mg/day = 552 (60)
Perretta et al.53Iowa
Gambling Task
n = 32 (16 early PD: Hoehn and Yahr score < 3; 16 later PD Hoehn and Yahr score ≥ 3)
UPDRS early PD = 11.3 (1.1)
UPDRS later PD = 27.2 (1.3)
IGT performance impaired in both early and later PD groupsLevodopa early PD = 265.6 (33.1) Levodopa later PD = 350.0 (38.5)
Ropinirole (15 patients), amantadine (7 patients), Pramipexole (3 patients), Pergolide (3 patients), Selegiline (2 patients)

For the general overview in the introduction we further used ‘anxiety’, ‘depression’, ‘sleep disorders’, ‘apathy’, ‘anhedonia’ in Parkinson's disease.

Studies on reward pathways in PD patients submitted to deep brain stimulation were not included in this review.

Clinical manifestations
Hedonistic homeostatic dysregulation

In parkinsonian patients treated with dopaminergic drugs (dopamine or dopamine agonist) behavioral disturbances have been observed. In particular, reward-seeking, compulsive or impulsive behaviors such as pathological gambling, compulsive shopping and eating, disinhibition are described.50,54 Moreover, in some patients a compulsive dopaminergic addiction has been detected, ascribable to a dopamine dysregulation syndrome.24 All these disorders, occurring in a restricted sample of the whole parkinsonian population, may be underpinned by common alterations in the reward circuitry. The present attempt is to better conceptualize them in the theoretical framework of ‘dopamine dysregulation syndromes’, considering especially the HHD syndrome36 and underlying similarities between PD patients and drug addicts. The HHD clinical syndrome in PD has been conceptualized by Giovannoni et al.36 as a disorder consequent to dopamine replacement therapy (DRT). Diagnostic criteria of HHD require the patient to be affected by PD and responsive to levodopa therapy and needing ever increasing doses of therapy in excess of the dose normally required to relieve parkinsonian symptoms (for detailed diagnostic criteria see Giovannoni et al.36). DRT (with the dopamine precursor levodopa or synthetic dopamine agonist) is the commonest treatment for the motor symptoms of PD. A few patients take their medications beyond the quantity sufficient to relieve their motor disabilities.24,36 These patients, in spite of the motor and behavioral consequences, demand ever-increasing doses of therapy, report being able to move only when clearly dyskinetic, deny symptoms and refuse to reduce daily doses.24 There are some characteristic phenomena that might occur associated with HHD such as punding (purposeless repetitive behavior) euphoria and hypomania, altered appetite, hypersexuality, pathological gambling and shopping, heightened aggression, craving, psychosis.24 Compulsive DRT use may become evident only when attempts are made to restrict the supply of DRT (e.g. during hospitalization) when additional doses are demanded or when exaggerated ‘off’ state occurs.

The major theories of psychostimulant addiction may help to explain manifestations seen in the dopamine dysregulation syndrome (for a complete review see Lawrence et al.24). Although the syndrome can partly be explained by psychostimulants addiction theories, HHD has lately received particular attention.36

The HHD theory suggests that addicts are pushed to take drugs not only for pleasure but also to avoid unpleasant withdrawal symptoms. Hence the drug activates a pleasant state through brain reward-related circuits, which, in turn, trigger the opposite unpleasant process, so that the homeostasis could be restored. The former activation (‘a’ process) causes euphoria, the latter (‘b’ process) its decay. After repeated use the b process is strengthened and manifests as tolerance to euphoria. When the pleasant effect of the drugs wears off, unpleasant withdrawal is initiated and lasts longer than the previous one. Because only the b process is thought to grow in magnitude and duration, even a small dose of drug will reinstate it and trigger withdrawal.55 Abstinence from the drug decays the b process and, once returned to normal, the individual is no longer addicted.

