Impulsive and Compulsive Behaviors in Parkinson’s Disease: The Norwegian ParkWest Study

INTRODUCTION

Impulsive-compulsive behaviors (ICBs) are recognized as serious neuropsychiatric complications in Parkinson disease (PD), with potentially devastating personal, social and financial consequences [1, 2]. These abnormal behaviors include the four major impulse control disorders (ICD) pathological gambling, compulsive shopping, binge eating and hypersexuality [1]. These behaviors are ego-syntonic and impulsive in nature, characterized by an effort to obtain arousal and gratification and cognitive biases [3–5]. In addition, a range of related behaviors have been described in PD, including punding, hobbyism, walkabout and compulsive dopaminergic medication overuse [6]. The related ICBs are ego-dystonic and compulsive in nature, associated with a calming or anxiolytic effect on the patient [5, 6].

Reported prevalence estimates of ICBs in PD vary considerably, ranging from 6% to almost 35% [1, 7]. Potential explanations include differences in the definition and assessment of ICBs, dopaminergic treatment, and patient selection, with most studies performed at highly-specialized movement disorders centers. In addition, since only few studies included normal control subjects, little is known about the risk of ICBs in PD relative to the general population [8–10]. Such information would, however, be important given that social, cultural and economic factors are likely to influence the prevalence of ICBs.

ICBs in PD have been associated most consistently with dopaminergic medication, and dopamine agonist (DA) treatment in particular [9, 11]. Other proposed determinants include premorbid personality traits, younger age, male gender, and depression and anxiety [12]. However, evidence in this respect is not unequivocal and even less clear for a range of other features within the spectrum of motor and non-motor symptoms associated with PD.

Against this background, we investigated the risk and determinants of ICDs and related impulsive-compulsive behaviors in a population-based PD cohort and normal controls (NCs) using comprehensive and standardized assessments of ICBs, as well as neurological, neuropsychiatric and cognitive functioning.

MATERIALS AND METHODS

RESULTS

DISCUSSION

The main finding of this population-based study was that patients with PD have a 3-fold increased odds of ICBs compared with age- and gender-matched controls. About 30% of our PD cohort screened positive on the QUIP for at least one ICB and almost 10% for multiple ICBs. Presence of ICBs in PD was strongly and independently associated with DA treatment and depressive symptoms, but not with other clinical or demographic variables, levodopa treatment or neuropsychological measures of attention, executive function, memory or visuospatial skills. These findings have implications for individualized patient management and follow-up.

All ICB subtypes were more common in patients than in NCs, particularly hypersexuality, compulsive eating and walkabout. These findings clearly underline the importance of screening for related behaviors beyond the major ICDs in patients with PD. Compared to our population-based estimate of 30.4% ICBs in PD, previous studies from Mexico, Finland and Denmark reported both lower and higher ICBs rates, ranging from 14.9% to 34.8% [10, 34]. However, these studies comprised convenient PD samples and no control group, making comparisons difficult. Indeed, we are aware of only one other controlled study reporting comparative data on the broad range of ICBs in PD. However, that study investigated drug-naïve patients and found similar ICBs rates compared to healthy controls, affecting about 20% in each group [8]. The frequency of ICBs in our control group was substantially lower, probably reflecting differences in sample recruitment and characteristics, as well as social, cultural and economic factors. These are important to consider when comparing the occurrence of ICBs between continents and countries.

ICBs in patients with PD have consistently been associated with dopaminergic medication, and DA use in particular. Our population-based data support and extend this observation, showing a more than 7-fold increased odds of ICBs among patients using DA but not levodopa, compared with NCs. In contrast, the association of ICBs with levodopa treatment has been less clear and a matter of debate. While some authors reported that levodopa treatment is associated with ICDs in PD [12], others argue that this finding may be an artefact of including patients with comorbid dopamine dysregulation syndrome who are taking high-dose levodopa [35]. Therefore, it has been claimed that levodopa remains a first-line choice in patients at high risk of ICDs and is essential to maintain antiparkinson efficacy in patients who need to reduce or stop DA treatment. Our findings seem to support this view, as the highest odds of ICBs was observed among DA only users, whereas the frequency of ICBs among patients not taking DAs was similar to that observed in NCs.

