Non-motor symptom burden in patients with Parkinson’s disease with impulse control disorders and compulsive behaviours: results from the COPPADIS cohort

The study was aimed at analysing the frequency of impulse control disorders (ICDs) and compulsive behaviours (CBs) in patients with Parkinson’s disease (PD) and in control subjects (CS) as well as the relationship between ICDs/CBs and motor, nonmotor features and dopaminergic treatment in PD patients. Data came from COPPADIS-2015, an observational, descriptive, nationwide (Spain) study. We used the validated Questionnaire for Impulsive-Compulsive Disorders in Parkinson's Disease-Rating Scale (QUIP-RS) for ICD/CB screening. The association between demographic data and ICDs/CBs was analyzed in both groups. In PD, this relationship was evaluated using clinical features and treatment-related data. As result, 613 PD patients (mean age 62.47 ± 9.09 years, 59.87% men) and 179 CS (mean age 60.84 ± 8.33 years, 47.48% men) were included. ICDs and CBs were more frequent in PD (ICDs 12.7% vs. 1.6%, p < 0.001; CBs 7.18% vs. 1.67%, p = 0.01). PD patients had more frequent previous ICDs history, premorbid impulsive personality and antidepressant treatment (p < 0.05) compared with CS. In PD, patients with ICDs/CBs presented younger age at disease onset, more frequent history of previous ICDs and premorbid personality (p < 0.05), as well as higher comorbidity with nonmotor symptoms, including depression and poor quality of life. Treatment with dopamine agonists increased the risk of ICDs/CBs, being dose dependent (p < 0.05). As conclusions, ICDs and CBs were more frequent in patients with PD than in CS. More nonmotor symptoms were present in patients with PD who had ICDs/CBs compared with those without. Dopamine agonists have a prominent effect on ICDs/CBs, which could be influenced by dose.

Concerning the PD features according to the presence of ICDs, the motor characteristics were similar in both groups, except for a higher score in the freezing of gait (FOG) scale in patients with ICDs (5.73 ± 5.73 vs. 3.38 ± 4.23; p < 0.001) ( Table 3). However, the differences in nonmotor features as well as in the performance and quality of daily living activities were notable. The score on the Non-Motor Symptoms Scale (NMSS) was higher in patients with ICDs (60.37 ± 42.48 vs. 41.33 ± 33.77; p < 0.001). In addition, when the NMSS subdomains were analysed, the cardiovascular, sleep/fatigue, mood/cognition, attention/memory, urinary, sexual and miscellaneous domains had a higher impact on the patients with ICDs (Fig. 2). Along these lines, the Neuropsychiatry Inventory (NPI) showed that, overall, patients with ICDs presented with a more prominent burden of these symptoms (8.6 ± 10.28 vs. 5.48 ± 7.32; p = 0.005). In terms of mood, patients with ICDs showed higher rates of depression (48.97% vs 65.38%, p = 0.004). Analyzing the subtype of depression according to the Beck Depression Inventory, patients with ICDs showed a higher ratio of subclinic and minor depression compared to those without ICDs. Anxiety, euphoria and irritability were the features specifically associated with patients with ICDs.
Apart from the NMSS sleep score, the Parkinson's Disease Sleep Scale showed that patients with ICDs presented with poorer sleep quality, as shown by a lower score (102.58 ± 25.84 vs 116.83 ± 25.63; p < 0.001).
The cognitive status did not differ between groups. After assessing the separate domains of the Parkinson's Disease Cognitive Rating Scale (PD-CRS), we did not find differences between the patients with and without ICDs.   (Fig. 3). ICDs were also associated with a poorer score on the performance of daily activities, as shown by a higher motor items score in the Unified Parkinson's Disease Rating Scale (UPDRS) part II in off condition (2.66 ± 1.79 vs. 3.34 ± 2.27; p = 0.009) ( Table 3).   Table 1). There were few exceptions for the age, age at onset, disease progression time and the score of the UPDRS part II in off that were similar in the group of CBs positive and negative.
Regarding dopaminergic treatment, we analysed patients receiving DA treatment, levodopa and/or monoamine oxidase-B inhibitors. After controlling for levodopa-equivalent dose, we found that ICDs were more frequent in patients on DA treatment and the agonist dose was related to the presence of ICDs (237.65 ± 194.72 vs. 174.72 ± 164.92; p = 0.02) ( Table 4).
To analyse the influence of the type of DA, we compared not taking any agonists with the risk linked to each type of agonist. For this analysis, we observed that Rotigotine had the lowest risk (OR 2.45; 95% CI 1.06-5.8; p = 0.03) and Ropirinole the highest (OR 2.82; 95% CI 1.25-6.57; p = 0.013) ( Table 5). Nevertheless, the analysis of ICDs risk among agonists did not show statistically significant differences.