Koob and Le Moal proposed an allostatic rather than homeostatic version of HHD.56 In HHD the experience of drug-induced pleasure motivates the individual, and addiction is defined as the presence of unpleasant emotional states, such as dysphoria, irritability and anxiety during abstinence.57 With repeated use of a substance, the b process, namely the process that causes the decay of euphoria, does not return to the baseline homeostatic normal state and often leads to psychiatric symptoms such as anhedonia. This state reflects a decreased baseline pleasure function, associated with reduced dopamine; pharmacological treatment augmenting dopamine release in limbic regions seems to alleviate such symptoms.58 Because baseline pleasure functions are lowered, an increase of drug intake is therefore needed to compensate for the shift in baseline reward. The altered reward-related brain activity found in some patients with PD20 may lead to increased vulnerability to DRT addiction.56,57 Hitherto the prevalence of HHD in PD has been estimated at around 4% (3.4–4%); and the disturbance seems to occur more frequently in men who have early onset of the disease and previous psychiatric history.36,42 In Table 1 recent studies on HHD are summarized.

The affective and behavioral changes seen in the HHD syndrome are similar to those found for other psychostimulants such as cocaine and amphetamines. It is likely that these similar effects are all mediated by the mesolimbic dopaminergic projections to the nucleus accumbens.36 It remains still unclear, however, why some patients develop the disorder and others do not. The hypothesis has been put forward that a previous psychiatric history, a personality disorder, temperament characteristics or familiarity for psychiatric diseases may predispose to addiction.29,33 Also, pathological changes to mesolimbic and mesocortical systems in PD may modulate the addictive potential of DRT. As previously seen, imaging studies have shown how the reward system could be dysfunctional in PD.59,60 Progressive dopaminergic denervation will reduce presynaptic dopamine storage mechanisms, triggering the pulsatile effects of DRT on limbic structures. Upregulation or changes in the sensitivity of dopamine receptors may also contribute to the development of HHD.36 An important role may also be played by the ascending dopaminergic and serotonergic projections. In particular D3 receptors are primarily localized in limbic regions in mammals and they may be involved in drug-induced reward.61 All the abused drugs such as alcohol, heroin, cocaine, tetrahydrocannabinol and nicotine increase dopamine levels in the shell of the nucleus accumbens,62 where the D3 receptors are expressed, and D3 antagonist may prove effective to manage drug dependence and addiction.61 Intracranial drug reward involves dopaminergic projections in the medial forebrain bundle63 ascending from the ventral tegmental area (VTA) to nucleus accumbens. D1 dopamine antagonist in VTA decrease the rewarding effect of cocaine; amphetamine reward is antagonized by either D1 or D2 dopamine antagonists.64 Studies suggest a more complex role of DA subtype receptors in reward, in particular a greater role for D1.65 Serotonin (5-HT) acts through several 5-HT receptors in the brain to modulate DA neurons in all three major dopaminergic pathways. The 5-HT2 receptor family is densely localized in the substantia nigra and VTA, as well as their terminal regions. Modulation of 5-HT2 receptor function is thought to be important in motor activation, motivation and reward; electrophysiological and biochemical data have shown that 5-HT2C receptor agonists decrease, while 5-HT2C antagonists enhance mesocorticolimbic DA function.66

Hence, the HHD theory constitutes a useful framework to conceptualize some peculiar neuropsychiatric complications of PD, underpinned by biological predisposition, neurotransmitters interactions, progression of the disease and likely to be triggered by dopaminergic therapy.

Pathological gambling and other impulse control disorders

Pathological gambling and other impulse control disorders such as compulsive buying and hypersexuality may occur in PD either as symptoms of a syndrome such as HHD or as isolated disorders. We have previously discussed HHD both as a clinical syndrome and a theoretical framework, which is useful to interpret some of the neuropsychiatric complications in PD. We will now analyze pathological gambling as an isolated disorder, with regard to its everyday life implication and relationship with DRT.