While ICBs were strongly associated with DA treatment, they were not related to DA dose in our cohort, suggesting a drug class rather than drug dose effect. A caveat of this conclusion is that our study did not differentiate between severe and less severe ICBs, making identification of a potential drug dose effect difficult. Although similar observations have been made previously [10], others report ICBs in PD to be associated with higher DA dosages [35–37]. This is also in line with the common clinical observation that down titration of DA dosage may alleviate ICB symptoms in patients with PD.

Despite conflicting findings, some studies argue that other antiparkinson drugs, such as monoaminoxidase-B (MAO-B) inhibitors and amantadine, may be associated with increased risk of ICBs [38–41]. While amantadine was not used in our cohort, we found no association between treatment with MAO-B inhibitors and frequency of ICBs.

We found strong associations between ICBs and depressive symptoms in our cohort, specifically symptoms of dysphoria and retardation. Although depressive symptoms were mild or even subclinical (i.e. under the cut-off for clinical significant depression in PD [42]) in most patients, the association with ICBs was consistent across several measures (NPI depression item and MADRS) and independent in multivariate analysis. Despite consistent evidence in multiple studies [33, 44], the relationship between ICBs and depressive symptoms in PD is not fully understood. However, it has been hypothesized that denervation of afferent dopaminergic neurons may result in sensitization of the subcortical motivational reward pathways. When exposed to exogenous DAs, affected patients may experience fluctuating symptoms of both depression and ICBs [44].

Presence of ICBs in our PD cohort was not related to other clinical or demographic measures. For example, we were unable to identify significant difference in gender between patients with and without ICBs. In line with previous studies [12, 43], patients with ICBs in our cohort were younger than those without. However, age was not independently associated with ICBs in multivariate analysis that also took into consideration the effects of DAs, which often are the preferred dopaminergic treatment in younger patients with PD. Furthermore, we were not able to demonstrate any association between ICBs and motor disability or disease duration, nor cognitive impairment. Although altered reward and stimulus valuation have been reported in patients with PD and ICBs [3, 9], global cognitive functioning has usually been demonstrated to be preserved [43, 45], in line with our findings. Indeed, a recent longitudinal study in PD even reported lower cognitive decline in patients with than without ICBs [45].

These observations argue against ICBs in PD being a consequence of more widespread brain pathology and rather suggest that the vulnerability to DAs in a substantial subset of patients reflects genetic susceptibility. In support of this, a recent study in PD found that 57% of the variance in ICB incidence was explained by common genetic variants, and that differentiation between patients at risk and patients without risk of ICB development may be possible using a broad candidate genetic panel [46]. These results are promising and highlight the involvement of premorbid genetic factors of multiple neurotransmitter systems in the pathogenesis of ICBs in PD. Continued research into potential genetic markers of ICBs might translate into clinical practice and make identification of at-risk patients possible.

Our study has both strengths and limitations. Major strengths include the population-based design, the well-characterized PD cohort, the comprehensive clinical, neuropsychiatric and cognitive assessments, and the age- and gender-matched control group from the same geographical area. We consider the use of the QUIP, a validated screening instrument covering a broad range of ICBs [20], another strength of our study, although we recognize that the short form version applied in this study does not allow to determine the severity of ICB symptoms, which is a relative limitation. We also recognize the assessment of ICBs by self-report as a potential limitation of this study, as affected individuals not always recognize ICBs as problematic [20]. Using QUIP, only moderate interrater reliability has been found between patients and their caregivers, arguing for a risk of false negatives in the absence of informant-based information on ICBs in PD [47, 48]. On the other hand, we are aware that the QUIP may overestimate the frequency of ICBs. However, this is most likely true for both patients and controls and should therefore not impact the odds ratios between these two groups. Another study limitation is the cross-sectional data presented here. However, these data extend previous evidence by demonstrating that ICBs are very common in the general PD population, with a more than 3-fold increased risk compared to matched control subjects. Importantly, this increased risk of ICBs in PD was driven by patients on treatment with DA, whereas levodopa per se was not associated with ICBs in our population-based cohort. Clinicians are therefore advised to demonstrate caution when administering DAs and to routinely screen for ICB symptoms in PD, as patients not always spontaneously report ICBs in daily clinical practice. Given the very limited long-term data in this field, collection of longitudinal data in our population-based study is ongoing and will hopefully provide further valuable insights into important aspects related to the causes, evolution and consequences of ICBs.

CONFLICT OF INTERESTS

The authors have no conflict of interest to report.