To examine the influence of ICDs severity according to treatment, we analysed the QUIP-RS total score with dopaminergic treatment characteristics. As expected, we observed that patients with PD undergoing DA treatment had a significantly greater QUIP-RS total score compared with the patients with PD without agonist treatment, controlling for age, age at disease onset and disease progression time (5.47 ± 9.24 vs. 2.26 ± 5.78, respectively; F = 19.39; p < 0.001). When we compared the type of DA with the non-use of DA and the severity of ICDs/CBs using the QUIP-RS total score, Ropinirole was associated with the greatest significant increase in  (Fig. 4b).
For the subanalysis of the group with an age at onset younger than 50 years, this group presented with a higher rate of ICDs (19.78%) and differences on the cultural level. Other features such as a premorbid impulsive personality, non-motor symptoms according to NMSS and BDI and poorer quality of life were similar to the   Tables 2 and 3).

Discussion
The objective of the present study was to assess ICDs and CBs in patients with PD and CS, as well as the clinical features linked to these disorders in both groups. We found that in the patients with PD, ICDs and CBs were more frequent compared with the CS (ICDs 12.7% vs. 1.6%, respectively; p < 0.001; CBs 7.18% vs. 1.67%, respectively; p = 0.01). The proportion of ICDs found in our sample was similar to that reported in the DOMINION study 2 , which is the largest PD prevalence study to date, in which they found ICDs in 13.2% of cases, in accordance with our study and with other European works 17 . This result is in contrast to the high rate of ICDs found in some other European studies, including in the Spanish population 3,10,11 . The prospective Italian study, ICARUS, had found a rate of ICDs at the basal evaluation of 29.3%, which was higher than in our baseline assessment in COPPADIS 3 . These discrepancies could partially be explained by the sociocultural differences in both groups and by the differing tools used for ICDs screening. In the ICARUS study, the Minnesota Impulsive Disorders Interview (MIDI) and QUIP were used, whereas in COPPADIS, the QUIP-RS was applied. Following the MDS recommendations, QUIP-RS has more accurate clinimetric properties for application in PD, being "recommended, " whereas MIDI lacks some criteria validity when compared with the QUIP-RS 18 . In addition, ICDs screening using QUIP could overestimate ICDs frequency since Papay et al. found that, up to 40% of patients who scored positive for ICDs on the QUIP did not fulfil the diagnosis of ICDs 19 . However, the QUIP-RS used in COPPADIS has stricter ICDs diagnostic criteria; thus, the proportion found in our study could be more accurate for the identification of ICDs. On the other hand and in contrast to two Spanish studies, Vela et al. had found an ICDs estimation that reached 58.3% in their cohort of early-onset PD; and in their study of patients undergoing DA treatment, García-Ruiz et al. found this rate increased up to 39.1% 10,11 . In COPPADIS, when we analysed these two groups separately, the prevalence of ICDs reached 15.87% in patients with PD undergoing DA treatment and 19.78% in patients with an age at disease onset younger than 50 years. Again, these differences could be justified by the different estimation of ICDs provided by QUIP as screening tool that could overestimate the prevalence of ICDs 19 .
Accounting for the features in PD and CS, we observed that a previous history of ICDs as well as premorbid impulsive personality were more frequent in patients with PD than in CS. A higher rate of antidepressant treatment was also shown in patients with PD, which could be an intrinsic feature of PD, given a higher burden of nonmotor symptoms is present compared with CS, and consequently, depression is also more frequent 7,8 .
In the PD group, as previous studies have reported, those who had ICDs were younger and their disease onset was earlier 2,3,10,11,20 . On the other hand, and in contrast to previous studies, in our cohort, ICDs were not related to a longer disease duration, nor a higher exposure to DRT 3,21,22 . This result is interesting, given previous studies had hypothesised that the presence of ICDs in young patients could be explained by a dysfunction of the mesocorticolimbic loop due to the influence of the DRT on a relatively preserved ventral striatum 23 . However,  www.nature.com/scientificreports/ recent studies have supported the contrary version given that, reduced dopamine synthesis capacity in the nucleus accumbens as well as reduced functional connectivity in this area have been shown 9,24 . Consequently, the higher rate of ICDs in young patients could not be fully explained by the neurodegenerative pattern progression. Genetic factors could instead be involved in the early disease onset and ICDs in these patients 5,25 .
In patients with PD who have ICDs, a premorbid history of ICDs and impulsive personality were more frequent, according to observations in previous studies developed in European countries and worldwide 2,17,26 .