Pathological gambling is a psychiatric disorder characterized by inappropriate, persistent and maladaptive gambling behavior that has repercussions on family, personal and professional life. It is classified as an impulse control disorder67 and widely considered as a non-pharmacological addiction.68 The term ‘behavioral addiction’ has been recently used to define compulsive non-drug behaviors, assuming that they lead to similar changes in reward circuits.69,70 Other impulse control disorders usually detected in PD patients are compulsive buying and sexual behavior.25,26,30,32,34,42,50,71 Two retrospective studies, investigating the prevalence of pathological gambling in a population affected by PD, found a rate of 4.8% (12/250) and 0.47% (9/1884), respectively.43,45 An epidemiological study carried out in Italy found a prevalence rate of 0.4%.49 The authors examined a PD population on DRT therapy, ascertaining a rate of pathological gambling of 6.1% (the prevalence in the sex-matched control group was 0.25%);48 Voon et al., using the South Oaks Gambling Scale, noted that lifetime prevalence of pathological gambling was 3.4% and on any dopamine agonist, 7.2%.72

A role of dopamine therapy has been hypothesized also by Weintraub et al., who described 18 patients (6.6% of their sample) who fulfilled criteria for an impulse control disorder (ICD) at some point during the illness (compulsive gambling and compulsive sexual behavior were found to be equally common).50 Predictive elements of an ICD were treatment with a dopamine agonist and a history of ICD symptomatology prior to PD onset. There were no differences between the dopamine agonists in association with ICD, and daily doses of dopamine agonists were higher in patients with an ICD than in dopamine agonist-treated patients without an ICD.

Between 2002 and 2004 Dodd et al. identified 11 patients with PD and pathological gambling; nine of them were taking pramipexole and two, vopinirole.47 In the Molina et al. study 10 out of the 12 pathological gamblers started gambling after the onset of disease and DRT. The pathological behavior was manifest in 11 of 12 patients when they were in the ‘on’ state.43 In their retrospective review Driver-Dunckley et al. associated pathological gambling and dopaminergic therapy with dopamine agonists: eight (1.5%) of 529 were receiving pramipexole, one (0.3%) of 331 was under pergolide and none of 421 patients receiving ropinirole had pathological gambling.45 But the total number of patients exposed to pramipexole or ropinirole was not mentioned by the authors45 and the exact incidence of gambling cannot be estimated.54 But the patients with pathological gambling in the Dodd et al. study were taking high doses of pramipexole (4.5–13.5 mg/day); hence the dose of agonist might play an important part in triggering the adverse effect.

Imamura et al. noted that pramipexole and cabergoline are more selective than ropinirole for D3 receptors than D2 receptors,48 with pramipexole exhibiting a stronger effect.73 A functional magnetic resonance imaging study carried out by Reuter et al. suggested that pathological gambling was linked to a reduced activation of the mesolimbic reward system, in particular pathological gamblers showed reduced activation of the ventral striatum and ventromedial prefrontal cortex.74 It is possible that because combined therapy of levodopa and dopamine agonist may underlie a greater risk for the development of pathological gambling, patients with advanced PD and greater need for combination therapy may have an increased propensity for the disease.48 In Table 2, recent studies on the subject have been summarized. Noteworthy, all patients developing pathological gambling were on high-dose therapy. As noted by numerous authors, reducing the dose was in most cases sufficient to ameliorate pathological gambling.45,47,48

