The most commonly reported pathological behaviors have been pathological gambling, hypersexuality, compulsive or binge eating, and compulsive shopping. Uncertainty remains regarding the overall frequency of DA-associated behavioral changes. Initial surveillance suggested very low rates—on the order of 2%–8% [19
]. Subsequent structured-questionnaire ascertainments found higher rates, with a recent large questionnaire-based assessment reporting a rate of 13.6% [20
]. This cross-sectional study assessed rates of pathologic gambling
(9.9%), compulsive sexual behavior
(4.4%), compulsive buying
(7.2%), and binge eating
(5.6%) among current DA users, with a total of 17.1% of current DA users exhibiting any pathological behavior. This compared to the significantly lower rate of pathological behaviors
(6.9%) among subjects not using a DA for at least 6 months prior to enrollment.
Some authors argue that reliance on impersonal questionnaires or spontaneous patient reports likely results in incomplete ascertainment due to the sensitive and/or potentially embarrassing nature of these symptoms. Another recent report utilized physician-directed symptom elicitation and found pathological behaviors in 24% of patients using DA at therapeutic doses and in 30% of patients using “target” DA dosing [21
]. Although involving a smaller population than some other reports, this paper highlights some difficulties in capturing behavioral changes with several patients exhibiting compulsive hobbying or computer use, and others having poor insight into their behavioral changes including a patient with compulsive gambling who perceived his behavior as “beneficial” due to net wins.
Emergence of pathological behaviors is very uncommonly seen among patients treated with L-dopa alone [22
]. A large study utilizing structured interview assessment found pathological behaviors in 6.9% of subjects not currently taking a DA, although prior exposure to DA was not reported [20
]. In previous reports, the DA with highest D3 affinity (pramipexole) appears to be more commonly implicated in pathological behaviors both in PD and in restless legs syndrome [23
], but a large cross-sectional study found no difference between current use and risk for pathological behaviors between DAs [20
]. Again, prior DA exposures and reasons for discontinuation were not reported.
The relationship between deep brain stimulation (DBS) of the subthalamic nucleus (STN) and impulse control disorders is complex, and it is the focus of several review papers [24
]. In general, a reduction in dopaminergic medication is seen after STN DBS, and with reduction or elimination of dopamine agonist therapy ICDs such as pathological gambling and others can improve [26
]. However, several studies have noted de novo ICDs after DBS [30
]. Interestingly, models of STN function [33
] suggest that the STN modulates decision thresholds in proportion to reinforcement and decision conflict. Patients with STN DBS showed typical conflict-induced slowing in “win-win” computerized decision-making tasks with their DBS off, but 10 minutes after turning the DBS on, they exhibited less slowing and increased impulsive decision making in these same tasks [34
]. Dopamine dysregulation syndrome (DDS) is a compulsive overuse of dopaminergic therapy. Preexisting DDS may or may not improve after STN DBS. Lim et al. found DDS remained unimproved or worsened in 12/17 patients after DBS, although this was a mix of STN and globus pallidus interna (GPi) DBS cases [32
]. In the remaining 5/17 patients, DDS improved or resolved.
Discontinuation of the DA or significant adjustment in dosage is the mainstay of treatment intervention and appears to be required to achieve full remission or significant reduction in behaviors [35
]. Even still, some patients exhibit persistent pathological behaviors. A study examining psychosocial outcomes in patents having exhibited pathological gambling found persistent financial and marital stress as a consequence of these behaviorsalthoughfull or partial resolution of the behaviors in all subjects followed [36
Some authors group DA-associated behavior changes as disorders of impulse control, but careful examination of the behavioral issues reported in the medical literature and by our patients suggests a more complex behavioral derangement than a general disorder of impulse control. Patients appear to demonstrate a circumscribed obsessive-compulsion for a particular behavior. Most commonly, patients exhibit one particular obsession, but even in cases where two or more obsessions manifest, the more widespread injudicious decision making and excessive spontaneity that characterize a general disorder of impulse control are absent [20
]. It may be that the neural systems mediating these pathologic behaviors are more closely aligned with punding (an intense fascination with meaningless movements or activities such as collecting, arranging, or taking apart objects), and one study suggested a strong relationship between punding and the expression of dyskinesias. Some studies suggest a D3 receptor-dependent response to L-dopa and dyskinesia, at least in monkeys [13
Several recent studies have documented the importance of the brain circuits involved in reward and risky decision making, including, thalamic, striatal, and ventromedial frontal regions. Using fMRI, Reuter and colleagues compared pathological gamblers and control subjects and found that activation in regions such as the ventral striatum is inversely related to their pathological gambling severity, as if risks and rewards were less salient to pathological gamblers except at high enough magnitudes [39
]. Another fMRI study had subjects play a game in which they decided to keep pumping up a virtual balloon or quit and collect reward points, with larger rewards associated with larger balloons [40
]. Increased activation levels in insular, thalamic, striatal, and dorsolateral prefrontal regions bilaterally and medial prefrontal cortex/anterior cingulate regions correlated with increases in active risk taking. Functional imaging studies in PD patients have implicated similar brain regions [41
Voon et al. [38
] studied PD patients with and without impulse control disorders (ICD) in a risk task involving a certain (e.g., +$100) or an uncertain outcome (e.g., 50/50 chance of winning either $200 or winning $0) for both gains (+$) and loss (−$) domains. PD patients without impulse control disorders behaved more similarly to healthy controls while they were on DA medications, making substantially more risky choices when they were confronted with losses than with gains, thereby showing “loss aversion” [43
]. These same patients made highly similar choices in the gain versus the loss domains without loss aversion when they were off DA medications. PD patients with ICD showed more risk taking in the gain domain whether on or off medication, a pattern that was opposite to those of the healthy controls and PD patients without ICD. Moreover, PD patients with ICD also showed higher sensitivity to risk when they were on DA medications, displaying a steeper drop in the number of risky choices as the value at stake became higher and higher. In another study [44
], PD patients without ICD were given the Iowa Gambling Task
(IGT) while they were on or off medications. In this task, subjects chose between four decks of cards with various risk reward payoffs
(i.e., risk disadvantageous
(RD) decks with larger and frequent rewards but also infrequent large losses leading to long-term net losses, versus risk advantageous
(RA) decks with smaller frequent rewards but also smaller infrequent losses leading to long-term net gains). PD patients off DA medications showed an appropriate decrease in choices for the risk-disadvantageous
(RD) decks over trials. In contrast, PD patients on DA medications failed to show such outcome-contingent learning; instead, they kept on choosing the RD decks.