Concerning motor performance, in our case and according to the vast majority of studies, the motor status did not differ in patients with ICDs 3,10,11 . However, it is interesting to note that, although the motor performance according to the UPDRS III was not different according to the presence of ICDs, the scores on the FOG scale were higher in patients with ICDs. This observation was similar in the subgroup of early disease onset, suggesting a shared physiopathological background or due to the perception of the patients since a poorer status was also found in the motor subdomain of the PDQ-39 in patients with ICDs.
The neuropsychiatric sphere is highly impaired in patients with ICDs, as is shown in multiple studies 3,14 . In COPPADIS, we found a high rate of depression as well as a poorer perception of mood disturbances according to the NMSS, PDQ-39 and NPI. The link between mood and impulsivity is not fully understood because it is not clear whether there is a predisposing factor or whether it is simply a consequence of the ICDs 6-8 . Indeed, it is difficult to elucidate by the comorbidity of both disorders. However, we must take into account that there are genetic forms of PD with prominent nonmotor manifestations, and a common genetic background could therefore underlie both, mood disorders and ICDs [27][28][29] . In COPPADIS, depression was more frequent in patients with PD compared with CS, which could suggest it as an intrinsic feature of PD rather than a consequence of ICDs. On the other hand, and also correlated with mood disorders, sleep disturbances are usually present in patients with ICDs [30][31][32] . The relationship could again be bidirectional, as in depression, but it also appears that there is a shared neurobiological substrate, given some genetic risk factors might justify the coexistence of both ICDs and sleep problems 33 . On the other hand, the intake of antidepressant drugs was not different between patients with ICDs positive or negative in the whole cohort. This observation is important since it could point out to a infradiagnosis of depression in patients with ICDs.
Apart from the impact of neuropsychiatric symptoms and sleep, in line with previous studies 3 , other nonmotor symptoms were more prominent in patients with ICDs. Urinary and cardiovascular symptoms were more frequent in patients with ICDs, in agreement with the Italian study ICARUS and other series. It is interesting to note that these patients present with a high burden of nonmotor symptoms from different spheres; thus, we should keep this into mind when performing an integral assessment in this particular group of individuals.
Concerning cognitive performance related to ICDs, we did not find any difference between the groups, even when analysing the subgroups of patients undergoing DA treatment or early-onset PD. It contrasts with the abnormalities noted in other studies in which executive dysfunction as well as impairment in the reward-related decision process and set shifting have been noticed 13,14,34 . In COPPADIS, the global cognition did not differ between groups. Analysing the frontal-subcortical domains of PD-CRS separately, no differences were found. To better understand the reasons for these discrepancies, future neuroimaging studies would be useful to determine the functionality of this area in our cohort.
Regarding the DRT and its influence on ICDs, we found that DA use had greater impact on ICDs; it was dose-dependent and related to severity, which was in line with the observation noted in multiple studies across populations 2,10,11,35 . However, the risks linked to the type of agonist can be different. We found, according to another Spanish study and a European multicentre study, that Rotigotine had a low risk of developing ICDs, and Ropirinole the highest 11,36 . As Rizos et al. explained, the association between Rotigotine and ICDs is poorly understood, but it could be associated with its mechanism of action from continuous dopaminergic stimulation 37 . This possibility is important, given we could consider this in patients with other risk factors for ICDs in the case that DA treatment is required.
On the other hand, it is interesting to highlight the fact that longer levodopa or DA exposure was not related to the appearance of ICDs; thus, ICDs could appear at any time in the disease course. Therefore, clinicians should be aware of and investigate ICDs over time.
This study has some limitations. The presented data correspond to the cross-sectional evaluation at baseline of COPPADIS cohort, nevertheless, the assessment of the ICDs/CBs behaviour over time would be more informative. This issue would be overcome in the future since the follow up of this cohort is on-going and we will obtain the longitudinal data 38 . On the other hand, following the methodology of the study, the UPDRS part II evaluation only included motor items to assess the phenotype (tremor dominant, postural instability and gait disorders or undetermined). It would be more accurate to evaluate the whole UPDRS part II to establish the handicaps in the performance of the daily activities according this scale. To partially compensate this issue, we assessed the PDQ-39 that also included activities of daily living.
conclusions ICDs and CBs are common nonmotor features in patients with PD compared with CS. Some premorbid features such as a previous impulsive personality act as risk factors for developing these disorders. DRT had a large impact on ICDs, and DAs were the drugs with the greatest effect. However, patients with ICDs had a high burden of nonmotor symptoms belonged to mood, cardio-vascular, sleep, urinary domains among others as well as a poor quality of life. Hence, apart from treatment, patients might have a genetic background that could influence the presence of a prominent non-motor phenotype.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.