Experimental studies
Impulsive decision making

Impulsive decision making is a condition of intolerance to delay of reward;73–77 it is exemplified by the reduced ability of an individual to choose a large delayed reward over small immediate ones. This behavior is observed in populations of drug addicts,78–82 although it is still unclear whether impulsivity is the cause or consequence of drug abuse.83 In this article we describe impulsive decision making because it has been frequently reported in PD patients probably due to an impairment of the mesocortical reward pathway. The condition of aversion to delay of reward can be often observed in experimental gambling tasks, but it is worth considering the clinical implications in ecologic decision-making. Furthermore, the relationship between poor decision-making and addiction in general has been emphasized.84 When a choice has to be made, information about expected outcomes has to be maintained in memory so that it can be compared and integrated with information about internal states and current goals.85 Such an integrative process generates an outcome expectancy, namely an internal representation of the consequences likely to follow a specific act.85 The orbitofrontal cortex (OFC) plays a crucial role in the generation and use of these outcome expectancies: impairments of this area lead to poor decision making,86 reversal deficits,87–90 and inability to guide behavior appropriately on the basis of the consequences of actions.91 In particular, Bechara et al. created the Iowa Gambling Task (IGT) to assess lesions to the OFC.91 In this task subjects must choose from decks of cards with varying rewards and penalties. In order to earn points the subject must be able to integrate the value of rewards and penalties. Larger rewards are associated with larger penalties. Hence, when subjects without impairments of the OFC begin to lose more than they earn, they change their strategy and turn to safer decks of cards. Individual with OFC damage instead fail to modify their responses. Czernecki et al. used the IGT in a PD population at an advanced stage of the disease and demonstrated deficits in performance, which did not change between the on medication and off medication conditions.51 Similar impairments have been also described in PD patients at earlier stages of illness.53 In order to assess the degree to which deficits in decision making are sensitive to the quality of information available, Rogers et al. introduced a novel decision-making task with contingencies presented in a readily comprehensible visual format.86 In this task boxes of two different colors were displayed at the top of a screen varying in ratio. The subject had to select where a yellow token was hidden and place a bet on their decision. Because the ratio of blue/red boxes varied from 9:1 to 1:9 there were conditions that were more and also less favorable. A normal betting strategy required a subject to select the color most likely to hide the token and modulate the amount of points to bet on the choice. To assess impulsivity two different conditions were set: in one case points to be bet increased as a function of time, in the other they decreased. An impulsive subjects do not modulate their choices but tend always to select the first conditions.52,86

It has been demonstrated that a group of PD patients was able to choose the most likely outcome, but tended when ‘on medication’ (l-dopa) to place impulsive bets, that is, to select the first choices in both the ascending and descending conditions.52 The orbitofrontal loop in mild PD is hypothesized to operate at a higher level of basal DA modulation, and so be more potentially subject to deleterious overdosing by additional levodopa-dopa medication.92 In PD subjects dopaminergic treatment causes an optimal performance in dorsolateral loop, but the same level of dopaminergic stimulation causes a decrease of performance in the orbitofrontal loop for an ‘overdosing effect’. Hence, dopaminergic therapy may help to ameliorate some cognitive skills related to the dorsolateral circuits, but has a detrimental effect on other components such as the ability to refrain from making impulsive choices. This ability is preserved when the patient is unmedicated (‘off’ state).52 Recent studies suggested a specific interaction of the ascending 5-HT system with the OFC,93,94 and one interesting speculation is that 5-HT mechanisms may contribute to different aspects of response inhibition,93 reward95 and the monitoring of negative feedback.96


In this paper we have reviewed recent literature on the clinical manifestations of impairments in the mesolimbic and mesocortical systems in PD, in order to better conceptualize clinical manifestations such as pathological gambling and impulsivity in the theoretical framework of HHD. The growing attention towards non-motor aspects of PD mirrors the desire of both patients and caregivers for improved quality of life and highlights the necessity to manage high psychosocial-impact clinical complications. Because there is no global accordance on whether and when an impairment of the reward system occurs in PD, we investigated the involvement of the limbic and cortical substrates of the pathway. Further studies are needed to clarify what could be described as a multifactorial model in which premorbid personality, progression of disease and adverse reaction to antiparkinsonian treatment may play a crucial role.


We thank Professor Geminiani Giuliano for his comments and for a first revision of this paper.