Prospective prevalence of pathologic gambling and medication association in Parkinson disease

Pathologic gambling (PG) in patients with Parkinson disease (PD) may be related to aberrant dopaminergic stimulation.¹ Retrospective studies report PG prevalences of 0.05 to 4.8% in PD. Similarly, retrospective studies have observed an association with pramipexole, a potent D3 receptor agonist.¹

We aimed to prospectively determine PG prevalence and the relationship with differing dopaminergic medications and doses.

Methods.

We screened consecutive follow-up patients with idiopathic PD (Queen's Square Brain Bank criteria) who attended a movement disorders clinic in Toronto, Ontario, Canada, over 3 months. Exclusion criteria were atypical parkinsonism, dementia within a year of motor symptom onset, and inability to complete the survey.

Patients were encouraged to complete the 10-minute patient-rated screen with the assistance of a spouse. The survey included a modified South Oaks Gambling Screen (SOGS)² validated in the general population.

Patients scoring ≥3 (indicating problem gambling in the general population)² were interviewed by phone by an experienced movement disorders neurologist (J.M.) and told that their scores suggested a potential medication-related behavioral symptom. Patients were referred for psychiatric assessment unless they had inappropriately completed the questionnaire.

An interview was conducted by a psychiatrist experienced with PD (V.V.). PG diagnosis was made using Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) criteria. PD characteristics, the Structured Clinical Interview for the Diagnosis of DSM Axis I psychiatric diagnosis, Mini-Mental State Examination, DSM-IV diagnosis of dementia, and first-degree family history were assessed. Medication types and doses concurrent with the behavior were recorded. Age, gender, age at PD onset, and medication types and doses of all screened patients were recorded. The unpaired t test and Fisher's exact test were used with threshold set at p < 0.05. The study was approved by the Research Ethics Board at the University Health Network.

Results.

Of 396 patients with PD, 297 (75%) adequately completed the survey (defined as at least 75% of questions answered). Surveys inadequately completed included patients with known cognitive deficits.

Sixteen patients were interviewed by phone; three of 16 completed the questionnaire but had misinterpreted the questions, 12 were psychiatrically assessed, and one patient endorsed gambling but distance prevented assessment.

Lifetime prevalence of PG was 10/297 (3.4%) and 3-month active prevalence was 5/297 (1.7%). Patient and PD characteristics, concurrent behavioral disorders, and personal and family psychiatric history are shown in table 1. Age at PD onset was earlier in PG (p < 0.0001). Two patients had hypomania following PG onset. All had PG onset after starting PD medication. Six of 10 (60%) had either a premorbid personal or family history of alcohol use disorder or a family history of bipolar disorder.

The percentage of PG patients within each medication class is represented in figure, A and B). PG was more frequent in patients on adjunctive agonist (p < 0.001) or as monotherapy (p < 0.05) compared to levodopa monotherapy. The rates of PG between patients on adjunctive pramipexole, ropinorole, or pergolide and between those on adjunctive agonist compared to agonist monotherapy were similar.

• Download figure
• Open in new tab
• Download powerpoint

Figure. (A) Percentage of Parkinson disease (PD) patients with pathologic gambling (PG) compared to all PD patients on levodopa or agonist monotherapy or adjunctive agonist.*p< 0.05 compared to levodopa monotherapy; **p< 0.001 compared to levodopa monotherapy. (B) Percentage of PD patients with PG compared to all PD patients on differing adjunctive dopamine agonists.

Medication doses for patients with PD screened with and without PG are shown in table 2. Statistical analysis could not be performed on the agonist monotherapy dose due to the small sample size. Total levodopa dose equivalent (LEDD) and LEDD calculated for dopamine agonist only was the same for patients with PD with and without PG.

Discussion.

Lifetime PG prevalence in a tertiary PD clinic was 3.4% and 3-month active prevalence was 1.7%. Lifetime prevalence on any dopamine agonist was 7.2%. The lifetime prevalence of DSM-defined PG in Ontario, Canada, is 1%.³

Study limitations included the lack of a clinician-rated questionnaire. The spousal assistance may have facilitated impairments in insight although may also have limited privacy. Cognitively impaired patients who inaccurately completed the survey may have been missed. Study strengths included its prospective nature, case-control comparison of medication doses, use of a psychiatric assessment, and rigorous behavioral definitions.

No patients had compulsive medication use. Dopamine agonists were more likely to be associated than levodopa monotherapy.¹ The rates of PG on agonist monotherapy compared to adjunctive agonist were not different. However, when compared to the rates on levodopa monotherapy, the degree of significance was greater for adjunctive agonist than for agonist monotherapy. This may be a feature of sample size, total dose, or a priming effect of levodopa. Similar to sensitization observed from psychostimulants⁴ or analogous to that of dyskinesias, pulsatile administration of levodopa may play a role in priming these behaviors.

That dose was not associated with PG suggests a potential underlying vulnerability in the patients who develop PG. Based on this study, younger age at PD onset is a vulnerability. Male gender was not associated, although this may be limited by sample size. Potential other vulnerabilities requiring case-control study corroboration include alcohol use or bipolar disorder diathesis.

In contrast to the retrospective reports of PG associations with pramipexole,¹ PG was not associated with a specific dopamine agonist. Both D1/D2 (pergolide) and D2/D3 (ropinorole and pramipexole) agonists were equally implicated. The retrospective literature may be biased due to differing prescribing practices between centers. However, this study does not rule out D3 mechanisms given that pergolide may have greater D3 than D1 receptor affinity.⁵

In the general PG population, some evidence points to dopaminergic mechanisms based on CSF dopamine level studies,⁶ dopamine receptor polymorphism associations,⁶ and a study demonstrating amphetamine priming of gambling motivation. Reduced ventral striatal and ventromedial prefrontal cortex activation in a functional imaging study was suggested to be related to aberrant reward and response inhibition.⁷

Despite potential vulnerabilities, these patients with PD did not express PG prior to medication onset in their seventh decade (compared to general PG onset commonly in the third decade), suggesting a protective role of PD. Similarly, ventral tegmental area (VTA) dopamine neurodegeneration has been suggested to mediate lowered risk-taking behavior rates of cigarette smoking and alcohol abuse in PD and the less consistent observations of decreased novelty-seeking.⁸

Nonphysiologic dopaminergic medication administration may unmask this underlying vulnerability in two potential ways. First, dopaminergic medication may affect the underlying physiologic role of dopamine in reward processing. In animal studies, phasic VTA neural activity occurs in response to unanticipated reward and particularly to reward prediction error.⁹ In contrast, tonic VTA neural activity was demonstrated to have the greatest response at maximal reward uncertainty.⁹ The study authors suggest that the reward uncertainty characteristic of gambling paradigms may be itself rewarding.⁹ Thus, how and when the medication is administered may affect reward processing.

Exogenous dopaminergic medication has also been demonstrated to affect frontostriatal functioning. The ventral prefrontal cortex (vpfc) and ventral striatum involves functions such as reward-based learning and response reversal and inhibition, measures that may be impaired in this population. In contrast, the dorsolateral prefrontal cortex (dlpfc) and dorsal striatum involves functions such as working memory, response initiation, and conditional associative learning. PD neurodegeneration and dopaminergic medications may differentially affect these functions.¹⁰ For instance, PD neurodegeneration predominantly affects dlpfc function. Levodopa administration in early PD impairs reversal learning, a relatively preserved vpfc function, but improves dlpfc functions.¹⁰ However, whether and which of these functions are impaired, how vulnerability interacts, or whether differential or progressive changes in PD neurodegeneration play a role in the expression of PG is not known.

The study patients differ from the general population given the preference for slot machines characterized by high reward uncertainty, limited need for higher cognitive processing, and a repetitive nature, suggesting pathophysiologic overlap with medication-related punding behaviors observed in PD. Whether reward processing may play a more prominent role in patients with gambling preferences of a less repetitive nature is not known.